The Howard Journal Vol 41 No 3.

July 2002
ISSN 0265–5527, pp. 245–262

Do Bail Information Schemes Really

Affect Bail Decisions?

MANDEEP K. DHAMI
Research Fellow, City University, London and University of Victoria, BC

Abstract: Bail Information Schemes gather and provide information on a defendant’s

community ties to the court. Past evaluations have concluded that schemes are effective because
more defendants are granted bail. However, the evaluations include possible confounds and
other limitations, thus threatening the validity of their findings. They are also incompatible with
research showing that magistrates do not always attend to community ties information. Here, an
experiment examined the effect of community ties information on magistrates’ bail decisions on
hypothetical cases. It was found that schemes do not significantly affect decisions, although they
do serve to significantly increase confidence in decisions. Implications for the future of schemes
are discussed.

The bail decision arises whenever a case is adjourned for trial, sentence, or
appeal. In the English criminal justice system, magistrates can bail a defen-
dant unconditionally, with conditions such as curfew, or remand a defen-
dant in custody, until the next hearing of the case. This decision has
significant consequences for defendants and society. For example,
compared to their bailed counterparts, remanded defendants are more
likely to lose their homes, jobs, and family ties (Hammond 1988), and they
are more likely to plead guilty, receive a custodial sentence, and less likely
to be acquitted (for example, Davies 1971). On the other hand, the public
is victimised by defendants who offend on bail (Brown 1998).
Magistrates’ bail decisions are governed by legislation (that is, the Bail
Act 1976) stipulating that bail should be granted, except in certain circum-
stances. These exceptions include the defendant’s failure to surrender to
previous court bail, or the court’s satisfaction that there are ‘substantial
grounds’ for believing the defendant would abscond, for example. In
theory, defendants with strong community ties (for example, dependants,
employment, and fixed address) have more to lose if they abscond (for
example, Home Office 1988). Thus, they pose a low risk of absconding and
may be ‘safely’ bailed.
The opportunity to make just and defensible decisions is, however,
hampered by the lack of relevant information (for example, Dhami 2001;
Hucklesby 1996; King 1971). Brief details of the defendants and cases
appearing in court on a particular day are presented on a court sheet, and
this may be supplemented with information provided by the prosecution,
defence, or defendant. King (1971) found that in 23 English courts,
245
 Blackwell Publishers Ltd. 2002, 108 Cowley Road, Oxford OX4 1JF, UK
and 350 Main Street, Malden, MA 02148, USA
community ties information was given in only 35% of cases, and that ‘it was
usually sparse and rather vague in nature’ (p. 91). Recently, Dhami (2001)
observed that the community ties information in two English courts mostly
concerned residence. Hucklesby (1996) reported that supplementary infor-
mation was available in only a third of the cases in three Welsh courts, and
in 41% of these, it was of limited breadth. Indeed, magistrates rarely probe
for information (Dhami 2001; East and Doherty 1984; Hucklesby 1996).
In order to overcome the lack of community ties information provided to
courts, Bail Information Schemes (BISs) were re-introduced in the English
system in 1987, after a brief period of operation in the mid-1970s. Probation
officers collect, verify, and provide largely positive information about a
defendant’s community ties to the prosecution and defence, who then relay
it to the court. In courts, BISs target defendants awaiting their first appear-
ance, who are likely to be remanded in custody because the police
remanded them in custody. In prisons, BISs target remanded defendants
awaiting their second court appearance. Officers operating schemes do not
make any recommendation regarding bail, although they do exercise discre-
tion in the cases selected, and the information collected, verified, and
presented. Past evaluations of BISs have concluded that magistrates were
more likely to bail defendants when provided with bail information (Fiddes
and Lloyd 1990; Godson and Mitchell 1991; HM Inspectorate of Probation
1993; Lloyd 1992; Mair 1988; Stone 1988). However, these evaluations are
methodologically limited, and their findings are incompatible with other
research on magistrates’ bail decision making (for example, Dhami and
Ayton 2001). In the following section, I shall review the past evaluations and
discuss these two criticisms, before outlining the aims of the present study.

Evaluating the Effect of Bail Information Schemes

BISs originated in America in the 1960s, when the Vera Institute of Justice
organised the Manhattan Bail Project (Ares, Rankin and Struz 1963). Ares et
al. (1963) evaluated this scheme by randomly assigning defendants to either
an experimental or control group. The recommendation (for or against
bail) was conveyed to the court for the former group and it was suppressed
for the latter. It was found that whereas 60% of the experimental group were
bailed, only 14% of the control group were bailed. The schemes subse-
quently spread to other courts in America (Tobolowsky and Quinn 1993).
By contrast, no quantitative evaluation was made of the first English scheme
(Pearce and Smith 1976). On the basis of interviews with magistrates and
observations of prosecution and defence practices, it was concluded that the
provision of community ties information ‘seems to encourage more
frequent granting of bail at first appearance’ (Pearce and Smith 1976, p.
48).
The agencies (that is, Home Office and probation service) responsible
for BISs in the English system, have now conducted quantitative evaluations
of the schemes. Stone (1988) evaluated BISs piloted over a year in eight
English courts. Using statistical models, it was predicted that over all courts,
29% of defendants who would otherwise have been remanded in custody,

246
 Blackwell Publishers Ltd. 2002
were bailed because of BISs. Defendants with no fixed abode were more
likely to be released on conditional bail to a hostel rather than be remanded
in custody. For some courts, defendants granted bail after the intervention
of a BIS were less likely to be arrested for a new offence while on bail, or due
to failure to appear, or for breach of bail conditions, than defendants who
were bailed in the absence of a BIS.
Lloyd (1992) evaluated BISs in three English courts and one prison. In
the former, he compared the defendants for whom information was
provided to the court with those for whom information was not provided.
Defendants for whom information was not provided for remand-related
reasons were excluded from analyses. For two courts, both groups were
comparable in terms of age, sex, offence, and reason for arrest. In each
court, the BIS group was less likely to be remanded in custody than the no-
BIS group. Over all courts, 23% of defendants were bailed because of the
BISs. There was no significant difference between the BIS and no-BIS
groups regarding the percentage of defendants who re-appeared for a new
offence, or for those who re-appeared for breaching bail conditions. In the
prison-based scheme, information was provided to the court in 63% of the
cases interviewed. Around 28% of defendants were granted bail due to the
intervention of the BIS.
Mair (1988) evaluated a prison-based scheme over a four-week period. The
BIS and no-BIS groups were similar in age. It was found that 17% of the
former group was bailed compared to only 3% of the latter. However, 15% of
the defendants bailed in the BIS group were bail failures because they
absconded from bail hostels, for example. Others similarly have reported that
prison-based schemes are successful in diverting defendants from further
custody (for example, Nottingham and Mitchell 1994; Williams 1992).
Finally, studies have also documented the effect of BISs upon the prose-
cution request and the defence application for bail. It is generally
concluded that in the presence of bail information, prosecutors are less
likely to request a remand in custody, and the defence is more likely to apply
for bail (Fiddes and Lloyd 1990; Godson and Mitchell 1991; HM
Inspectorate of Probation 1993; Lloyd 1992; Morgan and Henderson 1998;
Stone 1988).

Methodological Limitations
Although the extant literature comments on the ‘cause-effect’ relationship
between BISs and magistrates’ bail decisions, the reported studies lack
experimental control. Cases were not randomly assigned to the BIS and no-
BIS groups. Thus, differences in case characteristics may account for differ-
ences in the bail and custody rates between them. Lloyd (1992) revealed
that for one court, the two groups were different in terms of age, offence,
reason for arrest, and prior history with probation service. The directions of
the differences indicate that the no-BIS group would be less likely to be
bailed. The no-BIS group in Lloyd’s (1992) prison-based evaluation had
fewer previous applications for bail than the BIS group, which may reflect
the fact that cases in the former group stood little chance of gaining bail.
Stone (1988) did not distinguish between the two groups in terms of the
247
 Blackwell Publishers Ltd. 2002
reasons for bail information being withheld. There may have been cases in
the no-BIS group for whom these reasons may have affected the bail deci-
sion. Nor did Stone (1988) provide any details of the characteristics of the
cases in the two groups, and he did not examine whether the prosecution
requests, which have an important effect on magistrates’ decisions, varied
for both groups. The fact that BISs have been reported to also affect the
prosecution and defence requests means that their direct impact on magis-
trates’ decisions is unclear in the above studies, which do not disentangle
these variables. Around half of the defendants bailed in the BIS group in
Mair’s (1988) study, were bailed due to other factors such as the work done
by the defence solicitor. Finally, all of the evaluations suffered from prob-
lems associated with missing data. For instance, in Mair’s (1988) study, data
were missing regarding offence for around half of the no-BIS group, and it
is thus unclear whether they were charged with more serious offences,
making them less likely to be granted bail.
Incompatibility of Past Research
Studies attesting to the effectiveness of BISs imply that magistrates attend to
information concerning defendants’ community ties. Indeed, other
research supports this assumption. For example, defendants living outside
the geographical area in which the court was situated, and those with no
fixed abode, were more likely to be remanded in custody than those living
in the area and with a fixed address (Doherty and East 1985; Hucklesby
1996; Morgan and Henderson 1998). There is, however, considerable
evidence indicating that magistrates’ decisions are influenced by other
‘legal’ cues such as the nature and seriousness of the offence the defendant
is charged with, the defendant’s previous convictions, and past bail record
(for example, Dhami 2001; Dhami and Ayton 2001; Doherty and East 1985;
Jones 1985; Hucklesby 1996; Morgan and Henderson 1998). Furthermore,
magistrates attend to ‘extra-legal’ cues such as the defendant’s gender and
race (for example, Dhami and Ayton 2001; Fitzgerald 1993; Jones 1985;
Morgan and Henderson 1998), the police bail decision (which is made after
arrest and before first appearance at court), and the prosecution request
(for example, Dhami 2001; Dhami and Ayton 2001; East and Doherty 1984;
Jones 1985; Hucklesby 1996; Morgan 1994; Morgan and Henderson 1998).
Magistrates tend to underestimate the influence of such cues (Dhami 2001;
Dhami and Ayton 2001).
Recent experimental and observational studies have revealed that magis-
trates employ a simple ‘fast and frugal’ decision making strategy in which
they only search through a tiny fraction of the available information and
base their decisions on only one piece of information (Dhami 2001; Dhami
and Ayton 2001). An examination of the relative impact of different cues
upon magistrates’ decisions has shown that the community ties cue is not as
influential as other information (Dhami 2001; Dhami and Ayton 2001).
Thus, BISs may not be as effective as has been claimed. Finally, the fact that
magistrates have been found to be inconsistent in their decisions, and to
disagree with others’ decisions on the same case (Dhami and Ayton 2001),
may also impinge on the effectiveness of BISs.

248
 Blackwell Publishers Ltd. 2002
 The Present Study
Aim
In order to overcome the methodological limitations of past evaluations,
the present study used an experimental design to examine the effect of
BISs on: (i) magistrates’ decisions, (ii) magistrates’ consistency in making
decisions on repeated cases, (iii) agreement among magistrates’ decisions
on the same cases, (iv) magistrates’ post-decisional confidence, and (v)
their reports of the relative importance of different cues. The study also
examined (vi) the effect of different case characteristics on the effective-
ness of BISs.

Design
Hypothetical cases were constructed in which case characteristics were
disentangled, so for example, the seriousness of the offence the defendant
was charged with was unrelated to his or her age. Magistrates were
randomly assigned to either a BIS or no-BIS group. The cases presented to
both groups were identical except for the information on defendants’
community ties. The nature and amount of the information presented to
the BIS group was representative of that presented in a courtroom with a
BIS in operation. Similarly, the information presented to the no-BIS group
was representative of that available in a courtroom in the absence of a BIS.

Participants
Court managers in 70 randomly selected adult magistrates’ courts in
England and Wales were each mailed six booklets. The sampling frame
was obtained from Shaw’s 1997/98 Directory of Courts in the United Kingdom
(Morris 1997). Altogether, 132 magistrates, representing a 31.43%
response rate, fully completed and returned the materials within the time
limit. This is a reasonable response rate for a postal survey and consider-
ing the nature of the sample.
The BIS group comprised 77 magistrates from 35 courts, and the no-
BIS group comprised 55 magistrates from 31 courts. Of those who
provided their demographic details, there were 73 lay and 3 stipendiary
magistrates in the BIS group, and 50 lay and 3 stipendiary magistrates in
the no-BIS group. Magistrates in the BIS group had an average of 16 years
experience, compared to an average of 14 years experience in the no-BIS
group. In the BIS group, 62 magistrates were located in metropolitan
courts and 12 in provincial courts, and in the no-BIS group, 37 magis-
trates sat in metropolitan courts and 18 in provincial courts. Finally, a BIS
was in operation in the courtrooms of 28 magistrates in the BIS group,
and 22 magistrates in the no-BIS group. Therefore, both groups were
similar, and respondents were representative of the English magistracy in
terms of the relative proportions of lay and stipendiary magistrates,
magistrates with more and less experience, and magistrates sitting in
metropolitan and provincial courts (Lord Chancellor’s Department
1999a, 1999b).

249
 Blackwell Publishers Ltd. 2002
Construction of Cases
The procedure for constructing the hypothetical cases was the same as
that used by Dhami and Ayton (2001). The cases comprised a combina-
tion of nine cues, which are shown in Table 1. In addition, nine cues were
held constant, as a means of providing consistent background informa-
tion to the cases. The constant information referred to how the defen-
dant came to court, the number of charges against the defendant, the
defendant’s plea, the defence representation, the circumstances of
adjournment, whether the defendant was present in court, the length of
the adjournment, who requested the adjournment, and the number of
previous adjournments and bail applications. The nine cues in Table 1
were combined to form cases so that each cue was independent of every
other cue. The orthogonal design option in SPSS version 7.5 for Windows
was used to do this and it yielded a set of 27 cases. Seven cases were then
randomly selected from this set and duplicated (although the names of
the defendants were changed) to measure test re-test consistency. The
values of the nine cues and their distribution across the 27 cases are
shown in Table 1. The cue values were equally distributed among the cases
for the tertiary cues, and were distributed according to approximate real-
world distributions for the other cues. The properties of these cases were
not unrepresentative of those appearing in English courts (Dhami 2001),
and the realism of the cases was confirmed by a sample of 81 magistrates
(Dhami and Ayton 2001). The cues in each case appeared in the order
shown in Table 1.
In order to determine the amount and nature of the community ties
information to be presented to the BIS group, telephone interviews were
conducted with four probation officers, and the standardised forms used
to gather and disseminate bail information were analysed. It was found
that commonly BISs provide information on: (i) a defendant’s residen-
tial status, and the availability of a bail hostel place in cases where the
defendant has no fixed abode, (ii) a defendant’s participation in
employment or education, (iii) the nature of a defendant’s personal ties
such as being married and having children, and (iv) the availability of a
surety who will forfeit money if the defendant absconds. Therefore,
these elements were added to each case for the BIS group. For the
community ties information to be presented to the no-BIS group, the
observational data reported in Dhami’s (2001) study of two English
courts were analysed, as neither had a BIS in operation. Across both
courts, 97% of cases included information regarding the defendants’
residence, 9% included information about the defendants’ children,
11% included information about the defendants’ spouses, and 13%
included information regarding the defendants’ employment or educa-
tional status. In a minority of these cases the available information was
negative. Therefore, the proportion and nature of this information was
represented in the 27 cases in the present study. For example, in two of
the 27 cases, defendants were described as being married because 8% of
Dhami’s (2001) defendants were married.

250
 Blackwell Publishers Ltd. 2002
 TABLE 1
Cues, Their Values and Distributions in Hypothetical Cases

Cue* Values Distribution

Gender Male 18
Female 9
Race White 9
Asian 9
Black 9
Age 18–20 18
21+ 9
Offence Summary 9
Triable either-way 9
Indictable 9
Previous convictions and bail record None-none 6
(PCBR) None-good 6
Yes, dissimilar-good 6
Yes, similar-good 3
Yes, dissimilar-poor 3
Yes, similar-poor 3
Police bail decision (polbail) Unconditional bail 9
Surety 9
Remand in custody 9
Prosecution request (prosreq) Don’t oppose bail 9
Conditional bail 9
Oppose bail 9
Strength of prosecution case (proscase) Strong 18
Weak 9
Defence request (defreq) Apply for unconditional
bail 9
Suggest conditions 9
Don’t apply for bail 9

(Note: *Ages in each band were chosen randomly. Specific offences in each category
were selected because of their commonality according to the 1996 recorded crime
figures in England and Wales (which were the most recent at the time of design)
(Home Office 1997). A poor bail record is described in terms of absconding, offend-
ing or interfering with witnesses while on bail in the past. The prosecution case
against the defendant is strong where details of witnesses and/or forensic evidence
are presented at the bail hearing)

Measurement of Responses
Magistrates were first asked to make a bail decision on each case. The deci-
sion options were unconditional bail, conditional bail (with the conditions
specified), or remand in custody. Magistrates then were asked to indicate
how certain they were that they had made the appropriate decision based
upon the information provided, on an 11-point scale, ranging from
251
 Blackwell Publishers Ltd. 2002
absolutely uncertain (0) to absolutely certain (10). Finally, magistrates
ranked the available cues according to the relative importance attached to
them when magistrates made their decisions. The most important cue was
ranked 1.

Procedure
The 27 cases were placed in a booklet, and each booklet contained a different
random order of the cases to eliminate any order effects due to shifts in atten-
tion. The duplicate cases were placed at the end, to restrict the probability of
two identical cases being presented one after another. The ranking task
followed the hypothetical cases. The booklet also contained instructions that
listed the nine cues and described the tasks. Background information
common to all cases was provided. Participants were instructed to complete
the tasks individually, and not to return to cases that had already been
completed. Participants’ demographic details, namely, type of magistrate,
years of experience on the bench, location of court, and operation of BIS in
their court, were requested. The booklets were mailed to court managers who
were asked to distribute them to magistrates sitting in their courts. A covering
letter was included for magistrates, which introduced the study, guaranteed
respondents anonymity, and highlighted the four-week time limit for comple-
tion and return. A stamped, self-addressed envelope was provided.

Results
Decisions Made
The frequency with which magistrates in both the BIS and no-BIS groups
granted unconditional bail, conditional bail, and remanded defendants in
custody, was representative of the decisions made in the English system
(Home Office 1999). As Table 2 shows, there was no significant difference
between the BIS and no-BIS groups in the likelihood of making a punitive
decision (that is, conditional bail or remand in custody) on the 27 cases
(t[130] = –0.66, n.s.). The difference in the likelihood of remanding in
custody between the two groups was only marginally significant (t[130] =
–2.49, p = 0.03).1
Conditions Attached to Bail
For each magistrate, the mean number of conditions attached to bail on the
27 cases was calculated. There was no significant difference in the grand
mean number of conditions attached to bail by the BIS (M = 1.65, SD = 0.52)
and no-BIS group (M = 1.54, SD = 0.37) (t[129] = 1.42, n.s.). Figure 1 illus-
trates the percentage of each type of condition imposed in the cases granted
conditional bail by each group. The BIS group imposed a significantly
greater mean number of conditions of surety (M = 3.05, SD = 4.30), and bail
hostel (M = 0.78, SD = 0.60) than the no-BIS group. (For surety, M = 0.84,
SD = 1.62 and for bail hostel, M = 0.15, SD = 0.52) (t[102] = –4.11, p = 0.001,
t[125] = –6.38, p = 0.001.) There were no significant differences between the
two groups in terms of the mean number of the other types of conditions
attached to bail.

252
 Blackwell Publishers Ltd. 2002
 TABLE 2
Mean Proportions of Unconditional Bail, Conditional Bail and Remand in
Custody Decisions Made by BIS and No-BIS Groups

Decision Group* M SD

Unconditional bail BIS 13.73 5.21

No-BIS 13.20 4.24
Total 13.51 4.82
Conditional Bail BIS 10.94 4.69
No-BIS 10.60 4.76
Total 10.80 4.32
Remand in custody BIS 2.34 2.00
No-BIS 3.38 2.83
Total 2.77 2.43

(Note: *N = 77 for BIS group, N = 55 for no-BIS group, and N = 132 for both groups
together)
% of cases granted conditional bail

FIGURE 1
Conditions Imposed in Cases Granted Conditional Bail by BIS and No-BIS Groups

253
 Blackwell Publishers Ltd. 2002
Median rank order (9 = most important)

FIGURE 2
Self-reported Cue Importance of BIS and No-BIS Groups

Intra-magistrate Consistency
Each magistrate’s consistency in making decisions on the seven duplicate
cases and their original counterparts was computed using Cohen’s Kappa,
which ranges from 0 (that is, consistency is no better than chance) to 1 (that
is, perfect consistency). There was no significant difference in the mean
Kappa value between the BIS (M = 0.63, SD = 0.28) and no-BIS group
(M = 0.60, SD = 0.26) (t[130] = 0.50, n.s.).

Disagreement Among Magistrates

The number of cases each magistrate disagreed with the modal response was
calculated. There was no significant difference between the BIS (M = 7.57,
SD = 2.76) and no-BIS group (M = 7.47, SD = 3.52) regarding the mean
number of cases on which magistrates disagreed (t[130] = 0.18, n.s.).

Post-decisional Confidence
Each magistrates’ mean confidence ratings in the decisions made on the 27
cases was calculated. The BIS group reported significantly higher post-deci-
sional confidence (M = 8.54, SD = 0.91) than the no-BIS group (M = 8.20,
SD = 0.94) (t[130] = 2.05, p = 0.04), although the size of the difference is
moderate.
254
 Blackwell Publishers Ltd. 2002
 TABLE 3
Decisions Made by BIS and No-BIS Groups as a Function of Case Characteristics

Bailed (unconditionally/
conditionally)
Cue Values BIS group No-BIS group

Gender Male 88.74 89.47

Female 96.54 95.67
Race White 87.30 86.73
Asian 90.19 92.93
Black 96.54 94.05
Age 18–20 92.57 92.42
21+ 88.89 89.75
Offence Summary 95.09 95.06
Triable either-way 95.38 94.81
Indictable 83.55 83.84
Previous convictions None-none 98.48 95.67
and bail record None-good 93.07 92.42
(PCBR) Yes, dissimilar-good 97.40 94.81
Yes, similar-good 91.34 92.21
Yes, dissimilar--poor 91.77 93.07
Yes, similar-poor 61.04 72.73
Police bail decision Unconditional bail 95.82 94.37
(polbail) Surety 90.48 91.34
Remand in custody 87.73 88.89
Prosecution request Don’t oppose bail 95.53 93.94
(prosreq) Conditional bail 89.18 91.34
Oppose bail 89.32 89.32
Strength of Strong 90.55 90.69
prosecution case Weak 92.03 93.22
(proscase)
Defence request Unconditional bail 96.68 94.81
(defreq) Suggest conditions 90.33 92.50
Don’t apply for bail 87.01 87.30

Reported Cue Importance

In order to compare self-reported importance of cues between the BIS and
no-BIS groups, the median rank order attached to each cue was computed
across magistrates in each group. Any tied ranks were converted into
sequential unique values. Figure 2 illustrates the similarities and differences
between the self-reported importance of cues summarised across magis-
trates in each group. Note that the rank order of importance was reversed
for ease of illustration. According to Mann-Whitney U tests, there were no
significant differences between the groups in terms of the median ranks for
any of the cues, including the community ties cue.
255
 Blackwell Publishers Ltd. 2002
Effects of BISs as a Function of Case Characteristics
As Table 3 shows, there was virtually no difference in the granting of uncon-
ditional or conditional bail by the BIS and no-BIS groups as a function of
the characteristics of the case.2

Discussion
In a recent commentary on the effectiveness and future of pre-trial services,
Drakeford et al. (2001) state that BISs are effective in diverting defendants
from custody. They base their conclusion on the findings of past evaluations
of schemes and on their own cross-tabulation of the decisions made on
defendants for whom information was provided in three probation areas.
However, both sources of information are unreliable due to methodological
shortcomings. Contrary to these past findings, the present study revealed
that schemes do not significantly affect magistrates’ bail decisions.3
The findings of past evaluations are considered to be externally valid
because they were based on real decisions made in the courtroom, although
their internal validity was compromised by the lack of experimental control.
Whereas the present study can claim internal validity, its external validity
may be threatened because it lacked the social dimension in which BISs
operate, in two distinct ways. First, under real conditions the probation offi-
cer provides the completed bail information sheet to the prosecution and
defence rather than directly to the court (although for an exception see
Morgan and Henderson 1998), and it has been suggested that BISs work
indirectly through the prosecution request or defence request (Fiddes and
Lloyd 1990; Godson and Mitchell 1991; Lloyd 1992; Stone 1988). The fact
that in the present study, the BIS was independent of the prosecution and
defence requests is not a limitation because as Table 3 shows, there was virtu-
ally no difference in the bail decisions made by magistrates in the BIS and
no-BIS groups as a function of the prosecution and defence requests.
Second, the present study investigated decisions made by individual lay
magistrates who normally act as a bench. Although further tests are neces-
sary, the external validity of studying individual lay magistrates’ decisions on
hypothetical cases has been recently supported by Dhami’s (2001) observa-
tional study of benches of magistrates deciding on real cases. This finding
can be explained with reference to the psychological research on small
group decision making that demonstrates how, like individuals, groups are
inconsistent and simple in their decision making strategies (for example,
Janis 1982; Janis and Mann 1977; Petty, Harkins and Williams 1980; Weldon
and Gargano 1985).
Indeed, the findings of the present study are consistent with other
research on magistrates’ bail decision making. For example, the present
findings are compatible with the variations among courts reported in some
past evaluations of BISs (Godson and Mitchell 1991; Lloyd 1992; Stone
1988). Stone (1988) reported that in one court, only ten defendants had
been diverted from custody as a result of the scheme. Studies aggregating
over courts have tended to hide differences in the success of individual

256
 Blackwell Publishers Ltd. 2002
schemes. It is unclear why some schemes are more successful than others,
and a closer examination of the differences in the operation of more and
less successful schemes is warranted.
The present findings are also compatible with research demonstrating
that most magistrates do not use community ties information when it is avail-
able (for example, Dhami 2001; Dhami and Ayton 2001). In fact, as Figure 2
shows, magistrates in both the BIS and no-BIS groups did not explicitly
report community ties information to be particularly important to their
decisions. And why should they consider community ties to be important?
Although community ties information is considered theoretically to be an
important predictor of absconding on bail, there is as yet no empirical or
statistical data to support this. Clearly, the necessity and effectiveness of BISs
may rest on the results of such research.
Importantly, Drakeford et al. (2001) considered BISs from the perspec-
tive of the probation service that operates them. Their research included
interviews with bail information officers in three probation areas. The offi-
cers pointed to the lack of guidelines and training, and to operational weak-
nesses of BISs. For example, they highlighted the problem of inconsistent
targeting of defendants and collection of information, the lack of time avail-
able to interview defendants, the problem of providing information solely to
prosecutors who sometimes choose not to use it, poor communication
between prison and court-based schemes, and the problem of prioritising
other work higher than that of gathering bail information. These testi-
monies suggest that schemes would have a limited effect on bail decisions.
There are several other possible explanations for the ineffectiveness of
BISs in the courtroom. First, information pertaining to the defendant’s
community ties is not entirely absent in the courtroom where a BIS does not
operate (Dhami 2001; Hucklesby 1996). Information regarding residence is
contained in the court sheet and, as Hucklesby (1996) observed, defence
solicitors tend to present community ties information to the court. From
another point of view, Morgan and Henderson (1998) found that contrary
to recommended practice, BISs often provided the prosecution with nega-
tive information that came to light. Thus, BISs may not be as advantageous
to a defendant’s application for bail. Indeed, since the present study was
conducted, the Probation National Standards issued by the Home Office
(2000) allows BISs to provide ‘negative’ information. This change may result
in fewer defendants agreeing to be interviewed and thus further reducing
the impact of BISs.
Second, the information pertaining to community ties presented by a BIS
may be somewhat meagre and vague. The guidelines state that bail infor-
mation should be ‘verified’, ‘factual’, ‘favourable’, and there should be ‘no
discussion or comment on the alleged offence’, ‘no opinion expressed’ and
‘no recommendation’ (cited in Lloyd 1992, p. 51). The HM Inspectorate of
Probation (1993) investigated the quality of the information provided on
bail information sheets by examining 60 sheets in schemes run in eight
probation areas throughout England and Wales (which may include many
court-based BISs) and six prison-based schemes. Seventy-seven per cent of
the bail information sheets satisfied the criteria that they should include
257
 Blackwell Publishers Ltd. 2002
verified, factual, favourable data, and 20% were considered ‘acceptable’.
However, 3% were deemed poor. Many sheets were ‘one liners’ that
included information pertaining solely to the defendant’s living arrange-
ments. The problem of meagre or vague information could be overcome by
gathering and recording information on integrated computerised systems
(Drakeford et al. 2001).
Third, magistrates may not base their decisions primarily on the risk of a
defendant absconding. Information pertaining to community ties aims to
help magistrates assess the likelihood of a defendant absconding if released
on bail. However, this is only one of the risk judgments that magistrates must
make before deciding whether or not to bail a defendant. The Bail Act 1976
also requires magistrates to judge the likelihood of a defendant offending
while on bail, and interfering with witnesses/obstructing justice. The
strength of a defendant’s community ties are not considered pertinent to
judging these two risks (Home Office 1974), and the law does not provide
guidance as to whether the three risks are to be weighed equally, and what
is considered a high risk. Very little research has investigated magistrates’
bail risk judgments (with the exception of Dhami 2001; Morgan and
Henderson 1998). Future development of initiatives such as BISs should
target the aims of magistrates’ decisions.
The fact that experimental studies have found American BISs to be effec-
tive (for example, Ares et al. 1963), suggests that differences in the operation
of the schemes may account for the differences in their effectiveness. Unlike
BISs in the English system, BISs in the American system gather information
on specific elements of community ties, and then score it on an objective,
fixed scale. Schemes provide an explicit recommendation regarding the
defendant’s suitability for bail, and information is presented directly to the
court. The question of how these differences in the operation of BISs in the
American and English systems impacts on their effectiveness requires inves-
tigation. If these differences do result in the desired effect of leading magis-
trates to remand fewer defendants in custody, then changes must be made
to the current operation of BISs in the English system.
Although there were no differences between the BIS and no-BIS group
in terms of the decisions made, magistrates in the former group were signif-
icantly more likely to use sureties and bail hostels, as conditions of bail.
According to Raine and Willson (1995), the rationale for imposing such
‘locating’ conditions is to prevent absconding. It has been argued that
magistrates often do not impose these conditions because they are not
informed of the availability of sureties and bail hostel places. However, in
the present study, where the BIS group were so informed, these conditions
were no more popular than in previous research (for example, Dhami 2001;
Dhami and Ayton 2001; East and Doherty 1984; Morgan and Henderson
1998; Raine and Willson 1994).
Magistrates in the BIS group demonstrated significantly greater confi-
dence in their decisions, than the no-BIS group, although both groups were
highly confident. Such high confidence is inappropriate in the light of the
extent of magistrates’ inconsistency and disagreement, which was similar to
that reported by Dhami and Ayton (2001). According to Zakay (1997):

258
 Blackwell Publishers Ltd. 2002
‘ongoing feelings of confidence may determine whether or not the execu-
tion of an ongoing activity will be continued, and, if it is continued, whether
any changes in strategy will take place’ (p. 233). Research in decision
science that compares confidence in judgment with judgment accuracy, has
found that high and unrealistic confidence levels are related to sub-optimal
strategies (for example, Zakay 1985; Zakay and Glickshon 1992). Thus, not
only does high confidence mislead individuals into using ineffective strate-
gies, it also renders it difficult to alter their current practice. Rather than
improve magistrates’ decision making, BISs may provide them with a false
sense of confidence.

Conclusion
Evaluations should be carefully designed to test the cause-effect relation-
ships that policy makers and researchers believe to exist between policy
initiatives and magistrates’ decision making. Although probation officers
demonstrate on overwhelming commitment and support for pre-trial
services such as BISs, they also recognise the difficulties and constraints
involved in operating schemes. It should be no surprise therefore, that the
findings of the present study do not support those of past evaluations of
BISs, especially in light of the fact that the present findings are consistent
with other research on magistrates’ bail decision making. Since the present
study was conducted, the Home Office has explicitly allowed schemes to
provide negative information on defendants. This may result in fewer defen-
dants consenting to be interviewed, and so reducing the future impact of
BISs.
The future of BISs seems unclear in light of budgetary cuts affecting the
schemes in the mid- to late-1990s. Drakeford et al. (2001) argue that the
impact of schemes may be increased if, amongst other things, there is
greater consistency in targeting defendants and collecting bail information,
if this information is stored on a shared computerised system, if bail infor-
mation work is more closely connected to the provision of other pre-trial
services such as bail support, and if the work of the probation service is
better communicated to other criminal justice agents. In addition, the
discussion presented here implies that the impact of BISs may be increased
if legislation explicitly requires magistrates to use the community ties cue,
and requires them to weigh the risk of absconding as greater than other
risks posed by bailed defendants. Schemes could also provide recommen-
dations for bail directly to magistrates. Indeed, the main justification for not
providing information directly to the court was that magistrates might
presume there was little positive to say in cases where information was not
provided. Not only is it clear that the prosecution may likewise attach nega-
tive attributions to such cases, but also the recent change enabling the provi-
sion of negative information may counteract this problem. Further
recommendations for change may emerge from research that examines
carefully the differences in the operation of more and less successful
schemes both in the English system and in other jurisdictions.
Perhaps more importantly, the validity of the community ties cue in
259
 Blackwell Publishers Ltd. 2002
predicting absconding should first be established. There is need for more
research into whether the decisions based on BISs are the right ones. Are
defendants bailed due to BISs less likely to abscond? Only when such data
are available, can we be certain that the model of good practice is really that.
In the meantime, the present study demonstrates that BISs are merely
providing magistrates with a false sense of confidence.4

Notes
1 The means and variances are related and so a 2 arcsines transformation of the
proportions was first performed. To adjust for a type I error a Bonferroni correc-
tion was applied to the alpha level.
2 The author is not aware of any appropriate statistical test to apply to this data-set,
which involves a between-subjects analysis of repeated observations in each group,
and where the dependent variable is dichotomous.
3 The null findings cannot largely be attributed to a lack of statistical power to detect
an effect. Past evaluations (that is, Lloyd 1992; Stone 1988) suggest that BISs have
a small to medium effect on bail decisions. In the present study, in order to detect
a small effect of 0.35, where alpha = 0.05, with n1 = 77 and n2 = 55, power is 0.64 for
a one-sided statistical test. For a medium effect of 0.50, power is 0.88. Thus, we have
a 5% chance of detecting an effect when in fact it is not present, and a 64% or 88%
chance of detecting an effect if it is present.
4 Acknowledgements: The present research was funded by the Department of
Psychology, City University, London, UK. Portions of the present research were
reported in the author’s doctoral thesis. I thank David Mandel, Peter Ayton, and two
anonymous reviewers for their comments on an earlier draft of the manuscript.

References
Ares, C. E., Rankin, A. and Struz, H. (1963) ‘The Manhattan bail project: an interim
report on the use of pre-trial parole’, New York University Law Review, 38, 38–67.
Brown, D. (1998) Offending on Bail and the Police Use of Conditional Bail (Research
Findings No. 72), London: Home Office.
Davies, C. (1971) ‘Pre-trial imprisonment: a Liverpool study’, British Journal of
Criminology, 11, 32–48.
Dhami, M. K. (2001) ‘Bailing and jailing the fast and frugal way: an application of
social judgement theory and simple heuristics to English magistrates’ remand
decisions’ (unpublished PhD thesis, City University, London).
Dhami, M. K. and Ayton, P. (2001) ‘Bailing and jailing the fast and frugal way’,
Journal of Behavioral Decision Making, 14, 141–68.
Doherty, M. J. and East, R. (1985) ‘Bail decisions in magistrates’ courts’, British
Journal of Criminology, 25, 251–66.
Drakeford, M., Haines, K., Cotton, B. and Octigan, M. (2001) Pre-trial Services and the
Future of Probation, Cardiff: University of Wales Press.
East, R. and Doherty, M. (1984) ‘The practical operation of bail’, Legal Action, 12–13.
Fiddes, C. and Lloyd, C. (1990) ‘Assessing the impact of Bail Information Schemes’,
RSD Bulletin 29, London: Home Office.
Fitzgerald, M. (1993) Ethnic Minorities in the Criminal Justice System (Research Study
No. 20), London: HMSO.
Godson, D. and Mitchell, C. (1991) Bail Information Schemes in English Magistrates’
Courts (RPU Paper), London: Inner London Probation Service.

260
 Blackwell Publishers Ltd. 2002
Hammond, D. (1988) ‘The effects of remand’, Prison Service, Journal, No. 69, 19.
HM Inspectorate of Probation (1993) Bail Information: Report of a Thematic Inspection in
1992, London: Home Office.
Home Office (1974) Bail Procedures in Magistrates’ Courts, London: HMSO.
Home Office (1988) Bail (Circular No. 25/88), London: Home Office.
Home Office (1997) ‘Notifiable offences, England and Wales, 1996’, Statistical Bulletin
3/97, London: Home Office.
Home Office (1999) ‘Cautions, court proceedings and sentencing, England and Wales
1998’, Statistical Bulletin 21/99, London: Home Office.
Home Office (2000) Probation National Standards, London: Home Office.
Hucklesby, A. (1996) ‘Bail or jail? The practical operation of the Bail Act 1976’, Journal
of Law and Society, 23, 213–33.
Janis, I. (1982) Victims of Groupthink, Boston: Houghton Mifflin.
Janis, I. and Mann, L. (1977) Decision Making, New York: Free Press.
Jones, P. (1985) ‘Remand decisions at magistrates’ courts’, in: D. Moxon (Ed.),
Managing Criminal Justice: A Collection of Papers, London: HMSO.
King, M. (1971) Bail or Custody?, Nottingham: The Cobden Trust.
Lloyd, C. (1992) Bail Information Schemes: Practice and Effect (RPU Paper 69), London:
Home Office.
Lord Chancellor’s Department (1999a) Judicial Appointments, London: Lord
Chancellor’s Department.
Lord Chancellor’s Department (1999b) Judicial Statistics, England and Wales 1997,
London: Lord Chancellor’s Department.
Mair, G. (1988) Bail and Probation Work: The ILPS Temporary Bail Action Project (RPU
Paper 46), London: Home Office.
Morgan, R. (1994) ‘An awkward anomaly: remand prisoners’, in: E. Player and M.
Jenkins (Eds.), Prisons After Woolf: Reform Through Riot, London: Routledge.
Morgan, P. M. and Henderson, P. F. (1998) Remand Decisions and Offending on Bail:
Evaluation of the Bail Process Project (Home Office Research Study No. 184), London:
Home Office.
Morris, G. (1997) Shaw’s 1997/98 Directory of Courts in the United Kingdom, Kent: Shaw
and Sons.
Nottingham, S. and Mitchell, C. (1994) Bail Information Schemes in Prison, Review of Prison
Service Agency Action Plan, London: Home Office.
Pearce, W. and Smith, M. (1976) ILPAS/Vera Bail Project: Report of the First Year and
Proposal for the Second Year, London: Inner London Probation and After-Care Service.
Petty, R. E., Harkins, S. and Williams, K. (1980) ‘The effects of diffusion of cognitive
effort on attitudes: an information processing view’, Journal of Personality and Social
Psychology, 38, 81–92.
Raine, J. W. and Willson, M. J. (1994) Conditional Bail or Bail with Conditions? The Use and
Effectiveness of Bail Conditions, London: Home Office.
Raine, J. W. and Willson, M. J. (1995) ‘The imposition of conditions in bail decisions:
from summary punishment to better behaviour on remand’, Howard Journal, 35,
256–70.
Stone, C. (1988) Bail Information for the Crown Prosecution Service, London: Vera Institute
of Justice.
Tobolowsky, P. and Quinn, J. (1993) ‘Pretrial release in the 1990s: Texas takes another
look at nonfinancial release conditions’, New England Journal on Criminal and Civil
Confinement, 19, 267–327.
Weldon, E. and Gargano, G. M. (1985) ‘Cognitive effort in additive task groups: the
effects of shared responsibility on the quality of multiattribute judgments’,
Organizational Behavior and Human Decision Processes, 36, 348–61.

261
 Blackwell Publishers Ltd. 2002
Williams, B. (1992) Bail Information: An Evaluation of the Scheme at HMP Moorland,
Bradford: Haughton Publishing.
Zakay, D. (1985) ‘Post-decisional confidence and conflict experienced in a choice
process’, Acta Psychologica, 58, 75–80.
Zakay, D. (1997) ‘Post-decisional confidence – can it be trusted?’, in: R. Ranyard, W.
R. Crozier and O. Svenson (Eds.), Decision Making: Cognitive Models and
Explanations, London: Routledge.
Zakay, D. and Glickshon, J. (1992) ‘Overconfidence in a multiple-choice test and its
relationship with achievement’, The Psychological Record, 42, 519–24.
Date submitted: August 2001
Date accepted: January 2002

262
 Blackwell Publishers Ltd. 2002