|
Measure
3.3.4: Equality and Fairness in Bail Decisions
The
purpose of this measure is to provide information to the court concerning
the nature of the factors associated with bail, bond, and release on
recognizance decisions.29 In making these
decisions, a court should focus on factors permitted by law. One way to
measure the court’s reliance on appropriate factors is to determine
whether differences in bail decisions are linked more to factors recognized
in law or to extra-legal factors such as the defendant’s race or gender,
the judge assigned to the case, or the geographic location of the court.
According to Standard 3.3, the greater the degree to which the differences
in the bail status of defendants are consistent with factors permitted by
law, the better the court is performing on this measure. The remainder of
this discussion outlines a step-by-step procedure that courts can use to
measure and assess factors associated with bail decisions.30
Planning/Preparation.
The initial step is to identify the factors permitted by
law to shape the court’s bail decisions. Because States have different
bail guidelines, the list of factors will differ somewhat across
jurisdictions. However, most courts use a core set of factors in deciding
whether to release the defendant on recognizance and in setting the dollar
amount of the required surety bond if the defendant is not released.
Legally
relevant factors are as follows:
-
Defendant’s
Criminal and Court History
(1)
Prior record—Does the defendant have prior felony convictions? If so, how
many and for what offenses? The notion is that it is rational for the court
to set stiffer bond requirements for a more extensive prior record,
especially if the defendant has recent convictions for the same offense.
Some States may incorporate this rationale explicitly into bail guidelines
by limiting the release of "dangerous offenders." Finally, did the
defendant intimidate witnesses while on release for prior offenses? Such
behavior also is grounds for imposing a more restrictive bond.
(2)
Prior court appearances—Has the defendant missed prior court appearances?
How many times? Did the defendant leave the area on those occasions? Because
a rationale of bail is to ensure court appearance, previous
failures-to-appear also are reasonable grounds for imposing a more
restrictive bond.
(3)
Current legal status—Is the defendant on parole or probation? Are there
outstanding warrants? Parole or probation violations are considered sound
reasons for imposing stricter bond conditions. Similarly, an outstanding
warrant justifies stricter bond conditions.
Is
the defendant charged with a violent offense? Was there alleged bodily harm
caused to a victim? What is the length of the sentence on conviction of the
charged offense? It is often deemed appropriate to place more constraints on
individuals who are believed either to pose serious threats to the community
or face the possibility of severe sanctions.
Is
the defendant a resident of the jurisdiction? For how long? With whom does
the defendant live? Do family members live in the area? Is the defendant
employed? What is the defendant’s monthly income? Individuals with close
ties to the community are considered likely to appear in court when required
and are, therefore, regarded as appropriate candidates for release on
recognizance or low surety bonds.
Is
the defendant currently using drugs? Could the defendant’s mental or
physical condition be impaired by detention? Defendants free of drugs or
likely to suffer under detention should receive less restrictive bonds.
To
determine whether legally irrelevant factors affect bail decisions, data
also must be collected on these factors. Legally irrelevant factors include:
These
characteristics include race and gender.
Was
counsel available to the defendant? If so, when? What type of attorney
represented the defendant?
Each
judge need only be identified by an alphabetical character (e.g., Judge A,
Judge B, and Judge C, and so forth). The measure is included to see if the
composite effect of judge identity is greater than the effect of legally
relevant factors.
In
addition to identifying possible determinants of bail decisions, the
decisions themselves need to be outlined. Three of the most fundamental
issues are as follows:
-
Is
the defendant released on recognizance? Because of limited incomes, many
defendants cannot post even modest surety bonds. For these defendants,
release on recognizance may be the only avenue to pretrial release. It is
important therefore to see the relative frequency with which the court
decides to use this option.31
-
If
the decision is made not to release the defendant on recognizance, what is
the amount of the surety bond?
-
If
a different bail decision is made after the first appearance, should the
initial or subsequent decisions be counted? That is, if a surety bond is set
but the defendant is later released on recognizance, should the defendant be
considered to be released on recognizance? If the amount of the surety bond
is lowered or raised at a later proceeding, which figure should be recorded?
One approach to this question is to record bail status at the initial
appearance separately from the decision in place at 15 or 30 days after the
first appearance. This strategy captures more of the legal process without
elevating one decision over another.
Data
Collection. The
information necessary for this measure is available in closed court case
records and the records of local bail agencies, pretrial release
organizations, or probation departments. The process of selecting cases for
analysis involves drawing a sample of 1,000 closed court cases and tracing
those cases back to the other organization’s files. (The sample of cases
can be drawn from the pool of cases used for Measure 3.3.3, Equality and
Fairness in Sentencing).
Measurement
of bail decision determinants consists of a combination of quantitative
scales and classification schemes. An illustrative data collection form is
offered as a way of measuring the factors that determine bail decisions and
the decisions themselves. (See Form
3.3.4, Illustrative Bail Decision Data Collection Form.)
Generally,
the factors included on the data collection form are the same as those used
by researchers in the field.32 However, in some
jurisdictions, the court may never receive information on specific aspects
of the defendant’s community ties, character, or socioeconomic status.
Instead, these factors may be taken into account by a bail agency that
recommends bail decisions to the court. If this is the situation, the bail
agency’s recommendation should be considered a surrogate for those
factors.33
Data
Analysis and Report Preparation.
Quantitative techniques can be applied to the data and each data element
assessed for its effect on bail decisions after taking into account the
influence of all other factors. The results of the analyses will tell the
court whether and to what extent each legally irrelevant factor influences
bail decisions. Results also will tell the court whether and to what extent
legally relevant factors are more influential in decisionmaking than legally
irrelevant ones.
The
two types of bail decisions require different types of analyses: logit
analysis and regression analysis. They are discussed below.
In
the case of the decision to release or not to release on recognizance, an
appropriate technique is logit analysis. The basic results of logit analysis
are numbers, called coefficients. A coefficient is associated with each
factor. The sign (+) of the coefficient indicates whether there is a
positive relationship (e.g., the longer the defendant has lived in the
community, the more likely his or her release on recognizance) or inverse
relationship (e.g., the greater the number of past failures to appear, the
less likely his or her release on recognizance) between each factor and the
bail decision. A comparison of the magnitude of the coefficients will
indicate the relative importance of each factor in predicting the likelihood
of a defendant being released on recognizance.
The
issue of what factors predict the amount of surety bonds is examined by
another technique called regression analysis. Regression analysis is
designed to indicate the independent effects of variables on an interval
measure such as the dollar amount of bonds. Similar to logit analysis, in
regression analysis coefficients are generated by the technique. The sign (+)
of a coefficient indicates if there is a positive (e.g., the greater the
number of prior felony convictions, the larger the amount of the bond) or
inverse (e.g., if the offender has family members in the community, the
lower the bail amount) relationship between each factor and the bail amount.
The
coefficients bear upon the central purpose of the measure in two ways.
First, if legally irrelevant factors are not influencing decisions, the
coefficients associated with them should not be significantly greater than
zero.34 Secondly, the coefficients of all
legally relevant factors should be significantly larger than those of
irrelevant factors. An inspection of the coefficients should address these
issues.
After
the data have been gathered and analyzed, a key task is to present the
results to the court. Do the results make sense? For example, should the
court be concerned if the results indicate that having family members in the
community decreases a defendant’s chances of personal recognizance? What
does it mean if the presence of family members decreases the average surety
bond by a certain amount?
In
addition to reviewing the intuitive soundness of the results, the court
should assess their implications for court performance. Do the results
signal that irrelevant factors are not having undue influence? Or, do the
results confirm that irrelevant factors have emerged as unacceptably
powerful predictors of the court’s release decisions?
Finally,
the court must decide what to do with the results. Regardless of the level
of performance, what should be done? What sort of action is appropriate to
improve performance? For example, is a courtwide review of bail policies,
procedures and practices needed? Would special training programs for newly
appointed judges and programs on current developments in substantive and
procedural criminal law help? Although the court must make its own judgments
as to what is necessary and desirable, the empirical evidence should inform
the making of that judgment.
____________________________________
29
In most jurisdictions the majority of bonds are released on recognizance and
surety bonds. However, in some courts, case bonds also are prominent. In
this event, the court should consider what factors account for the amounts
of different cash bonds.
30 The measure proposed outlines a statistical
approach to assessing whether there is undue disparity and bias. However, it
is not a complete treatment of every aspect of particular techniques and
their improvement. For this reason, the court may wish to consult outside
experts when applying the measure.
31 The majority of defendants released on
recognizance typically have nonfinancial conditions placed on them, such as
third-party custody, prohibitions against returning to the scene of the
crime, and restrictions on residence, travel, associations, drug and alcohol
use, and weapons possession. These conditions are not crucial to determining
whether legally relevant or irrelevant factors explain who is released and
who is not. Hence, the court should collect data on these matters only if it
seeks to pursue other research questions concerning bail decisions.
32 See, for example, J. Goldkamp and M.
Gottfredson, Guidelines for the Pretrial Release Decision: Superior
Court of Arizona, Maricopa County; Circuit and County Courts, Dad County;
Boston Municipal Court; and Suffolk County Superior Court, Bail
Guidelines Project (Philadelphia: Temple University, 1985).
33 See, for example, I. Nagel, "The
Legal/Extra-Legal Controversy: Judicial Decisions in Pretrial Release,"
Law and Society Review 17 (1983):481.
34 A coefficient may be greater but not statistically
greater than zero because the factor under consideration (e.g., race) does
not have consistent, uniform effects on what is being measured (e.g., to
release or not to release on recognizance). However, a statistical test
performed by the software will indicate whether each coefficient is
significantly greater than zero.
Go
to Form 3.3.4
Go
to Measure 3.3.3
Go
to Standard 3.3
Go
to Table of Contents |
