THE LAW REFORM COMMISSION

AN COIMISIÚN UM ATHCHÓIRIÚ AN DLÍ

(LRC 50–1995)

REPORT ON

AN EXAMINATION OF

THE LAW OF BAIL

IRELAND

The Law Reform Commission

Ardilaun Centre, 111 St. Stephen's Green, Dublin 2

© Copyright The Law Reform Commission 1995

First Published August 1995

THE LAW REFORM COMMISSION

The Law Reform Commission was established by section 3 of the Law Reform Commission Act, 1975 on 20th October, 1975. It is an independent body consisting of a President and four other members appointed by the Government.

The Commissioners at present are:

The Hon. Anthony J. Hederman, former Judge of the Supreme Court, President; John F. Buckley, Esq., B.A., LL.B., Solicitor;

William R. Duncan, Esq., M.A., F.T.C.D., Barrister-at-Law, Professor of Law and Jurisprudence, University of Dublin, Trinity College;

Ms. Maureen Gaffney, B.A., M.A. (Univ. of Chicago), Senior Lecturer in Psychology, University of Dublin, Trinity College;

Simon P. O'Leary, Esq., B.A., Barrister-at-Law.

The Commission's programme of law reform, prepared in consultation with the Attorney General, was approved by the Government and copies were laid before both Houses of the Oireachtas on 4th January, 1977. The Commission has formulated and submitted to the Taoiseach or the Attorney General forty nine Reports containing proposals for the reform of the law. It has also published eleven Working Papers, nine Consultation Papers and Annual Reports. Details will be found on pp.199–203.

Alpha Connelly, B.A., LL.M., D.C.L., is Research Counsellor to the Commission.

Ms. Nuala Egan, B.C.L., LL.M. (Lond.), Barrister-at-Law, Ms. Sarah Farrell, LL.B., LL.M. (Lond.), and Mr. Niall Fitzgibbon, B.C.L., LL.M. (Cantab.), Barrister-at-Law are Research Assistants.

Further information from:

The Secretary,

The Law Reform Commission,

Ardilaun Centre,

111 St. Stephen's Green,

Dublin 2.

Telephone: 671 5699.

Fax No: 671 5316.

This Report was submitted on 17 August 1995 to the Attorney General, Mr. Dermot Gleeson, S.C., under section 4(2)(c) of the Law Reform Commission Act, 1975. It embodies the results of an examination of and research into the law of Bail which was carried out by the Commission at the request of the former Attorney General, Mr. Harold A. Whelehan, S.C. The request did not extend to the formulation of proposals for reform.

While this Report is being considered in the relevant Government Departments the Attorney General has requested the Commission to make it available to the public, in the form of this Report, at this stage so as to enable informed comments or suggestions to be made by persons or bodies with special knowledge of the subject to the said relevant Government Department.

INTRODUCTION

On the 1st day of February, 1994 the then Attorney General asked the Commission to undertake an examination of and conduct research in relation to the law of bail. The Commission was not asked to formulate and submit proposals for reform.

In referring the matter of bail to the Commission, the Attorney General drew the Commission's attention to a statement from the then Minister for Justice on the subject. In the course of the statement the following views were expressed:

Chapter 1 briefly summarises the law relating to bail in Ireland and related legal matters, such as the status of preventive detention generally in the Irish criminal justice system. Chapter 2 examines the empirical evidence concerning the extent of the problem of bail offending, and discusses a number of methodological difficulties associated with studies of bail offending. Chapter 3 examines a number of prediction studies, that is to say, studies which attempt to predict future criminal behaviour among charged or convicted individuals. Chapter 4 examines the law relating to bail under the United States Federal Bail Reform Act 1984, the Canadian Criminal Code, the law in Scotland, England and Wales and a number of Australian jurisdictions. There is also a brief examination of the case-law under the European Convention on Human Rights and Fundamental Freedoms, and of the law in some Continental European jurisdictions. The criteria upon which bail decisions are made in those jurisdictions are given primary emphasis. Chapter 5 draws together much of the material from previous chapters relating to pre-trial preventive detention and concludes by setting out an alternative approach to the present position in Irish law, which emerges from the examination and research into the law here and in other jurisdictions. Finally, Chapter 6 examines methods of addressing bail offending other than pre-trial preventive detention. Research for this Paper was carried out by Úna Ní Raifeartaigh, Reid Professor of Criminal Law, Trinity College, Dublin.

CHAPTER 1: THE LAW RELATING TO BAIL IN IRELAND

General

1.1

Irish law is unusual among common law jurisdictions in having failed to introduce comprehensive bail legislation in recent decades. The jurisdiction of the courts to grant bail is governed by the common law, the Constitution, some legislative provisions and the case-law of the Irish courts. Irish law is unusual in having maintained fewer restrictions on the granting of bail than obtain in other jurisdictions. Other jurisdictions, as we shall see, have considerably widened the grounds on which bail may be refused. On two occasions, an attempt has been made to persuade the Irish Supreme Court to widen the criteria for refusal of bail, but on both occasions the Court rejected the proposed extension, which would have permitted the courts to refuse bail on the ground that the accused might commit further offences. However, the question of offending on bail has regularly been a source of public anxiety, and in 1984 the Oireachtas introduced provisions for mandatory consecutive sentences in respect of offences committed while on bail. This does not appear to have allayed public anxieties, however, and certain sectors have more recently called for the introduction of pre-trial preventive detention, through the medium of a Constitutional amendment if necessary. Ireland has not witnessed developments such as the introduction of bail information schemes or bail hostels. This may be due to the fact that, by international standards, Ireland remands few persons in custody pending trial, relying instead on sureties or, very occasionally, on cash deposits.¹ In this chapter, we summarise the law in Ireland concerning the matter of bail.

Jurisdiction To Grant Bail

The Garda Siochana

1.2

Members of the Garda Siochana have the power to release the accused on recognisance before appearance in court following arrest. Section 31 of the Criminal Procedure Act, 1967 provides that whenever a person is brought in custody to a Garda station by a member of the Garda Siochana, the sergeant or other member in charge of the station may “if he considers it prudent to do so and no warrant directing the detention of that person is in force, release him on bail and for that purpose take from him a recognisance, with or without sureties, for his due appearance before the District Court at the appropriate time and place.”

1.3

This is usually referred to as “station bail”. A sum of money may be accepted in lieu of a surety or sureties.² The money is deposited by the member of the Garda Siochana with the District Court Clerk in the relevant area. The recognisance may be estreated in the same manner as a recognisance entered into before a justice, if the accused fails to appear.³

1.4

Where a District Justice issues a warrant directing the arrest of any person and the warrant contains an endorsement that the person named therein is to be released on bail upon his entry into a recognisance, the Garda in charge of the station must admit the person to bail in accordance with the endorsement.⁴

1.5

The District Court Rules of 1948 originally provided that the recognisance for station bail had to provide for an appearance at the next sitting of the District Court for the court area in which the arrest had taken place. The District Court (Criminal Procedure Act, 1967) Rules 1985 altered this rule to provide that the recognisance might be conditioned for his appearance at the next sitting of the Court or any subsequent sitting to be held not later than 30 days after such sitting. This rule was struck down in The State (Lynch) v. Ballagh.⁵ In this case, the accused was arrested and taken to a Garda station where he was charged with a summary offence. He was released on station bail and the station sergeant made the recognisance returnable to the 6th March, although the next sitting of the relevant District Court was the 4th March. When the prosecutor appeared before the District Court, his solicitor objected to the jurisdiction of the court on the ground that he had not been brought before the court at the earliest possible sitting. The case was adjourned and the prosecutor obtained an absolute order of certiorari in the High Court on the ground that the 1985 Rules were ultra vires the District Court Rules Committee, in that they dealt with a function of the Garda Siochana rather than with the practice and procedure of the District Court. A majority of the Supreme Court held that the

Rules were ultra vires the Rules Committee for that reason. However, the Supreme Court allowed the appeal and discharged the order on the basis that the fact that the accused had appeared before the District Court on the 6th March was sufficient to give jurisdiction to the District Judge. In The State (D.P.P.) v. Ruane,⁶ Blayney J. held that s.15 of the Criminal Justice Act, 1951 (as inserted by s.26 of the Criminal Justice Act, 1984), which provides that the accused must be brought promptly before the court if in custody, applies only where station bail has been refused.

1.6

Formerly, bail could be granted by a peace commissioner. In State (Lynch) v. Ballagh,⁷ Walsh J. took the view that the functions purportedly conferred on peace commissioners with respect to bail were judicial functions and their constitutionality was therefore in doubt. In O'Mahony v. Melia,⁸ Keane J. held that the power of a peace commissioner to grant bail under s.15 of the Criminal Justice Act, 1951, as inserted by s.26 of the Criminal Justice Act, 1984, was unconstitutional because the power to grant bail was an aspect of the administration of justice and could only be exercised by members of the judiciary.

The District Court

1.7

Bail applications are more numerous in the District Court than any other court, because all cases begin there, unless directed to the Special Criminal Court. Jurisdiction to grant bail is conferred on the District Court by section 22 of the Criminal Procedure Act, 1967 which provides that where a District Court remands a person or sends him forward for trial or sentence, the Court may (a) commit him to prison or other lawful custody, or (b) release him conditionally on his entering into a recognisance, with or without sureties. Section 28 deals with remand periods. Where the Court remands a person on bail, it may remand him for a longer period than eight days if he and the prosecutor consent, but where the Court remands a person in custody on an occasion other than his first appearance in court, it may remand him for a period exceeding eight days but not exceeding thirty days if he and the prosecutor consent. On his first appearance, the Court may not remand him for a period exceeding eight days.

1.8

If the accused has entered into a recognisance to appear at every time and place to which the hearing will be adjourned, he will be on continuing bail and need not enter into fresh recognisances after every District court appearance. However, if the accused is returned for trial, both he and his sureties must enter into fresh recognisances for the appearance of the accused before the trial court. Where the accused has been returned for trial in custody, the District Judge may admit him to bail at any time before the first day of the sitting of the court to which he has been returned.

1.9

Section 28 of the Act provides that a District Justice shall admit to bail

a person charged before him with an offence, other than an offence to which section 29 applies, if it appears to him to be a case in which bail ought to be allowed. Section 26 of the same Act provides that a District Judge may direct that a sum of money equivalent to the amount of bail be accepted in lieu of a surety or sureties. Subsection 28(2) provides that refusal of bail at a particular appearance before the District Court shall not prevent a renewal of the application for bail at a subsequent appearance or while the accused is in custody awaiting trial.

1.10

The offences in respect of which the District Court has no jurisdiction under s.29 of the Act are treason, an offence under sections 2 or 3 of the Treason Act, 1939, an offence under sections 6, 7 or 8 of the Offences Against the State Act, 1939, a grave breach such as is referred to in s.3.1.i of the Geneva Conventions Act, 1962, genocide (added by the Genocide Act, 1973, s.7), an offence under s.9 of the Official Secrets Act, 1963 or an offence prejudicial to the safety or preservation of the State, murder, attempted murder, conspiracy to murder, or piracy. An application for bail in respect of such offences must be made to the High Court. Where a District Judge returns an accused to the Special Criminal Court for trial, he may not be admitted to bail by the District Judge without the consent of the Director of Public Prosecutions.

Courts of trial

1.11

A court of trial has jurisdiction to grant bail to persons being tried before it, after the accused has been arraigned or during the course of trial. The Special Criminal Court has jurisdiction to grant bail to a person charged before it or sent for trial before it by a District Judge, or transferred to it for trial from an ordinary court, pending trial or during trial, but not after conviction.⁹

The High Court

1.12

The Constitution vests the High Court with full original jurisdiction in all matters whether of law or fact, civil or criminal. The High Court has the power to grant bail prior to trial for any offence, even if the District Court or any other court has already dealt with the question of bail. Under section 28(3) of the Criminal Procedure Act, a bail applicant has a right of appeal from the District Court to the High Court against a refusal of bail or on the basis of the amount of bail required or the conditions imposed. The High Court will also hear bail applications in cases where statute precludes the District Court from admitting persons to bail, e.g. s.29 of the Criminal Procedure Act, 1967. It may also admit an appellant to bail where the District Court has failed to fix recognisances for appeal to the Circuit Court from the District Court. The High Court may also grant bail where a person convicted in the District, Circuit or Special Criminal Court is seeking certiorari, habeas corpus or prohibition, pending the making absolute of any such order.¹⁰ The High Court

would appear to retain jurisdiction to grant bail following conviction on indictment, although the point is at present before the Supreme Court in the case of Vincent Connell.¹¹

1.13

The High Court has a discretion to grant bail in all cases except misdemeanour cases falling within section 2 of the Habeas Corpus Act, 1781, in which case a High Court Judge must grant bail if he is satisfied with regard to the sufficiency of the bail.

1.14

Applications to the High Court for bail may be renewed on an indefinite number of occasions, although of course, in practical terms, subsequent applications are unlikely to succeed unless there are new grounds. Any applicant dissatisfied with a High Court decision on bail may appeal to the Supreme Court.¹²

Non-Custodial Bail

1.15

Admission to bail in Ireland usually involves the accused and others (his sureties) acknowledging that they will be liable to forfeit a sum of money should the accused disobey the conditions of his bail. The principle that bail should not be excessive was part of the common law and was enshrined in the 1688 Bill of Rights and the Eighth Amendment to the United State Constitution. That bail should not be excessive was acknowledged by the Irish Supreme Court in Attorney General v. O'Callaghan.¹³ In fixing the amount of bail, the courts have regard to the gravity of the offences as well as the circumstances of the accused.

1.16

The judge granting bail must satisfy himself as to the sufficiency of the bailsman.¹⁴ Bailsmen are usually drawn from the friends or family of the accused and professional bondsmen are not found in Ireland.

Non-Monetary Conditions Of Bail

1.17

Irish courts often attach conditions when granting bail, such as reporting to a Garda station daily or surrendering one's passport.¹⁵ However, no statutory authority sets out the precise conditions which may be imposed or the purposes for which conditions may be imposed. Moreover, unlike other jurisdictions, there is no direction to the courts that they must impose the least restrictive form of release possible to ensure the accused's attendance at trial, and that bail shall only be refused where no form of conditional release shall secure the attendance of the accused at trial.

Criteria To Be Considered When Deciding Whether Or Not To Admit To Bail

1.18

Irish law is unusually restrictive in respect of matters which may constitute grounds for refusing to grant bail. It may be noted that the criteria are the same in the case of a person detained in custody for extradition purposes as a person detained in custody for trial.¹⁶ No criteria are set out in the Criminal Procedure Act, 1967. The criteria employed by the courts prior to the Attorney General v. O'Callaghan¹⁷ were listed by Hanna J. in State v. Purcell.¹⁸ He said that the fundamental test was whether the accused would stand trial, and the matters a court could take into account in forming a view on the probability of his evading justice were:

1.19

In Attorney General v. Duffy,¹⁹ he added a further criterion to the list, namely, whether the accused was likely to interfere with the course of justice. In People (Attorney General) v. Crosbie,²⁰ O'Dalaigh C.J. made it clear that the only consideration for the court was whether the accused would abscond before his trial or interfere with witnesses. The power to refuse bail on the ground that the accused might commit further offences was not included.

1.20

In People (Attorney General) v. O'Callaghan,²¹ the defendant had been

returned for trial on charges of larceny, breaking and entering, malicious damage, receiving, resisting arrest and assault, which were alleged to have been committed while he was on bail in respect of other charges. His application for bail was refused first by the District Court and then in the High Court by Murnaghan J.. In the course of his judgment, Murnaghan J. listed the factors which he thought to be relevant in a bail application:

1.21

Murnaghan J. gave three reasons for refusing to grant bail in this particular case. First, the superintendent in charge of the case had told him the applicant was an aggressive type and that he was of the opinion that he would interfere with witnesses if admitted to bail; second, that the offences were committed while the applicant was on bail in respect of earlier charges and there was a serious risk that the applicant would commit further offences if he were granted bail; and third, that because of the applicant's record, he would be facing a substantial sentence. The defendant appealed against the refusal of bail to the Supreme Court and the Supreme Court granted bail. Judgments were delivered by O'Dalaigh C.J. and Walsh J..

1.22

O'Dalaigh C.J. observed that counsel for the Attorney maintained that the applicant, whom he conceded was likely to stand trial, should be refused bail because the offences in respect of which he was seeking bail were alleged to have been committed while he was on bail in respect of earlier charges. The Chief Justice characterised this as a suggestion that the applicant should be held as a preventive measure because if granted bail, it was feared that he might commit

further offences. In response to this argument, the Chief Justice said:

1.23

He emphasised that the single question in all bail applications is whether the applicant is likely to stand trial.

1.24

The Chief Justice then referred to Part II of the Offences Against the State (Amendment) Act, 1940, which permits a Minister of State to order the detention of a person without trial, and observed that a number of conditions have to be satisfied before the provisions of this Act can be operated:

1.25

Walsh J. also emphasised that the sole purpose of bail was to secure the attendance of the accused at his trial:

1.26

He also rejected the likelihood of committing further offences as a reason for refusing bail, saying:

1.27

Walsh J. noted that over the years, the likelihood that the accused would interfere with witnesses or jurors or destroy evidence had been added to the grounds for refusing bail, but said that this was because they were all “different aspects of the evasion of justice already referred to.”²⁶ He said that this should be a ground for refusal of bail only where it was reasonably probable that such interference would occur. He rejected the notion of protective custody for the accused on the basis that if an accused wanted protective custody he would not ask for bail and that a bail motion “cannot be used a vehicle to import into the law the concept of protective custody for an unwilling recipient.”²⁷ He also commented on ground (9) of Murnaghan J's list, saying that while a court should pay attention to the objections of the Attorney General or the police authorities, the fact that these authorities object was not itself a ground for refusing bail. He also said, with regard to ground (11), that the possibility of a speedy trial was only relevant in the sense that if there was no prospect of a speedy trial the court may very well allow bail where it might not otherwise have done so, but that the prospect of a speedy trial is not a ground for refusing bail where it ought otherwise to be granted. Regarding the substance of bailsmen, he referred to the common law principle that the amount of bail must not be so high as to be effectively unprocurable for this particular defendant.

1.28

Some difference of opinion emerged between the two judges concerning the issue of previous convictions. O'Dalaigh C.J. thought that these should never emerge in court until they became relevant to the imposition of punishment after conviction. However, Walsh J. felt that although it was undesirable to mention previous record, it might sometimes be relevant to the likely sentence, which in turn affected the likelihood of the accused absconding. However, he thought that it was most undesirable that the tribunal hearing such information would be the same tribunal to try the accused.

1.29

A second opportunity to rule on the likelihood of further offences as a ground for refusing bail arose in Ryan v. D.P.P..²⁸ The applicant had sought bail pending trial on a charge of burglary. He was brought to the District Court and remanded in custody on the ground that he might commit further offences. His application for bail in the High Court was opposed by the Director of Public Prosecutions on this ground. As there were no grounds for believing that the applicant would not turn up for trial if granted bail, the President of the High Court, Hamilton P., granted bail and refused to rely on the ground that the applicant might commit further offences. The applicant brought a further application in the High Court seeking a reduction in the amount of bail fixed by the District Court granting him bail on another set of charges, involving other offences under the Larceny Act, 1916. Barron J. ordered a reduction in the amount of bail. The Director of Public Prosecutions appealed against both High Court orders to the Supreme Court. The Supreme Court dismissed the appeal.

1.30

Finlay C.J. delivered a judgment with which Walsh, Griffin and Hederman JJ. agreed. He said that the sole issue in the case was whether the decision of the Supreme Court in People (Attorney General) v. O'Callaghan²⁹ was correct in deciding that the probability that a person awaiting trial would, if released, commit crime could never be a ground for refusing bail. The respondents argued, first, that the common law recognised a discretion to refuse bail on this ground, and second, that the constitutionally protected rights to life, bodily integrity, and property which would be affected by the commission of crime confers a right and duty on the Court to avail of its discretion to prevent the commission of crime. The Chief Justice rejected the constitutional argument, quoting extensively from the judgments of O'Dalaigh C.J. and Walsh J. in O'Callaghan with approval. Moreover, he said that pre-1922 Irish and English decisions did not lay down likelihood of future crime as a criterion for refusing bail; on the contrary, they established that attendance at trial was the sole criterion. He referred to the English decision in R. v. Phillips³⁰ as extremely unsatisfactory, being clearly obiter as part of a judgment in an appeal against severity of sentence, in which counsel did not appear. Although followed by an Irish court in Attorney General v. McCann,³¹ which was followed in Attorney General v. McEvoy,³² the decision must be taken to have been overruled by Attorney General v. O'Callaghan.

1.31

Finlay C.J. went on to say that quite apart from the constitutional objection to any form of preventive detention, there were a number of other reasons for refusing to permit it. Given the purpose of bail, an attempt to refuse bail on the suggested ground would be an abuse of power, namely, an exercise of it for a purpose which was outside its scope. He pointed out that an intention to commit a crime is not itself a crime unless it is furthered by overt acts of preparation or converted by agreement with others into a conspiracy. As the courts could not create offences, they would certainly not detain a person on suspicion of an intention, which even in a full trial could not lead to his punishment. Moreover, if such a power could be exercised, he asked, why should it be confined to cases where the person is an applicant for bail? He referred to certain practical problems. How should such an intention be proved and by what standard of proof? Would the accused require notice of the evidence against him and an opportunity to contest the charges being brought against him, as would be the case in a criminal trial? Would every bail application in which this ground was advanced take on the requisites of a criminal trial? He said that these queries “not only indicate practical problems but more importantly highlight the nature of the jurisdiction which it is sought to invoke without legislation.”³³ He continued:

1.32

He added that the fact that the Oireachtas had so relatively recently availed of this method to curb the mischief of crimes on bail confirmed his view that it could not be appropriate for the Court to assume the jurisdiction argued for. He concluded by saying that references to the legal position of other jurisdictions were of no value as the fundamental objections to the jurisdiction contended for derived from the Constitution, although he intimated that the decision in U.S. v. Salerno³⁵ would be of assistance if the court were ever called upon to decide upon the constitutional validity of legislation which dealt with the sort of extraordinary circumstances mentioned by Walsh J. in O'Callaghan.

1.33

McCarthy J. delivered a judgment with which Walsh J. also agreed. He agreed that the cases cited fell far short of establishing the existence of any common law principle supporting the jurisdiction contended, and whatever support there may be found for a court in England holding that such a principle forms part of the English common law, there was no authority to suggest it was part of the common law in 1922. It offended the fundamental principle of the presumption of innocence. McCarthy J. referred to three other legislative bases for preventive detention: the Prevention of Crimes Act, 1908, the Mental Treatment Acts, and the Offences Against the State (Amendment) Act, 1940. He pointed out that under the 1940 Act, the Minister had to be satisfied that the person concerned “is engaged” in the relevant activities, not that he “will engage” in such activities. However, he did not make any other comment regarding these Acts, save to comment in passing that the Mental Treatment Acts concerned a “different context.” He went on to say that if a person were to be denied his liberty because of a well founded suspicion, there was:

1.34

The judgments in O'Callaghan and Ryan emphatically reject the notion of preventive detention in the context argued for. However, it may be worth briefly pausing to disentangle the various elements and arguments employed, in particular to identify which of these are anchored in the Constitution and which are not. We will return to a full discussion of the arguments in this area in Chapter 5.

1.35

Both judges in O'Callaghan characterised pre-trial detention to prevent offending as a punishment.³⁷ They either stated or proceeded on the assumption that the presumption of innocence applies before trial and said that the true purpose of bail was to ensure that the accused would stand trial. Pre-trial detention to prevent offending violated the presumption of innocence. None of the judges discussed the issue of whether such violation of the presumption of innocence could be justified in any circumstances except Walsh J. who mentioned that liberty might be restricted before conviction in extraordinary and highly limited circumstances. However, some of the arguments did not appear to derive from Constitutional values. In Ryan, Finlay C.J.'s reference to certain practical problems does not appear to raise a constitutional matter and might be capable of being addressed by legislation providing for procedural safeguards, and his argument concerning the absence of a crime based on intention was fundamentally an argument about the inability of the Courts to legislate. We may note that Walsh J. expressed serious scepticism about the ability to predict future offending from either past bail record or current charge. We may also note that both judges in Ryan suggested that this form of detention would open the way for preventive detention more generally. Finally, we may note that Walsh J. in O'Callaghan and both judges in Ryan said that the accepted method of dealing with offences was based on deterrence rather than prevention.

Sentencing In Respect Of Bail Offences

1.36

Prior to the coming into force of the Criminal Justice Act, 1984, a judge sentencing an accused in respect of several offences, some of which were committed while on bail, was entitled to exercise his discretion as to what penalties were appropriate. Section 11 of the Criminal Justice Act, 1984 attempts to curb this discretion significantly by requiring a judge to make a sentence of imprisonment in respect of an offence committed on bail consecutive to any other sentence. We may contrast this section, for example, with the equivalent English provision which requires the court to treat the fact that the offence was committed on bail as an aggravating factor, but does not require sentence to be consecutive.³⁸ The wording of section 11 is as follows:

1.37

The two year limit to the jurisdiction of the District Court is an exception to the normal one year limit. Section 12(1) of the Criminal Justice Act, 1984 amends section 5 of the Criminal Justice Act, 1951, so that where two or more sentences passed by the District Court are ordered to run consecutively, the aggregate term of imprisonment shall not exceed two years. It may be noted that the offence of “failure to answer bail” created by s.13 of the 1984 Act is to be treated as an offence committed while on bail for the purposes of s.11.³⁹

1.38

It is arguable that the Legislature might have been surprised by the judicial interpretation of their intentions in section 11, first, by means of the so-called totality principle, and second, in that one of the consecutive sentences may be suspended. This was perhaps inevitable given the Irish tradition of judicial sentencing discretion and hostility to any diminution of such discretion. In People (D.P.P.) v. Noel Healy,⁴⁰ the applicant was charged in Limerick Circuit Court with a series of offences arising out of a conspiracy to commit false imprisonment and conspiracy to rob and was sentenced to concurrent terms of eight years imprisonment on each count. The applicant was on bail when these offences were committed. When sentencing, the President of the Circuit Court took into account the total number of years that would be served by the applicant and adjusted the sentence downwards. The applicant appealed against severity of sentence on the basis that he had received a considerably greater sentence than his co-accused. In the course of the appeal, counsel for the Director of Public Prosecutions submitted that a sentencing judge should disregard the duration of the sentence imposed for a previous offence and fix a sentence appropriate to the second crime, and should not follow the English practice of taking into account the total number of years to be served for all offences. The Court of Criminal Appeal dismissed the application holding that the sentence imposed on the applicant was not excessive or founded on any error of principle, and a comparison with the sentences imposed on the other accused was inappropriate. McCarthy J., delivering the judgment of the Court, also took the opportunity to comment that the sentences imposed on the other accused were based on a misconstruction of the 1984 Act, and that the sentencing Court should determine the sentence appropriate to the offence without having regard to the fact that it must be consecutive under s.11 on any sentence for a previous offence. However, in the case of grave offences, the Court should adjust the sentence downwards where not to do so would impose a manifestly unjust punishment on the accused.

1.39

The obligation to consider the sentence in its entirety arguably dilutes any punitive intent of the Legislature in enacting s.11. As against that, the legislation has not been amended.

1.40

The question of suspending one of the consecutive sentences was addressed in People (D.P.P.) v. Thomas Dennigan.⁴¹ The applicant pleaded guilty to a series of offences contained on two bills of indictment. The second

series of offences were committed while on bail in respect of the first. With respect to the first series of offences, he was sentenced to concurrent terms of imprisonment, the longest being four years. He was given a five year sentence for the second series, to run concurrently with the other sentences imposed. Counsel for the applicant pleaded for leniency on the basis of the personal circumstances of the accused, and argued that the Court was entitled to suspend one of the sentences as this had not been expressly prohibited by s.11 of the 1984 Act. The Court held that it did have power to suspend a consecutive sentence in appropriate circumstances, although the instant case was too serious to warrant such suspension. However, it reduced each of the sentences on the basis that insufficient attention had been paid to the accused's circumstances. The Court exercised its power to suspend a sentence in People (D.P.P.) v. Michael Farrell.⁴² The applicant had been sentenced to two terms of two years detention to run concurrently on various charges relating to larceny and burglary, to which he had pleaded guilty. One of the sentences related to an offence committed while on bail. The applicant applied for leave to appeal against severity of sentence. It was held by the Court of Criminal Appeal that s.11 was a mandatory direction to the Court to pass a consecutive sentence in respect of any sentence committed while on bail, but it said that it would suspend the first sentence of two years in respect of the offence which was not committed while on bail.

1.41

Although the suspended sentence is in theory, a significant penalty, it is often viewed as no sentence at all, because the offender is not required to suffer any immediate hardship. Although, in theory, the suspended sentence may be re-activated upon future misconduct, in practice, it appears that this rarely, if ever, occurs.

Failure To Appear And Forfeiture

1.42

The traditional remedy for breach of a bail order by failure to appear was estreatment or forfeiture of the money secured by the recognisance of the accused or his surety. This power of entreatment continues to operate,⁴³ although it is relatively rarely enforced. Often, even when the power is invoked by the Director of Public Prosecutions, the court will be reluctant to estreat all of the money, particularly if the surety is a financially hard-pressed spouse struggling to make ends meet in the absence of the criminal spouse.

1.43

Section 13 of the Criminal Justice Act, 1984 created the offence of failing to answer bail. Subsection (1) provides that if a person who has been released on bail in criminal proceedings fails to appear before a court in accordance with his recognisance, he is guilty of an offence carrying a maximum penalty on summary conviction of £1,000 or imprisonment of twelve months or both. Subsection (2) provides that it is a defence for the accused to show that he had

a reasonable excuse for not so appearing. Subsection (4) confers a power of arrest without warrant upon the Garda Siochana in respect of a person who has failed to appear in answer to his bail, where the court has issued a warrant by reason of such failure to appear. The Garda must produce and serve the warrant on the arrestee as soon as practicable.⁴⁴ Under s.33 of the Criminal Procedure Act, 1967, where a person has been admitted to bail, a District Judge may issue a warrant for the arrest of the accused upon the application of the surety or any of the sureties or of a Garda and upon information in writing, that the accused is about to abscond for the purposes of evading justice. Upon being brought before the court, the court may commit the accused to prison to await his trial or until he enters into a fresh recognisance or, if he is on remand, remand him.

Confiscation Orders

1.44

Proponents of pre-trial detention to prevent future offences often refer to the phenomenon of career criminals who, when facing trial and conviction in respect of a set of charges, make use of their period on bail to re-offend and develop a financial fund to await them upon their release from the potential prison sentence they face. If this be the case, it may be partly addressed by the recently introduced legislation providing for the confiscation of the proceeds of crime, namely the Criminal Justice Act, 1994.

Related Issues: The Status Of Preventive Detention In Irish Law And Prediction

Preventive detention

1.45

Preventive detention before trial is the main issue in this Paper, but the issue of preventive detention after conviction is a closely related concept. The proponents and opponents of each rely on similar arguments. It may be noted that many of the jurisdictions which, as we shall see, permit pre-trial preventive detention also permit post-conviction preventive detention. Irish law has kept the use of preventive detention to a minimum. One example of a preventive detention provision in the criminal context is contained in the Prevention of Crime Act, 1908, which provided for a double-track sentence for convicted persons found to be “habitual offenders” the first part of the sentence was to be proportional to the offence, and the second part merely to prevent future offences. The offender was to serve the second portion of his sentence in a non-punitive but secure regime.⁴⁵ This legislation was implemented in England, but it is not clear that it was ever put to use in this jurisdiction. Certainly, it is not currently employed by the courts and no facilities exist in which an accused could serve the second part of his sentence. In England, similar types of legislation

have been enacted at different points throughout the century,⁴⁶ but this example has not been followed in this country.

1.46

More recently, the Court of Criminal Appeal has, at least on two occasions, rejected the use of selective incapacitation as a valid method of sentencing. In The People (D.P.P.) v. Carmody,⁴⁷ one of the applicants had a list of convictions stretching back to 1968 and the other had 120 convictions. Both had served a series of prison sentences. They were sentenced to six years' imprisonment in respect of burglary and malicious damage. The trial judge said that they were not amenable to the ordinary constrictions of society and that he had a duty to protect the public. On appeal to the Court of Criminal Appeal, McCarthy J. quoted from the O'Callaghan case on the constitutional prohibition on the refusal of bail because of the likelihood of future offending. He referred to the Prevention of Crimes Act, 1908, noting that there were no facilities in the State for such sentences. He said that the radical departure from the previous measures of imprisonment in this case was an attempt to secure reforms by prevention, but that in the absence of appropriate statutory provisions, this was not an acceptable basis for sentencing. In this brief judgment, it is not made clear whether a statute could render such sentencing valid, or whether, which is more likely, any such legislative provision would run into constitutional difficulty. In The People (D.P.P.) v. Jackson,⁴⁸ the trial judge imposed life sentences in respect of two rapes, saying that he did so to protect women against the accused until such time as in the judgment of the authorities he was fit to be released. On appeal to the Court of Criminal Appeal, Hederman J. said that preventive detention was not known to our judicial system, and the court reduced the sentences to fifteen years and eighteen years respectively.

1.47

Internment provisions also represent legislative provision for pre-trial detention.⁴⁹ Internment has been permitted in Ireland by successive legislative enactments: the Public Safety (Emergency Powers) (No. 2) Act, 1923,⁵⁰ the Public Safety (Powers of Arrest and Detention) Temporary Act, 1924,⁵¹ the Public Safety (Emergency Powers) Act, 1926, the Offences Against the State Act, 1939, the Offences Against the State (Amendment) Act, 1940, the Emergency Powers Act, 1939, and the Emergency Powers (Amendment) Act, 1940. Part VI of the Offences Against the State Act, 1939 provided for the power of internment, and to bring this Part into effect it was necessary for the Government to publish a proclamation to the effect that the powers conferred by Part VI were necessary to secure the preservation of public peace and order. In The State (Burke) v. Lennon,⁵² Gavan Duffy J. heard a habeas corpus application in which the

central issue was whether the Minister who had signed a warrant ordering the applicant's arrest and detention was purporting to administer justice. Gavan Duffy J. in allowing the application, cast doubt on the constitutionality of Part VI of the 1939 Act. The Supreme Court refused to hear an appeal by the State against the granting of habeas corpus by the High Court.

1.48

The Oireachtas then passed the Offences Against the State (Amendment) Bill, 1940, which was referred to the Supreme Court by the President under Article 26 of the Constitution, thus enabling the Supreme Court to give its opinion on internment, since the new Act was similar to the 1939 Act. The Supreme Court upheld the constitutionality of the Bill.⁵³ Sullivan C.J. laid emphasis on the fact that several pre-1937 Acts provided for internment and the Constitution did not expressly prohibit it. He characterised internment as a “precautionary measure taken for the purpose of preserving the public peace and order and the security of the State.” This decision granted immunity to the Act from further constitutional challenge, and a challenge to the Act on the ground that it was repugnant to the Convention for the Protection of Human Rights and Fundamental Freedoms was rejected in In Re O'Laighleis.⁵⁴ While it might be argued that these internment provisions represent a significant precedent for preventive detention in Irish criminal law, it must be noted that these concern political offences in emergency situations. They were never extended to mainstream criminal law. Moreover, most if not all of these powers were dependant upon the relevant Minister forming the opinion that the individual was already engaged in certain types of behaviour.⁵⁵ Finally, it may be noted that what was primarily at issue in In Re Article 26 and the Offences Against the State (Amendment) Bill, was whether the Minister was purporting to exercise justice, not whether preventive detention breached the presumption of innocence. We will return to these matters below.

1.49

It is clear, therefore, that preventive detention is not unknown in Irish criminal law.

Prediction

1.50

In all criminal justice systems, prediction of future behaviour plays a role, but again its presence is limited in the Irish judicial arena. It is seen primarily in the context of bail decisions, when a judge must take a view on whether an accused is likely to abscond, interfere with witnesses or tamper with evidence. It also plays a role when a sentencing judge is choosing between, for example, a

custodial sentence and community service or some other method of disposal. It is also employed by non-judicial persons in relation to imprisoned persons, when deciding whether a prisoner is suitable for temporary release. In other systems, prediction is employed in a range of additional contexts. For example, prediction of future behaviour is crucial to a parole decision and attempts to employ scientific methods of predicting parole risks have long been in operation in the United States. Again, many jurisdictions in the United States require predictions regarding future dangerousness to be made prior to the imposition of a capital sentence. Many jurisdictions permit preventive sentencing in respect of terms of imprisonment based on predictions about future dangerousness. Moreover, in the civil context, commitment of mentally ill persons often requires prediction of future behaviour. Again Ireland is unlike many other jurisdictions in its greatly limited scope for prediction and in its failure to employ any scientific methods of prediction in the limited areas to which it operates.

Criteria For The Grant Of Bail To Young Persons And Children

(i) Station bail

1.51

For the purposes of the law of bail, remand and detention, “young person” and “child” are defined as follows in the Children Act, 1908:⁵⁶

In effect, however, the law of bail applies only to children above the age of criminal responsibility, i.e. those aged between 7 and 14.⁵⁹

1.52

Section 94 of the Children Act, 1908 as amended, sets out the law as it applies to the grant of station bail to children and young persons:

1.53

The Criminal Procedure Act, 1967 deals with the law of remand of “persons”, a term which clearly includes children and young persons. Thus many of the provisions of this Act apply to children and young persons on remand. In the event, however, of conflict between the terms of the Acts of 1908 and of 1967, the rules of statutory interpretation require that the terms of the former, being a “specific” Act, takes precedence over those of the latter “general” enactment. Thus the authority which section 31(1) of the Act of 1967 grants to a sergeant or other member in charge of the station to release a person brought in custody to a Garda Siochana station, and in respect of whom no warrant ordering his detention exists, whenever he or she considers it prudent to do so, is inapplicable in the case of children and young persons.

1.54

However, the obligation to detain or remand which section 94 of the Children Act, 1908 imposes upon a superintendent or inspector of police or officer of equal or superior rank, applies whether the young person or child is detained with or without warrant. This provision may, therefore, and unlike its counterpart in the case of detention on remand of those of 17 years of age or more, permit a superintendent or inspector of police or officer of equal or superior rank to superimpose his or her decision to detain over that of the District Justice.⁶¹ A District Justice, when issuing a warrant, may only insert a provision ordering detention in custody if the facts of the case meet the O'Callaghan criteria. Thus, if he or she is satisfied that the release on bail of the person to be arrested would not interfere with the course of justice, a warrant of arrest may not contain any restriction: a member of the Gardaí of appropriate rank may nonetheless be obliged to remand the young person/child in custody if the charge is one of homicide or other grave charge or if it is necessary in the interest of such person to remove him or her from association with any reputed criminal or prostitute. Similarly, it is arguable that the terms of a warrant which

orders the release on bail of a child or young person may be altered by a superintendent or equivalent, for he or she is empowered by section 94 to release “with or without sureties, for such an amount as will, in the opinion of the officer, secure the attendance of such person upon the hearing of the charge, being entered into by him or his parent or guardian.”

1.55

The constitutionality of the powers granted by section 94 of the Children Act, 1908 to Gardaí of the requisite rank to disregard the order of a District Justice as found in a warrant to arrest must surely be open to debate.

(ii) Court bail

1.56

The Children Act, 1908 does not specify the conditions under which a court may grant or refuse to grant bail to a young person or child. In those circumstances, the “general law” criteria as set out by the Supreme Court in O'Callaghan for the grant or refusal of court bail appear to apply. It may, however, be arguable that the matter of the welfare of the child or young person is a factor which must also be taken into account by the court in determining whether bail should or should not be granted. This suggestion arises from a literal interpretation of section 3 of the Guardianship of Infants Act, 1964 which states that:

1.57

“Infant” is defined as a person under eighteen years of age other than a person who is or has been married.⁶²

1.58

Arguably, those “proceedings” to which the section relates and the term “custody” itself may be limited by reference to the context in which those words are found. On the other hand, the section clearly refers to any proceedings in which custody is in issue, a category into which a bail application falls. If, indeed, the welfare of the child or young person must be considered, then there may clearly be situations in which he or she may be detained by the courts even though no question of interference with the administration of justice is at issue.⁶³

1.59

The common-law power traditionally enjoyed by the courts to determine the terms on which an accused may be released on bail – and affirmed in the Criminal Law (Bail) Bill, 1995– applies in the case of children and young persons also. The question of “welfare” may also be of relevance in this regard.

1.60

The various provisions of section 24 of the Criminal Procedure Act, 1967 dealing with the length of detention on remand also apply,⁶⁴ in the absence of more specific provisions relating to children and young persons.

PLACES AND CONDITIONS OF DETENTION OF YOUNG PERSONS AND CHILDREN

(i) Places of detention

1.61

The law provides for the detention of young persons and children outside of the main-stream adult prison system; only in every extreme circumstances – to be discussed below – may a young person be detained in an adult prison. In no circumstances may a child be so detained.

1.62

Section 95 of the Children Act, 1908, as amended,⁶⁵ provides that a child or young person who is not released on recognisance by a superintendent or officer of equivalent or superior rank shall be detained in a place of detention as provided under the Act, until he or she can be brought before a court of summary jurisdiction. Section 97 of the Act provides that a court of summary jurisdiction, on remanding a person who is not released on bail, shall also commit him or her to custody in a place of detention, for the period of his or her remand. The Act does not set out or define “places of detention.”

1.63

The Prevention of Crime Act, 1908 established a Borstal system for males aged between sixteen and twenty-one years. Following the enactment of the Criminal Justice Act, 1960, the term “Borstal” was discontinued and that age group was subsequently detained in St. Patrick's Institution.⁶⁶The Criminal Justice Act, 1960 provides for the detention in St. Patrick's of males aged between seventeen⁶⁷ and twenty-one years and, if the court considers that none of the other methods by which the case may be dealt with are suitable, of sixteen year olds. Thus it is possible to detain a sixteen year old under the Children Act, 1908 system or under the system established in 1960. As an almost invariable rule, however, sixteen year olds are detained in St. Patrick's.

1.64

Section 9 of the Act of 1960 provides that a power to remand in custody a person between the ages of sixteen and twenty-one shall include a power to

remand to a remand institution. No such remand institutions have, however, been established.

1.65

List of Places of detention: At the moment, there are 4 centres in operation which have been certified, pursuant to the Act of 1908, as places of detention for the purposes of the remand of children and young persons.

1.66

The enactment of the Children Act, 1908 led to the establishment of all

of the above centres and they are now under the auspices of the Department of Education. The philosophy underpinning these centres is one of minimum-intrusiveness, permitting to the greatest extent feasible, the continuation of enjoyment of civilian life. The law does, however, permit the detention of a young person⁷² in prison on remand.

1.67

Section 97 of the Children Act, 1908, which states that a court shall, if it remands or commits for trial a young person, commit him to custody in a place of detention, adds that:

1.68

“Certificates of character” resulting in transmission to prison are issued only in respect of young persons detained in accordance with the provisions of the Children Act, 1908. Thus, sixteen year olds detained on remand in St. Patrick's may not be transferred to prison under the terms of this provision.⁷³

1.69

Section 25(1) of the Criminal Procedure Act, 1967 provides that a person may be remanded by the courts into the custody of a member of the Garda Siochana for a period not exceeding four days.⁷⁴ It has been judicially accepted, however, in a number of unreported judgments that as the Children Act, 1908 specifically provides that a child or young person shall be detained in a place of detention certified under that Act, no child or young person may be remanded in a Garda station unless such station is certified as a place of detention in accordance with the Act of 1908.⁷⁵

1.70

St. Patrick's Institution: St. Patrick's Institution offers, as we saw, accommodation to sixteen to twenty-one year olds, whether on remand or already convicted. Its maximum capacity in 1992, the latest year for which figures are available, was purportedly in the region of 120 people, although the average number detained on any given day in that year was 145.⁷⁶ There are no statistics available as to the numbers remanded in custody in St. Patrick's Institution nor as to the length of time spent by remand prisoners therein. However, the Governor has stated that a period of one to two weeks remand in custody would not be so unusual.

1.71

Section 7 of the Prevention of Crime Act, 1908, as amended, empowers the Minister for Justice to transfer a prisoner detained in St. Patrick's Institution to a prison where the Visiting Committee certifies such person as incorrigible or declares that the prisoner in question is exercising a bad influence over the other inmates of the Institution. This provision has not been used in recent years.

(ii) Conditions of detention

1.72

The Children Act, 1908 was seen as “a great and fundamental step in child protection”⁷⁷ which showed, for its time “an enlightened approach to juvenile offenders and their treatment by the courts”.⁷⁸ The welfare approach encompassed in the Act is reflected in the form of care afforded to young offenders detained in centres established in accordance with its provisions. The centres aim to ensure for the young people and children detained, as far as is practicable, the continuance of a “normal” family, educational, recreational and social life. When applying this philosophy, no distinctions are drawn between young persons/children on remand and those already convicted other than those necessitated by the very nature of remand detention. The families and adult friends of both categories of detainee, for example, have unlimited visitation access. The centres are very clearly concentrated in the Dublin region; consequently, and in order to make these unlimited visiting rights a reality, travel vouchers are available to the families from beyond this catchment area to cover expenses incurred. No distinction is drawn between the families of different categories in the allocation of these vouchers. On the other hand, young persons/children from the Dublin area who have been remanded by the courts for assessment, may avail of weekend home-leave thereby continuing contact with the family and, in the case of remand assessment, involving families in the assessment process; home-leave, however, may not be afforded to those on straight remand. In all other respects, children and young persons on remand may avail of the facilities and services provided in these centres.

1.73

In Trinity House and in Oberstown Boys Unit there is no physical separation between those who are convicted and those who are on remand. The other two places listed above contain only young persons and children on remand – there is no physical segregation in those centres between children and young persons on “straight” remand and those who are remanded by the courts for assessment.

1.74

St. Patrick's Institution is, as previously noted, outside the scheme for detention of young offenders established by the Children Act, 1908. Its regime is, consequently, more akin to that which exists in adult prisons, involving a stricter and more regimented regime than is found in a “Children Act centre.”

In 1985, the Whitaker Report stated that the accommodation provided therein was so unsuitable as to require immediate replacement.⁷⁹

1.75

The Institution, along with all of the prisons⁸⁰ in the State, is governed by the Rules for the Government of Prisons, 1947. Draft Prison Rules to replace the Rules of 1947 have been prepared by the Minister for Justice and were published in June 1994.⁸¹ The 1947 Rules contain a number of special provisions governing the regime of unconvicted prisoners under the age of 17⁸² and further provisions in respect of those aged 17–21. A distinction is drawn in the rules relating to those under 17 between (a) those unconvicted prisoners who have not been in prison before and those who are well-conducted in prison and (b) convicted prisoners who have been in prison before and those who misbehave in prison. The Rules envisage that remand prisoners shall be kept apart from those who have been convicted and they provide for a further segregation between those remand prisoners under 17 who fall into category (a) above and those who fall into category (b).⁸³ The authorities in St. Patrick's confirm that the remand prisoners are, where possible, detained in separate sections of the Institution but concede that in circumstances of extreme overcrowding, “mixed accommodation” cannot be avoided. Overcrowding can on occasion also result in the integration of remand and convicted prisoners when engaging in recreational, exercise, educational and training activities although, again where possible, segregation is maintained. The authorities in St. Patrick's add that, in reality, no distinction is drawn on any occasion between remand prisoners under 17 years of age of class (a) or (b).

1.76

Those facilities – educational and training, recreational and exercise – which are available to young convicted prisoners are also available to unconvicted young persons, albeit separately.⁸⁴ Unconvicted persons are entitled to receive visits from not more than three persons at one time for a quarter of an hour every day except Sunday; this entitlement compares with a right on the part of convicted prisoners to receive such visitors for thirty minutes once a week.⁸⁵

THE LAW GOVERNING THE HOLDING OF UNCONVICTED PRISONERS ON REMAND

1.77

In accordance with the legislative code governing prisons,⁸⁶ male remand prisoners may be committed by the courts to Mountjoy, Cork or

Limerick Prisons. Remanded females may be committed to Mountjoy and Limerick Prisons. Those of 17 years of age or more may be detained therein. Seventeen to twenty-one year olds may also be remanded in St. Patrick's Institution.⁸⁷

1.78

Only a limited number of official statistics are available in respect of the remand prison population. Statistics from the Department of Justice show that the numbers remanded, at some point in the criminal process, in St. Patrick's Institution more than doubled from the figure of 2,257 in 1984 to 5,628 in 1992. Of that 2,257, 234 were remanded in custody for trial in 1984 and 550, likewise, in 1994. During the same period, there was an increase of 50% in the entire prison population, growing from 6,462 in 1984 to 11,485 in 1992.⁸⁸ As of 31st December 1992, the total capacity of the three committal prisons was 921 whilst the greatest number detained during that year in all three prisons was 1,053. The authorities in the three prisons speak of increasing administrative difficulties posed by the increase in numbers and concede that many convicted persons are being released in order to provide space for newly-convicted prisoners and remand prisoners. It is acknowledged that no methodical approach is used in determining which convicted prisoners ought to be released to accommodate new prisoners, and the situation is governed purely by the “science of space.”

1.79

No official figures can be obtained relating to the length of time spent by prisoners on remand. The Governors of the three prisons stated that it is usual to be remanded in detention for a period of one to three weeks. In extreme circumstances, however, detention or remand may persist for three to six months.

1.80

Adult prison life is regulated by the Rules for the Government of Prisons, 1947. Part III thereof contains special rules for particular classes of prisoners: one such class are prisoners on remand. The Draft Prison Rules published in 1994 also make special provision for unconvicted prisoners.

1.81

The rules state that prisoners awaiting trial shall be kept apart from unconvicted prisoners and “while attending church and at other times shall, if possible, be placed so that they may not be in view of the convicted prisoners.”⁸⁹ The 1994 Draft Prison Rules also envisage, where possible, the separate accommodation of remand prisoners from those who have been convicted.⁹⁰

1.82

At the end of 1992, the latest date for which statistics are available, Mountjoy Prison, the largest prison in the State, could officially provide accommodation for 551 prisoners – both male and female –,⁹¹ although its prisoner population often reaches close to 700. The remand population ranges,

on average, from 70–100, although it may reach 140. The female remand population has fluctuated from 3 to a maximum of 20 prisoners. As a matter of course, male remand prisoners are kept separate from those who are convicted, but when numbers swell to the numbers indicated above, total segregation is not possible. In that event, remand and convicted prisoners may indeed share a cell. A similar situation exists in Cork Prison but in Limerick, “mixed accommodation” is the norm, due to lack of facilities. There is no segregation in Mountjoy or in Limerick, between female remand prisoners and females who have been convicted.

1.83

In Mountjoy and Cork Prisons, segregation is maintained, where possible, when unconvicted prisoners are undertaking physical exercise and recreational activities. This physical separation of the convicted and unconvicted also exists in Cork Prison in relation to educational courses, at which all categories of prisoners may attend. In Limerick Prison, again due to lack of facilities, educational courses are mixed. In Mountjoy Prison, those on remand may not avail of the educational courses on offer to convicted prisoners. The reasons offered are lack of facilities and resources, and the fact that the high “turnover” of remand prisoners, obviously, means that little benefit may be obtained from educational courses in the course of their short period of detention, particularly when weighed against the sizeable amount of administrative work which apparently must be undertaken in order to facilitate attendance at courses.

1.84

A remand prisoner, unlike a convicted prisoner, is not obliged by the 1947 Rules⁹² to perform any labour. It is, however, an option and one which, according to the Governors of the three prisons, is often availed of. This provision is repeated in the Draft Prison Rules.⁹³

1.85

Like their counterparts under the age of 17, unconvicted adult prisoners may receive a fifteen minute visit from no more than three visitors every day except Sunday, an entitlement which compares favourably with the right of convicted prisoners to weekly visits of 30 minutes.

1.86

Visits are supervised in the same manner as those involving convicted prisoners and no physical contact may be maintained between visitor and prisoner.⁹⁴ The Rules provide that a prisoner awaiting trial shall be allowed to see his or her legal adviser on any day except Sunday at any reasonable hour within the sight but out of earshot of a prison officer.⁹⁵ Such visits are allowed under similar conditions from a medical practitioner appointed by friends or a legal adviser where such visits are necessary for the purposes of a defence.⁹⁶ A prisoner who is in prison in default of bail shall be permitted to see any person at any reasonable hour, Monday to Saturday inclusive, for the bona fide purpose

of securing bail.⁹⁷ Written communications between unconvicted persons and others are “inspected” but not read, as in the case of convicted persons.

1.87

A number of other provisions of the Rules of 1947 emphasise the fact that remand prisoners by the very nature of their status as unconvicted persons shall not be treated as fully integrated into the prison regime. Matters such as the right to wear one's own clothes, or to wear clothes specifically for remand prisoners⁹⁸ and freedom from compulsion to obtain a standard hair-cut,⁹⁹ as set out in the Rules of 1947, maintain distinctions between the unconvicted and convicted. There is also a right to daily access to one's money which is maintained in the hands of the Governor.¹⁰⁰

1.88

Limited provision is also made for the detention of adults outside of the committal prison system: section 25(1) of the Criminal Procedure Act, 1967 provides that a court may, when it remands a person in custody for a period not exceeding four days, commit him or her to the custody of a member of the Garda Siochana. The detention of a person so detained must be in accordance with the terms of the Criminal Justice Act, 1984 (Treatment of Persons in Custody in Garda Siochana Stations) Regulations, 1987.¹⁰¹ In addition to setting down standards in relation to the conduct of interviews, maintenance of custody records and records of arrest and detention, the regulations require a member of the Garda Siochana to inform the detained person of his or her right to notify another person reasonably named by him or her of the fact of his or her custody¹⁰² and of the entitlement to consult a solicitor.¹⁰³ They also permit a visit from a relative, friend or other person with an interest in his or her welfare,¹⁰⁴ and also from a solicitor. Consultation with a solicitor may take place in sight but out of earshot of a member of the Garda Siochana.

Treatment of untried prisoners: international standards

1.89

There is only one directly relevant international treaty provision, Art. 10(2) of the International Covenant on Civil and Political Rights. When Ireland became a party to this treaty, it entered a reservation to this provision. There is however another treaty to which Ireland is party and which is of general relevance although it contains no specific provision regarding the treatment of untried prisoners. This is the European Convention for the Prevention of Torture and Inhuman or Degrading Treatment or Punishment which entered into force on 1st February 1989. There has been established under this treaty a European Committee for the Prevention of Torture and Inhuman or Degrading

Treatment or Punishment, and it is the role of the Committee “by means of visits, [to] examine the treatment of persons deprived of their liberty with a view to strengthening, if necessary, the protection of such persons from torture and from inhuman or degrading treatment or punishment” (Article 1 of the Convention). The Committee visits a number of places of detention in States Parties each year and draws up a report on the treatment of detainees in the places it visits in each State (Article 10(1)). It visited Ireland in the autumn of 1993. Publication of a report is not automatic and, under the Convention, only follows upon a request by the State Party concerned to this effect (Article 11(2)). While it is, therefore, possible for Ireland legally to block publication of the Committee's report on its visit to Ireland, it has not been unusual to date for such reports to be published. The Committee may include in its report “any recommendations it considers necessary” (ibid.) and “may consult with the Party with a view to suggesting, if necessary, improvements in the protection of persons deprived of their liberty” (ibid.). Although the Committee's recommendations are not legally binding, the Convention states:

1.90

Apart from these treaty provisions, there are two other international texts dealing with the treatment of prisoners, and both of them contain specific provisions on the treatment of untried prisoners. There are the U.N. Standard Minimum Rules on the Treatment of Prisoners and the European Prison Rules. Both were adopted within the context of intergovernmental organisations of which Ireland is a member, in the former case, the United Nations, in the latter, the Council of Europe. They have the status of recommendations and are not therefore legally binding as such but may be taken into account in diplomatic correspondence and by international bodies such as the European Human Rights Commission in interpreting the standards laid down in other international texts which are legally binding.

1.91

In addition to the international provisions specifically dealing with the treatment of untried prisoners, it should be remembered that the international human rights standards also apply in general to such persons even if their enjoyment of these rights may be curtailed somewhat owing to their detention.

1.92

The Prison Rules of 1947 are in the process of being changed and the current draft of the new Rules contains the following provision:

1.93

The relevant provisions of the International Covenant on Civil and Political Rights, the U.N. Standard Minimum Rules for the Treatment of Prisoners and the European Prison Rules are to be found in the Appendix.

CHAPTER 2: OFFENDING ON BAIL – THE EXTENT OF THE PROBLEM

Introduction

2.1

In recent decades, most if not all common law jurisdictions have experienced anxiety about the phenomenon of bail offending, that is to say, the commission of offences by a person who is, at the time of offending, on bail in respect of other offences. Irish law deals with this problem through the method of sanction, as we saw in Chapter 1, whereas many other jurisdictions, as we shall see in Chapter 4, have also chosen to employ pre-trial detention as a device. This Chapter examines the empirical evidence concerning the extent of the problem of bail offending, and various problems associated with drawing conclusions from the available statistics. We will look at a number of studies of the problem conducted in England and Scotland, primarily with a view to highlighting the different picture which emerges depending on the methodology adopted. Finally, we will examine a number of figures bearing on the position of pre-trial detainees more generally.

2.2

It is of interest to note that at different times and in different places, the problem of offending on bail, or at least of what category of offenders are perceived to be a problem, has varied. Recent Gardai statements appear to target the recidivist burglar who commits offences on bail to build up a “nest egg” for his or her release from imprisonment, secure in the knowledge that even if detected and convicted, these offences will not add substantially or at all to his or her sentence. In Australia, a number of jurisdictions have voiced concern about the granting of bail to perpetrators of domestic violence who may return to the home and abuse the complainant. Several Australian jurisdictions have also reacted legislatively to concerns about granting bail to drug offenders. At various times, concerns have been raised about the granting of bail to persons accused of highly violent behaviour, on the basis that in the case of such a serious crime, their right to liberty should cede to the higher public interest in preventing such behaviour in the future. Accordingly, the phenomenon of bail offending or

“bail bandits” is often used as a generic term, which may conceal the fact that different concerns lie underneath the surface.

Rates Of Offending On Bail In Ireland

2.3

A number of studies have been conducted in various countries to estimate the rate of bail crime in particular areas. These studies employ a number of different methodologies and we may note some of the relevant differences at this point. First, some count the number of persons who commit bail crimes, while others count the number of offences committed during the bail period. So, if a small number of persons are committing a high number of offences, the respective figures produced will be quite different. Second, and most importantly, some studies count the number of persons who are on bail at the time of arrest (or offences in respect of which persons were arrested while on bail), while other studies count the number of persons convicted of offences who were on bail at the time of the offences. This will be discussed further below. Third, some studies concern court bail only, while others are concerned with court and police bail. Fourth, in some studies, the fact that a person was on bail when charged with an offence is taken as meaning that he committed that offence while he was on bail, although he could have committed both offences prior to bail. It is crucial to know the kind of study one is dealing with when quoting the figures.

2.4

The figures used in this Paper for Ireland are based on Garda Siochana Statistics.¹ These figures indicate the percentage of offences in respect of which persons were “detected” while on court bail.² The figures do not include summary offences.

2.5

The figures are reproduced in two tables. Table A lists for the years 1988 to 1993 offences detected as having been committed in a particular year by a person on bail and the number of convictions ultimately recorded for those offences, not necessarily in the year the offence was committed, because of delays inherent in the system. Table B consists of a more detailed breakdown of detected offences for the years 1992 and 1993.

2.6

In these tables, a “detected” offence means an offence which the Gardai were satisfied was committed by a person on bail for which that person was arrested while still on bail. There is an obvious discrepancy between figures for detection and conviction. The discrepancy arises for many different reasons, including:

2.7

Every person is presumed innocent until proven guilty. So a conviction for the relevant offence is the only absolute proof that it was committed by a person on bail. However, a “detection” in this context while, obviously, not a conviction can be said to be a more significant item than a simple arrest on suspicion. When an offence is “detected” in this context, the case is concluded, as far as the Gardai are concerned. The crime is solved and no further Garda investigation takes place. While the Gardai may be unable to prove the offence, the reality is that they know who committed it and for their internal administrative purposes, they re-direct their resources to unsolved cases.

2.8

Rather than advance this detection rate as an unshakeable platform for law reform, we would simply remark that one relies on the Gardai to apply their own criteria in keeping the peace as best they can on a day to day basis. They are the only professional investigators of crime in the State.

2.9

The average clearance rate is approximately 33% in respect of reported crimes. Accordingly, one might argue that if a certain number of persons are detected offending while on bail, this represents only a third of bail offenders, the other two thirds not being solved. This would seem to suggest that the detection figures understate the rate of bail crime.

2.10

This argument assumes that the average rate of clearance for crime may be applied to bail offenders. There are a number of reasons why this may not be appropriate. First, it is possible that there are some offenders who are more likely to be detected than others; thus, if detected once, they are more likely to be detected a second time while on bail. Moreover, it is possible that a person who is on bail and who is known to the police is more likely to be arrested for a subsequent crime either because of his high visibility or because his methods of operation are known to the police. It seems likely, therefore, that while the detection rate does understate the amount of bail crime because not all bail crime is either reported to or solved by the police, one cannot assume that the relevant proportion is 33%. It is not possible to know.

2.11

Just as not all persons charged with an offence allegedly committed on bail will be convicted of that offence, not all persons will be convicted of the original offence for which they were granted bail. In a case where the offender

is ultimately convicted of the offence during the bail period but not of the original offence, it is true to say that the second offence was on bail, but it is not possible in law to say that he re-offended. Accordingly, one should bear in mind that one should not equate the rate of offending during the bail period with the rate of re-offending.

2.12

Despite all the caveats entered above, it will be seen that the Irish figures on offending on bail are lower than the English figures, despite the fact that in England, courts may refuse bail on the ground of possible commission of further offences. This would be expected to remove from circulation at least a proportion of those likely to do so. This suggests either (1) that the English rate of offending on bail is much higher than that in Ireland or (2) both that the English rate of offending on bail is higher than that in Ireland and that the court's power to refuse bail on this ground is ineffective in selecting the offenders who would commit offences while on bail.

2.13

Table A lists, for the years 1988 to 1993, the total number of indictable offences recorded, detected, detected as having been committed by persons on bail and convictions for the latter offences recorded to date.

TABLE A

2.14

Table B consists of a breakdown by offence of some of the figures for 1992 and 1993.

TABLE B

Comments

2.15

Conversely, in 1992, the rate for rape was low at 1% while the detection rate was 80%, the second highest; the rate for assaults was low at 2% while the detection rate was 74.5%, the third highest; the larceny rate for cash was low at 3% and the detection rate was 44%; the larceny rate for larceny (pedal cycles) was 3% and the detection rate was 45%. Similarly, in 1993, the highest rates were for armed robbery at 27% while the detection rate was 23%; robbery at 24% while the detection rate was 31%; larceny (pickpocket) at 25% while the detection rate was 18%; and larceny (mugging) 24% while the detection rate was 17.5%. In 1993, assaults were at 1% while the detection rate was 75%; rape was at 2% while the detection rate was 77%; larceny (pedal cycles) was at 4% while the detection rate was 54%; and larceny (shop stalls) was at 5% while the detection rate was 84%.

2.16

Finally, we may note that the rates for homicides are not separated out from the category of “other offences” nor are sexual assaults falling short of rape. Moreover, it is not possible to distinguish the gravity of assaults within the “assault” category. Accordingly, it is not possible to estimate the rates for serious violent offences.

Rates Of Offending On Bail In England

2.17

It should be noted at this point that English courts already have the power to refuse bail on the ground that the offender might commit offences if released on bail.⁵ Nonetheless, the issue of bail offending or “bail bandits”, as such offenders have been dubbed, has been and is a source of substantial controversy in recent years as a result of a number of studies conducted by the police and the Home Office. These studies employ a range of the methodologies outlined at the outset of this section and were recently summarised in a 1992 Home Office Report by Morgan.⁶ Morgan argues that the rate of bail abuse is largely dependent upon the methodology adopted, and advocates a common methodology to allow reliable estimates of the extent of offending on bail. We discuss this study partly out of comparative interest, but also to illustrate the dramatic differences in figures which result from employing different methodologies.

2.18

The following studies were examined by Morgan:

(i) Home Office Study 1978⁷

This involved a study of 7,400 defendants given bail by the magistrates' courts in England and Wales over the first six months of operation of the Bail Act, 1976. Their records were checked with the National Identification Bureau at Scotland Yard to see if they were subsequently convicted of offences committed during that period of bail. Convictions were defined to include offences taken into consideration and cautions. It was found that 9% of the defendants outside London were convicted of offences committed while on bail, and 12% of defendants in London were so convicted.

Home Office Research and Planning Unit (R.P.U.) studies 1986 and 1988⁸

2.19

This was a study of bailed defendants charged with violence, burglary, fraud, forgery, theft, and handling at Brighton magistrates' courts in 1988, Birmingham magistrates' court in 1986, and Bristol magistrates' courts over a period of six months in 1986. It was found that 10% were convicted of an offence committed while on bail.

(iii) Metropolitan Police⁹

2.20

This was a study of persons arrested and charged with certain offences during the first quarter of 1988 and then granted bail. Convictions did not include offences taken into consideration or cautions. It was found that 16% were on court bail when charged, and 12% were convicted of an offence committed while on bail.

(iv) Northumbria Police study¹⁰

2.21

This study examined the number of persons arrested for bail crime and the number of convictions for bail crime, including cautions and offences taken into consideration. The study also examined the volume of crime accounted for by the police (by charge, summons, offences taken into consideration, caution, or prison admission) and related this to bail. It was found that:

(v) Avon and Somerset Constabulary¹¹

2.22

This was a study of persons charged and arrested and it was found that 23%–34% of defendants were on police or court bail when arrested, and 28% of people who were charged were on police or court bail at the time.

(vi) Greater Manchester Police

2.23

This study reported that 26–29% of offenders charged were on police or court bail when charged.

2.24

The Metropolitan and R.P.U. studies counted the number of separate occasions on which a defendant was found guilty of at least one offence committed while on bail and found 1.3 occasions on average. The 1978 Home Office study had showed a 1.5 average.

2.25

Morgan, having reviewed these studies in the 1992 Home Office Report, commented on the difficulties of comparing studies measuring different things and recommended that in the future, only persons convicted should be measured. However, she reached the following conclusions:

2.26

A Scottish study published in 1994 examined bail decisions in three court areas (Doon, Braid and Tweed) in 1991–92 over a 36 week period.¹² A census of bail decisions was conducted in the study courts, and a detailed examination was made of a sample from all arrests where police made a decision to detain people in custody or release them on an undertaking to appear in the sheriff court. There were significant variations in bail practice in the different areas, not related to the seriousness of cases. Police in Doon had a high use of custody and release on undertaking, and the court had a high use of bail and low use of ordain (unconditional release) or remand. Police at Braid had the lowest use of custody and undertakings and tended to release most people for summons, and the court had a high use of remand. It was found that one in four accused in Doon was ultimately not convicted, whereas in Braid, one in ten was not convicted.

2.27

The study came to the conclusion that the disparity in the rates of bail offending between the different areas had less to do with real differences in recidivism and more to do with the criminal justice culture of a particular area. If the police use the summons procedure more than release on undertaking or custody, a summonsed accused will not have been on bail prior to trial, so that subsequent offences will not be recorded as bail offences; whereas in another area, a similar offender would be released on undertaking, the court might then grant bail, and his subsequent offending would show up as bail offending. Accordingly, a high level of recorded bail abuse was associated with a high use of bail by the court, but was not necessarily indicative of higher rates of recidivism. The study found that accused in an area with a lower use of bail were just as likely to have cases pending as were accused in an area with a high use of bail. The study also pointed out that figures based on arrests for alleged offences committed on bail overstated the true figures of bail offending because not all persons arrested would be convicted. When these two points were combined, it was found that there was no evidence to suggest that the public were being placed at increased risk from large scale serious further offending in the areas with a high use of bail. The study also pointed out that even proven

bail offending may not be directly equated with recidivism, because there will not always be a conviction in respect of the original offence in respect of which the person was on bail.

Irish Figures Concerning Persons Remanded In Custody

2.28

It is worth noting that the number of persons remanded in custody prior to trial in Ireland is extremely low by international standards. A greater proportion are remanded in custody after a plea of guilty or conviction to await sentence. But even the combined total of untried and unsentenced remandees is still low.

2.29

The Council of Europe Prison Information Bulletins collect statistics from member states every six months, including information on the percentage of unconvicted prisoners in the total prison population and the rate of unconvicted prisoners to the total population of the member state, expressed as a rate per 100,000. “Unconvicted prisoners” include prisoners who have not yet received a final sentence. The figures show that Ireland has low rates of pre-trial detention, on both measures, relative to other countries.

2.30

The following two tables illustrate Ireland's position in relation to other Council of Europe countries as of 1/9/1988, 1/9/1990 and 1/9/1991 in respect of the two matters referred to above, i.e. percentage of remand prisoners in total prison population, and rate of detention per 100,000.

TABLE C

TABLE D

2.31

The following tables illustrates Ireland's trend on these two indicators since 1983.

TABLE E

Disposition Of Persons Remanded In Custody

2.32

An interesting issue is the number of persons detained in custody before trial or sentence who subsequently do not receive a custodial disposition. Some writers and law reform bodies have taken the view that a person who does not ultimately face a custodial sentence should not be detained prior to trial.¹⁴ However, it might be argued that as the purpose of pre-trial detention is to ensure attendance at trial, whether or not a custodial sentence would later be ensure attendance at trial, whether or not a custodial sentence would later be

appropriate is irrelevant. A person's offence might not warrant a custodial sentence, but the fact that he might abscond before trial for that offence warrants his custody before the trial. Also, it seems likely that in many cases the fact that a person spent time in custody before trial is an important reason why he ultimately received a non-custodial sentence. In England, where protection of the public is a ground for remanding in custody, it has been argued that it is particularly inappropriate for a person to be remanded in custody supposedly to protect the public when his subsequent non-custodial disposition shows this not to have been necessary. Moreover, it has been said that it is objectionable that the criteria for imposing custodial sentences under the Criminal Justice Act, 1971 and 1993 are narrower than the criteria for imposing pre-trial custody under the Bail Act, 1976.

2.33

According to the 1991 Annual Report on Prisons and Places of Detention, 65 persons were in custody on remand on 1st January 1991 and 3,517 were committed on remand during that year, yielding a total of 3,582. Out of this total, 1,435 were “brought up and not recommitted”. Thus it seems that well over half the persons held in custody prior to final disposition were ultimately released, either because they were acquitted, the case was not proceeded with, or they received a non-custodial disposition.

2.34

In England a figure of 60% has been given for detainees who do not ultimately receive a custodial sentence. NACRO (The National Association for the Care and Resettlement of Offenders) examined the figures for 1990, and say that of the 59,620 remanded in custody before trial or sentence for which outcome is recorded, 25% received a non-custodial sentence, 3% were found not guilty or not proceeded against, and 72% received a custodial sentence, i.e. 28% of defendants for whom the outcome was recorded did not receive a custodial sentence. However, they say that this probably significantly understates the proportion of acquittals and non-custodial sentences. The “outcome not recorded” group comprised 27,215 persons who probably were not sentenced to custody, because if they had been so sentenced, the prison service would have had records of them. On recalculation NACRO say that about 60% of male defendants, 77% of female defendants, and 6% of all defendants did not receive a custodial sentence although they were remanded in custody.¹⁵ Cavadino and Gibson examined the figures for 1991, which show that 62,604 people were remanded in custody before trial or sentence. The outcome was not recorded in 23,937 cases, but of the remainder 6% were found not guilty, or the case was not proceeded with, 30% received a non-custodial sentence, and 64% received a custodial sentence. 36% of all defendants remanded in custody in respect of whom the outcome was recorded did not subsequently receive a custodial sentence. Again, the authors say that this figure understates the proportion of acquittals and non-custodial sentences because those in the “outcome not recorded” group are not likely to have been sentenced to custody because that would have been recorded by the Prison Service. The authors say

that a recalculated more accurate percentage of all defendants who did not receive a custodial sentence would be 60%.¹⁶

Average Length Of Custody

2.35

It has not been possible to obtain figures from the Department of Justice in respect of the average length of time spent in custody on remand nor the average length of time awaiting trial while on bail.

Conclusion

2.36

An examination of the Garda figures for the years 1992–93 shows that the recorded level of detections of bail offences is an average 9%. There is considerable variation in respect of different offences, however. In 1993 the highest rates were for armed robbery 27%, larceny (pickpocket) 25%, robbery 24% and larceny (mugging) 24%. The lowest rates in 1993 were assaults 1%, rape 2%, drugs 4% and larceny (pedal cycles) 4%. Even in respect of the highest rates, about three-quarters of the detections were not ultimately translated into convictions. All bail offending figures suffer inevitably from the fact that the majority of crimes are undetected by the police. In addition, the Garda figures quoted suffer from the fact that what is recorded is the number of detections which is invariably greater than the number of convictions for bail offences. Moreover, there will not always be convictions in respect of the original offences in respect of which the offender was charged.

2.37

The argument can be made that statistical evidence should never form the basis for a refusal of bail in that a refusal could only be based on the individual characteristics of a particular accused.

CHAPTER 3: PREDICTION STUDIES

3.1

In Chapter 2, we examined figures in an attempt to estimate the number of detected offences which were committed on bail. If one had determined, albeit in a rough way, the level of offending on bail, a further question would then arise. This question concerns not how many offences are committed on bail, but whether and to what extent we can identify which offenders given bail will commit crimes during that period. For example, assuming that a particular percentage, e.g. 9% of detected offences are committed by an offender who is on bail, how accurately can we predict whether a particular offender given bail will be one of those bail recidivists? This question has been addressed through the medium of prediction studies.

3.2

Prediction plays a role in a number of areas of the criminal justice system, particularly in countries other than Ireland. For example, parole decisions are based on predicting how a person will behave upon release, and the sentencing of “dangerous” persons, as provided for in some countries, involves assessing their likely future behaviour. In the United States, decisions about the use of the death penalty may depend upon assessments of persons as dangerous. In Ireland, for example, prediction plays a role in bail decisions when the court assesses whether an offender might abscond before trial. Traditionally, the predictions involved in such decisions have been made on the basis of common sense, taking into account factors which would seem to be obviously correlated with the behaviour being predicted. However, since the 1920s, attempts have been made to predict certain types of behaviour in certain populations in a more scientific way, by conducting research to discover which factors are in fact correlated to those behaviours and then applying the results of that research in the decision-making process. We will examine some of the leading prediction studies below. However, it should be noted that most studies have concentrated on the prediction of serious violent behaviour rather than re-offending generally, and most have concerned violent re-offending over a number of years rather than

THIS IS AN ORIGINAL PAGE-BREAK:

PAGE NUMBER=48

in a lesser period which would correspond to a pre-trial period.

Methods Of Prediction

3.3

A distinction is usually drawn between two different methods of prediction, the actuarial or statistical method, on the one hand and the clinical method, on the other. According to the statistical method, the data relied on to make the prediction are set out explicitly, usually in tables, and the method tends to involve reasonably simple concepts such as age, number of previous convictions, offence charged and so on. The information is then converted into a prediction using an automatic or mechanistic set of rules. Under the clinical method, the choice of data relied on may vary somewhat with the individual case and is not so rigidly determined. For example, the clinician might decide that the offender's place in the family is not relevant in one case although it is in another. The categories of data can also be more abstract; for example, the concept of “ego strength” might be used. Also, the method of converting the data into a prediction tends to be intuitive rather than mechanistic, and involves the clinician relying on a subjective combination of factors deemed by him to be relevant.¹

3.4

It is said that studies show that actuarial tables are more accurate than clinical predictions in predicting the same events.² Despite this, legal system have generally preferred to employ clinicians to make predictions about offenders than to rely on statistical surveys. This preference appears to be based on a number of factors:

3.5

The preference for the clinical method of prediction was explicitly articulated by the authors of the 1981 English Floud Report.³ They discussed the two methods of predicting dangerousness, and said that while the actuarial method was relatively cheap and involved less errors, it was objectionable because it derived the risk attaching to an individual from the fact of his being a member of a class, which was not acceptable in the administration of justice. The case study prediction method, they said, has the advantage that it predicts

on the basis of understanding and can take into account every possible item of relevant information, even though it may be less exact than the actuarial method. The Report's authors assessed that there was an important difference between:

3.6

In contrast, Monahan, who published a leading text on predictions of dangerousness in 1981, suggests that better prediction would involve a combination of clinical and actuarial expertise. For example, he says that clinical studies should direct their attention to base rates⁵ and should confine themselves to predicting dangerous behaviour in high base rate populations, that they should ensure that the base rate population is as specific and relevant as possible,⁶ and that clinicians should always search for factors which decrease the propensity for violent behaviour as well as factors which increase that propensity.

The base rate

3.7

The base rate is the statistical prevalence of the behaviour under study (e.g. violent behaviour) in a given group over a given period of time, usually one year. The group under study will vary, e.g. the population generally, persons charged with criminal offences, offenders on bail, persons convicted of crimes. The accuracy of the possibilities of prediction will vary with the base rate of the behaviour in the group under study. More specifically, the lower the base rate, the lower the rate of accuracy. It has been said that the best population on which to apply clinical predictions of violence is one with a base rate of 50%.⁷ As the base rate differs substantially from 50%, clinical differentiation becomes more difficult. If 90% of the group is non-violent, the best prediction in the

individual case is to predict them all non-violent. If another group has a base rate of 90% for violent behaviour, the most accurate prediction would be to predict all of them violent. It is often pointed out that dangerousness predictions studies are at a significant disadvantage because of the low base rate of dangerous or violent behaviour, even among small sub-groups of offenders.

The prediction process

3.8

The prediction process requires that a person be assessed at two stages. At a particular stage, data are collected about him in respect of items that are believed to relate to the behaviour one is interested in predicting, e.g. number of convictions for violence, age, present charge and so on. These are prediction variables. Some studies select items that research has already designated as important, while others employ a vast range of potentially relevant items and later identify which were the important variables after having allowed a certain period to elapse and having seen how the offenders in fact behaved. Some of the studies discussed below collected data from a variety of sources about the person, sometimes on hundreds of items. Using the information collected, the offender will be placed into a category, e.g. dangerous, non-dangerous, or into one of several categories of high to low risk, or as in the Harvard study discussed below, will be assigned a score on a “dangerousness” scale. This classification represents the prediction, based on information known about the individual. In some studies, the researchers themselves do not gather the information nor make the prediction, but examine the diagnosis made by someone else, such as a psychiatrist.

3.9

At a later stage, an assessment will be made to see if the prediction was validated. This may be based on a conviction, an arrest, or some other factor, which were referred to as criterion variables. If it was predicted that certain behaviour would occur, and it did occur, the outcome is referred to as a true positive, and if it was predicted that certain behaviour would not occur, and it did not, the outcome is a true negative. A false positive refers to a person who was predicted to engage in certain behaviour who did not so engage and a false negative is a person who was predicted not to engage in certain behaviour who did so engage. The overall rate of prediction success would be measured by looking at all outcomes.⁸ However, in the criminal justice system, there is particular concern about “false positives” because these are persons who would have been incarcerated for potential risk, who have since been shown not to pose that risk. Most studies concentrate on the false positive rate rather than the overall prediction success rate.

3.10

The importance of the false positive rate is illustrated by the following example:

Assume that one person out of a thousand will kill. Assume that an exceptional test is created which differentiates with 95% accuracy those who will kill from those who will not. If 100,000 people are tested, out of the 100 who would kill, 95 would be isolated. But out of the 99,900 who would not kill, 4,995 would be isolated as potential killers.

Prediction Studies

3.11

As we shall see in Chapter 4, some jurisdictions permit pre-trial detention of offenders who might commit further offences of any kind, while other jurisdictions only permit pre-trial detention of offenders who might commit further offences of a particular kind, e.g. serious or violent offences. The prediction studies discussed here are divided into two categories. First, studies which specifically attempted to predict bail offending are discussed. Second, studies which attempted to predict further serious violent or “dangerous” behaviour are examined. The first category is of course most relevant to the purpose of this Paper, but unfortunately, less studies of this kind appear to have been conducted. Certainly none appear ever to have been conducted in Ireland. The second category of study is primarily of relevance to a consideration of the introduction of pre-trial preventive detention concerning violent offences. However, even then, the studies conducted have certain limitations for our purposes, which will be discussed below.

1. Prediction studies concerning offending on bail

Harvard Study

3.12

In 1970, the District of Columbia in the United States introduced a statute authorising pre-trial detention for 60 days in respect of persons who might commit certain dangerous crimes while on bail awaiting trial. An empirical study was carried out by a team from Harvard University to examine the problem of bail offending and to see whether the application of the Act would make any difference to that problem.⁹ The Harvard team conducted a study of persons arrested in Boston during six months in 1968. They identified 427 persons arrested for violent or dangerous crimes who would have qualified for preventive detention had the Act been in force at the time of their arrest. The average pretrial period for these defendants was 112 days. Under the Act, as noted, detention could only be ordered for up to 60 days. Of these 427 persons, they found that:

3.13

However, since the Act's period of detention was 60 days, many of the offences would not have been caught because they occurred outside that period. Only 30% of the definite, identifiable bail crimes occurred during the period provided for by the Act. The Act would not, for example, have protected against a murder or any of the rapes or robberies committed.

3.14

We may note in passing that, regarding the overall effect on crime, the researchers concluded that had preventive detention been applied to all of the defendants studied, 10% would have been prevented from committing crimes for which they were later convicted, which represented approximately 0.75% of the total number of persons arrested for violent or dangerous crimes in Boston during 1968, and 2.5% of those convicted for those offences. The team concluded:

3.15

The research team then proceeded to the prediction aspect of their research. As the Act directs certain factors to be taken into account before pre-trial detention is imposed,¹² the team collected information relating to those factors, consisting of twenty-six items of personal data in all.¹³ They then used a number of methods to determine how accurate prediction of dangerous offending based on those items could be.

3.16

First, the research team assessed how individual variables correlated with violent recidivism during the bail period. The conclusions showed that the individual criteria were very poor predictors. For example, the initial charge, which is often considered to be one of the most important factors in setting bail, proved “little better than a random indicator of recidivism”.¹⁴ They also found that the “variables used for predicting default are of little value in predicting recidivism. None of the variables investigated had a correlation with conviction of bail crime higher than .229. For most, it was considerably lower.”¹⁵ They

concluded that the unweighted criteria did not have the potential predictive power of combined data items, using multivariate statistical analysis, to which they then turned.

3.17

In order to test the predictive power of combined variables, the research team created two “scales” of dangerousness. The first scale was created by assigning points to each of the 26 data items on the basis of the researchers' subjective views of the respective importance to be given to each item. They called this scale DS-1. This would be similar to the way in which a judge subjectively assigns weight to items of information about the defendant. The second scale was created by assigning weight to each of the factors, but this time according to correlations between recidivism and certain factors established by research. This they called DS-2. As DS-2 was based on objectively tested correlates of recidivism rather than intuition, one would expect it to have a better success in predicting. The purpose was to identify a cut-off point on each scale, above which everyone would be detained and below which everyone would be released, and in doing so, to maximise the number of recidivists while minimising the number of non-recidivists. Each offender was then given a score on each scale by totalling his points on each of the twenty-six data items. The researchers then looked at the results to see the predictive success of each scale, by comparing the predictions on each scale with the actual behaviour of the offenders in question.

3.18

If either scale were a perfect predictor, one would find a certain cut-off point such that all offenders with a score above it did in fact violently recidivate, while none of those below it would have violently recidivated. However, as it emerged, both of the scales were somewhere between perfection and complete randomness. The scale DS-2 was more accurate than DS-1, as expected. However, no matter which cut-off point was adopted on either scale, the non-recidivists were always in the majority. For example, to detain all recidivists, the cut-off score would have had to be set so low, that the false positive rates would have been 88.5% on DS-1 and 84.2% on DS-2. The most promising cut-off score would have yielded false positive rates of 74.3% (DS-1) and 58.7% (DS-1). Non-recidivists were distributed both above and below the cut-off point. Moreover, many of the more serious offences were committed by defendants with a score below that cut-off point, and the five highest DS-2 scores proved to be non-recidivists. The authors of the study commented:

3.19

The chart overleaf illustrates the full results of the Harvard study by showing the number of defendants who must be detained to prevent a given number of recidivist crimes according to:

3.20

Importantly, the authors also commented that judicial bail decision-making would be even less accurate than these scales for a number of reasons:

3.21

The authors concluded that preventive detention inevitably suffers from both under-inclusion (of recidivists) and over-inclusion (of non-recidivists).

The Melvin and Didcott Study

3.22

In a Scottish study published in 1976,¹⁸ the researchers examined a total of 10,500 bail decisions in 1972–3. Of those released on bail, 6% failed to answer bail, while 3% were recorded as having re-offended on bail, although the authors of the study estimated that the true rate might be around 8%, based on rates in less busy rural areas. Some individuals had abused bail in both ways and therefore appeared in both categories (i.e., 29 out of each group). The authors examined the characteristics of each group. Regarding those who absconded only, they found that there was less to distinguish between the absconders and the people who did not abuse bail than between persons granted and refused bail. The characteristics most closely associated with absconding were bad criminal record, especially one involving custodial penalties, no fixed abode, and unemployment. They found a slightly increased risk of absconding where the charge was theft. None of the other factors seemed to be related to absconding. The four factors identified were combined, but it was found that even the group with the highest absconding rate had an absconding rate of only 15%.

3.23

The researchers then examined the combined group of absconders and re-offenders (although they did not examine re-offenders separately). Again they found that the facts which distinguished between bail abusers and non-abusers were the greater number of previous convictions, but also lack of fixed abode, unemployment, where the charge was theft, and where there were several charges. The researchers then combined the factors and found that the very worst group had an abuse rate of 31.4%:

The Philadelphia Research²⁰

3.24

Voluntary and presumptive guidelines for decision-making have become common in the United States in the areas of parole and sentencing.²¹ In the late 1970s a project was carried out in Philadelphia in order to gather

information about bail decision-making so that bail guidelines might be developed for judges. It was felt that judicial decision-making in the area of bail was conducted under pressure of time with little information about defendants and with little feedback about the correctness of a decision, unless a particular case received negative publicity because a judge had released a person on bail who then offended. These factors tended to foster a conservative and unstructured approach to bail. The purpose of the project was to develop guidelines for judges in bail decisions so that such decisions would be structured, rational, consistent and visible. The purpose of the project was not specifically to predict pre-trial offending with a view to selecting the bad risks, although this matter was addressed in the course of the project.

3.25

The project was carried out by a research team collaborating with a steering committee comprised of judges from the Municipal Court of Philadelphia. The first phase was descriptive and consisted of attempting to uncover the factors judges were currently employing to make bail decisions. Data were collected for more than 200 items of information for each of 4,800 bail decisions between 1977 and 1979, and the factors that were currently influencing judges were identified. However, the research team also conducted research into the factors or combination of factors that were in fact predictive of bail abuse, i.e. failure to appear and pre-trial crime. In the course of its research, the project developed four separate prediction instruments: one for failure to appear, one for rearrest, one for serious rearrest, and one for combination of failure to reappear and rearrest. A scoring system was developed which allocated weight on the various factors found to be predictive of bail abuse. On the basis of these factors, groups could be identified that had greater or lesser probabilities of bail abuse in its various forms. For example, for failure to appear, the research showed the following factors to be related to likelihood of flight – current charge, prior arrest, history of behaviour on pre-trial release, and age. No information is given, unfortunately, about the factors found to be correlated to pre-trial offending. The final prediction instrument developed, which sought to predict pre-trial failure generally (failure to appear and rearrest), was validated on a sample of defendants who secured release. Defendants classified in Group 1 failed on release 12% of the time, Group 2 18% of the time, Group 3 23%, Group 4 38% and Group 5 54%.

3.26

Following discussion of the results with the steering committee, the project moved into its prescriptive phase, i.e. the formulation of guidelines. Guidelines were constructed on the basis of two axes; the first based on a measure of charge severity (15 levels), and the second based on risk of bail abuse (failure to appear or rearrest) (5 levels), leading to a grid of 75 cells. Each cell would have a presumptive bail decision, and as is the case with guidelines, the judge would be free to depart from the presumptive bail decision only in exceptional cases and would have to give reasons for the decision to do so.

3.27

A crucial matter in the formulation of guidelines is whether they are based on current practice or represent a departure from practice. The Philadelphia guidelines were neither drawn up to reflect current practice nor to

introduce a whole new practice (e.g. based on what were identified by the research as good predictors of bail abuse), but rather a hybrid model which combined judges' intuitive ideas about factors which are important and factors identified by the research as predictive of bail abuse. The information necessary for the judge to make a bail decision was presented to the court by the pre-trial service on a complete worksheet. The guidelines system was then tried out on a group of judges making bail decisions in 1981 and 1982, with a control group for the purposes of comparison, and was cautiously deemed to be a success in terms of producing greater uniformity of decision (76% of the guidelines judges' decisions fell within the presumptive guideline ranges, while only 57% of the non-guideline judges did), rationality, visibility and equity. In 1983 the judges moved to adopt the guidelines for use by the full court, and in 1984, the National Institute of Justice founded the Bail Guidelines Project to test the approach in another three jurisdictions.

3.28

However, within 90 days of the first court appearance, 12% of the non-guideline cases had failed to appear and 11% had been arrested for crimes as compared with failure to appear of 13% and alleged further offending of 10% of the experimental cases. These figures are not encouraging about the use of guidelines to reduce bail abuse.

3.29

In a 1983 article,²² Goldkamp, one of the researchers in the Philadelphia bail project, criticises previous bail abuse prediction studies for failing to include detained defendants; these studies, he says, are based on defendants who achieved pre-trial release, thus ignoring a sizeable population of defendants, in particular that section of the population most likely to abuse bail. Prediction instruments derived from such a narrow base of defendants, he argues, are inevitably more flawed than those based on all defendants:

3.30

In this article, Goldkamp describes a study carried out by him to determine whether pre-trial detention in Philadelphia was chaotic and arbitrary, as some critics suggested, or appropriately selective. He demonstrated that pre-trial detention in Philadelphia was selective by examining three matters:

3.31

Goldkamp concluded that the results of this study underscored the need to examine the entire population of defendants seeking bail (and not merely those who obtain bail) when constructing prediction instruments. However, he went on to say that while the study showed that detention was not chaotic and that a large share of the population appeared to be very high risk, this did not dispose of the problem of false positives. A sizeable minority of the detained population did not fall into high risk categories (13% were assessed as very low risk).

3.32

Thus, while Goldkamp believes that prediction rates can be improved if studies examine detainees as well as persons granted bail in order to develop prediction instruments, he accepts that the problem of error remains and urges caution accordingly. He expresses the hope that future predictive efforts would take into account the characteristics of detained defendants, thereby enhancing the strength of predictions.

3.33

The studies cited contain some important lessons. First, those who think that accurate prediction simply involves a common sense evaluation of certain obvious variables such as past record of offending, of bail abuse and of current charge, greatly underestimate the difficulty of accurately predicting bail abuse of any kind, and bail offending in particular. Rates of false positives, which remain hidden when offenders predicted to abuse bail are detained for the pre-trial

THIS IS AN ORIGINAL PAGE-BREAK:

PAGE NUMBER=61

period, emerge in significant quantities when studies test predictive accuracy on released defendants. Second, the factors which actually correlate with bail recidivism are not necessarily those thought to be important by judges making bail decisions. Third, sophisticated studies employing statistical techniques have been unable so far to eliminate substantial numbers of false positives. Fourth, judicial bail decision-making is unlikely to reach the prediction rates of the studies because of certain-factors inherent to the bail process (e.g. speed of decision-making, lack of information about the defendant). Fifth, to achieve even the limited predictive accuracy of the studies mentioned, one would need to conduct research in this country to develop prediction instruments which could then be employed by judges in bail decision-making.

3.34

Ashworth has recently commented:

3.35

We may conclude this section by noting some important comments and conclusions by Goldkamp, one of the authors of the Philadelphia experiment,²⁶ writing in 1985. First, regarding the ability of judges to predict pre-trial crime, he says:

3.36

Regarding the ability of preventive detention to reduce crime, he continues:

2. Prediction studies concerning the identification of dangerous or violent offenders – “Dangerousness Studies”

3.37

Many studies have attempted to predict violent behaviour in a post-conviction context. These studies are sometimes cited as examples of our poor ability to make correct predictions. While it is true that these studies illustrate the difficulty of achieving accurate predictions, there are a number of reasons why one should hesitate to assume that the accuracy achieved in these studies would be similar to those which could be achieved in the bail context:

3.38

Dangerousness studies may be sub-divided into those which employed the clinical method and those which employed the statistical method.

(a) Clinical studies

3.39

In 1966, the United States Supreme Court held in Baxstrom v. Herold³⁰ that an offender detained in a hospital for the criminally insane had been denied equal protection of the law under the Fourteenth Amendment by being detained beyond his three-year sentence in this institution without a new hearing to determine his dangerousness, and held that he must be released or granted a civil commitment hearing at which the State would have to prove his dangerousness. Following this decision, Baxstrom and 966 other patients in the same legal class were transferred from hospitals for the criminally insane to civil mental hospitals for re-evaluation. The majority were eventually released to the community. This provided a rare opportunity for a “natural experiment” in the accuracy of predictions. That the offenders were diagnosed as “dangerous” was presumed from the fact that they had been held in a hospital for the criminally insane, and would have continued to have been so held had it not been for the Supreme Court decision. It was found that only 20% of the patients were assaultive to persons in the civil hospital or the community at any time during the four years following their transfer, and only 3% were sufficiently dangerous to be returned to a hospital for the criminally insane during the four years after the decision. Of 121 patients released into the community during an average two and a half years of freedom, only 8% were convicted of a crime, and only one of those convictions was for a dangerous act.³¹

3.40

However, it must be noted that no psychiatric reports were available for inspection; as noted above, the judgment of “dangerousness” was inferred from their presence in the secure hospital.

3.41

Thornberry and Jacoby (1979)³² studied a group of offenders who had been released following a successful petition for their release in Dixon v. Pennsylvania in 1971.³³ Patients at a mental hospital for the criminally insane whose criminal commitments had expired were retained without judicial action under legislative provisions which allowed commitments to be based on two physicians' certificates and only required administrative review of each case. The

relevant statutory provisions were held unconstitutional for breach of due process and equal protection (i.e. vis-á-vis the civilly committed mentally ill). Some of the patients were civil commitments transferred from mental hospitals because they were disruptive, while some were criminal commitments (a) transferred from prison awaiting trial, (b) transferred from prison while under sentence, or (c) committed after conviction. The immediate outcome of the case was that the patients were transferred to civil mental hospitals where after re-evaluation, decisions concerning their release to the community would be made.

3.42

The hospital staff predicted that a majority of the Dixon patients would be dangerous if released from maximum security confinement. The authors of the study found that this view was based not so much on either clinical or statistical prediction but rather on political prediction, in that the prediction was based on assumed characteristics of the group rather than individual prediction.³⁴ They were predicted to be dangerous by virtue of their status as mentally ill offenders. The authors justified evaluation of these admittedly unscientific predictions on the basis that they are the most common type of prediction used in decisions concerning the release of the criminally insane.

3.43

The characteristics of this group of subjects were described by Thornberry and Jacoby as follows:

3.44

Upon examination of the career of these patients in the civil mental hospitals, the researchers found that 65% were released into the community. Thus the psychiatrists at the transfer hospitals did not agree with the earlier view that the majority of them required maximum security or even institutionalised care.

3.45

Upon examination of the patients after transfer to the civil mental hospitals, it was found that only 19% of the subjects were involved in serious violence; only 5% were involved as initiators or victims of more than two violent incidents and fewer than 3% were so frequently involved in violence as to be considered persistently violent.³⁵ The follow-up period was, however, only an average of one year at the hospital.

3.46

Upon study of those released into the community over an average four-year follow-up period (414 men), they found that only 23.7% were re-arrested for any offence at least once, and only 14.5% could be classified as dangerous after their release, “dangerousness” being illustrated by an arrest for a violent offence or any rehospitalisation for a violent act.

3.47

Moreover, the researchers analysed the variables that might be able to predict membership of the 14.5% who proved dangerous, or the 23.7% who were re-arrested for any offence, e.g. previous sex offences, race, original type of commitment, type of seriousness of offence, and age. However, it was found that while statistically significant differences existed between the dangerous subjects and the non-dangerous subjects, and the re-arrested and the non-arrested group, even in combination, these variables did not provide the basis for accurate statistical predictions.³⁶ Using multivariate analysis with the above-mentioned and other variables, the arrested/not arrested analysis yielded a rate of false positives of 60%; and the dangerousness analysis yielded a false positive rate of 70%:

3.48

Cocozza and Steadman³⁸ conducted a study of indicted felony defendants found incompetent to stand trial, for whom a court psychiatric report on dangerousness was made between September 1971 and August 1972 under the New York Criminal Procedure Law 1971, which made provision for a determination of dangerousness in respect of such defendants. The determination as to dangerousness was based upon the oral and written testimony of two psychiatrists. A court finding of dangerousness permitted placement of the incompetent defendant in a facility run by the Department of Correctional Services, whereas a finding of not dangerous led to commitment in a civil mental hospital run by the Department of Mental Hygiene. The study examined 257 defendants. Of these, 154 (60%) were evaluated as dangerous by the reporting psychiatrists.

3.49

The researchers found that there was very little difference between those evaluated as dangerous and those evaluated as not dangerous according to three sets of variables:

3.50

The factor which seemed to differentiate significantly between the groups was the current charge. The data revealed that the more serious the crime the defendant was charged with, the more likely was the psychiatrist to reach a finding of dangerousness. More important than the legal category of crimes was the actual or perceived level of violence associated with the crime. The researchers found that with this single piece of information, one could have successfully predicted the findings of the psychiatrists in over 73% of the 108 cases in which the defendant was charged with a violent crime. However, on exploration of the reports written by the psychiatrists, the researchers found that they rarely overtly justified their predictions on the basis of the current charge.

3.51

Legally, the decision with respect to dangerousness rested with the judges, but in 87% of cases, the psychiatrist's recommendation as to the defendant's dangerousness was accepted by the court.

3.52

Although the statute provided for commitment in a civil hospital in respect of the group evaluated as non-dangerous, the Department of Mental Hygiene established a new, high security, civil facility for those patients. The two forms of hospital were therefore almost identical during the first six months of the study. Also, the lengths of hospitalisation prior to their release to the community or return to court were very similar for the two groups.

3.53

By examining hospital files and arrest records, the researchers were able to compare the two groups on the following indicators:

3.54

They found that the patients evaluated as dangerous were no more dangerous than the other group; all of the indicators revealed only slight differences between the two groups and none were statistically significant. Indeed, in respect of one of the most important indicators, the number subsequently arrested for violent crimes, only 14% of the dangerous patients released to the community were subsequently rearrested for a violent crime, while 16% of those evaluated as not dangerous released to the community were subsequently arrested for a violent crime. The authors concluded:

3.55

The Floud Report, however, said that the study provided “better evidence of bad practice than of the state of the art of assessing dangerousness” and that the psychiatric reports were based on “routine and slovenly procedures.”⁴⁰

3.56

Kozol et al (1972)⁴¹ engaged in a ten year first-hand study involving 592 male offenders, most of whom had been convicted of violent sexual offences. Each offender, legally sane, was referred by the Court to the Massachusetts Centre for the Diagnosis and Treatment of Dangerous Persons where he was examined by at least two psychiatrists, two psychologists, and a social worker and was subjected to clinical examinations, psychological tests and a reconstruction of family history from the offender, family, friends, neighbours, teachers, employers, the court, and the correction and mental hospital record. This was

done under a 1958 Massachusetts law which provided for psychiatric examination of a convicted offender for offences in which sex was an element and under which any person diagnosed as “dangerous” would become subject to life-long detention under a special mental health commitment and would receive treatment.

3.57

Of the 592 admitted for observation, 435 were released. Of these, release had been recommended by the Centre for 386, and release opposed for 49. Of the 386 released with approval, 304 were recommended for release at the initial stage; and of these, 8.6% committed a serious assaultive crime. After treatment for 43 months, a further 82 patients were recommended for release; of these 6.1% committed serious assaultive crimes; i.e. of the 386 recommended at some stage for release, there was a recidivism rate of 8%. Concerning the 49 who were released against recommendation, there was a recidivism rate of 34.7%. This illustrated some predictive ability, although in the group deemed dangerous, there was a false positive rate of 64%. It may be noted that the follow-up period was five years. The authors of the study concluded:

3.58

However, the Floud Report pointed out that no account was taken in this study of the differences in length of time during which the individuals were at liberty. They also point out that while the false positive rate in the “dangerous” group was high, this was only one-fifth of the total group judged dangerous, and it is plausible to assume that as the court ordered their release, they were borderline cases in any event.⁴² The Floud Report also cites a later review of this study by Kozol et al,⁴³ in which they took the view that 29 of the 49 cases in question had been misclassified as dangerous, in light of later experience of diagnosis. Had these men been assessed differently with the benefit of experience, the number of disputed cases would have been 35 instead of 49, and the chances of a correct diagnosis of dangerousness would have been 50:50, instead of 2:1 against.

3.59

Another first-hand study, concerning the first ten years of operation of the Patuxent Institution in Maryland, concerned 421 patients, each of whom had received at least three years of treatment. The psychiatric staff opposed the release of 286 patients on grounds of danger who were nonetheless released by the court, i.e. a much greater number than in the above study. The recidivism rate was 39–46% (46% if patients were released directly from the hospital and 39% if a conditional release was imposed). The 135 which the staff had recommended as safe who were released had a recidivism rate of 7%. However,

it is important to note that recidivism was measured by any new offence during the first three years after release, not merely offences of violence.⁴⁴

3.60

In 1981 having studied a number of the American dangerousness studies, Monahan concluded that:

3.61

Monahan acknowledged that criticisms had been made of the research for testing things such as bureaucratic inertia rather than prediction accuracy, and pointed out that much violence which occurred on release may have escaped detection.⁴⁶ He concluded that:

3.62

The authors of the 1981 Floud Report came to a similar conclusion in respect of American and British studies, saying:

Moreover, they added:

3.63

In a 1985 article, Ewing reviewed some further studies attempting to predict violence among juveniles.⁴⁹ The first study examined the accuracy of psychiatric consultants' assessments of the potential for violence of 511 wards of the California Youth Authority. Of the 118 youths in this group clinically evaluated as posing a moderate or high potential for violence, only nine (7.6%) recidivated by committing a violent offence during a fifteen month post-release follow-up period, i.e. yielding a false positive rate of over 92%. The second study combined the clinical and research literature regarding violent behaviour among juveniles and isolated thirty factors thought to be predictive of violent behaviour among children and adolescents. In addition, a survey of juvenile court psychiatric clinics yielded an additional sixteen factors said to be predictive of violent behaviour in juveniles. The researcher then examined the court and clinical records of 122 juveniles referred for psychiatric evaluation before the disposition of their case, and followed the juveniles for one year. Only 7 of the 122 committed a violent offence, and the researcher could find no statistically significant relationship between the commission of violent offences and any of the predictor variables.⁵⁰

3.64

Whether or not this is the best that clinical predictions can do continues to be debated. Some say that with better techniques, accuracy could be improved. For example, it is suggested that clinicians should combine their own expertise with actuarial knowledge,⁵¹ and pay attention to situational as well as dispositional factors.⁵² It has also been suggested that much more detailed information should be collected about the persons whose behaviour is being predicted.⁵³ It is unlikely that we have heard the last word on ability to predict violent behaviour. As the authors of one text say:

3.65

Floud and Young said, in similar terms:

(b) Actuarial/statistical studies of dangerousness

3.66

Among the better known statistical studies of predicting violence are three studies reported by Wenk, Robison and Smith (1972) undertaken in the California Department of Corrections.⁵⁶ The first in 1965 attempted to develop a violence prediction scale to aid in parole decision making. The population under observation had a base rate of 5% violent behaviour. The prediction items included the offence for which the person was committed, the number of prior commitments, opiate use and length of imprisonment. The scale was then validated by examining the number of parolees who engaged in violence. They were able to identify a small class of offenders (less than 3% of the total), who were three times more likely than other offenders to commit acts of violence while on parole for fifteen months. However, only 14% of this small group in fact engaged in violence, leaving a false positive rate of 86% even for this group. The second study concerned a group with a lesser base rate of 2.5% violent behaviour. The individuals in a group of 7,712 parolees were assigned to one of six categories of risk for violent behaviour. One in five were assigned to a “potentially aggressive” category, and the rest to a less aggressive category. During a one year follow-up, the rate of conviction and imprisonment for crimes involving violence was 3.1 per thousand for the “potentially aggressive” group, and 2.8 per thousand for the less aggressive group. For every correct identification of a potentially aggressive individual, there were 326 wrong ones and only five of the 1,630 parolees put in the high risk category were convicted of crimes of violence.

3.67

The third study examined 4,146 California Youth Authority wards with a base rate of 0.3% violent behaviour. The subjects were followed for 15 months after their release from juvenile institutions. Data on 100 variables were analysed retrospectively to see which items predicted a violent act of recidivism. The study concluded that a parole decision-maker who used a history of actual violence as his sole predictor of future violence would have false positives of 19 in every 20 predictions. Several multivariate regression equations were developed from the data, but none achieved a better rate than an eight to one false positive to true positive ratio. After analysing data on 100 variables and using multivariate techniques, the researchers were never able to attain better than 88% rate of false positives. It may be noted that these studies used conviction for violent behaviour as a measure of recidivism.

3.68

In 1978, the Department of Corrections of the State of Michigan undertook a study of a sample with a base rate of 10.5% violent behaviour.⁵⁷ Data on 350 variables was collected for 2,200 male inmates released on parole in 1971 and an actuarial table relating to arrest for violent crime while on parole was developed. The study was then validated on a portion of the sample. Persons were classified into categories according to very high risk, high risk, middle risk, low risk, or very low risk. It was found that among those designated as being at very high risk (4.7% of the sample) the rate of violent recidivism during a fourteen month follow-up was 40%. The violent recidivism rates for the high risk and middle risk group were only 20.7% and 11.8% respectively. Michigan replicated the study using a sample of 1,200 inmates released in 1974; with a follow-up period of 14 months parole and a base rate of violent crime of 16%. The recidivism rates were found to be similar to the previous study, namely, 33% for the highest risk category.

3.69

The Floud Report mentioned three British studies, in which it was found that one variable alone (having three or more previous convictions for violence) identified groups of offenders of whom more than 1 in 2 were subsequently reconvicted of violence, including robbery. Walker, Hammond and Steer (1970)⁵⁸ reported studies of two samples. The first involved a study of 264 Scots who incurred their first conviction for violence in 1947. For the 11 with a fourth or subsequent conviction for violence, 55% incurred a further such conviction. The study was cut off at 1958 convictions. The second involved 401 Londoners convicted of indictable violence in 1957. For the 21 who incurred a fourth or subsequent such conviction, 52% incurred another one in the next five years.

3.70

Phillpots and Lancucki⁵⁹ described a sample of adults convicted of a standard list of offences in 1971. Of the 828 men who had been either convicted of an offence with violence or robbery or had at least one such previous conviction, 43 had three or more convictions of the same kind, and of these, 54%

were re-convicted of further such offences within six years of sentence.

3.71

In 1980, a British Home Office Study was published, the purpose of which had been to see if it were possible to identify a group of dangerous offenders who were said to be disproportionately responsible for dangerous offences.⁶⁰ Details about every one in ten prisoners in the south east of England were collected from a wide variety of sources in 1972. Out of 811 men, the researchers Brody and Tarling identified a group, 38% of the sample, which included “men who specialised in lucrative and successful robberies, burglaries or frauds, who had badly wounded or killed their victims or would have carried out threats to do so, and who had sexually assaulted children or adults. Life sentence prisoners were automatically included in the serious group”.⁶¹ From this group they then selected groups of dangerous offenders using categories of “serious danger”, “possible danger”, “no serious threat”, and “not dangerous”. They rate 77 (9.5% of the sample) of the prisoners as dangerous. According to the authors:

3.72

Early in 1979, most prisoners in the sample had been released from their original sentence for at least five years. Information was available in respect of 700 out of the original sample. Out of these 700 offenders, 77 were convicted of robbery, physical or sexual assault, and only 18 were re-convicted of “dangerous” offences. Of the 77 re-convictions mentioned, only 13 were committed by the 52 “dangerous” offenders free to re-offend.⁶³ However, dangerous offenders were responsible for about half of the dangerous offences.

3.73

The authors of the study commented:

3.74

The authors concluded:

Prediction And Procedural Safeguards

3.75

As we shall see on our section on the United States, it has been asserted by courts and writers that procedural safeguards at detention hearings can reduce the level of false predictions by courts. In the bail context, among the procedural safeguards suggested are:

3.76

Reflection on the above studies will show that it is highly unlikely that such procedural safeguards will reduce the level of false predictions below those on the studies, because these safeguards were already built into the research designs which produced such low levels of prediction success. For example, many of the studies only concerned offenders charged with or convicted of the limited category of serious crimes. Most if not all involved extensive and sometimes exhaustive research into the person's background, usually far more extensive than a court could or would ever engage in. Regarding the suggested “safeguard” of limiting the prediction to certain types of crime, one writer has said:

3.77

Similarly, the standard of proof “safeguard” will not improve accuracy. As the empirical predictions are more often wrong than right, this would hardly satisfy even the lowest standard of proof. It seems meaningless to ask the risk to be proved as greater than 50% in court, when studies have demonstrated that this kind of precision has never yet been attained. While the absence of procedural safeguards would no doubt make predictive accuracy even worse, it would be foolish to expect procedural safeguards to improve on the figures so far reported.

Conclusion

3.78

This chapter has illustrated the complex nature of studies which test predictive accuracy. It shows the discouraging results which have emerged from such studies about our ability to predict, at least when such studies concentrate on violent crime. However, it would be worth undertaking a study in this country to establish the degree to which one could predict bail offending generally. Such

a study could learn from the methodological pitfalls seen above and might concentrate on crime that tends to be serial, i.e. child sexual abuse, burglary, shoplifting, larceny from motor-vehicles, mugging or handbag snatching.

3.79

It is important to realise, however, that the figures provided by prediction studies are the beginning, not the end, of discussion about the acceptability of pre-trial preventive detention. As Monahan has said:⁶⁷

CHAPTER 4: THE LAW RELATING TO BAIL IN OTHER JURISDICTIONS

4.1

This Chapter examines the law relating to bail in a number of other common law jurisdictions. The jurisdictions examined are the United States, Canada, England and Wales, Scotland and Australia. Although each jurisdiction has focused to a greater or lesser extent on certain aspects of bail, certain common concerns and themes emerge, as we shall see. The United States and Canada approaches are particularly interesting in that they are countries which have had to forge their bail laws within the constraints of constitutional imperatives. As we shall see, they have interpreted those constitutional constraints in a different manner from the Irish Supreme Court. We also address the European Convention for the Protection of Human Rights and Fundamental Freedoms, and very briefly, a number of Continental European systems.

UNITED STATES

Introduction

4.2

The topic of bail has been on the reform agenda since at least the 1960s in the United States. Different trends at different periods are, however, discernible. The first wave of interest in bail reform concentrated on the perceived inequalities and injustices to defendants, arising in large part out of the dominance of money-based bail, which was felt to discriminate against economically deprived defendants who would be nonetheless good bail risks. It was also felt that unlimited judicial discretion in the area of bail led to arbitrary and irrelevant factors being relied on to refuse bail. These concerns led to, for example, the celebrated Manhattan Bail Project developed by the Vera Institute for Justice which sought to assist judicial decision-making by providing information to the court about the defendant's background, and the Federal Bail Reform Act, 1966, which sought to decrease the emphasis on money bail. There have also been developments in relation to the edging out of bail bondsmen, the

introduction of bail deposit schemes, and supervised release programmes. However, from at least the 1970s onwards, interest in the bail topic started to focus on the issue of bail offending and public safety from persons charged with offences who were released pending trial. In 1970, the District of Columbia enacted a statute authorising preventive detention in respect of defendants deemed to be at risk of offending while on bail. This example was followed by most other states in the years between the enactment of this statute and the Federal Bail Reform Act, 1984 which authorised pre-trial detention for similar reasons and was, indeed, modelled on the District of Columbia Act. The constitutionality of the Federal Act was upheld in 1987 in the Supreme Court decision in U.S. v. Salerno,¹ but it would be inaccurate to say that this decision has put to rest concerns about the issue of preventive detention. Numerous writers have criticised the decision in trenchant terms, and there seems to be agreement that the application of the Act in some circumstances will prove unconstitutional. Each of these developments will be discussed in more detail below. For the moment, it will suffice to say that the present climate of opinion regarding the bail issue is marked by the residue left by these two waves of opinion. Two of the leading authors in this area have said:

4.3

Goldkamp has also noted the changing political climate providing a backdrop to these concerns:

The Bail Reform Act, 1966 And The Manhattan Bail Project

4.4

The Federal Bail Reform Act, 1966 was introduced on a wave of concern about the discriminatory use of bail and the conditions of bail. Its primary purpose was to reduce reliance on money bail by encouraging release on recognisance rather than using financial bonds. However, the ground upon which restrictive conditions of bail, financial bail, or refusal to grant bail could be decided was the question of whether the accused would appear for trial. The Act did not expressly permit the court to consider the question of whether the accused might pose a danger to the community or commit further offences while on bail. At the time of its enactment, Congress considered the question of preventive detention and refused to endorse it.

4.5

It has been generally accepted that in the years which followed, judges were influenced by the question of public safety and bail offending, and achieved preventive detention sub rosa by setting money bail at high levels to prevent an accused meeting it in order to prevent his release where he was felt to be a potential bail offender. Thus preventive detention existed de facto although without the legitimacy of supporting legislation. Indeed, one of the reasons offered to support the introduction of the Federal Bail Reform Act, 1984 was that preventive detention decision-making would be brought out into the open and regulated.

4.6

The purpose of the Manhattan Bail Project, as stated above, was to improve the quality of information reaching the court about the accused in order to enable him to release “good risk” defendants who might otherwise have been detained. The following Table illustrates the point scoring system of the New York Release on Recognizances Project, formerly the Manhattan Bail Project. The project officer collects information on all of these matters and gives scores to the defendant, following which process he is deemed a good or a bad bail risk.

To be recommended the defendant needs:

4.7

A more recent project of this type is the Nassau Bail Project, also sponsored by the Vera Institute. This scheme targets those who have already been refused bail. Again the target population is subject to an investigation process, in the course of which information about community ties, employment, and likely outcome of the case are obtained. There is intensive monitoring of persons released to ensure that they comply with the conditions of bail.

The Introduction Of Preventive Detention Prior To The 1984 Act

4.8

As early as four years after the reforming 1966 Act, the District of Columbia introduced the first Act permitting pre-trial detention of potential bail offenders, the Court Reform and Criminal Procedures Act, 1970. It is perhaps of interest to note that the Act was passed in the wake of a National Bureau of Standards study which showed that the amount of serious felony pre-trial crime was about 5%, that there was no correlation between the type of crime for which the first arrest was made and the severity of the second offence, that most bail recidivism did not occur in the immediate post-arrest period, and that prediction rates were low.⁴

4.9

The statute contained certain narrowing provisions included to satisfy what were thought to be constitutional pre-requisites for a preventive detention scheme, namely:

4.10

The District of Columbia Act was challenged in U.S. v Edwards⁵ and was held not to be unconstitutional on its face by the District of Columbia Court of Appeals. Following the decision in U.S. v. Edwards, attention shifted in many quarters from the question of the constitutionality of preventive detention per se, to the question of what procedural safeguards would render it constitutional.

4.11

By 1984, thirty-four states had laws addressing danger to the public as an aspect of bail decision-making, and ten states had revised their laws through constitutional amendments.⁶ These statutes, while similar in orientation, differed in detail. They referred variously to the “safety of the community”, the “danger to the public” and “danger to any other person or the community”. Dangerousness was variously explained; some statutes created categories of defendants charged with particular offences, e.g. felonies, or felonies in particular

circumstances, e.g. on bail, on probation, on parole; others eschewed the categorical approach and provided general phrases, such as whether liberty would constitute an unreasonable danger to the defendant or the public. The eligibility of defendants for pre-trial detention was linked to various factors; the nature of the offence charged, current charge and past record, current charge and whether on probation, parole, or bail, assessments of risk of danger. The current charge was inevitably important, although often combined with other factors.

The Federal Bail Reform Act, 1984

4.12

This development towards the enactment of preventive detention statutes culminated in the enactment of the Federal Bail Reform Act, 1984, enacted as Chapter One of the Comprehensive Crime Control Act of 1984 and replacing the 1966 Federal Bail Reform Act. It was similar in many respects to the original District of Columbia Act, although one essential difference was that no time limits on detention were included. The Act retained the emphasis of its predecessor in providing that a judicial officer should impose restraints in an ascending order of necessity, culminating in pre-trial detention only if no condition or combination of conditions would serve to prevent bail abuse. The “ladder” runs through release on recognisance or execution of an unsecured appearance bond, through release on increasingly restrictive conditions, to detention. At each stage, the judicial officer must impose the minimum restraint possible.

4.13

However, the key provision, introducing preventive detention on the hitherto federally unrecognised ground is s.3142(e), which provides as follows:

4.14

Subsection (g) lists the factors to be considered in determining whether there are conditions of release that will reasonably assure the appearance of the person as required and the safety of any other person and the community as follows:

4.15

These factors are to be considered when addressing flight and safety of the community. The Act does not weight them in any way, nor does it indicate that certain factors are more relevant to flight than to safety, and vice versa.

The presumptions

4.16

The Act also created two rebuttable presumptions “that no condition or combination of conditions will reasonably assure the safety of any other person and the community”. The first is the “previous violator presumption”, which applies where the defendant has, while on pre-trial release in the preceding five years, been convicted of certain offences. The second is the “drug and firearm offender” presumption. In practice, the latter is most often relied on. Persons in these categories are presumed to be flight and/or danger risks and the burden of proof is shifted to the defendant to show otherwise. Under the Act, the prosecution can invoke the rebuttable presumption only if the judicial officer finds that “probable cause” exists to believe that the defendant committed the crime charged. It is generally agreed that the potential protection offered by this safeguard has been diluted, if not lost, by the fact that most courts have held that an indictment itself establishes probable cause to believe that the defendant committed the offence charged, and therefore triggers the presumption.

Procedure

4.17

Section 3142(f) provides that the judicial officer shall hold a hearing to determine whether any condition or combination of conditions will reasonably assure the appearance of the person as required and the safety of any other person and the community:

> upon motion of the attorney for the government, in cases involving

> upon motion of the attorney for the Government or upon the judicial officer's own motion, in a case that involves:

4.18

S.3142(f) also provides that at the hearing, the person has the right to be represented by counsel, to have counsel appointed if his or her means are inadequate, an opportunity to testify, to present witnesses, to cross-examine witnesses who appear, and to present information by proffer or otherwise. The rules concerning the admissibility of evidence in criminal trials do not apply. The facts used by the judicial officer for a finding in favour of detention “shall be supported by clear and convincing evidence”.

4.19

The judicial officer must include in the detention order written findings of fact and a written statement of the reasons for detention. He must direct that the person be committed to the custody of the Attorney General for confinement in a corrections facility separate, “to the extent practicable”, from persons awaiting or serving sentences or being held in custody pending appeal and direct that the person be afforded “reasonable” opportunity for private consultation with counsel.

4.20

To meet the concern that use of high money bail to achieve pre-trial detention would circumvent these procedural formalities, the Act states that the judge may not impose a financial condition that results in the pre-trial detention of the person.⁸

Breaches of bail

4.21

A defendant who wilfully fails to appear before a court as required by a condition of his release may be subject to forfeiture of any security given or pledged for his release, and is liable to be guilty of an offence, commonly known as “bail jumping”. Under the Federal Act, punishment is geared to the relative seriousness of the offence with which the defendant was charged or convicted. For example, if the original offence is punishable by death, life imprisonment or over fifteen years imprisonment, the appropriate penalty is a maximum ten years imprisonment, whereas if the penalty for the original offence is five or more years imprisonment, the appropriate penalty is a maximum ten years imprisonment. If the penalty for the original offence is five or more years imprisonment, then

the penalty for bail jumping is also five years. It is a defence that uncontrollable circumstances prevented the defendant's appearance and that he did not contribute to the creation of such circumstances in reckless disregard of the requirement to appear or surrender, and that he appeared or surrendered as soon as such circumstances ceased to exist. It may be noted that this defence is more restrictive than the Irish equivalent under s.13 of the Criminal Justice Act, 1984.⁹

4.22

Under s.3147 of the Act, if a defendant is convicted of an offence committed while on bail under the Federal Act, his punishment may be enhanced by a term of imprisonment of not more than ten years if the offence was a felony, and one year if a misdemeanour, and must be consecutive. If he violates a condition of release, he is subject to a “revocation of release, an order of detention, and a prosecution for contempt of court”, under s.3148(a).

Relevant Pre-Salerno Caselaw

4.23

The first significant challenge to the Bail Reform Act to reach the Supreme Court was the case of U.S. v. Salerno,¹⁰ discussed below. However, it may be useful at this point to draw attention to a number of cases which preceded that decision and were drawn on by the Supreme Court in Salerno to reach its conclusions. The main themes in these cases are the distinction between a regulatory and a punitive measure, the validity of predictions of dangerous behaviour, and the constitutionality of preventive detention provisions relating to juveniles.

4.24

In Kennedy v. Mendoza-Martinez,¹¹ the Supreme Court addressed a challenge to an immigration law providing for the automatic forfeiture of citizenship upon draft evasion. A majority of the Court held that the measure was unconstitutional because the sanction of deprivation of citizenship had been employed by Congress as a punishment without affording the procedural safeguards of the Fifth and Sixth Amendments.¹² The Court described the test to be employed to determine whether a governmental act serves regulatory or penal purposes. It said that in the absence of an express legislative intent to punish, the courts must consider seven factors in their review of a governmental act:

4.25

The first factor is congressional intent, and it is only if the congressional intent is non-punitive that the other factors must be examined. In total, therefore, it is an eight-pronged test. The legislative history of a statute is relevant in analysing its character as penal or regulatory. In this particular case, it was clear that the congressional intent had been punitive, and the Court held that loss of citizenship for draft evasion was punishment and therefore required a conviction.

4.26

The distinction between a regulatory and a punitive measure again came under Supreme Court scrutiny in Bell v. Wolfish,¹³ where a number of pre-trial detainees challenged certain conditions of their confinement. The majority opinion delivered by Rehnquist C.J. discussed the Mendoza test in its entirety, but went on to apply a three-pronged test; in the absence of showing an express intent to punish, the determination of whether a statute is punitive or regulatory turns on whether an alternative purpose to which the restriction may be rationally connected is assignable for it and whether the detention appears excessive in relation to the alternative purpose. This abbreviated test was subsequently to be relied on by the majority in Salerno. Another interesting aspect of the Bell decision was that the Court held that the presumption of innocence had no application to a determination of the rights of a pre-trial detainee during confinement before the trial has even begun. The Supreme Court in Salerno relied on this view, although the Bell challenge involved the conditions of confinement rather than the fact of confinement itself.

4.27

The question of predicting dangerousness in adults has arisen on a number of occasions before the Supreme Court. In Jurek v. Texas,¹⁴ the defendant was convicted of murder in the course of committing and attempting to commit kidnapping and forcible rape. Under Texas law, a separate pre-sentence hearing was required to be held before the jury, at which one of the questions the jury must answer in the affirmative was whether there was a probability that the defendant would commit criminal acts of violence that would constitute a continuing threat to society. The jury answered this question in the affirmative and the defendant was sentenced to death. The only testimony on the defendant's future dangerousness before the court was lay testimony. On certiorari, the Supreme Court affirmed. The main question in the case was whether this method of deciding upon a death penalty removed the death penalty from the category of “cruel and unusual punishment” under the Eighth and

Fourteenth Amendments, and a majority of the Court held that it did. However, seven justices of the Court also expressed the opinion that the question with regard to dangerousness was not so vague as to be meaningless merely because of the difficulty of predicting future criminal conduct. Stevens J. said:

4.28

In Estelle v. Smith,¹⁶ another Texas defendant was convicted of murder and sentenced to death after a sentencing hearing. At the hearing, a psychiatrist for the State who had conducted an examination of the accused testified that, among other things, the defendant was going to commit other similar criminal acts if given the opportunity to do so. On appeal to the United States District Court for the Northern District of Texas, the Court vacated the death sentence, and on certiorari, the Supreme Court affirmed. The reason for the invalidity of the death sentence in this case was that the defendant had not been advised that he had the right to remain silent at the pre-trial psychiatric examination or that any statement he made could be used against him, and thus his Fifth Amendment privilege had been violated. His Sixth Amendment right to the assistance of counsel was also violated since defence counsel were not notified in advance that the examination would encompass the issue of their client's future dangerousness. However, the Court referred with approval to its view of prediction in Jurek v. Texas as expressed in the quotation above, and went on to say that they were in “no sense disapproving the use of psychiatric testimony bearing on future dangerousness”. It may be noted that the American Psychiatric Association had filed a brief as amicus curiae taking the view that clinical predictions relating to future violence are of very low reliability and that psychiatrists had no special qualifications in making such forecasts.

4.29

In Barefoot v. Estelle,¹⁷ a defendant was convicted of capital murder in a Texas court and sentenced to death following a sentencing hearing. The jury had found that there was a “probability that the defendant would constitute a continuing threat to society”. The State introduced into evidence the defendant's prior convictions and his reputation for lawlessness. The State also called two psychiatrists, who testified that the defendant would probably commit further acts of violence and represent a continuing threat to society, although they had not personally examined the defendant and gave their evidence in response to hypothetical questions. At the sentencing hearing, no evidence was offered by the defendant to contradict the evidence of these witnesses. One of the State's witnesses at trial testified that a psychiatrist could predict the future dangerousness of an individual if given enough background information. On appeal to the Texas Court of Appeals, the defendant argued that the use of psychiatrists in this regard was unconstitutional because psychiatrists are not competent to predict future dangerousness, and that their predictions are so likely to produce erroneous sentences that their use violated the Eighth and Fourteenth Amendments of the Constitution. He also argued that psychiatrists should not be permitted to testify about future dangerousness in response to hypothetical questions and without having examined the defendant personally. The Court affirmed the conviction and sentence. A petition for certiorari to the Supreme Court was denied, and subsequently he lodged a petition for habeas corpus in the District Court for the Western Court of Texas. The District Court rejected the claims put forward by the defendant. It may be noted that at the District Court hearing, the two expert witnesses called by the State agreed that accurate predictions of future dangerousness can be made if enough information is provided. The defendant called two experts, one of whom criticised the diagnosis based on hypothetical questions, while the other said that psychiatric predictions of future dangerousness were wrong two out of three times. The defendant appealed to the Court of Appeals for the Fifth Circuit and then to the Supreme Court. The American Psychiatric Association supported the defendant's arguments in an amicus brief. In an opinion by White J., joined by Burger C.J., Powell, Rehnquist and O'Connor JJ., the Supreme Court affirmed the judgment of the District Court, and held that it had not been shown that psychiatric testimony regarding future dangerousness was almost entirely unreliable or that the fact-finder and adversarial system are not competent to test the competence of the testimony's shortcomings. It was proper to base such testimony on hypothetical questions and without the psychiatrists' personal examination of the defendant, and there was no due process violation on the basis that the doctors were permitted to give an opinion on the ultimate issue before the jury. White J. referred with approval to Jurek v. Texas and Estelle v. Smith and added:

4.30

White J. discussed the testimony given by the various experts in the case, and commenting on Monahan's study and conclusion,¹⁹ he said that although Monahan originally believed that it was impossible to predict violent behaviour, by the time he had finished his text he felt that there may be circumstances in which prediction is both empirically possible and ethically appropriate.

4.31

Blackmun J., joined by Brennan and Marshall JJ., dissented on the ground that psychiatric testimony regarding future dangerousness was too unreliable to be admissible as evidence.

4.32

The question of preventive detention was addressed in relation to juveniles on a number of occasions. At issue in a number of cases was a New York statute permitting pre-trial detention of juveniles on the basis that there was a serious risk that they might commit further offences before their return date. In People ex rel Wayburn v. Schupf,²⁰ the New York Court of Appeals rejected a challenge to the statute based on the argument that the statute was unconstitutionally vague because of the inability to predict future criminal behaviour accurately. The Court said that this element was necessarily present in bail, sentencing and parole decisions and any other procedure in which discretionary authority for differing criminal dispositions is vested in a court or administrative body. Judge Fuchsberg, dissenting, agreed that the statute did not violate equal protection (between juveniles and adults), but took the view that the statute violated due process because it offered no criteria for determining which juveniles pose a serious risk of future crime. He said that the absence of such criteria was due to the fact that there are no criteria upon which accurate predictions of future crimes may be made. He concluded that choice among juveniles as to who is at risk and is not are essentially random and inexplicable. The same statute was challenged in the Federal Courts in U.S. ex. rel Martin v. Strasburg,²¹ where the United States District Court for the Southern District of New York held that while the statute did not violate equal protection, it violated due process: first, because it lacked guidelines and the judicial decision would be based on each judge's subjective views and bias and was therefore arbitrary, and second, after reviewing the results of a number of empirical studies of the prediction of dangerousness, because predictions could not be made accurately:

4.33

In 1982, the Court of Appeals for the Second Circuit affirmed the District Court's decision on the ground that pre-trial detention under the New York statute was primarily “to impose punishment before adjudication of the alleged criminal acts”.²³ The Court attached weight to statistics proferred by the juvenile appellants showing that the vast majority of juveniles detained under the statute either had their petitions dismissed before adjudication or were released after adjudication.

4.34

In 1984, however, the U.S. Supreme Court reversed this judgment in Schall v. Martin,²⁴ a habeas corpus class action challenging the New York statute on the basis of violation of the due process and equal protection clauses of the Fourteenth Amendment. The evidence, as in the previous case, showed that the vast majority of juveniles held under the statute either had their petitions ultimately dismissed or received a non-custodial disposition. The majority judgment, in designating the measure as regulatory, applied a two-prong test, whether the statute served a legitimate State objective and whether the procedural safeguards adequately protected the rights of the detainee. It held that the first prong was satisfied by reference to the State's legitimate interest and parens patriae power in both preventing crime and protecting juveniles and society from the consequences of their criminal activities. It also found the procedural safeguards adequate; detention was limited to seventeen days, the juvenile was given full notice of the charges and was given a hearing at which he was informed of his rights, he could be accompanied by a parent or guardian, he could be represented by counsel chosen by him or a law guardian assigned by the court, he could call witnesses and offer evidence, and probable cause had to be established to believe that the juvenile had committed the offence. In response to a claim that it was impossible to predict criminal behaviour, Rehnquist C.J., writing for the majority, said:

4.35

Marshall J. dissenting, criticised what he saw as a dilution and misapplication of the Mendoza test and said that due to 'overwhelming' evidence that even highly trained criminologists cannot accurately predict criminal behaviour, the New York statute did not meet its purpose of preventing crime.

4.36

These cases show a significant degree of divergence among judges on a

number of matters. Widely diverging views are offered on the issue of prediction in particular, some judges asserting that it is impossible to do accurately, others that prediction is a legitimate exercise. As to the Federal Bail Reform Act, 1984 itself, although the Court of Appeals for the Second Circuit found s.3142(e) unconstitutional in Salerno, every other Court of Appeals that had considered it had upheld its constitutionality prior to the Supreme Court decision in Salerno.²⁶

U.S. v. Salerno²⁷

4.37

Salerno and Cafaro were arrested in 1986 after being charged in an indictment alleging various violations of the Racketeer Influenced and Corrupt Organizations Act, mail and wire fraud offences, extortion, and a number of criminal gambling violations. The RICO counts included fraud, extortion, gambling, and conspiracy to commit murder. At the detention hearing, the Government case was that Salerno was the boss of the Genovese Crime Family of La Cosa Nostra and that Cafaro was a captain in the family. According to the government's proffer, which was based in large part on conversations intercepted by a court-ordered wiretap, they had participated in wide-ranging conspiracies to aid their illegitimate enterprises through violent means. The Government also offered the testimony of two of its trial witnesses, who would assert that Salerno personally participated in two murder conspiracies. Salerno opposed the motion for detention, challenging the credibility of the Government's witnesses and offered the testimony of several character witnesses as well as a letter from his doctor stating that he was suffering from a serious medical condition. Cafaro presented no evidence at the hearing, but instead characterised the wiretap conversations as merely “tough talk”.

4.38

Their detention was ordered by the District Court, and they appealed to the Court of Appeals for the Second Circuit, which vacated the order of the District Court and remanded the case for setting conditions of bail, holding that the Act violated the due process clause of the Fifth Amendment. On appeal to the Supreme Court, the Supreme Court granted the Government's petition for a writ of certiorari and reversed the judgment of the Court of Appeals. In an opinion by Rehnquist C.J. joined by White, Blackmun, Powell, O'Connor and Scalia JJ., a majority of the Court held that the contested provisions of the Bail Reform Act did not on their face violate substantive due process under the Fifth Amendment, procedural due process under the Fifth Amendment or the Eighth Amendment guarantee against excessive bail.

4.39

Rehnquist C.J. commenced by observing that a facial challenge to an Act is the most difficult challenge to mount successfully, because it must show that there are no circumstances in which the Act could be constitutional. In response

to the respondent's argument that the Act imposed punishment before trial and therefore violated the due process guarantee, he acknowledged that substantive due process prevents the Government from engaging in conduct that shocks the conscience or interferes with rights implicit in the concept of ordered liberty. However, he held that the Act in question was regulatory and not punitive. The test of whether a restriction on liberty constitutes impermissible punishment or permissible regulation involves looking, first, to legislative intent, and thereafter, to:

4.40

He continued:

4.41

The Chief Justice went on to say that the Court had repeatedly held that the government's interest in community safety can outweigh the individual's liberty interest in appropriate cases. He gave examples, including the detention of individuals believed to be dangerous in times of war or insurrection, the detention of dangerous aliens pending deportation proceedings, the detention of mentally unstable individuals who present a danger to the public, the detention of dangerous defendants who become incompetent to stand trial and the pre-trial detention of persons who present a risk of flight. He pointed out that the

respondents characterised all these cases “as exceptions to the 'general rule' of substantive due process that the government may not detain a person prior to a judgment of guilt in a criminal trial”. However, although he freely conceded such a general rule, he thought that these cases “show a sufficient number of exceptions to the rule that the congressional action challenged here can hardly be characterised as totally novel”.²⁹

4.42

He then addressed the issue of procedural due process, holding that the extensive safeguards in the Act were even more exacting than those in Schall. Regarding prediction he said:

4.43

The Chief Justice then turned to the excessive bail clause, which the respondents had contended granted them a right to bail grounded solely upon considerations of flight and threats to judicial process by intimidation of witnesses, because these matters related to the function of ensuring the integrity of the judicial process. He rejected this argument:

and

4.44

In a vigorous dissent, Marshall J. opened by saying:

4.45

Marshall J. commenced his argument by raising a serious question as to whether the cases of Salerno and Cafaro presented issues. The issue of pre-trial detention in respect of Salerno had effectively become moot when he was convicted in 1986 after a jury trial on other charges, for which he was sentenced to one hundred years imprisonment. The District Judge, Marshall J. felt, had artificially kept the present matter alive by releasing him on bail pending appeal, with the Government's consent, despite having just sentenced him to one hundred years imprisonment. Meanwhile, Cafaro became a co-operating witness and was released, ostensibly temporarily for medical care and treatment, with the Government's consent. These facts, he thought, raised a substantial question as to the Court's jurisdiction as it was far from clear that there was an actual controversy between the parties.

4.46

Marshall J. proceeded to attack the majority's analysis of due process. According to their test, if Congress determined after investigation that a large proportion of violent crime was committed by unemployed persons after dark and chose to impose a dusk to dawn curfew on the unemployed, the majority would find that this was a legitimate regulatory non-punitive measure. This ludicrous result, he said, resulted from the majority's:

4.47

He also criticised the majority's view on the excessive bail clause:

4.48

Marshall J. felt that the ultimate irony was that the essence of the case was something to which the majority did not refer, namely the presumption of innocence. He gave the example of a defendant indicted, shown to be dangerous and detained, but ultimately acquitted at trial. The Government in such a case would clearly not be permitted to detain him based on the showing of dangerousness alone, which illustrates that the indictment is being taken as proof of guilt. However, the presumption of innocence declares that he is as innocent

on the morning of trial as on the morning after his acquittal. Marshall J. says that the conclusion is inescapable that the indictment has been turned into evidence that the defendant is guilty of the crime charged, and if left to his own devices, he will soon be guilty of something else. He discussed the legal import of the indictment further. It establishes probable cause that the accused committed the offence and from this flows the power to try, which in turn gives rise to the power to assure that the processes of justice will not be evaded or obstructed. However, preventive detention to prevent the commission of further offences bears no relation to the power to try, and thus the interests it serves are outside the scope of interests which may be considered in weighing the excessiveness of bail under the Eighth Amendment. Marshall J. concluded:

4.49

Stevens J. also dissented, expressing the view that there might be times when the Government's interest in protecting the safety of the community would justify the brief detention of a person who had not committed any crime, but that the provision of the Act which allowed pre-trial detention on the basis of future dangerousness to the community was unconstitutional.

4.50

It is noteworthy that the majority of the Supreme Court in Salerno placed reliance on the procedural safeguards in finding that the Act was not unconstitutional. In this regard, a more recent Supreme Court decision seems to suggest that failure to observe the procedural safeguards will not necessarily result in the invalidity of the detention. It may be that the decision is limited either to its facts or the particular safeguard in question – the requirement to hold a hearing at first appearance – but it may illustrate that the procedural safeguards will not be stringently enforced. In U.S. v. Montalvo-Murillo,³⁶ a magistrate in New Mexico failed to hold a hearing at the defendant's first appearance before him because the pre-trial services report was not ready. There had been some delay because the accused had been brought from New Mexico to Illinois, and then back to New Mexico. The magistrate decided to order the release of the accused on certain conditions. The prosecutor appealed to the District Court, which refused to order the accused's detention on the basis that the hearing had not been held upon his first appearance as required by s.3142(f) of the Bail Reform Act, although it found that he would have been a candidate for detention. The Court of Appeals for the Tenth Circuit affirmed, but the Supreme Court reversed, holding that the failure to comply with the section did not entitle him to pre-trial release, because nothing in the section indicated that compliance with the first appearance requirement was a condition

precedent for the hearing. Automatic release would contravene the object of the Act, which is to provide fair bail procedures while protecting the safety of the public and assuring the defendant's appearance at trial. The Court said that where there were circumstances involving several districts, which contributed to a missed deadline for which no real blame could be fixed, there was no reason to bestow a windfall on the defendant and visit upon the Government and citizens a severe penalty by releasing him every time a minor deviation from the Act occurred. The detention in this case was harmless, because if the hearing had been held earlier, he would have been detained. Stevens J., joined by Brennan and Marshall JJ., dissented, saying that the magnitude of the injury inflicted by pre-trial detention requires adherence to strict procedural safeguards that cannot be sacrificed in the name of community safety.

Use Of Preventive Detention Under The Bail Reform Act, 1984

4.51

At the time of the enactment of the Bail Reform Act, 1984, Congress appears to have thought that the power to detain dangerous defendants before trial would be sparingly used. Research has since been conducted into the use of pre-trial detention generally, and detention of dangerous defendants particularly, to test whether this is the case.³⁷ This research shows that detention ordered wholly or partly on the basis of danger as a percentage of all detentions ordered after hearing in all districts of the eleven circuits was as follows: