The Impact of Aggravating and Mitigating Factors on the Sentence Severity of Sex Offenders

Abstract

The aggravating and mitigating circumstances that contribute to increased, or decreased, sentence severity for sex offenders have largely been unexplored. Although previous studies have evaluated offending groups who have targeted adult-only, or children-only victims, the current study compares the sentencing outcomes of both offending groups. Using a sample of 519 federally sentenced sex offenders in the province of Quebec the current study explores the extent to which the Canadian criminal justice system penalizes offender- and offense-based characteristics. The results indicate that offense-based characteristics increased sentence severity for offenders who victimized adults and offender-based characteristics influenced sentence severity for offenders who victimized children. Findings are discussed within the context of previous studies to empirically explore sex offender sentencing and compare differences that aggravating and mitigating circumstances have on sentence outcomes.

Keywords

sentence severity sex offenders aggravating factors mitigating factors recidivism premium

The sentencing of an offender is one of a multitude of decisions that are made as an individual proceeds through the criminal justice system (Simon, 1996). Although the length of an offender’s sentence is normally fixed within sentencing guidelines, aggravating and mitigating circumstances function to either increase, or decrease, sentence severity (Ruby et al., 2008). Mitigating circumstances (those that can decrease sentence severity) can include any of the following: being a first-time offender, having prior good character, submitting a guilty plea or expressing remorse, being impaired at the time of the offense,¹ having a stable employment record, an offender’s willingness to participate in rehabilitation, attempts to provide reparation or compensation, the presence of provocation or duress at the time of the criminal event, having prior convictions in the distant past but not having committed crimes for an extended period of time, and coming from a disadvantaged background (Ruby et al., 2008). Conversely, aggravating factors (those that can increase sentence severity) can include the threat or use of violence and/or a weapon, cruelty, inflicting significant physical or psychological damages, targeting more vulnerable victims, the planning and purposeful execution of an offense, and finally having a previous criminal record (Ruby et al., 2008).The goal of the current study is thus to explore the potential impact of aggravating and mitigating circumstances on sex offender sentence severity and to compare any differences that exist between offenders who target adults and offenders who target children.

The Role of Aggravating andMitigating Factors in Sentence Severity

One of the most important aggravating factors in the sentencing of any offender is the extent of his or her prior criminal activities. The role of prior criminal convictions in the process of offender sentencing is commonly referred to as “the recidivist sentencing premium” (Bennett, 2010; Roberts, 2008, 1997, 1996, 1994; von Hirsch, 2010; Welch, Gruhl, & Spohn, 1984). A central tenant of this perspective is that offenders who have committed multiple criminal offenses, or the same offense multiple times, have not been deterred from criminal involvement by previous incarceration, and thus, are less apt to be deterred from future criminal involvement by legal sanctions (Roberts, 1997). As such, previous convictions can meaningfully influence sentencing severity to the extent that they not only affect decisions to incarcerate or issue community-based sanctions but also to the extent that they can affect the length of time that an offender must serve (Maxfield, 2002; Roberts, 2008, 1997; Welch et al., 1984). Notably, an offender’s prior record, or criminal career, has been identified as being the second most important factor, after the current offense, in determining an offender’s sentence length (Frase, 2010; Levesque, 2000; Roberts, 1997, 1996, 1994; Simon, 1996; Spohn & Welch, 1987; Welch et al., 1984). However, although prior convictions are the most commonly used measure of an offender’s criminal career, Spohn and Welch note that different measures of prior criminal activity (i.e., prior periods of incarceration, prior convictions, and prior arrests) can differently affect sentence severity with the number of prior periods of incarceration being associated with the most increased severity.

In addition to prior criminal activities, the extent to which legal and extra-legal factors (or aggravating and mitigating circumstances) contribute to sentence severity, or to a judge’s decision to incarcerate, have been explored for general and violent offending populations. A myriad of factors, including social context, geographic locations (urban vs. rural), judge’s discretion, offender characteristics, the offender’s relationship to the victim, political conservatism, as well as race and gender, have been identified and empirically examined. The findings of these studies suggest that being female, expressing remorse, being under the influence of alcohol at the time of the crime event, being a first-time offender, having targeted a victim known to the offender, any negative characteristics associated with the victim (i.e., the victim being under the influence at the time of the offense, being a sex trade worker, being alone public places at night), being older, and being sentenced in an urban location have been found to be associated with reduced sentence severity. However, being black, being male, being young, being sentenced in a rural location, and victimizing a female have all been associated with increased sentence severity (Crawford, 2000; Curry, Lee, & Rodriguez, 2004; Daly & Tonry, 1997; Harrell, 1981; Johnson, Van Wingerden, & Nieubeerta, 2010; Kingsnorth et al., 1999; Steffensmeier, Ulmer, & Kramer, 1998; Ulmer & Johnson, 2004; Wooldredge, 2010). However, the extent to which the intersection of these factors contributes to increased sentence severity for sex offenders remains relatively unexplored.

The Sentencing of Sex Offenders

Although much research has been devoted to examining disparities among gender and race sentencing outcomes few studies have empirically explored sex offender sentencing outcomes (Kingsnorth et al., 1999; Levesque, 2000; Walsh, 1984, 1994). What can be garnered from the literature that exists pertaining to sex offenders and sentence severity centers mostly upon offenders who target children as the victims of their sexual crimes. In his 2000 study exploring how factors such as offender’s prior crimes and offense characteristics can affect sex offender sentencing outcomes, Levesque presents that sentencing severity was more related to the offender’s criminal career than victim characteristics or experiences. On average, the length of the prison term served by the sex offenders in Levesque’s sample was 3.5 years or 42 months.²

However, Patrick and Marsh (2011) offer differing conclusions in their study of 815 cases of child sexual abuse for which the offender was sentenced to either prison or probation. While examining offense, offender, and victim characteristics they found that increases in the age of the offender, the total number of charges, and the total number of prison sentences all increased the odds of being sent to prison. However, the closeness between the victim and the offender, and amending the charge to a nonsexual crime, both decreased the odds of going to prison. Moreover, they found that for those sent to prison, offenders with younger victims, with more charges, and with a higher income received longer sentences, suggesting that offense-based characteristics are more salient in the sentencing of offenders than the characteristics of the victim or the offender (Patrick & Marsh, 2011).

When comparing child sex offenders to violent offenders, Champion reports that offenders who sexually abused children were sentenced more severely (on average 16.4 months) than offenders who committed other violent offenses (on average 13.3 months), such as aggravated assault or attempted murder. Furthermore, regardless of their prior criminal histories, sex offenders were more likely to be incarcerated than to receive alternative measures such as probation (Champion, 1988). Moreover, when exploring potential differences in the sentence severity of heterosexual and homosexual child molesters, Walsh (1994) reports that after controlling for factors such as the offender’s prior record, the offender’s relationship with the victim, and the victim’s cooperation during the commission of the offense, the only factors that significantly increased sentence severity were having targeted a victim of the same sex and having denied responsibility for one’s actions. Walsh (1984, 1994) further notes that mitigating factors that commonly function to reduce sentence severity for general offending populations (i.e., stable employment records, being of a higher socioeconomic standing, being in a committed relationship, less frequently utilizing weapons, or excessive violence in the commission offenses) had little, or no, impact on sentencing decisions for sex offenders as they are frequently punished more severely than nonsex offenders. For example, the average sentence length for rapists reported in Walsh’s 1984 study was 5.54 years, or 66.5 months, whereas the average sentence length for those convicted of voluntary manslaughter was 4.49 years, or 53.8 months.

As demonstrated victim–offender relationships play an important role in sentence severity. As with the operationalization of an offender’s criminal career difficulties in establishing, and defining, an offender’s prior relationship with a victim have been reported (Simon, 1996). In light of this limitation, Simon (1996), and Kingsnorth et al. (1999), report that sex offenders who target strangers are sentenced to longer periods of incarceration. Kingsnorth et al. (1999) note that after controlling for aggravating circumstances, any prior relationship between the offender and the victim reduced the offender’s sentence length by 35 months. They further report an average sentence length of 203.61 months for offenders who sexually assaulted a stranger, while the average sentence length for offenders who victimized someone who was known to them was 93.07 months.

Included in the models tested by Kingsnorth et al. (1999) were indicators of ‘negative victim characteristics’ which included alcohol or drug consumption at the time of the offense, acts of prostitution, being alone in a public place at night or being alone at a bar, having no fixed address, assisting the offender in removing clothing, and hitchhiking or riding in a stranger’s car. They report that the presence of at least one negative victim characteristic removed approximately 17 months from the offender’s sentence length with additional negative factors reducing sentencing severity even more (Kingsnorth et al., 1999). Furthermore, they note that offenders who target children are more likely to be prosecuted and that as the age of victims increased (in 1-year increments) sentence lengths also increased by 1.5 months (Kingsnorth et al., 1999). The reported patterns of increased sentence length for offenders who were unknown to their victims and the impact that victim’s perceived risk taking behaviors can have on the sentence severity of sex offenders reported by Kingsnorth et al. have also been reported by Spohn and Spears (1996).

The Current Study

Given the relative absence of empirical studies exploring the role that aggravating and mitigating circumstances have on sex offender sentence severity, the current study explores the extent to which offense and offender characteristics are predictive of sentence severity. Whereas previous studies utilized samples of offenders who have targeted adult-only, or children-only, victims the current study will first analyze the sample in its entirety and subsequently identify offenders who targeted adults and those who targeted children, to determine whether different factors are more or less predictive for these offending groups. Drawing on the aggravating and mitigating factors identified by the empirical studies of Levesque (2000) and Kingsnorth et al. (1999), six groups of predictors will be tested (i.e., offender characteristics, victim characteristics, offense characteristics, offender remorse, criminal career parameters, and any charges in addition to the current sexual offense). Given the importance of the recidivist sentencing premium, it is expected that the criminal career parameters included in this study will be an important factor in determining sex offender sentencing severity. It is also anticipated that although offense characteristics may be more important determinants of sentence severity for offenders who victimize adults, offender characteristics are expected to be important determinants of sentence severity for offenders who victimized children.

Method

Sampling and Procedures

From April 1994 through to June 2000 all offenders convicted of, and incarcerated for, a sexual offense in the province of Quebec, Canada, were asked to participate in a longitudinal research project evaluating recidivism. Of the offenders admitted during the 6-year research period, 93% of the offenders consented to participate in the project (n = 553). This sample was comprised of all male sex offenders having received a prison sentence of at least 2 years meaning that these individuals had committed more serious sexual crimes and/or were offenders with more extensive criminal backgrounds than those sentenced to a provincial prison or other community-based measures (e.g., probation, fine, etc.).³ Of the 553 offenders admitted 34 cases (6%) were excluded due to incomplete criminal histories (i.e., missing nature of previous charges) or incomplete file information (i.e., victim characteristics, offense characteristics). Taken together, the current study utilized 519 offenders.

At the time of contact with the research team all participants were incarcerated at a maximum-security facility operated by the Correctional Service of Canada. Offenders are incarcerated at this institution for the duration of their correctional assessment procedures, an average of 8 weeks, after which they are transferred to an institution that can best accommodate their level of risk and treatment needs. Participation in the study was completely voluntary. All participants included in the study signed a consent form allowing the information gathered to be used solely for research purposes. Participants also consented to allow researchers to access their correctional files, which included their criminal history data. Criminal history data were subsequently compiled by trained research assistants, graduate students in criminology or psychology, using a computerized questionnaire (QIDS; Proulx, St.-Yves, & McKibben, 1994).

Sample Description

The information pertaining to the sociodemographic information of the offending sample is presented in Table 1. The sample was composed exclusively of adult male offenders (n = 519) who were predominantly Caucasian (87.9%) and who at the time of their prison admission were on average 40 years of age (SD = 13.13, range 18-78). At the time of their index offense almost two thirds of the sample participants were single (63.6%) and unemployed (59.8%), more than half had not completed high school (58%), and 52% of offenders reported having learning disabilities. The sample was characterized by a large portion of child molesters who predominantly targeted female victims, as 297 of the victims targeted were aged 13 and younger (65%) and 81% were female. However, the nature of the offenses for which the offenders were incarcerated varied greatly as 60.7% (n = 315) were charged with committing a sexual assault, 11.8% (n = 61) were charged with committing a sexual assault with a weapon, 19.8% (n = 103) were charged with committing an indecent assault, 26% were charged with committing a sexual contact offense (n = 136), and 9.8% were charged with committing an incest offense (n = 51).

Table 1.

Sample Description.

Sample (n = 519)
Sociodemographic information
Offender age at prison admission	Mean age = 40 years (SD = 12.13), range 18-78
Ethnicity (Caucasian)	87.9% (n = 456)
Civil status (no partner present)	63.6% (n = 328)
Educational achievement (not completed high school)	58% (n = 293)
Employment status (unemployed)	59.8% (n = 310)
Learning difficulties	52% (n = 263)
Sentence length (in months)	Mean age = 46.27 years (SD = 23.77), range 24-180
Offender characteristics
Authority figure over victim	2.9 % (n = 15)
Family member of victim	54.2% (n = 282)
Offender victim of sexual assault or contact (prior to 18)	41% (n = 214)
Victim characteristics
Female	81% (n = 419)
Male	19% (n = 98)
Age	Mean age = 15.91 (SD = 11.50) years, range 1-61
Victim in poor environment at the time of the offense	39% (n = 194)
Suffered physical injuries from offense	26.4% (n = 137)
Offense characteristics
Victim unknown to the offender	14% (n = 72)
Weapon used during the offense	17.4% (n = 90)
Victim forced into completing one or several sexual acts	63.7% (n = 330)
Offender remorse
Offender does not admit to charges	27.2% (n = 141)
Offender does not admit to damages caused to victim	59.6% (n = 308)
Offender does not admit responsibility	33.7% (n = 177)
Offender does not admit to having sexual issues	58.8% (n = 305)
Criminal career
At least one prior nonsexual violent charge	59.9% (n = 307)
At least one prior nonsexual nonviolent charge	60.3% (n = 313)
At least one prior sexual (contact and noncontact) sexual charge	32.6% (n = 169)
Additional current charges
At least one prior nonsexual violent charge	17 % (n = 88)
At least one prior nonsexual nonviolent charge	25.9% (n = 134)

Measures

Sentence length

As all of the offenders utilized in the current study were sentenced to a federal penitentiary in Canada, the minimum sentence length possible was 24 months, or 2 years, whereas the maximum prison sentence allowable for a sexual offense is 14 years or 168 months (Criminal Code, 1985). On average, the offenders were sentenced to 46.27 months, or 3.83 years (SD = 23.77). Offenders who targeted adult victims (here those above the legal age of sexual consent: aged 14 or older) were sentenced to an average of 50.44 months, or 4.20 years (SD = 28.51), whereas those who targeted children as their victims (aged 13 or younger) were sentenced to an average of 42.87 months or 3.57 years (SD = 18.95).

Victim characteristics

The legal age of sexual consent in Canada between 1994 and 2000 was 14 (Bill C-2, 2008).⁴ Victims ranged in age from 1 to 61 with a mean age of 15.91 (SD = 11.50). Approximately 57% of the victims targeted were aged 13 or younger, and the remaining 43% were aged 14 or older. About one third of the victims suffered physical injuries as a result of the offense (26.4%, n = 137), which was coded as (0) no physical injuries sustained and (1) physical injuries sustained, and 39% (n = 194) were deemed to be in a dysfunctional or poor environment at the time of the event, which was coded as (0) not in a poor or dysfunctional environment and (1) in a poor or dysfunctional environment. The victim from a poor or dysfunctional environment variable was scored yes when at least one of the following items was identified: the environment in which the victim was living in did not possess sufficient resources to meet the basic needs of the victim (e.g., sleeping, eating, clothing, housing, security) or the other family members living with the victim; the victim was coming from a dysfunctional background (e.g., physical, psychological, sexual abuse, prostitution), or where alcohol and/or drug abuse were present.

Offender characteristics

In conjunction with the sociodemographic factors (civil status, educational achievement, and employment status at index offense) three additional offender characteristics were included in the models: if the offender had learning disabilities, if the victim was a family member of the offender, and if the offender himself had been the victim of a sexual assault, or a sexual contact offense, prior to the age of 18. Approximately half of the participants reported having learning disabilities (52%, n = 263), a characteristic that has been taken into account in the sentencing of some offenders who have targeted victims below the age of legal sexual consent (Ruby et al., 2008). Learning disabilities were coded as (0) no learning disabilities and (1) learning disabilities. Just more than half of the sample participants (54%, n = 282) victimized a member of their family. Familial relations included: biological father, stepfather, adoptive father, a former stepfather, brother, half-brother, a member of the extended family, and a known intimate partner of a family member. This factor was coded as (0) victim was not a family member of the offender and (1) victim was a family member of the offender. Furthermore, 41% (n = 214) of the offenders incarcerated had been the victim of a sexual assault, or a sexual contact offense, prior to the age of 18.

Offense characteristics

Three factors pertaining to the criminal event were included in the model: if the offender victimized a stranger, the use of a weapon during the commission of the offense, and if the offender forced the victim into completing one, or multiple, sexual acts. In 14% of cases the offender victimized a stranger. The variable was coded in the following way: (0) offender known or had previously seen the victim and (1) offender unknown to the victim. In 17.4% of cases a weapon was used during the commission of the offense. The variable was coded as (0) no weapon used during the commission of the offense and (1) weapon used during the commission of the offense. Finally, in 63.7% of the cases the offender forced the victim into completing one or several sexual acts, which was coded as (0) did not force the victim into one or several sexual acts and (1) forced the victim into one or several sexual acts.

Offender remorse

Four facets of offender remorse were included: if the offender admitted to the charges, if the offender admitted to causing damages to the victim, if the offender admitted responsibility for the offense, and finally if the offender admitted to having sexual issues. For each of the four facets of remorse the variables were coded as (0) admits or partially admits remorse and (1) does not admit remorse. Offenders who admitted partial responsibility are those who acknowledged that a sexual act had occurred, but provided any one of the following as a justification for their behaviors: the victim wanted it, the victim deserved it, the victim was consenting, the victim explicitly asked for it, it was educative for the victim or that it was beneficial or therapeutic for the victim. More than half of the sample participants (59.6%, n = 308) did not admit to causing damages to their victim, 27.2% (n = 141) did not admit to the offense they were charged with, and 33.7% (n = 177) did not admit responsibility for the offense. Finally, 58.8% (n = 305) did not admit to having any sexual issues.

Criminal career

Given the significance placed on prior criminal activity in sentencing discretion, three criminal career parameters were included in the study: having at least one prior nonsexual violent charge as an adult, having at least one prior nonsexual/nonviolent charge as an adult, and finally having at least one prior sexual (including both contact and noncontact offenses) charge as an adult. Each variable was coded as (0) no prior charge and (1) at least one prior charge. One third of the sample participants (32.6%, n = 169) had previously been charged with at least one prior sexual offense, 59.2% (n = 307) had previously been charged with at least one prior nonsexual violent offense, and 60.3% (n = 313) had previously been charged with at least one prior nonsexual/nonviolent offense.

Additional charges in conjunction with the current sexual offense

In order to control for the potential effect that multiple charges may have had on the sentence severity the presence of nonsexual violent charges and nonsexual/nonviolent charges, in addition to the current sexual charge, were included in the models. Approximately one fifth of the participants (17 %, n = 88) had at least one additional nonsexual violent charge, and 25.9% (n = 134) had at least one additional nonsexual/nonviolent charge. Both variables were coded as (0) no additional charges and (1) at least one additional charge.

Analytic Strategy

The outcome variable of interest here is sentence length, which in the current study has been calculated in months. Given that the minimum sentence length for the sample is 24 months the dependant variable is truncated, and thus, it is not normally distributed. One solution to address a nonnormally distributed dependant variable is to transform it making useable in ordinary least squares or multiple linear regression (Tabachnick & Fidell, 2007). However, a more appropriate modeling strategy for nonnormally distributed count data is negative binomial regression, whereby any dependencies among the count data is accounted for by the inclusion linear combinations of linear effects in the linear predictor (Booth, Casella, Friedl, & Hobert, 2003; Land, McCall, & Nagin, 1996). As such, negative binomial regression with a log-link function was employed to explore the impact of aggregating and mitigating factors on sentence severity in the three offending groups.⁵ The first modeling strategy included the entire sample (both offenders who had targeted adults and children; n = 519) and explored five separate models to evaluate the effect of each group of covariates independently on sentence severity. Next a hierarchical model was computed to evaluate the impact that each covariate had on the model when controlling for additional aggregating and mitigating factors. The second modeling strategy identified only those offenders who had targeted victims aged 14 or older (the legal age of sexual consent; n = 217), and as before the covariates were tested independently as well as in a hierarchical model. Finally, the same sample modeling strategy was employed for offenders who had targeted victims aged 13 or younger (n = 297). The selection of offenders who had targeted victims above, and below, the legal age of sexual consent was conducted to allow for comparisons of any potential differences in the predictors of sentence severity for these offenders.

Results

The results of the negative binomial regression for the full sample are presented in Table 2. Of the offender characteristics the only covariate to emerge as significant was targeting a family member (b = −.17, SE = .05, p < .001) indicating that sentence severity decreased for those who victimized a family member. Next, the victim characteristics were considered. Of the four covariates two were significant: the age of the victim (b = .005, SE = .002, p < .05) and whether physical damages were caused during the offense (b = .16, SE = .06, p < .001). Accordingly, having an older victim and inflicting more severe damages to the victim were predictive of increased sentence severity. Each of the three offense characteristics tested were predictive of an increased sentence length: targeting a stranger victim (b = .18, SE = .07, p < .05), using a weapon during the offense (b = .34, SE = .06, p < .01), and forcing the victim into completing one or more sexual acts (b = .07, SE = .04, p < .10). Of the four facets of offender remorse only one factor was indicative of sentence severity as admitting to causing damages to the victim was predictive of a shorter sentence length (b = −.13, SE = .05, p < .05). Of the three criminal career parameters tested, only having prior nonsexual violent charges increased sentence severity (b = .17, SE = .05, p < .01).

Table 2.

Negative Binomial Regression.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)
Offender characteristics
Age of offender	.001 (.002)
Civil status	−.02 (.05)
Educational achievement	.01 (.05)
Employment status	.05 (.05)
Learning disabilities	−.02 (.05)
Family member of victim	−.17 (.05)****
Victim of sexual abuse or contact (−18)	−.04 (.04)
Victim characteristics
Gender		.02 (.05)
Age		.005 (.002)**
Environment at time of offense		.003 (.05)
Physical damages caused during offense		.16 (.06)***
Offense characteristics
Victim a stranger to offender			.18 (.07)**
Weapon used			.34 (.06)***
Victim forced into one or more sex act			.07 (.04)*
Offender remorse
Admits to charges				.10 (.06)
Admits to damages				−.13 (.06)**
Admits responsibility				.10 (.06)
Admits sexual issues				−.03 (.05)
Criminal career (prior charges)
Prior nonsexual violent					.17 (.05)***
Prior nonsexual/nonviolent					−.03 (.05)
Prior sexual (contact and noncontact)					.05 (.05)
Additional charges at current offense
Nonsexual violent						.10 (.07)
Nonsexual/nonviolent						.01 (.06)
Likelihood ratio χ²(df), p value	19.41 (7)***	28.67 (4)****	73.81 (3)****	10.21 (4)**	17.80 (3)****	3.49 (2)
Deviance (value/df)	0.999	1.00	0.998	0.997	0.999	0.998
Intercept	3.89 (.10)****	3.70 (.04)****	3.69 (.04)****	3.89 (.04)****	3.71 (.04)****	3.81 (.04)****

Note: n = 519.

p < .10. **p < .05. ***p < .01. ****p < .001.

Next, a hierarchical model was performed to investigate the impact that each covariate had upon its entry into the model, while controlling for all of the aggravating and mitigating factors (Table 3). The addition of the victim characteristics in the second block somewhat tempered the effect of victimizing a family member (b = −.15, SE = .05, p < .01), whereas the offender’s civil status (b = .01, SE = .05, p < .05), educational achievement (b = −.01, SE = .05, p < .05), and learning difficulties (b = −.15, SE = .05, p < .01) emerged as significant predictors of sentence severity. Furthermore, physical damages caused during the offense remained a significant predictor of sentence length (b = .20, SE = .05, p < .01). With the addition of the offense characteristics the offender’s civil status, educational achievement and learning disabilities no longer increased sentence severity. Offender age at prison admission emerged as a significant predictor (b = .01, SE = .001, p < .01), in conjunction with targeting a victim who was a stranger (b = .16, SE = .08, p < .05), using a weapon during the offense (b = .32, SE = .07, p < .001), and causing physical damages to the victim during the offense (b = .10, SE = .06, p < .10).

Table 3.

Hierarchial Negative Binomial Regression.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)
Offender characteristics
Age of offender	.001 (.002)	.004 (.002)	.01 (.001)***	.01 (.002)***	.01 (.002)***	.01 (.002)***
Civil status	−.02 (.05)	.01 (.05)**	.01 (.04)	.01 (.04)	.01 (.04)	.01 (.04)
Educational achievement	.01 (.05)	−.01 (.05)**	−.03 (.05)	−.03 (.05)	−.03 (.06)	−.04 (.05)
Employment status	.05 (.05)	.03 (.04)	.04 (.04)	.04 (.04)	.03 (.04)	.03 (.04)
Learning disabilities	−.02 (.05)	−.001 (.05)****	−.01 (.05)	−.01 (.05)	−.02 (.05)	−.01 (.05)
Family member of victim	−.17 (.05)****	−.15 (.05)***	−.10 (.05)**	−.10 (.05)	−.08 (.05)	−.08 (.05)*
Victim of sexual abuse or contact (−18)	−.04 (.04)	−.006 (.05)	−.01 (.04)	−.03 (.04)*	−.04 (.04)	−.04 (.04)
Victim characteristics
Gender		−.05 (.06)	−.04 (.06)	−.03 (.04)	−.04 (.07)	−.04 (.06)
Age		.002 (.002)	.00 (.002)	.00 (.00)	.00 (.002)	.00 (.06)
Environment at time of offense		−.002 (.05)	.03 (.04)	.03 (.04)	.03 (.04)	.02 (.04)
Physical damages caused during offense		.20 (.06)***	.10 (.06)*	.10 (.06)*	.10 (.06)**	.09 (.06)
Offense characteristics
Victim a stranger to offender			.16 (.08)**	.14 (.08)*	.15 (.08)**	.14 (.08)*
Weapon used			.32 (.07)****	.33 (.07)****	.31 (.07)****	.31 (.07)****
Victim forced into one or more sex. act			.05 (.05)	.05 (.05)	.05 (.05)	.06 (.04)
Offender remorse
Admits to charges				.02 (.06)	.03 (.06)	.02 (.06)
Admits to damages				−.04 (.05)	−.05 (.05)	−.05 (.05)
Admits responsibility				.06 (.06)	.05 (.06)	.06 (.06)
Admits sexual issues				−.10 (.05)*	−.08 (.05)	−.08 (.05)
Criminal career (prior charges)
Prior nonsexual violent					.07 (.05)	.06 (.05)
Prior nonsexual/nonviolent					−.01 (.05)	−.004 (.05)
Prior sexual (contact and noncontact)					.05 (−.05)	.05 (.04)
Additional charges at current offense
Nonsexual violent						.09 (.06)
Nonsexual/nonviolent						−.03 (.05)
Likelihood ratio χ²(df), p value	19.41 (7)***	45.01 (11)****	91.76 (14)****	96.96 (18)****	100.47 (21)****	104.33 (23)****
Deviance (value/df)	0.999	0.999	1.00	0.998	0.997	0.998
Intercept	3.89 (.10)****	3.67 (.12)****	3.52 (.11)****	3.56 (.11)****	3.52 (.11)****	3.49 (.10)****

Note: n = 519.

p < .10. **p < .05. ***p < .01. ****p < .001.

When the four facets of offender remorse were added in the next block offender age at prison admission (b = .01; SE = .002; p < .01), causing physical damages to the victim during the offense (b = .14; SE = .08; p < .10), and using a weapon during the offense (b = .33; SE = .07; p < .001) all remained predictors of sentence severity; however, being the victim of sexual abuse or a sexual contact offense prior to age 18 (b = –.03; SE = .04; p <.10), and admitting to having sexual issues (b = –.10; SE = .05; p < .10) nearly emerged as predictors of sentence severity. None of the criminal career parameters were significant predictors of sentence length, but the addition of these covariates strengthened the effect of causing physical damages during the offense (b = .10, SE = .06, p < .05) and targeting a stranger (b = .15, SE = .08, p < .05). With the addition of the final block of covariates controlling for additional current charges, only four covariates emerged as significant predictors of sentence length: offender age at prison admission (b = .01, SE = .002, p < .01), targeting a stranger victim (b = .14, SE = .08, p < .10), using a weapon during the commission of the offense (b = .31, SE = .07, p < .001), and targeting a family member (b = −.08, SE = .05, p < .10).

The same set of models were run selecting only offenders who had targeted victims aged 14 or older (the legal age of sexual consent). The results are presented in Table 4. Of the offender characteristics both employment status (b = .15, SE = .07, p < .05) and victimizing a family member (b = −.20, SE = .08, p < .01) emerged as significant predictors of sentence severity. The only victim characteristic to emerge as close to a significant predictor of sentence length was causing physical damages to the victim during the offense (b = .15, SE = .08, p < .10). Of the offense characteristics, both victimizing a stranger (b = .26, SE = .08, p < .01) and using a weapon during the commission of the offense (b = .33, SE = .07, p < .001) increased sentence severity. Of the four facets of offender remorse, the only facet significantly related to sentence length was admitting to damages caused to the victim, which decreased sentence severity (b = −.35, SE = .10, p < .001). Finally, having prior nonsexual violent charges was the only criminal career parameter predictive of sentence length with neither of the additional current charges being predictive of sentence length.

Table 4.

Negative Binomial Regression With Offenders Who Targeted Victims Above the Age of 14.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)
Offender characteristics
Age of offender	.00 (.004)
Civil status	.01 (.08)
Educational achievement	−.05 (.10)
Employment status	.15 (.07)**
Learning disabilities	−.08 (.10)
Family member of victim	−.20 (.08)***
Victim of sexual abuse or contact (−18)	−.02 (.07)
Victim characteristics
Gender		.01 (.10)
Age		.004 (.004)
Environment at time of offense		.02 (.08)
Physical damages caused during offense		.15 (.08)*
Offense characteristics
Victim a stranger to offender			.26 (.08)***
Weapon used			.33 (.07)****
Victim forced into one or more sex. act			.10 (.07)
Offender remorse
Admits to charges				.13 (.10)
Admits to damages				−.35 (.10)****
Admits responsibility				.06 (.10)
Admits sexual issues				−.01 (.08)
Criminal career (prior charges)
Prior nonsexual violent					.17 (.08)**
Prior nonsexual/nonviolent					.01 (.09)
Prior sexual (contact and noncontact)					.04 (.08)
Additional charges at current offense
Nonsexual violent						.17 (.11)
Nonsexual/nonviolent						.13 (.08)
Likelihood ratio χ²(df), p value	13.41 (7)**	8.10 (4)*	45.95 (3)****	19.50 (4)***	8.01 (3)**	11.12 (2)***
Deviance (value/df)	1.00	1.00	1.00	1.00	0.999	1.00
Intercept	3.99 (.14)****	3.73 (.10)****	3.67 (.05)****	4.01 (.07)****	3.79 (.09)****	3.83 (.04)****

Note: n = 217.

p < .10. **p < .05. ***p < .01. ****p < .001.

Employment status (b = .12, SE = .07, p < .10) and victimizing a family member (b = −.20, SE = .07, p < .01) remained significant predictors with the addition of the victim characteristics in the second block of the hierarchical model (Table 5). Physical damages sustained by the victim also emerged as a significant predictor in the second block of the model (b = .16, SE = .08, p < .05), but the effect was lost in all subsequent blocks. With the introduction of the offense characteristics, the effects of employment status (b = .13, SE = .07, p < .10) and victimizing a family member (b = −.13, SE = .08, p < .10) were mediated and lost in all subsequent blocks. However, victimizing a stranger (b = .22, SE = .02, p < .05) and using a weapon during the commission of the offense (b = .26, SE = .08, p < .001) both increased sentence severity. With the introduction of the four facets of offender remorse, victimizing a stranger (b = .19, SE = .09, p < .05) and using a weapon during the commission of the offense (b = .34, SE = .07, p < .001) remained predictive, whereas admitting to causing damages to the victim (b = −.20, SE = .09, p < .05) was the only facet of remorse to influence sentence length. The effects of victimizing a stranger, using a weapon, and admitting damages to the victim remained the same in the next block, with the introduction of the criminal career parameters, none of which were predictive of sentence length. Finally, with the addition of the last block of covariates (additional current charges), offender age at admission (b = .01, SE = .004, p < .05), educational achievement (b = −.17, SE = .09, p < .05), and having additional current nonsexual violent charges (b = .17, SE = .10, p < .10) emerged as significant whereas victimizing a stranger (b = .18, SE = .08, p < .05), using a weapon during the offense (b = .34, SE = .07, p < .001), and admitting to damages caused to the victim (b = −.21, SE = .09, p < .05) remained significant predictors of sentence length.

Table 5.

Hierarchical Negative Binomial Regression With Offenders Who Targeted Victims Aged 14 or Older.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)
Offender characteristics
Age of offender	.00 (.004)	.001 (.004)	.004 (.003)	.01 (.003)	.01 (.004)	.01 (.004)**
Civil status	.01 (.08)	.02 (.08)	.02 (.07)	.01 (.07)	.01 (.07)	.002 (.07)
Educational achievement	−.05 (.10)	−.07 (.10)	−.10 (.09)	−.11 (.09)	−.12 (.09)	−.17 (.09)**
Employment status	.15 (.07)**	.12 (.07)*	.13 (.07)*	.10 (.08)	.10 (.07)	.08 (.07)
Learning disabilities	−.08 (.10)	−.04 (.10)	−.05 (.09)	−.03 (.09)	−.03 (.09)	.03 (.08)
Family member of victim	−.20 (.08)***	−.22 (.07)***	−.13 (.08)*	−.11 (.08)	−.10 (.08)	−.11 (.07)
Victim of sexual abuse or contact (−18)	−.02 (.07)	−.01 (.08)	−.04 (.07)	−.04 (.07)	−.05 (.07)	−03 (.07)
Victim characteristics
Gender		−.10 (.11)	−.02 (.12)	−.04 (.12)	−.01 (.12)	−.04 (.12)
Age		.001 (.004)	.001 (.003)	.001 (.003)	.00 (.003)	−.002 (.003)
Environment at time of offense		−.002 (.08)	.07 (.07)	.10 (.08)	.08 (.07)	.04 (.07)
Physical damages caused during offense		.16 (.08)**	.04 (.07)	.03 (.07)	.02 (.07)	.04 (.07)
Offense characteristics
Victim a stranger to offender			.22 (.09)**	.19 (.09)**	.19 (.09)**	.18 (.08)**
Weapon used			.26 (.08)****	.34 (.07)****	.34 (.07)****	.34 (.07)****
Victim forced into one or more sex. act			.06 (.07)	.04 (.07)	.04 (.07)	.07 (.07)
Offender remorse
Admits to charges				.06 (.09)	.05 (.09)	.05 (.09)
Admits to damages				−.20 (.10)**	−20 (.09)**	−.21 (.09)**
Admits responsibility				.09 (.10)	.09 (.10)	.09 (.10)
Admits sexual issues				−.04 (.08)	−.02 (.10)	−.04 (.10)
Criminal career (prior charges)
Prior nonsexual violent					.006 (.09)	−.02 (.09)
Prior nonsexual/nonviolent					.09 (.09)	.10 (.09)
(continued)
Prior sexual (contact and noncontact)					.04 (.07)	.07 (.07)
Additional charges at current offense
Nonsexual violent						.17 (.10)*
Nonsexual/nonviolent						.13 (.08)
Likelihood ratio χ²(df), p value	13.41 (7)**	19.80 (11)**	56.51 (4)****	62.62 (18)****	63.96 (21)****	78.37 (23)****
Deviance (value/df)	1.00	1.00	1.00	0.999	1.00	0.997
Intercept	3.99 (.14)****	3.83 (.16)****	3.50 (.15)****	3.61 (.196)****	3.58 (.16)****	3.42 (.16)****

Note: n = 217.

p < .10. **p < .05. ***p < .01. ****p < .001.

In the final set of models, offenders who had targeted victims aged 13 or younger (n = 297) were selected (Table 6). The only offender characteristic found to be predictive of sentence length was offender age at prison admission (b = .01, SE = .002, p < .05). None of the victim characteristics or the offense characteristics were predictive of sentence length, and the only facet of offender remorse that emerged as significant was admitting to sexual issues (b = −.10, SE = .06, p < .10). Furthermore, none of the criminal career parameters or additional current charges were predictive of sentence length.

Table 6.

Negative Binomial Regression With Offenders Who Targeted Victims Aged 13 or Younger.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)
Offender characteristics
Age of offender	.01 (.002)**
Civil status	−.01 (.05)
Educational achievement	.004 (.06)
Employment status	−.04 (.06)
Learning disabilities	.03 (.06)
Family member of victim	−.07 (.06)
Victim of sexual abuse or contact (−18)	.03 (.05)
Victim characteristics^a
Age		.03 (.06)
Environment at time of offense		−.01 (.06)
Physical damages caused during offense		.13 (.11)
Offense characteristics
Victim a stranger to offender			−.08 (.14)
Weapon used			.26 (.21)
Victim forced into one or more sex. act			.05 (.06)
Offender remorse
Admits to charges				.04 (.08)
Admits to damages				.06 (.06)
Admits responsibility				.08 (.07)
Admits sexual issues				−.10 (.06)*
Criminal career (prior charges)
Prior nonsexual violent					.08 (.06)
Prior nonsexual/nonviolent					−.06 (.06)
Prior sexual (contact and noncontact)					.08 (.05)
Additional charges at current offense
Nonsexual violent						−.10 (.05)
Nonsexual/nonviolent						−.23 (.06)
Likelihood ratio χ²(df), p value	8.08 (7)	5.05 (4)	5.70 (3)	6.62 (4)	4.39 (3)	17.09 (2)****
Deviance (value/df)	1.00	1.00	1.00	1.00	1.00	0.999
Intercept	3.57 (.13)****	3.81 (.08)****	3.71 (.05)****	3.75 (.04)****	3.74 (.04)****	3.80 (.03)****

Note: n = 297.

Gender was removed from the model as all of the victims below the age of 15 were females.

p < .10. **p < .05. ***p < .01. ****p < .001.

In the final hierarchical model, offender age at admission (b = .01, SE = .002, p < .001) was predictive of sentence length in the first block, and with the addition of the victim characteristic covariates, only physical damages caused to the victim was marginally predictive of sentence length (b = .19, SE = .11, p < .01; Table 7). Offender age at prison admission was the only significant predictor in the third block of the model (b = .01, SE = .002, p < .001). With the addition of the offender remorse covariates, offender age at admission remained significant (b = .01, SE = .002, p < .001), physical damages caused to the victim emerged as nearly significant (b = .18, SE = .10, p < .10), and admitting to having sexual issues also increased sentence severity (b = −.14, SE = .06, p < .05). Each of the three predictors remained significant with the addition of the criminal career parameters, none of which increased sentence severity. With the addition of the last block of covariates, offender age at admission (b = .004, SE = .002, p < .10) remained close to significant, whereas the effect of causing physical damages and admitting to having sexual issues was lost. Finally, having prior nonsexual violent charges (b = .11, SE = .06, p < .05) and having additional nonsexual/nonviolent current charges (b = −.21, SE = .05, p < .001) also increased sentence severity.

Table 7.

Hierarchical Negative Binomial Regression With Offenders Who Targeted Victims Age 13 or Younger.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)	b (SE)
Offender characteristics
Age of offender	.01 (.002)**	.01 (.002)***	.01 (.002)****	.01 (.002)**	.01 (.003)**	.004 (.002)*
Civil status	−.01 (.05)	.02 (.06)	.02 (.05)	.01 (.05)	.001 (.05)	.003 (.–05)
Educational achievement	.004 (.06)	.02 (.06)	.02 (.06)	.02 (.06)	.0016 (.06)	−.02 (.05)
Employment status	−.04 (.06)	−.03 (.05)	−.01 (.05)	−.02 (.05)	−.03 (.05)	−.02 (.05)
Learning disabilities	.03 (.06)	.03 (.06)	.02 (.06)	.02 (.06)	.01 (.06)	−.01 (.06)
Family member of victim	−.07 (.06)	−.07 (.06)	−.10 (.06)	−.07 (.06)	−.05 (.07)	−.04 (.07)
Victim of sexual abuse or contact (−18)	.03 (.05)	.004 (.05)	.01 (.05)	−.007 (.05)	−.02 (.05)	−.01 (.05)
Victim characteristics^a
Age		−.003 (.08)	−.004 (.008)	−.001 (.008)	.00 (.008)	.005 (.008)
Environment at time of offense		.01 (.06)	.01 (.05)	.006 (.05)	.01 (.05)	.004 (.05)
Physical damages caused during offense		.19 (.11)*	.15 (.10)	.18 (.10)*	.18 (.10)*	.17 (.10)
Offense characteristics
Victim a stranger to offender			−.19 (.15)	−.20 (.15)	−.23 (.14)	−.19 (.14)
Weapon used			.17 (.24)	.15 (.24)	.13 (.20)	.11 (.22)
Victim forced into one or more sex. Act			.04 (.06)	.05 (.05)	.07 (.05)	.08 (.05)
Offender remorse
Admits to charges				−.03 (.07)	−.03 (.07)	−.02 (.07)
Admits to damages				.05 (.06)	.06 (.06)	.05 (.06)
Admits responsibility				.10 (.07)	.09 (.07)	.08 (.07)
Admits sexual issues				−.14 (.06)**	−.11 (.06)*	−.08 (.06)
Criminal career (prior charges)
Prior nonsexual violent					.10 (.06)	.11 (.06)**
Prior nonsexual/nonviolent					−.06 (.06)	−.02 (.06)
Prior sexual (contact and noncontact)					.11 (.07)	.10 (.07)
Additional charges at current offense
Nonsexual violent						−.03 (.05)
Nonsexual/nonviolent						−.21 (.05)****
Likelihood ratio χ²(df), p value	8.08 (7)	521.17 (11)	20.74 (13)*	27.06 (17)*	33.92 (20)**	44.71 (23)***
Deviance (value/df)	1.00	0.998	0.997	1.00	0.999	1.00
Intercept	3.57 (.13)****	3.45 (.16)****	3.50 (.16)****	3.49 (.17)****	3.43 (.17)****	3.45 (.16)****

Note: n = 297.

Gender was removed from the model as all of the victims below the age of 15 were female.

p < .10. **p < .05. ***p < .01. ****p < .001.

Discussion

The sentencing of an offender is the culmination of a multitude of decisions made by the criminal justice system (Simon, 1996). Although much attention was given to the topic of sentencing in United States in the late 1970s and 1980s, sentencing, as an area of criminological inquiry, has mainly focused on race and gender sentencing disparities (Crawford, 2000; Curry et al., 2004; Mitchell, 2005; Steffensmeier et al., 1998). Although factors that contribute to sentence severity have been explored for general offending populations (e.g., Daly & Tonry, 1997; Engen & Gainey, 2000; Johnson et al., 2010; Ulmer & Johnson, 2004; Wooldredge, 2010) very little is known about the impact that aggravating and mitigating factors have on the sentence severity of sex offenders.

Of the limited studies that have empirically explored factors that contribute to the sentencing severity of sex offenders, the samples utilized tend to focus either on offenders who have exclusively targeted adults, or those who have exclusively targeted children, but not both (Champion, 1988; Levesque, 2000; Kingsnorth et al., 1999; Walsh, 1984, 1994). Two of the most rigorous of these studies are those by Levesque (2000) and Kingsnorth et al. (1999). While Levesque (2000) utilized a sample comprised of offenders who victimized children only, conversely Kingsnorth et al. (1999) utilized a sample comprised of offenders who victimized adults only. The current sample was comprised of 519 adult sex offenders who had been convicted of a federal sexual offense in the province of Quebec, Canada, between 1994 and 2000, and included both offenders who had victimized children and adults. Offenders incarcerated for federal terms are those who have committed more serious offenses and often those who have more extensive criminal backgrounds. As previously noted, the sample was characterized by a large proportion of heterosexual child molesters, which in part accounts for the mean age of the victims as well as for the fact that 81% (n = 419) of the victims were female. Despite this finding, the sample participants studied committed a varied range of sexual offenses, including sexual assault, sexual assault with a weapon, indecent assault, sexual contact offenses, and incest offenses.

As the offenders were sentenced to federal terms, the minimum sentence length for this offending population was 24 months or 2 years. The average sentence lengths reported in the current study for offenders who victimized adults are in line with those reported by Walsh (1984) but less than those reported by Kingsnorth et al. (1999). For offenders who victimized children the reported sentence length here is similar to the sentence length noted by Levesque (2000) but greater than those reported by Champion (1988). The disparity between the sentence lengths of those who victimized adults and children was smaller than expected but still worth noting. In the current study, as well as those cited here, offenders who victimized children were incarcerated for shorter periods of time. While the effect of aggravating and mitigating factors on sentence severity will be discussed in detail, this distinction raises questions as to how the criminal justice system interprets harm caused by sexual offenders who victimize adults and children.

Negative binomial regression models were employed to test the impact of each of aggravating and mitigating factors on sex offender sentence severity (i.e., offender characteristics, victim characteristics, offense characteristics, offender remorse, criminal career parameters, and additional charges). When considering the full sample and the offending group who victimized adults only, the only offender characteristics to predict sentence severity was victimizing a family member—those who did so received shorter sentences. These results are in line with those reported by Simon (1996) and Kingsnorth et al. (1999) whereby offenders who target victims who are known to them receive shorter sentences. However, the absence of any effect of this mitigating factor in the model evaluating offenders who victimized those below the age of 13 is contrary to the case law presented by Ruby et al. (2008) where offenders who victimized young family members were sentenced less severely. One possible explanation for this finding may be the absence of a covariate indicating whether or not the offender was in a position of authority over the victim—this may have had a mediating effect. Including whether or not the offender was in a position of authority would have been ideal; however, the low occurrence of this prevented its inclusion in multivariate analyses. Regardless, the differences reported between the models testing adult-only and children-only offending groups suggest that offenders who target adult familial victims are being treated differently by the criminal justice system.

When considering the victim characteristics for the full sample, both victim age (in line with the findings reported by Kingsnorth et al., 1999) and the physical damages caused during the offense were significantly related to increased sentence severity; however, in models testing the adult-only and children-only offending groups, the effect of age disappears. Physical injury is only nearly predictive of sentence length for offenders who target adult-only victims and the effect disappears entirely in the model evaluating offenders who targeted children only. The impact of physical damage was expected given its identification as an aggravating factor (Ruby et al., 2008).

Victimizing a stranger was predictive of increased sentence severity for both the full sample and for those who victimized adults only. The same effects were found for utilizing a weapon during the commission of the offense for the full sample and in the adult-only sample. Again, the findings regarding victimizing a stranger and utilizing a weapon reported here are in line with those reported previously (Kingsnorth et al., 1999; Simon, 1996; Walsh, 1984, 1994). The effect of these factors on increased sentence severity indicates the importance of offense characteristics in sentencing decisions made for offenders who victimize adults.

When considering the four facets of offender remorse there were differences between the three groups tested. In the full sample, and adult-only offending group, admitting to causing damages to the victim reduced sentence severity. However, in the model testing offenders who victimized those below the age of 13, admitting to having sexual issues nearly reduced sentence severity. Offender remorse is a mitigating factor identified by Ruby et al. (2008) as having the potential to reduce sentence length. Again, the differences found here between the two groups of offenders are of note. In the previous models tested for the adult-only offending group, inflicting physical damages during the commission of the offense was indicative of increased sentence severity whereas here admitting to these damages decreased sentence severity. Furthermore, the only factor indicative of sentence severity for the offending group who victimized children only was admitting to sexual issues. This evidence further supports the premise that perhaps the criminal justice system places more emphasis on offense characteristics for those who victimize adults and more emphasis on offender characteristics for those who target children.

The results of the criminal career parameters were somewhat surprising given the emphasis that has been placed on the recidivist sentencing premium (Frase, 2010; Levesque, 2000; Roberts, 1997, 1996, 1994; Simon, 1996; Spohn & Welch, 1987; Welch et al., 1984). In the full-sample model, the only criminal career parameter to increase sentence severity was having at least one prior nonsexual violent charge, whereas prior sexual charges surprisingly did not increase sentence severity. A similar pattern was found for those who victimized adults only, whereas no effect was found in the model of offenders who victimized those below the age of 13. Although the use of a recidivist sentencing premium is commonplace, its use has been criticized by scholars who contend that increasing sentence severity for offenders based on their previous criminal behaviors is equivalent to convicting, and subsequently sentencing, offenders of the same offense twice, or the legal principle of double jeopardy (Frase, 2010; Roberts, 1997, 2008). Finally, having additional current charges did not increase sentence severity across any models.

The hierarchical models offer somewhat of a different picture. When considering the full-sample models, especially with respect to the offender and victim characteristics, moderating effects are evident. The final block of the full-sample model illustrates that of the 23 mitigating and aggravating factors utilized only offender age, victimizing a family member, victimizing a stranger, and using a weapon during the commission of the offense were predictive of sentence severity. None of the victim characteristics, offender remorse, criminal career parameters, or other current charges increased sentence severity. Similar results are found in the hierarchical model tested for the adult-only group of offenders. As reported in the full-sample model, in the final block tested for the adult-only offenders, offender age at prison admission, victimizing a stranger, and using a weapon during the offense increased sentence severity. In addition, being less educated and admitting to causing damages to the victim decreased sentence severity and having additional nonsexual violent charges in addition to the current sexual charge increased sentence severity.

Finally, in the hierarchical model with the children-only group of offenders, offender age at prison admission, having prior nonsexual violent charges, and having no additional current nonsexual/nonviolent charges in addition to the current offense affected sentence severity. The results of the hierarchical models again indicate that offense characteristics may have a larger effect on the sentencing severity of offenders who victimize adults than those who target children. Interestingly, the only criminal career parameter that significantly increased sentence length in the hierarchical models was having prior nonsexual violent charges in the group of offenders who victimized those below the age of 13. Also of interest was the impact that offender age of admission had on increased sentence length. If a recidivist sentencing premium effect did exist here, one could argue that as offenders age they are more likely to have more extensive criminal careers, and this relationship could account for the role of offender age at prison admission on sentence severity. However, this is not the case in the current study and the effect of offender age at prison admission on sentence severity remains unclear.

Perhaps one explanation for the differences in the effects that the aggravating and mitigating factors tested here had on sentence severity can be attributed to judicial discretion. Clear differences exist in the factors that increased sentence severity for the group of offenders who targeted adults and the group of offenders who targeted children. Offense characteristics seemed to play a greater role in the sentence severity of offenders who targeted adults. This raises questions about the criminal justice system’s perceptions of the harm inflicted by sexual offenses. Although child molestation or incest cases may not result in the physical injuries that can occur in sexual assault cases, the lasting negative effects of being sexually victimized as a child are documented (e.g., Beitchman et al., 1992; Browne & Finkelhor, 1986; Oddone Paolucci, Genuis, & Violato, 2001). Currently, the criminal justice system appears to be more concerned with damages that are caused during the commission of the offense and not the long-term damages that result from being the victim of sexual abuse in childhood.

Implications and Future Directions

The results presented here suggest that there may be differences in the aggravating and mitigating circumstances that affect the sentence severity of sex offenders. One of the most puzzling findings reported here is the absence of any affect of a sentencing premium. As the offenders utilized in the study were on average 40 years of age at the time of their prison admittance there was sufficient time in their criminal careers to accumulate prior charges. Given this, and the nature of the offenses under examination here, it was expected that prior criminal involvement would have been one of the strongest predictors of sentence severity; however, this was not the case. Future studies should further explore the extent to which prior criminal activities increase sentencing severity for both general and sex offending populations, and specifically, for sex offenders who target adult and child victims.

Furthermore, differences in how the criminal justice system responds to sexual offenses committed against adults and those committed against children are evident. The findings here indicate that for offenders who targeted adults (rapists) the seriousness of the crime rather than offender or victim characteristics was most indicative of sentence severity. If the assertion that “justice is blind” is true we would expect this to the be reality for offenders who targeted children (child molesters) as well; however, the findings here indicate that the criminal justice system responds differently to this offending group. The reasons for this differential treatment remains unclear. Perhaps it is the result of judge’s more stringently following sentencing guidelines in child molestation cases, or perhaps it is because they base their decisions on factors that were not included in this study. Regardless of these uncertainties, the ways in which the criminal justice system responds to various sexual offenses requires further clarification. Of particular importance in the sentencing of offenders who victimize children is the role of the victim–offender relationship, and in particular, whether or not the offender was in a position of authority over the victim. It would be interesting to see whether this factor had a meaningful effect on sentence severity and whether this effect was stable for offenders who victimized boys and girls.

Limitations

Although the criminal career parameters tested here did not contribute to an increased sentence length, questions about what constitutes the best measure for prior criminal activities have been raised (Welch et al., 1984). Here, the only measure available was prior charges; however, prior periods of incarceration followed by prior convictions have been empirically shown to better predict sentence severity (Welch et al., 1984). Furthermore, the extent to which plea-bargaining processes may have reduced sentence severity was not available in the data used for this study. The distance between prior charges and the current sexual charge for which the offender was included in the sample was not controlled for. This might have influenced the impact that these factors had on sentence severity. The only measure of sentence severity available was the amount of time the offender was sentenced to spend incarcerated; however, the actual amount of time that the offenders spent in prison was unknown. The exclusion of variables accounting for offenders in positions of authority over their victims as well as negative victim characteristics may have mediated some of the results presented in the statistical models. The differences in the results presented in the independent-block models and hierarchical models may have been a function of statistical power especially when modeling the between group differences.

A further limitation that should be noted is the failure to include female sex offenders. Although they represent only a fraction of the overall sex offender population, there may be important differences in the extent to which aggravating and mitigating circumstance may influence sentence severity between male and female offenders that are not captured by the current study. A final limitation is related to our relying on the length of prison sentence as the sole measure of sentence severity. Other possible expressions of sentence severity could have included offenders receiving sanctions in addition to a prison sentence (e.g., split-sentence, additional fines) or examining differences between offenders who received prison sentences as opposed to less severe sanctions such as probation.

Conclusion

The results of this study indicate that aggravating and mitigating circumstances differentially affect the sentencing severity of sex offenders who target adults and those who target children. Differences were found in the extent to which aggravating and mitigating factors influenced sentence severity for the offending group who targeted adults and the offending group who targeted children. The results indicate that the criminal justice system may place greater emphasis on offense-based characteristics for offenders who target adults and offender-based characteristics for those who target children. However, the results of this study are largely exploratory and require further investigation.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received the following financial support for the research, authorship, and/or publication of this article: Funding for this study was provided by the Social Sciences and Humanities Research Council of Canada.

Notes

Author Biographies

Joanna Amirault is a PhD candidate in the School of Criminology at Simon Fraser University. She is a recipient of the Joseph-Armand Bombardier Canada Graduate Scholarship administered by the Social Sciences and Humanities Research Council of Canada. Her research interests lie in the criminal careers of sex offenders, desistance and criminal justice responses to sexual crimes. She has co-authored articles on offending trajectories and the predictability of prior criminal charges on future offending outcomes.

Eric Beauregard is an Associate Professor in the School of Criminology at Simon Fraser University and he is the Director of the Centre for Research on Sexual Violence. His research interests focus on the criminal event, decision making, and the offending process of criminals. Dr. Beauregard’s work has been published in Criminal Justice and Behavior, Journal of Research in Crime and Delinquency, Sexual Abuse: A Journal of Research and Treatment, Violence and Victims, and the Journal of Criminal Justice.

References

Beitchman

J. H.

Zucker

K. J.

Hood

J. E.

DaCosta

G. A.

Akman

Cassaiva

(1992). A review of the long-term effects of child sexual abuse. Child Abuse & Neglect, 16, 101-118.

Bennett

(2010). “More to apologize for”: Can we find a basis for the recidivist premium in a communicative theory of punishment. In Roberts

J. V.

von Hirsch

(Eds.), Previous convictions at sentencing: Theoretical and applied perspectives (pp. 73-90). Portland, OR: Hart Publishing.

Bill C-2. Tackling Violent Crime Act (2008). Assented to 28th February 2008, 39th Parliament, 2nd Session. Retrieved from http://lawslois.justice.gc.ca/eng/AnnualStatutes/2008_6/page-1.html

Booth

J. G.

Casella

Friedl

Hobert

J. P.

(2003). Negative binomial loglinear mixed models. Statistical Modelling, 3, 179-191.

Browne

Finkelhor

(1986). Impact of child sexual abuse: A review of the research. Psychological Bulletin, 99(1), 66-77.

Champion

(1988). Child sexual abusers and sentencing severity. Federal Probation, 53, 53-57.

Crawford

(2000). Gender, race and habitual offender sentencing in Florida. Criminology, 38(1), 263-280.

Criminal Code, R.S. c. C-46. (1985). Retrieved from http://laws.justice.gc.ca/en/C-46/index.html.

Curry

T. R.

Lee

Rodriguez

S. F.

(2004). Does victim gender increase sentence severity? Explorations of gender dynamics and sentencing outcomes. Crime & Delinquency, 50, 319-343.

10.

Daly

Tonry

(1997). Gender, race, and sentencing. Crime and Justice, 22, 201-252.

11.

Engen

R. L.

Gainey

R. R.

(2000). Modeling the effects of legally relevant and extralegal factors under sentencing guidelines: The rules have changed. Criminology, 38(4), 1207-1230.

12.

Frase

R. S.

(2010). Prior-conviction sentencing enhancements: Rationales and limits based on retributive and utilitarian proportionality principles and social equality goals. In Roberts

J. V.

von Hirsch

(Eds.), Previous convictions at sentencing: Theoretical and applied perspectives (pp. 117-136). Portland, OR: Hart Publishing.

13.

Johnson

B. D.

Van Wingerden

Nieuwbeerta

(2010). Sentencing homicide offenders in the Netherlands: Offender, victim and situational influences in criminal punishment. Criminology, 48(4), 981-1018.

14.

Kingsnorth

MacIntosh

R. C.

Wentworth

(1999). Sexual assault: The role of prior relationship and victim characteristics in case processing. Justice Quarterly, 16, 275-302.

15.

Land

K. C.

McCall

P. L.

Nagin

D. S.

(1996). A comparison of Poisson, negative binomial and semiparametric mixed Poisson regression models: With empirical applications to criminal careers data. Sociological Methods & Research, 24, 387-442.

16.

Levesque

R. J. R.

(2000). Sentencing sex crimes against children: An empirical and policy analysis. Behavioural Sciences and the Law, 18, 331-341.

17.

Maxfield

L. D.

(2002). Prior dangerous criminal behavior and sentencing under the federal sentencing guidelines. Iowa Law Review, 87, 669-696.

18.

Mitchell

(2005). A meta-analysis of race and sentencing research: Explaining the inconsistencies. Journal of Quantitative Criminology, 21(4), 439-466.

19.

Oddone Paolucci

Genuis

M. L.

Violato

(2001). A meta-analysis of the published research on the effects of childhood sexual abuse.The Journal of Psychology, 135(1), 17-36.

20.

Patrick

Marsh

(2011). Sentencing outcomes of convicted child sex offenders. Journal of Child Sexual Abuse, 20, 94-108.

21.

Proulx

St.-Yves

McKibben

(1994). CQSA: Computerized Questionnaire for Sexual Aggressors. Unpublished manuscript.

22.

Roberts

J. V.

(1994). The role of criminal record un the federal sentencing guidelines. Criminal Justice Ethics, 13, 21-30.

23.

Roberts

J. V.

(1996). Public opinion, public record, and the sentencing process. American Behavioral Scientist, 39(4), 488-499.

24.

Roberts

J. V.

(1997). The role of criminal record in the sentencing process. Crime and Justice, 22, 303-362.

25.

Roberts

J. V.

(2008). Punishing persistent offenders: Exploring community and offender perspectives. Oxford, UK: Oxford University Press.

26.

Ruby

C. C.

Davies

Doucette

Loosemore

Orkin

J. R.

Wawzonek

(2008). Sentencing (7th ed.). Markham, ON: LexisNexis.

27.

Simon

L. M. J.

(1996). Legal treatment of the victim-offender relationship in crimes of violence. Journal of Interpersonal Violence, 11, 94-106.

28.

Spohn

Spears

(1996). The effect of offender and victim characteristics in sexual assault case processing decisions. Justice Quarterly, 13, 649-679.

29.

Spohn

Welch

(1987). The effect of prior record in sentencing research: An examination of the assumption that any measure is adequate. Justice Quarterly, 4(2), 287-302.

30.

Steffensmeier

Ulmer

Kramer

(1998). The interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, black and male. Criminology, 36, 763-798.

31.

Tabachnick

B. G.

Fidell

L. S.

(2007). Using multivariate statistics (5th ed.). Boston, MA: Pearson.

32.

Ulmer

J. T.

Johnson

(2004). Sentencing in context: A Multilevel analysis. Criminology, 42, 137-177.

33.

von Hirsch

(2010). Proportionality and the progressive loss of mitigation: Some further reflections. In Roberts

J. V.

von Hirsch

(Eds.), Previous convictions at sentencing: Theoretical and applied perspectives (pp. 1-16). Portland, OR: Hart Publishing.

34.

Walsh

(1984). Differential sentencing patterns among felony sex offenders and non-sex offenders. The Journal of Criminal Law and Criminology, 75, 443-458.

35.

Walsh

(1994). Homosexual and heterosexual child molestation: Case characteristics and sentencing differentials. International Journal of Offender Therapy and Comparative Criminology, 38(4), 339-353.

36.

Welch

Gruhl

Spohn

(1984). Sentencing: The influence of alternative measures of prior record. Criminology, 22(2), 215-227.

37.

Wooldredge

(2010). Judge’s unequal contributions to extralegal disparities in imprisonment. Criminology, 48(2), 539-567.