Understanding the relationship between self-reported offending and official criminal charges across early adulthood

Abstract

Background

There has been very little research examining criminal careers in adulthood using both self-report data and official records.

Aims

The aims of this paper are to use self-reports and official criminal records to explore (1) the prevalences and frequencies of offending behaviour in adulthood; (2) continuity in offending behaviour across the life course; and (3) predictors of official court charges in adulthood.

Method

Data are drawn from the Seattle Social Development Project (SSDP), a longitudinal study of 808 participants followed from childhood into early adulthood. Data from ages 21 through 33 are used to examine criminal careers.

Results

Prevalences of offending behaviour decreased with age, while frequency among offenders remained stable or increased. There was significant continuity in offending from adolescence to adulthood in both self-reports and official records, especially for violence. Violent offences were most likely to result in a court charge. Even after controlling for self-reported frequency of offending, demographic variables (gender, ethnicity, and poverty) were significantly related to a court charge.

Conclusions

Self-report and official records, both separately and together, provide valuable information for understanding criminal careers in adulthood, especially with regard to offending continuity across the life course and predicting the likelihood of a court charge.

Introduction

Criminal career research, defined as examining within-individual changes in criminal behaviour across the life course (Piquero et al., 2007), has become one of the most well-studied and visible areas of criminology (DeLisi and Piquero, 2011; Farrington, 1986; Steffensmeier et al., 1989; Tittle and Grasmick, 1997). However, there are very few longitudinal studies that allow for a thorough examination of criminal careers over time and across developmental periods (Sampson and Laub, 2003). Furthermore, most studies rely only on official records to study criminal careers (Benson, 2013; Farrington, 1997). The current study is especially valuable as it uses longitudinal data (drawn from a community sample from the Seattle Social Development Project (SSDP)), and uses both self-reported offending and official court records from across adolescence and early adulthood to study criminal careers in adulthood. Farrington et al. (2003) used SSDP data to compare adolescent criminal careers in self-reports and court records. The current study serves as a follow-up to that paper, addressing some of the same questions, but focusing on early adulthood.

We address three research questions in this paper. First, what are the prevalences and frequencies of self-reported offending behaviour compared to official charges across early adulthood (through age 33)? When studying prevalence of offending, criminologists have long established that the “age-crime curve” generally peaks in adolescence and declines quickly into adulthood (Blumstein and Cohen, 1987; Moffitt, 1997). We examine whether this remains true in our sample in both self-reports and official records. There is a debate in criminology about whether frequency of offending also follows this same pattern (Piquero et al., 2007; Sampson and Laub, 2003). Moffitt (1993, 1997), has argued that there is a small group of “life-course persistent” offenders who continue to offend at high rates into adulthood, while Gottfredson and Hirschi (1990) assert that levels of offending decrease for all offenders as they grow older. Using official arrest data, both Piquero et al. (2007) and Sampson and Laub (2003) found that frequency of offending decreased as the adult sample aged, generally supporting Gottfredson and Hirschi’s model. We also address this empirical question by examining the frequency of offending in both self-report and official records across early adulthood.

Second, how much continuity is there in self-reported offending behaviour and court charges across the life course? Researchers have found substantial continuity in offending across age ranges, including continuity in general offending in official records from adolescence to early adulthood (Farrington, 1992; Piquero et al., 2007), and continuity in specified offences in both self-reports and official records (Blokland and Nieuwbeerta, 2010; Farrington, 1990). Considering how few longitudinal studies are equipped to answer these questions, in the current study we contribute to the sparse body of literature by testing whether specified offending in adolescence predicts similar offending in adulthood, in both self-reports and official records.

Finally, who is most likely to be charged for criminal behaviour and for which types of crimes? That is, what factors are related to concordance or discordance between self-reported offending and official charges in adulthood, including type of offence, ethnicity, gender, and poverty? Using the same sample as the current study, Farrington et al. (2003) found that property crimes were most likely to lead to a court referral in adolescence. We examine whether this remains true in adulthood. Furthermore, we examine the relationship between demographic factors and court charges, after controlling for self-reported offending, to determine the amount of disproportional representation by gender, ethnicity, and poverty.

Methods

Sample and procedures

This study uses longitudinal data from the Seattle Social Development Project (SSDP). SSDP consists of a multi-ethnic sample of males and females followed prospectively from 1985, when participants were in the fifth grade (mean age = 10.3 years), into adulthood. Participants were recruited in the fall of 1985 from all fifth-grade students attending 18 Seattle elementary schools serving high-crime neighbourhoods (N = 1,053), 808 of whom (77%) consented to participate in the study. Due to mandatory school bussing at the time, the sample also included youths from higher income neighbourhoods. Of the 808 students, 396 (49%) were female, 381 (47.2%) were European American, 207 (25.6%) were African American, 177 (21.9%) were Asian American and 43 (5.3%) were Native American. About 5% of all students self-identified as Hispanic. More than half of the student sample (52%) had participated in the National School Lunch/School Breakfast Program in the fifth, sixth, or seventh grade.

Data were obtained from self-report surveys and official court records. Survey data were collected annually through adolescence (with the exception of one year) and every three years in adulthood. Juvenile and adult court records were obtained for each year through age 33. Only records of official court charges that occurred within Washington State were available. Thus, for the present paper, respondents who lived outside of Washington State were excluded from analyses during the year(s) in which they did not live in the state. Due to mobility of participants, the analysis sample size varied at each data collection time point, ranging from 678 in 1996 (age 21) to 573 in 2008 (age 33). All data collection procedures have been approved by the University of Washington Human Subjects Review Board.

Measures

In order to accurately compare prevalences and frequencies of offending behaviour, only offences that were measured in both self-reported surveys and in official records were used. A total of eight offences were available from both sources. Due to low prevalences in adulthood of specific offences, offences were grouped into property, violent, and drug-related categories.

Self-reported property offences included property destruction (“How many times in the past year have you purposely damaged or destroyed property or things that did not belong to you?”), burglary (“how many times in the past year have you broken into a house, store, school or other building without the owner’s permission?”), theft (past year frequency of theft of items worth under $50, theft of items worth over $50, using illegal checks or fraudulent money, cheating someone by selling worthless goods, and credit card fraud) and auto theft (“how many times in the past year have you taken a motor vehicle, such as a car or motorcycle, for a ride without the owner’s permission?”). Violent offences included robbery (“how many times in the past year have you used a weapon or force to get money or things from people?”) and assault (“how many times in the past year have you hit someone with the idea of seriously hurting them?”). Finally, drug-related offences including drug selling (“how many times in the past year have you sold illegal drugs such as marijuana, cocaine, LSD, or heroin?”) and drug possession (past year use of illegal drugs, including marijuana, crack cocaine, powder cocaine, amphetamines, sedatives, psychedelics, and narcotics.). Frequencies and prevalences for all self-reported behaviours were only reported for the past year at each survey wave in adulthood (age 21, 24, 27, 30, and 33). Thus, these analyses represent a series of “snapshots” of self-reported offending behaviour and criminal charges at three-year intervals across adulthood.

To parallel self-report measures, past-year frequency of court charges were computed for the eight offences mentioned above, and also categorized into either property, violent, or drug-related offences. Finally, demographic variables, including gender, ethnicity, and welfare receipt (used as a measure of poverty) were self-reported.

Results

Prevalence and frequency of offending

Table 1 provides the past year prevalences and frequencies (mean number of offences per offender) of both self-reported offending behaviour and official court charges in the SSDP sample from ages 21 to 33. Overall, the prevalence of self-reported offending for any crime decreases over time from 52.8% at age 21 to 30.9% at age 33. This pattern is consistent for violent and drug-related crimes. For self-reported property crime, the prevalence decreases from 13.2% at age 21 to 7.0% at age 33, though the prevalences fluctuate across time. Changes in official court charges over time mirror those seen in self report, for the most part. The prevalence of a court charge for any crime decreases from 6.2% at age 21 to 3.3% at age 33, with the exception of one uptick at age 27.

Table 1.

Prevalence and Frequency of Offending Behaviour Ages 21–33

	Prevalence		Frequency
	Self-Report	Official	Self-Report		Official
			Mean	Median	Mean	Median
Age 21 (n=678)
Property	13.2%	2.4%	6.2	3	1.8	2
Violent	11.8%	3.2%	4.4	2	1.6	1
Drug	46.9%	2.1%	22.7	12.5	1.7	1
Any offence	52.8%	6.2%	22.7	11	2.1	2
Age 24 (n=639)
Property	7.8%	1.9%	8.8	3	2.4	2
Violent	8.8%	2.3%	3.7	1.5	1.5	1
Drug	37.3%	1.4%	24.6	16.5	1.2	1
Any offence	41.6%	4.4%	24.5	13.5	2.3	2
Age 27 (n=622)
Property	11.3%	2.7%	7.1	2	1.4	1
Violent	6.6%	3.7%	3.4	1	2.3	2
Drug	36.0%	2.1%	26.4	21	2.2	2
Any offence	42.0%	6.4%	25.1	12	2.6	2
Age 30 (n=603)
Property	5.8%	1.2%	7.1	2	1.3	1
Violent	4.3%	1.2%	2.6	1	1.3	1
Drug	32.3%	1.2%	25.7	20	1.3	1
Any offence	35.9%	3.3%	24.8	16	1.4	1
Age 33 (n=573)
Property	7.0%	1.4%	5.9	2	2.1	1.5
Violent	4.2%	1.9%	2.7	1	1.6	1
Drug	26.6%	0.9%	29.1	40	1.6	2
Any offence	30.9%	3.3%	26.8	20	2.3	2

Table 1 also presents the mean and median frequencies of past year self-reported offending behaviour and official charges per offender at each age. To account for possible measurement error and reduce the skewness of the self-reported data, frequencies were set to a maximum of 50 per year per offence. The median number of self-reported offences (any type) increases over time, from a low of 11 at age 21 to a high of 20 at age 33, though it appears that drug-related offences largely account for this increase. The frequencies of property and violent offences remain relatively stable. The mean number of official charges for any offence type fluctuates across time, reaching a high of 2.6 at age 27 and a low of 1.4 at age 30.

Continuity in the prevalence of offending from adolescence to adulthood (ages 21–33)

When comparing adolescent offending to adult offending, there is considerable continuity. Table 2 shows the odds of offending in adulthood (age 21–33) for adolescent offenders compared to non-offenders for each offence type in both self-report data and official charges. For example, Column 1 shows that 17.4% of individuals who never reported a property offence in adolescence (adolescent property non-offenders) went on to self-report at least one property offence in adulthood. These individuals can be thought of as adult onsetters. On the other hand, 32.1% of individuals who did report at least one property crime in adolescence (adolescent property offenders) are adult persistent offenders. That is, they also self-report property offences in adulthood. In self-reporting, property offenders are 2.2 times more likely to self-report offences in adulthood than adolescent non-offenders. For violent and drug offences, adolescent offenders are 7.5 and 4.1 times more likely, respectively, to offend in adulthood. In official charges, the odds of being charged in adulthood for adolescent offenders (official) versus non-offenders are 4.2 times higher for property offences, 12.3 times higher for violent offence, and 12.2 higher for drug offences.

Table 2.

Continuity in the Prevalence of Offending from Adolescence (Ages 10–18) to Adulthood (Ages 21–33)

Self-Report		Official
	% Adult Property Offenders		% Adult Property Offenders
Property		Property
Adolescent Property Non-offenders	17.4%	Adolescent Property Non-offenders	4.0%
Adolescent Property Offenders	32.1%	Adolescent Property Offenders	14.8%
(Odds Ratio)	2.2***	(Odds Ratio)	4.2***
	% Adult Violent Offenders		% Adult Violent Offenders
Violent		Violent
Adolescent Violent Non-offenders	6.3%	Adolescent Violent Non-offenders	4.4%
Adolescent Violent Offenders	33.5%	Adolescent Violent Offenders	35.9%
(Odds Ratio)	7.5***	(Odds Ratio)	12.3***
	% Adult Drug Offenders		% Adult Drug Offenders
Drug		Drug
Adolescent Drug Non-offenders	45.0%	Adolescent Drug Non-offenders	3.8%
Adolescent Drug Offenders	77.2%	Adolescent Drug Offenders	32.6%
(Odds Ratio)	4.1***	(Odds Ratio)	12.2***

^***p<.001.

Probability of a court charge given a self-report ages 21–33

Table 3 shows the percentage of self-reported offending individuals who were charged, as well as the percentage of offences leading to a charge, for each offence type at each age. For example, at age 21 18.4% of self-reporting property offenders were charged for a property offence, while only 5.4% of all property offences lead to a court charge. The percentage of self-reporting offenders (any offence type) who were charged fluctuates across time, showing no distinct increasing or decreasing patterns. It is important to note, however, that the overall low prevalence of court charges in the sample could make it difficult to detect reliable patterns. Column 2 of Table 3 shows similar results when examining the percentage of offences leading to a court charge.

Table 3.

Probability of a Court Charge Given a Self-Report Ages 21 – 33

	% Per Offender	% Per Offence
Age 21 (n=678)
Property	18.4%	5.4%
Violent	28.2%	10.6%
Drug	4.5%	0.3%
Any offence	12.0%	1.1%
Age 24 (n=639)
Property	25.0%	6.9%
Violent	27.8%	11.7%
Drug	3.9%	0.2%
Any offence	10.9%	1.0%
Age 27 (n=622)
Property	25.0%	5.0%
Violent	57.5%	39.6%
Drug	6.0%	0.5%
Any offence	15.8%	1.7%
Age 30 (n=603)
Property	20.0%	3.6%
Violent	26.9%	13.2%
Drug	3.8%	0.2%
Any offence	9.9%	0.5%
Age 33 (n=573)
Property	20.0%	7.3%
Violent	45.8%	28.1%
Drug	3.4%	0.2%
Any offence	11.1%	0.9%

Violent offenders have the highest probability of being charged (ranging from 26.9% to 57.5% over time), followed by property (ranging from 18.4% to 25.0%). Drug offenders are the least likely to be charged. Similarly, in any given year, less than 1.0% of drug-related offences result in a criminal charge. In comparison, at least 10%, and up to nearly 40%, of violent offences result in a criminal charge.

Differences in court charges in adulthood by gender, ethnicity, and poverty

Finally, we examined the extent to which gender, ethnicity, and poverty are associated with the probability of a court charge in adulthood, controlling for the self-reported frequency of each crime type. For example, in the model predicting the likelihood of a property offence charge, we controlled for the self-reported frequency of property offending. Analyses drew on data across all five adult time points (age 21 – 33), so that all individuals who ever lived at least one year in Washington State during that time were included in the analysis, resulting in a sample size of 733. If a respondent ever had a court charge he/she was coded as (1), otherwise (0). Self-reported frequencies of offending behaviours were averaged across the years in which the respondent was residing in Washington State.

Results from these analyses are shown in Table 4. For all types of offences, self-reported offending behaviour significantly predicts a court charge at the zero-order level (model 1), and continues to predict court charges after controlling for gender, ethnicity and poverty (model 2). However, even after controlling for self-reported offending behaviour, demographic variables significantly predict a court charge. Specifically, males are about four times more likely than females to have a court charge for a violence-related court charge (odds ratio=3.91) and a drug-related charge (odds ratio=4.09). There are also large, significant associations between ethnicity and official charges. African Americans are roughly three times more likely to be charged with a violent offence and nearly three times more likely to be charged with a drug-related offence compared to European Americans. Native Americans are 4.2 times more likely than European Americans to be charged with a violent offence. Significant ethnic differences are not evident with regard to property offences. Finally, those individuals receiving welfare, an indicator of poverty, are 3.1 and 2.3 times more likely than those not in poverty to be charged with violent or drug-related offence, respectively, after controlling for self-reported frequencies of these offences.

Table 4.

Predicting Court Charge by Demographics

	Property offence charge		Violent offence charge		Drug offence charge
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
	OR		OR		OR
Self-reported frequency of offence	1.15***	1.12**	1.33***	1.23**	1.05**	1.05**
Gender (male)		1.55		3.91***		4.09**
African American (vs. European American)		1.75		3.27***		2.77*
Native American (vs. European American)		1.09		4.20**		1.92
Asian American (vs. European American)		0.40		0.38		0.54
Poverty (welfare receipt)		3.11**		2.30**		2.1

^*p<.05;

^**p<.01;

^***p<.001

Conclusions and Discussion

With regard to the first research question, results for changes in prevalence were as expected. The overall prevalence of offending behaviour decreased as the sample aged in both self-report and official records. However, the frequencies of offending behaviour (mean number of offences per offender) remained steady for property and violent offences, and actually increased for drug-related offences. This finding is consistent with other studies that have found that drug-related offending tends to peak later than property or violent offending (Day et al., 2008; Sampson and Laub, 2003). The theoretical implications for these findings are also significant. While Blumstein et al. (1988) contended that it is more likely the decrease in prevalence, rather than frequency, that accounts for the age-crime curve, Hirschi and Gottfredson (1983) argued that prevalence and frequency should both decrease with age, as they represent the same construct. We found evidence to support the former assertion, that variations in prevalence and frequency have different paths across early adulthood. However, it should be noted that Sampson and Laub (2003), who followed participants into late adulthood (age 70), found that eventually prevalence and frequency decreased significantly for all offenders.

An examination of the continuity in offending across the life course revealed that an offence in adolescence (either self-reported behaviour or an official charge) significantly predicted offending in adulthood for all three types of crimes examined. Violent offending showed the highest continuity in both self-report and official charges, followed by drug and property offending. Other researchers have also found that violent offending shows significant continuity from adolescence to adulthood (Jennings et al., 2011). Thus, prevention efforts should specifically target those at risk for violent offending, so as to potentially interrupt a lifelong trajectory of violent behaviour. Overall, there was higher continuity in official charges than in self-reporting, though the difference was most pronounced for drug offences. This could represent a bias in the court system against known offenders, especially drug offenders. Interestingly, while there was significant continuity in drug offending from adolescence to adulthood, drug offenders also showed high rates of adult onsetting, which is consistent with the findings discussed above, regarding the relatively late onset of drug offending.

The third research question was answered with several interesting findings. First, violent offences were more likely than either property or drug-related offences to lead to a criminal charge in adulthood. This is especially interesting given that Farrington et al. (2003) found that in adolescence property offences were the most likely offence type to lead to a court charge in this sample. One consistent finding in both adolescence and adulthood for this sample was that drug-related offences were least likely to lead to a court charge.

Demographic factors, including gender, ethnicity, and poverty were also significantly related to the likelihood of a court charge even after controlling for self-reported offending behaviour. It is well documented that men of colour from impoverished neighbourhoods are overrepresented in the US criminal justice system in general (Brame et al., 2014; Pettit and Western, 2004; Snyder, 2011). Huizinga et al. (2013) demonstrated that this was particularly true in this SSDP sample during adolescence. The present study shows that this disproportionality continues to persist at present into adulthood. Future research should examine the relationships between race and other demographic factors and the depth and type of justice system involvement, as well as the extent to which other factors, not accounted for in this study (e.g., system bias), affect disproportionality.

This study had limitations that should be discussed. First, we only examined individuals who live and Washington State. These results should be compared with those from similar analyses in other samples. Second, our data allowed for “snapshots” of offending behaviour every three years in adulthood. Finally, researchers (Farrington et al., 2003; Sampson and Laub, 2003) have suggested that criminal careers should be examined into late adulthood to understand the full trajectory of onset, career length, and desistence. Despite these limitations, this study successfully answered important questions for criminal careers research with a very unique data set that includes self-report and official offending from adolescence to adulthood.

By extending analyses of criminal careers in both self-report and official records from adolescence into adulthood, several important insights are gained. First, we established that there is continuity in the likelihood of offending from adolescence to adulthood, particularly for violence and drug offences. Given that, we also see a good amount of discontinuity. For example 45% of those who did not self-report drug use in adolescence did report onset in adulthood, and 67.9% of self-reporting adolescent property offenders desisted in adulthood. Second, on both a per offender and per offence basis, the type of offence that most likely leads to a court charge changes from property in adolescence to violence in adulthood. Finally, the racial disproportionality in official charges observed in adolescence (Huizinga et al., 2013) persists into the adult justice system as well. Hsia and colleagues (2004) describe policy and procedural changes that were implemented in Washington State to reduce racial disparities in juvenile prosecutions since this sample was in adolescence. Their report noted that these efforts had significantly reduced racial disparities in the State juvenile justice system from 1990–1999. The present study demonstrates that similar policy and procedural changes may be warranted in the adult justice system as well. Overall, the results of this paper expand our understanding of criminal careers across the life course, and point to important policy implications. Future work will continue to examine these questions into late adulthood.