Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
B. Keith Crew
To cite this article: B. Keith Crew (1991) Sex differences in criminal sentencing: Chivalry or
patriarchy?, Justice Quarterly, 8:1, 59-83, DOI: 10.1080/07418829100090911
B. K E I T H C R E W
University of Northern Iowa
The effects of legal and extralegal variables on sentences are compared for
separate samples of male and female felony imprisonment cases from a
Downloaded by [University of Calgary] at 22:58 22 November 2017
had different sentencing laws for m e n and for women: sex was a
legal, not an extralegal, factor (Clements 1972; Temin 1973).
The tendency for women to receive more lenient treatment,
w h e t h e r observed empirically or merely assumed, traditionally has
been interpreted as an expression of either chivalry or paternal-
ism. Moulds (1978:280) distinguishes the two terms as follows:
chivalry refers to men's unwillingness to inflict h a r m on a woman,
combined with a certain amount of disbelief that a woman could
be really criminal; paternalism, on the other hand, refers to the at-
titude that women are childlike and therefore 1) in need of protec-
tion and 2) are not fully responsible for their actions, criminal or
Downloaded by [University of Calgary] at 22:58 22 November 2017
Predicted Sentencing
of Female Offenders Intervening or Condi-
Hypothesis Compared to Males tional Factors
Prior Research
Studies of the effects of extralegal factors on sentencing gen-
erally have deployed some version of a basic discrimination or dis-
parity model. In these analyses, zero-order or total associations
between defendants' attributes and criminal sanctions are re-
ported; t h e n controls are introduced for legally relevant variables.
A n y residual or "direct" effects of association between the extrale-
gal variables and the dependent variables are taken as evidence of
discriminatory treatment. The dependent variable is often a di-
chotomy representing the "in~out" decision (imprisonment or pro-
bation). Nagel and Hagan (1982) reviewed five studies of this type
(Hagan and Bernstein 1979; Hagan, Nagel, and Albonetti 1980; My-
ers 1979; Nagel, Cardascia, and Ross 1982; Rhodes 1977). They con-
cluded that there is a small but statistically significant gender
B. KEITH CREW 63
processes of judges and prosecutors differ for male and for female
offenders. In other words, the relationship between sex, sanction
severity, and intervening legally relevant factors is interactive
r a t h e r than additive. Thus it may be more appropriate to con-
struct a separate multivariate model for each sex than to "partial
out" a zero-order correlation between sex and sentencing. For the
present study, I conducted cross-sex analyses on a sample of males
and females sentenced to prison in a state system.
An essential part of any study of sentencing discrepancies is
the inclusion of controls for legally relevant factors. The most im-
portant of these factors are assumed to be prior criminal record
Downloaded by [University of Calgary] at 22:58 22 November 2017
Sample
I collected data on all women (N--108) convicted of felonies
and sentenced to a t e r m of imprisonment in K e n t u c k y in 1980.
For comparison, I selected a 10 percent random sample of male fel-
ony offenders sentenced in the same year. The final sample size
was 228 men. Although the small sample sizes presented some dif-
ficulty for the statistical analyses, t h e y allowed for the inclusion of
more detailed information t h a n is typical in sentencing studies.
The nature of the sample places some important limitations
Downloaded by [University of Calgary] at 22:58 22 November 2017
Variables
The variables included in this study are summarized in Table
2. The data were collected from inmates' official "jackets," on file
at the state correctional department's central office. All variables
except sentence length were coded from the pre-sentence investi-
gation (PSI) report. I coded sentence length from the commitment
order, the legal document remanding the offender to the custody
of the corrections department. In addition to the quantitative vari-
ables listed in Table 2, I kept narrative descriptions of other infor-
mation relevant to the processing of the cases in the form of field
notes. These notes included, among other things, unusual features
of some cases that did not fit into the a priori coding scheme, com-
ments by prosecutors and other officials, and instances where the
defendant's version of events differed from the official story.
Means*(Standard
Deviations) for Each
_ Variable Coding Description Variable, by Sex
Males Females
Sex 0 = female 1----male (N=228) (N----108)
Race 0=white 1=black .21 .23
(.41) (.43)
Priors Prior felony convictions .42"* .30"*
(interval) (.50) (.46)
Offense Sellin-Wolfgang offense 11.52 9.79
seriousness score (11.60) (12.74)
Downloaded by [University of Calgary] at 22:58 22 November 2017
(interval: r a n g e = 0 to 71)
Charges Charge severity 14.50 10.37
interval: m a x i m u m (11.36) (7.85)
statutory
penalty in years
Codefendant Codefendant named? .37 .52
0 = n o 1---yes (.48) (.50)
Children Number of dependent .26** .68**
children (.44) (.47)
Married Marital status .53 .22
1 = married 0 = n o t (.50) (.42)
married
Employed Employed full time at 46 .22
arrest?
1 = y e s 0--no (.50) (.42)
Sentence Sentence length in years 6.81 5.42
(10.73) (10.68)
* F o r d i c h o t o m o u s variables, t h e m e a n e q u a l s t h e p e r c e n t a g e f a l l i n g into t h e
c a t e g o r y coded 1.
** P r o p o r t i o n w i t h o n e child or m o r e .
STATISTICAL ANALYSIS
I performed the first stage of the analysis on the pooled sam-
ple, with sex entered as one of the independent variables. Sex was
coded so that O--female and l = m a l e ; therefore positive coeffi-
cients indicate longer sentences for men, whereas negative coeffi-
cients indicate longer sentences for women. I regressed sentence
length on all legal and extralegal variables, including two-way and
three-way interactions between sex and the other independent
variables. The results, including significant interaction effects, are
displayed in Table 3.
As expected, the most important determinant of sentence
length is charge severity (variable CHARGES): the zero-order
correlation between charging and sentencing is .610 for the pooled
sample. Charge severity is also the variable closest to sentencing
in the causal order; i.e., all other independent variables are tempo-
rally prior to charge severity. It is possible, then, that some of the
other independent variables affect sentence length indirectly via
charge severity. To illustrate the role of charge severity as an in-
tervening variable, Table 3 reports two sets of regression coeffi-
cients. In the first set, charge severity is excluded from the model;
t h e n the equation is recalculated with charge severity included.
Indirect effects via charge severity can be calculated as the differ-
ence between the coefficients in the first equation and those in the
second equation (Alwin and Hauser 1975). Thus, for example, the
first equation in Table 3 shows a standardized coefficient (beta) of
.124 for the variable PRIORS. In the second equation, the beta for
PRIORS is only .077, showing that charge severity accounts for
part of the effect of prior convictions on sentence length.
On the average, male offenders received longer sentences
(mean=6.8) than female offenders (mean=4.5). Although the lin-
ear relationship between sex and sentencing is significant and is in
the expected direction, it is not very strong (r---.101) and is ex-
plained largely by offense seriousness and charge severity. When
Downloaded by [University of Calgary] at 22:58 22 November 2017
Table 3. Regression Coefficients (b), Beta Weights (B), and Zero-Order Correlation Coefficients (r) for Sentence Length with Sex
and Other Variables
Table 4. Regression Coefficients (b), Beta Weights (B), and Zero-Order Correlations (r) for Sentence Length, by Offender's Sex
Women Men
Independent Variables r b (s.e.) ........B r b (s.e.) B
Race .042 .214 (1.917) .008 .184"* -7.486 (2.831)** -.283
Married .223** 1.255 (1.933) .050 -.073 -1.034 (1.620) -.041
Children -.158" -1.593 (1.690) -.069 -.061 -1.173 (1.537) -.048
Employed .155" 1.133 (.474)* .175 -.062 .183 (.366) .027
Codefendant -.216" -.979 (1.615) -.045 -.075 -.965 (1.130) -.022
Priors -.011 .261 (1.752) .011 .083 .888 (1.256) .041
Offense .526*** .301 (.070)*** .358 .312"** .015 (.059) .016
Charges .608*** .603 (.113)*** .438 .614"** .519 (.059)*** .550
Interaction Terms:
Race*Priors 7.758 (2.996)** .210
Race*Offense .501 (.155)** .288
R2 .475 .405
* p < .05 ** p < .01 *** p < . 0 0 0 1
Note: S t a n d a r d e r r o r s a r e in p a r e n t h e s e s f o l l o w i n g r e g r e s s i o n c o e f f i c i e n t s .
72 SEX DIFFERENCES IN SENTENCING
ing that black offenders receive longer sentences. In the full re-
gression model shown in Table 4, the effect is negative (beta = -
.283), showing that blacks receive shorter sentences. This reversal
occurs because of the significant interactions between race and
prior convictions (RACE*PRIORS) and between race and offense
seriousness (RACE*OFFENSE). Neither of these interaction
terms is significant for female offenders. A test for difference of
slope shows that the RACE*OFFENSE effect is significantly dif-
ferent across the subsamples. This finding is consistent with the
finding of the significant three-way interaction
(SEX*RACE*OFFENSE) in Table 3. It appears that being a black
male increases the effects of offense seriousness (and, to a lesser
degree, the effects of prior convictions) on sentence length.
Table 5 illustrates the clustering of the most severe sanctions
on black males. The mean offense seriousness scores are not sig-
nificantly different for the four race/gender classifications. Never-
theless, on the average, black m e n are convicted of more serious
charges and sentenced to longer prison terms than the other three
groups. The average sentence for black men is almost twice that
of any of the other categories.
legally relevant variables. This is not the case for two of the extra-
legal variables in the present analysis, namely employment status
of female offenders and race of male offenders. Furthermore, the
legal model cannot account for the sex-linked differences observed
in the effects of offense seriousness on sentence severity.
We find sex-linked differences in sentencing and charging that
are not predicted by the legal model. The overall pattern of re-
sults illustrates the necessity for a combined model of gender and
sanctioning because neither a general pattern favoring women
over m e n (chivalry/paternalism) nor one of special punitiveness
toward w o m e n (evil woman) is apparent. Also, the data provide
Downloaded by [University of Calgary] at 22:58 22 November 2017
QUALITATIVE ANALYSIS
In order to explore further some of the relationships reported
in the statistical analysis, I analyzed qualitatively the narrative ac-
counts of the offenses and the social histories of the offenders con-
tained in the pre,sentence investigation reports. Of particular
concern was the likelihood that men's and women's crimes dif-
fered in ways not captured by the operationalization of the key va-
riable of offense seriousness. For example, it is commonly held
that female robbers and burglars typically act in an accessory or
subordinate role with male partners (Bowker 1978:277). Before
conducting the qualitative analysis, I sorted all cases by computer
according to the general legal category of the most serious offense
(e.g., nonviolent property crimes, burglary, assault, robbery, homi-
cide). Then I was able to compare the actions of men and of wo-
m e n convicted of legally similar crimes.
The tendency of w o m e n to play a subordinate role is particu-
larly evident in sexual assaults. In the present sample, 14 men and
B. KEITH CREW 75
DISCUSSION
Although it is commonly assumed that female offenders re-
ceive more lenient sentences than their male counterparts, the
present study provides little support for this hypothesis. Similarly,
there is no consistent evidence of an "evil woman" effect whereby
w o m e n are penalized more harshly t h a n men, unless one inter-
prets as such the stronger effect of offense seriousness for women.
78 SEX DIFFERENCES IN SENTENCING
REFERENCF~
C~
O0