The Australian Self‐report Delinquency Scale: A revision

Abstract

Objectives

The Australian Self‐Reported Delinquency Scale has been a widely adopted measure of delinquency, yet requires updating to ensure appropriate content coverage and consistency with contemporary language. The aim of this research was to revise the Australian Self‐Reported Delinquency Scale and derive a measure of contemporary delinquent behaviour among Australian adolescents with satisfactory psychometric properties.

Method

In Study 1, we conducted focus groups with 16 adolescents and 20 professionals (youth workers, police officers, school teachers/counsellors) in Canberra to update items included in the measure. Contents of the scale were revised accordingly, yielding a 56‐item checklist of contemporary delinquent activities. In Study 2, 312 students (57.7% male, aged 13–17) from government and independent schools in Canberra completed the 56‐item Delinquency Checklist.

Results

Subsequent item analysis and exploratory and confirmatory factor analyses resulted in a 30‐item Australian Self‐Reported Delinquency Scale‐Revised with eight subscales (Driving/Vehicle, Theft, Cheat, Disturb, Fight, Drugs) including two new subscales (Alcohol and Media), with evidence of reliability and validity.

Conclusions

The revision of the Australian Self‐Reported Delinquency Scale‐Revised, complemented by a 56‐item Delinquency Checklist, may prove useful in educational, rehabilitation, and research settings, and aid in evaluating clinical interventions with increased specificity.

Key words:

• adolescence
• alcohol
• delinquency
• focus group
• scale development

Underage alcohol use and juvenile delinquent activity are two problem behaviours that are particularly prominent among Australia's youth. In Australia, juvenile offense rates triple that of the adult cohort (Australian Institute of Criminology, 2011), with approximately half of juvenile crimes committed under the influence of alcohol (Prichard & Payne, 2005). These statistics cause significant concern and justify the need for further research to understand how and why these behaviours occur. To better assist research and intervention efforts, the development of a valid instrument of contemporary delinquent activities, including alcohol use, is required for Australian adolescents.

Because delinquent activity is often concealed, many teenagers are never apprehended by the police. As a result, official police records likely underestimate adolescent involvement in crime (van Batenburg‐Eddes et al., 2012), and measures often require a self‐report format to capture the extent of delinquency (Kivivuori, 2011). However, cost‐effective, self‐report measures that examine a range of marginally deviant to seriously criminal acts among adolescents are limited, especially those modelled on Australian samples (Carroll, Durkin, Houghton, & Hattie, 1996). The majority of measures used to assess delinquent behaviours are based on overseas samples and many are used to assess diagnostic categories (e.g., Achenbach, 1992; Frick & Hare, 2002). This is problematic as these scales may reflect cultural or legal considerations that do not apply in Australia (Carroll et al., 1996), and reveal an absence of measures used to assess delinquent activities, representing a range of prevalence and severity, within clinical practice and the community.

Since its publication in 1993, the Australian Self‐reported Delinquency Scale (ASRDS; Mak, 1993), containing 34 delinquency items spanning nine areas of delinquent activities (Driving, Vehicle, Status, Theft, Cheat, Disturb, Fight, Harm, and Drugs), has proven utility in assessing adolescent delinquency with varying populations utilising a 12‐month time frame (e.g., Carroll et al., 2009; Leas & Mellor, 2000; Mak, 1996). Altogether, the ASRDS contains 40 items, with two validity items pertaining to police cautions and Children's Court appearances, and four lie items included to combat response bias such as social desirability. Despite evidence of validity and reliability upon construction, the ASRDS is now over two decades old and uses language and terms that appear irrelevant to today's youth. Furthermore, with the recent advancement of social media and online networking sites, cyberbullying is fast becoming a problem among youth (Hemphill et al., 2012) and may be a relevant form of delinquent behaviour not included in this measure.

For these reasons, we feel that it is necessary to update the ASRDS to ensure that its contents and wording are consistent with the contemporary youth culture and language. Moreover, the use of the ASRDS in the past two decades has suggested that for efficient research and screening purposes, there is the need for a concise scale with relatively prevalent delinquent activities. Furthermore, a shorter time frame (6 months) may allow for better recollection of delinquent involvement and may be more useful than the original 12‐month time frame if the updated instrument is to be used to assess any change in delinquent involvement over an academic year.

In addition, there is also the need for an expanded, more comprehensive checklist of contemporary adolescent delinquent activities, including some serious although possibly less prevalent acts, for purposes of clinical assessment, decisions around interventions, and treatment evaluation.

The present research

To address these identified needs for scale revision, we conducted a preliminary study involving focus groups with relevant stakeholders to revise and update descriptions of delinquent activities from the ASRDS. This would provide a checklist of an expanded range of contemporary delinquent activities that could be used to assist clinical assessment and treatment evaluation. In Study 2, we administered the checklist to a sample of adolescents in a cross‐sectional survey. Data analysis performed would provide (1) psychometric properties of the expanded clinical checklist and (2) the basis for deriving a shorter, psychometrically sound revision of the ASRDS and its subscales, a research instrument based on principles of prevalence, severity, and content coverage of contemporary activities.

Study 1: Revising and Adding Contemporary Items to the ASRDS

In order to capture descriptions of contemporary adolescent delinquent behaviours representing a range of severity and prevalence and to ensure content validity of items, we consulted focus groups comprising law enforcement officers, professionals who work with delinquent youth, and potential offenders themselves (aged under 18 years) who could provide insight into current terminology and contemporary delinquent activities for inclusion in the updated measure. Consultations with police officers also ensured that the subsequent measure would represent law‐breaking activities for juveniles in the jurisdiction where the research was located.

Method

Participants

Purposive sampling was used to recruit adolescents, school teachers/counsellors, police officers, and youth mental health workers in Canberra, for stakeholder consultations. Five focus groups were conducted with 36 participants comprising secondary school students (n = 6), ‘at‐risk’ adolescents (defined as students at risk of disengaging from school who have had previous mental or behavioural problems; n = 10), youth workers (n = 7), police officers (n = 6), and school teachers/counsellors (n = 7). For a summary of participants' demographic information, see Appendix A.

Design and procedure

Following ethical clearance by the appropriate boards, a registered psychologist facilitated the focus groups at the participants' school or work (each taking approximately 1-hr). Participants were informed that the study was voluntary and that completed informed consent forms prior to participating were required. The focus groups were audio recorded and transcribed.

Participants were provided with a definition of delinquency (illegal or risky behaviour for adolescents under the age of 18 years) before being read each item from the ASRDS. The facilitator then invited specific feedback regarding the relevance of each item in terms of language and terminology, descriptions of commonly occurring types of delinquent behaviours and substance use (adding prevalent types as indicated by focus groups), and applicability to contemporary adolescents.

Results

All groups highlighted the absence of more recent forms of delinquent activity, such as cyberbullying, online stalking, ‘sexting’ (the act of sending sexually explicit messages or photographs primarily by mobile phones or social networking sites), and online trolling (primary intent of harassing others and provoking emotional reactions online). Stakeholder groups also mentioned that substance use was limited and outdated in language and failed to mention over‐the‐counter medication. The adolescent group reported that they did not know what barbiturates were and reported that abusing prescription medication ‘happens a lot’. Items such as purchasing cigarettes, cruelty to animals, obtaining or forging fake identification, and illegally downloading music and movies were suggested by the stakeholder groups and endorsed by the adolescent groups.

In addition to data collected from adolescent and relevant stakeholder groups, published studies that had previously adapted the ASRDS to suit their particular sample were examined for relevant items. The following items suggested by Carroll et al. (1996) for Western Australian youths living in a metropolitan area were considered relevant for inclusion: Ignored a red light while driving a car, joyriding in a stolen car, stolen and driven a car, sold drugs, used hard drugs, taken part in a robbery, been involved in a hit‐and‐run accident, and been suspended or expelled from school. Subsequently, a checklist of 56 delinquency items was developed from consultations (see Appendix B). In addition to Mak's (1993) four lie items and two validity items (relating to police cautions and appearances in the children's court) that are embedded within the scale, an additional validity item (being suspended or expelled from school) was incorporated into the checklist as a result of consultations.

Study 2: Development of a Revised Delinquency Scale and Subscales

Based on the existing literature and stakeholders' input in Study 1, we constructed a 56‐item checklist with contemporary coverage of delinquent activities, for trial administration to a sample of adolescents aged under 18 years in Study 2. From this pool of items, we aimed to develop a revised version of the ASRDS that would be more concise than the checklist, would comprise subscales, and could be used for research and screening purposes. Items will be selected and subscales derived based on item and factor analyses.

Method

After obtaining ethics clearance by the relevant ethical boards, we approached government and independent high schools and colleges in Canberra for permission to recruit student participants. Owing to ethics protocol, students from government schools required opt‐in parental consent, whereas students from independent schools were allowed opt‐out parental consent. The sample comprised 318 students aged 13–17 years from eight schools (four government, four independent) in Canberra. Four cases were excluded for being 18 years of age and a further two were excluded because of potentially biased responding (failing to respond to at least two of the four lie items). Of the remaining 312 participants, 180 were male (57.7%). The overall mean age was 15.64 years (15.49 years for males and 15.84 years for females).

Participants were requested to indicate their engagement (Yes/No response) in the 56‐item delinquent activities checklist over the past 6 months. Participants responded anonymously to an online (122; 39.1%) or paper (190; 60.9%) survey. An independent samples t‐test revealed no differences in delinquency responses by survey type, t(308) = 1.46, p = .146. Participation was voluntary and participants were assured of the anonymity and confidentiality of individual responses. Given the recommended sample size of 300 participants for factor analytical studies (Norušis, 2005), the current sample size (312) suggested adequate statistical power.

Results

Psychometric analysis of 56‐item checklist

Appendix B presents the percentages of respondents reporting participation in each of the 56 delinquent activities, males and females separately, and item‐level chi‐square (χ²) tests for gender differences. The 56‐item format contains items representing a range of severity, from marginally deviant behaviour (e.g., ‘Not attended classes or wagged school’) to very serious forms of criminal behaviour that are unlikely to be committed by general populations and require targeted and specialized early intervention services (e.g., ‘Taken part in a robbery, using a weapon or force’). Furthermore, the 56‐item checklist comprises items pertaining to contemporary delinquent activities, including those involving new media. For example, while piracy was prevalent across both genders, only a small percentage of both genders ‘sexted’. Males were more likely than females to report accessing pornography (62.6% and 17.4%, respectively), threatening others via mobile phone (11.7% and 3.8%, respectively), and committing violent and destructive acts. Females, on the other hand, were more likely than males to report committing online identity theft (17.4% and 9.5%, respectively), getting tattoos or piercings without parental permission (15.2% and 5.6%, respectively), wagging school (47% and 33.3%, respectively), and illegally obtaining or abusing prescription/non‐prescription medication (15.2% and 6.1%, respectively).

Scores on the delinquency checklist were obtained by collating the number of types of self‐reported delinquent acts committed over the previous 6 months, with possible scores ranging from 0 to 56. In this study, the obtained range of scores was 0–56 for males and 0–47 for females. The mean number of delinquent acts committed was 9.61 by males and 7.99 by females.

Coefficient alpha was .95 for males and .94 for females (.95 overall), demonstrating a high level of internal consistency reliability for the checklist. Point‐biserial correlations between the 56‐item checklist and embedded validity items indicated that those with official police warnings (.52 for males, .48 for females), court appearances (.50 for males, .40 for females), or who had been suspended or expelled from school (.48 for males, .25 for females) had higher delinquency scores than students who did not, providing some evidence of construct validity.

Derivation of the revised delinquency scale and subscales

In order to determine which items should be included in the updated delinquency instrument primarily intended for research purposes, we conducted item analysis on the 56‐item checklist. To reduce the number of items and ensure that the scale represented prevalent delinquent activities, guidelines for inclusion in the research instrument were: (1) selected items were required to have initial item–total correlations of .20 or above, and (2) items required positive responses by more than 5% of overall participants. The 46 items found to meet these criteria were subject to exploratory factor analysis (EFA) using maximum likelihood factoring following oblimin rotation in PASW (version 18; SPSS Inc., Chicago, IL, USA) for further data reduction purposes. Eight factors with eigenvalues greater than unity and accounting for 65.08% of the total variance were extracted.

Derived from the findings of Study 1, and based on previous literature (Mak, 1993) and the initial EFA, we constructed theoretical subscales underlying the items in the updated scale to consider newer categories of items and grouping together intercorrelated items appearing to derive from the same domain of offences. Eight factors were proposed: Driving/Vehicle (8 items), Alcohol (3 items), Theft (4 items), Cheat (4 items), Disturb (5 items), Fight (4 items), Drugs (6 items), and Media (10 items; see Appendix B for proposed factors). Two items (run away from home and gotten piercings/tattoos) did not appear to cluster with the proposed factors and were removed from the subsequent analysis.

To test the suitability of the eight‐factor structure, we conducted a confirmatory factor analysis (CFA) on the remaining 44 items using maximum likelihood estimation and analysing the covariance matrix in AMOS (version 18; SPSS Inc.). The model was then respecified, removing items that had factor loadings below .5 or that weakened scale reliability. This resulted in 30 items across 8 factors: Driving/Vehicle (five items), Alcohol (three items), Theft (three items), Cheat (four items), Disturb (four items), Fight (three items), Drugs (five items), and Media (three items). Table 1 presents the CFA results following the recommendations by Floyd and Widaman (1995), including item loadings, various parameter estimates, and reliability coefficients. All items in the final model had factor loadings above .50. Internal consistency reliabilities for the factors ranged from .62 to .86.

Table 1. Factor loadings and additional estimates for the 30‐item Australian Self‐reported Delinquency Scale‐Revised (n = 312)

Items	Factor loading	Estimate	S.E.	C.R.	SMC
Factor 1: Driving/Vehicle (α = .80)
Engaged in any other illegal behaviour in a car	.77	2.02*	.20	10.09	.59
Raced with other vehicles	.75	1.77*	.18	9.95	.56
Stolen things/parts out of a car/motorbike/bike	.69	1.17*	.12	9.41	.47
Driven a car/bike when drunk	.59	1.00			.35
Taken and driven a car/bike	.57	.94*	.12	9.41	.47
Factor 2: Alcohol (α = .80)
Drunk alcohol in a public location (e.g., park, school)	.84	1.07*	.08	13.55	.71
Bought or illegally obtained alcohol	.74	1.00			.54
Drunk alcohol in an 18 + venue	.71	.74*	.06	11.69	.50
Factor 3: Theft (α = .70)
Stolen money of more than $US20 in one go	.70	.97*	.11	9.16	.49
Stolen money of less than $US20 (in one go)	.68	1.21*	.13	.901	.46
Shoplifted	.64	1.00			.40
Factor 4: Cheat (α = .65)
Got into a facility without paying the proper fee	.70	1.00			.49
Cheated or stolen food, drinks, or other goods	.54	.68*	.08	8.71	.29
Pulled a runner without paying the proper fee	.54	.52*	.06	8.67	.29
Not attended classes or wagged school	.53	.82*	.10	8.55	.28
Factor 5: Disturb (α = .78)
Purposely damaged property in public places	.75	.95*	.08	11.65	.56
Purposely messed up other people's property	.71	1.00			.50
Purposely put graffiti on walls, toilet doors, bus panels	.70	1.06*	.10	10.95	.48
Purposely damaged school property	.60	.84*	.09	0.41	.35
Factor 6: Fight (α = .62)
Used or threatened to use force	.74	.92*	.12	7.73	.55
Used or threatened to use a weapon	.68	.97*	.13	7.49	.47
Purposely bashed someone up or got involved in a fist fight	.50	1.00			.24
Factor 7: Drugs (α = .86)
Sold or dealt any illegal drugs	.79	.77*	.06	14.17	.63
Illegally obtained or consumed cigarettes	.77	1.00			.59
Bought, possessed, or consumed ecstasy or amphetamines	.77	.73*	.05	13.79	.60
Bought, possessed, or consumed cannabis	.75	1.08*	.08	13.36	.57
Bought, possessed, or consumed LSD	.70	.60*	.05	12.32	.49
Factor 8: Media (α = .64)
Threatened or harassed someone via mobile phone	.72	1.17*	.12	9.49	.51
Threatened, harassed, or intimidated someone online	.58	1.24*	.15	8.23	.34
Designed, obtained, sold, or used a fake ID	.58	1.00			.34

Note. S.E. = standard error; C.R. = critical ratio; SMC = squared multiple correlations.

*p < .001.

The model was statistically significant, χ²(377) = 1,088.11, p < .001; however, the sample size was relatively large. We also conducted goodness of fit tests to determine model fit. Indices of comparative fit (comparative fit index (CFI) = .83; Tucker Lewis index (TLI) = .80) were below the recommended cut‐off of .95 (Hu & Bentler, 1999; Schermelleh‐Engel, Moosbrugger, & Müller, 2003), whereas absolute indices examining closeness of fit suggested a reasonable level of fit between the hypothetical model and the sample data (χ²/df ratio = 2.89 (within the suggested range of 1–3; Carmines & McIver, 1981), and root mean square of approximation (RMSEA) = .07 (lowest (LO) = .07. highest (HI) = .08), below the recommended .08; Browne & Cudeck, 1993). Lower information theoretic indices (Akaike Information Criterion (AIC) = 1,324.11) were also found for the hypothetical model, suggesting better parsimony and comparative fit than the independence model (AIC = 4,811.65). Fit indices should be interpreted with caution; however, as Heene, Hilbert, Draxler, Ziegler, and Bühner (2011) illustrate that the chi‐square model and cut‐off values for fit indices, such as the RMSEA and CFI, cannot be interpreted independently of the size of unique variances and complexity of a given model, and if cut‐off values are interpreted as routinely recommended, can lead to erroneous and invalid results. However, fit indices should be interpreted with caution. For example, Heene, Hilbert, Draxler, Ziegler, and Buhner (2011) illustrate that the chi‐square model and cut‐off values for fit indices, such as the RMSEA and CFI, cannot be interpreted independently of the size of unique variances and complexity of a given model. Furthermore, if cut‐off values are interpreted as routinely recommended, erroneous and invalid outcomes may result. Nevertheless, model fit indices are still useful in the evaluation of models, and the current factor structure appears theoretically significant. The contents of the factors reflect groupings of activities that are comparable in nature, making it possible to derive subscales that are both internally homogenous and meaningful in their contents. Table 2 presents intercorrelations among the eight factors, consistently showing positive associations of medium to large effect sizes.

Table 2. Intercorrelation matrix of the eight‐factor solution of the 30‐item Australian Self‐reported Delinquency Scale‐Revised

	Driving/Vehicle	Alcohol	Theft	Cheat	Disturb	Fight	Drugs	Media
Driving/Vehicle	–	.60	.52	.61	.54	.54	.63	.52
Alcohol		–	.44	.59	.56	.39	.63	.45
Theft			–	.56	.53	.46	.45	.46
Cheat				–	.63	.48	.52	.51
Disturb					–	.48	.52	.55
Fight						–	.46	.58
Drugs							–	.58
Media								–

Note. All correlations are significant at p < .001.

Following Byrne's (2004) procedures, the factor structure was also tested for measurement invariance between genders (summarised in Appendix D). Within the measurement model, genders only differed significantly on the Theft and Media factors, and on Items 21 (using or threatening to use force) and 25 (buying or consuming lysergic acid diethylamide (LSD)) from the Fight and Drugs factors, respectively. Given the complexity of the current model, testing the factor structure for males (180) and females (132) independently should be interpreted with caution, as it may result in convergence failures, improper solutions, and lowered accuracy of parameter estimates (Loehlin, 1992). Independent sample t‐tests with Bonferroni correction were further used to examine gender differences across factors, with a significant difference found only for Fight (t(289.31) = 4.44, p < .001), with males more likely to report engagement in violent behaviours than females.

Delinquency scores were obtained by collating the number of types of delinquent acts the respondent admitted to committing over the previous 6 months. Scores ranged from or 0 to 30 (males ranged from 0 to 30, females from 0 to 26). The mean number of delinquent acts reported was 5.25 by males and 4.32 by females. Scores based on the 30‐item delinquency scale correlated strongly with scores from the 56‐item checklist, r = .98, p < .001 for males and females.

Coefficient alphas for the 30‐item delinquency scale were .94 for males and .92 for females (.93 overall), similar to the high internal consistency reliabilities of the 56‐item checklist. Content validity of the research instrument was ensured given that the 30 items were selected from the larger checklist of 56 items regarded as current juvenile delinquent activities by law enforcement officers, professionals who work with delinquent youth, and potential offenders themselves.

To obtain an estimate of the 30‐item scale's construct validity, the relationship between the adolescents' scaled scores and their self‐reported contacts with the police, the Children's Court for breaking the law, and the extent of suspension or expulsion from school were examined. Point‐biserial correlations between the 30‐item scale and the three validity items indicated that those with official police warnings (.51 for males, .47 for females), court appearances (.46 for males, .39 for females), and who had been suspended or expelled from school (.49 for males, .35 for females) had higher delinquency scores than students who did not, indicating construct validity of the revised scale.

The final revised delinquency scale, as listed in Appendix C, includes the validity item referring to police cautions (this attracted a higher participation rate compared with appearance in the Children's Court or expulsion from school). Also included are three of the four lie items with the highest response rates (failed to keep a promise, told a lie to someone, and did something your parents did not want you to). These lie items are used to detect an unusually high level of social desirability; the authors recommend that records with ‘no’ responses to at least two of the lie items be excluded from research data analysis. The revised ASRDS has altogether 33 items, fewer than the total of 40 items in the original ASRDS.

Discussion

Our overall research aim was to update Mak's (1993) ASRDS to ensure applicability to contemporary adolescents. Consultations with potential offenders, law enforcement officials, mental health professionals, and school teachers/counsellors were used to revise and update items, covering a range of marginally deviant to seriously delinquent activity known to occur with frequency among contemporary Australian adolescents. Results of consultations provided a valid checklist of 56 contemporary delinquent activities that could be used to assist clinical assessment and treatment evaluation. From this larger pool of 56 items, item analyses EFA and CFA were used to reduce items and assess the suitability of an eight‐factor structure for the delinquency instrument. Subsequently, a 30‐item scale of contemporary delinquent activities was developed with eight specific subscales (Driving/Vehicle, Alcohol, Theft, Cheat, Disturb, Fight, Drugs, and Media). The resulting scale provides an update of Mak's (1993) original measure and has meaningful and homogenous subscales. Consultations with stakeholders (including law enforcement agents), the initial EFA, the subsequent CFA, and analysis based on validity items embedded within the scale have provided evidence of content and construct validity.

The eight factors comprise distinct but related types of delinquent offending and are useful for characterising an individual's involvement in specific areas. The Driving/Vehicle factor refers to illegal behaviours related to a vehicle (e.g., intoxicated driving, racing with other vehicles, stealing a vehicle or items from a vehicle). The Alcohol factor refers to illegally purchasing or obtaining alcohol and consuming alcohol in public locations. The Theft factor refers to shoplifting and stealing money of less than or more than $US20. The Cheat subscale refers to misdemeanours (e.g., wagging school, not paying the proper fee to enter or use facilities/services). The Disturb factor pertains to public disturbance such as damaging public or private property (e.g., egging houses, vandalising school property, graffiti, etc.). The Fight factor refers to violent or threatening acts (e.g., using or threatening to use force or a weapon, engaging in fights). The Drugs factor involves consumption of licit and illicit substances (e.g., cigarettes, cannabis, LSD, amphetamines) and selling or dealing drugs. Finally, the Media factor refers to using various forms of media to threaten or harass another (e.g., online or by mobile phone) and includes designing, obtaining, or using fake identification. Slightly different factor structures were found for males and females, with males more likely to endorse items on Theft, Media, and Fight factors.

These eight factors are similar to the nine original domains found by Mak's (1993) scale; however, the ASRDS‐R provides a more parsimonious factor structure with higher factor reliabilities than the original ASRDS. In the ASRDS, Driving (illegal driving behaviours) and Vehicle (stealing items or parts from a vehicle) were considered separate factors, whereas the current factor structure (Driving/Vehicle) combined all illegal behaviours related to a vehicle. Similarly, the ASRDS separated Fight (engaging in fist fights or using weapons) and Harm (beating others or using blackmail) factors, whereas the ASRDS‐R Fight subscale combines bashing others, using a weapon, and threatening others. While the ASRDS Status subscale included alcohol items (among other status offences), the ASRDS‐R contains a three‐item subscale specific for underage alcohol use, which is useful in reflecting and measuring the community and professionals' concerns around youth problem drinking. The new Media subscale of the ASRDS‐R is also valuable, as it takes into consideration how technology and social media may be used to commit covert acts of delinquency, such as threatening others online. Furthermore, internal consistency reliabilities for the ASRDS‐R (.93 overall) and its factor structure (ranging from .62 to .86) were much higher than those found for the original ASRDS (.88 overall) and its factor structure (ranging from .44 to .73).

The ASRDS‐R can be used to measure an individual's overall level of delinquency (by collating the number of types of delinquent acts the respondent admitted to committing), ranging from 0 to 30, with higher scores indicating a greater variety of delinquent activities engaged in. As with the original ASRDS, frequency scores can be derived by summing an individual's frequency of committing each of the delinquent acts in a specified time period. The shorter time frame (6 months in the current study) may allow for better recollection of delinquent involvement and may be more useful than the original 12‐month time frame if the updated instrument is to be used to assess any change in delinquent involvement over an academic year. Furthermore, the eight meaningful subscales allow measurement of an individual's involvement in specific areas of offending (with higher scores on each of the subscales indicating higher levels of involvement in that specific area).

The new 56‐item delinquency checklist can be used to assist clinical assessment and treatment evaluation by providing prompts to delve into specific areas of clinical interest and can help gauge frequency, severity, recidivism, and treatment prospects. Given the length of the 56‐item checklist and literacy levels for at‐risk adolescents, it may be used as part of a semi‐structured clinical interview by an experienced clinician who can build rapport with disengaged youths. The 56‐item checklist can provide insight into an individual's presentation, as it covers a range of marginally delinquent to very serious behaviours that would not be seen in the general population (e.g., items that were committed by less than 5% of respondents) and likely require specialized intervention services (e.g., arson, sex offending).

The revised 30‐item delinquency scale is likely to be useful in various applications and populations. The scale can be used as a research instrument in educational, counselling, and rehabilitations settings and may assist in evaluating different theories of delinquency. Additionally, it could be used in private or group settings for screening adolescents who may be at‐risk of criminal behaviour and may be used in schools, youth services, or clinical practice. The scale may also aid in evaluating intervention methods with increased specificity as the scale can be used as a succinct assessment of an individual's overall level of delinquency as well as characterise their involvement in specific areas of offending using the subscales derived. Finally, the updated scale may prove useful in determining discrepancies between official crime‐related statistics and self‐reported involvement in law‐breaking activities.

As with Mak's (1993) scale, the updated delinquency measures were developed in one city, Canberra, and the data obtained may vary across urban/rural locations, different jurisdictions, and over time. For these reasons, it is imperative that instruments are refined to maintain appropriate language and coverage of contemporary delinquent behaviours that are culturally relevant. As demonstrated in the present article, the advancement of technology and social media has introduced new types of delinquent activities (e.g., sexting, cyberbullying), and common types of problematic substances being used may also change over time.

Despite the current evidence for reliability and validity, there are limitations inherent within the self‐report method. Adolescents may refrain from reporting accurately and are subject to biases such as social desirability, acquiescent responses, inconsistent reporting, and memory distortions or failures (Paulhus, 1991; Sibley et al., 2010). In the case of the present scale, biases due to social desirability can, to some extent, be detected by unusually low scores on the lie items. Uncommonly low scores on these items reflect a tendency for respondents to portray an idealistic and unrealistic picture of themselves, which is likely to be associated with underreporting involvement in delinquent behaviours (Mak, 1993). The authors recommend that records with ‘no’ responses to at least two of the lie items be excluded from research data analysis because of the potential for biased responding.

A limitation in Study 2 pertains to adolescent response rates and representativeness of the sample. Ethics protocol required Australian Capital Territory (ACT) government students to obtain opt‐in parental consent before completing the survey, whereas independent schools allowed an opt‐out approach. While we were unable to determine the exact response rates for government versus independent students, we observed that it was harder to recruit student participants from government schools compared with recruiting from independent schools. The opt‐in parental consent procedure in government schools may also have led to the selection of adolescents with lower levels of risk‐taking behaviour in government students. While differential response rates and selection procedures may potentially compromise the representativeness of the sample and caution against using the reported participation rates as being normative, the resulting scale items closely match those described in Mak's (1993) original study and are enhanced by the use of qualitative data obtained from both government and independent school students and various stakeholders experienced at working with at‐risk youth. Furthermore, the items demonstrate high levels of validity as evidenced by factor analyses and validity items embedded within the scale.

Implications for future research

The current research has updated the ASRDS, resulting in a reliable and valid 30‐item self‐reported delinquency scale for contemporary Australian adolescents, covering a range of frequently occurring marginally deviant to highly serious criminal activities. In addition, this research has provided a reliable and valid 56‐item checklist of contemporary delinquent activities that can assist with clinical intervention and treatment evaluation. Future research could focus on collecting more forms of psychometric information, such as test–retest reliability and additional validity estimates, including concurrent and predictive validation with official delinquency records, as well as cross‐validation in different states and countries and using larger samples. Furthermore, owing to variation in laws and types of delinquent behaviours occurring over time, the scale contents may have to be adapted for particular purposes and should be revised periodically.

Future research can also identify adolescents' perceptions regarding the seriousness or severity of each of the delinquent behaviours. Furthermore, many items include law‐breaking behaviour that may hold relevance for adults (see Appendix B for items that may be potentially relevant for adult samples), and the ability of the measure to be adapted for adult samples should be further investigated. Future research could additionally test the factor structure of the current model for males and females separately, using a larger and more heterogeneous sample. Finally, although the current measures use a Yes/No format to determine participation in delinquent behaviours, future research could expand the response format to take frequency of various delinquent activities into consideration, which may also be useful for purposes of cross‐cultural validation.

The updated ASRDS assesses individual differences in participation in a list of 30 types of contemporary delinquent activities. The updated instrument contains eight subscales (Driving/Vehicle, Alcohol, Theft, Cheat, Disturb, Fight, Drugs, Media), facilitating assessment of engagement in specific areas of offending. The revision and shortening of the ASRDS, complemented by a 56‐item delinquency checklist, may prove useful in educational, rehabilitation, and research settings and aid in evaluating clinical interventions with increased specificity.

Acknowledgements

The authors would like to acknowledge the participating school principals and teachers for their assistance in data collection.

Appendix Appendices

Appendix A

Summary of focus group participants (N = 36).

Table

Stakeholders	Male (n)	Female (n)	Age range	M age	Years of experience range	M years of experience
Secondary school students	6	0	17	17	–	–
‘At‐risk’ adolescents	2	8	16–17	16.9	–	–
Youth workers	3	4	25–51	35.6	2–22	11.9
Police officers	2	4	25–64	39.6	2–22	12.8
School teachers School counsellor	0	6 1	30–63	48.9	15–35	26.1

Appendix B

The 56‐item delinquency checklist with item‐level participation rates (in percentages) and proposed factors.

Table

	Participation rates			Proposed factor
	Male	Female	Item level χ^2 (p)
*1. Driven an unregistered car	11.7	9.1	.465	Driving/Vehicle
*2. Driven a car or a motor bike on the road without a legal driver's license or a learner's permit	25.0	18.2	.152	Driving/Vehicle
*3. Driven a car or a bike when drunk or over the legal alcohol limit	6.1	6.1	.985	Driving/Vehicle
*4. Driven a car or a bike when under the influence of any illicit substances	5.6	3.8	.471	<5% total respondents
*5. Ignored a red light while driving a car	10.0	6.1	.214	Driving/Vehicle
*6. Stolen and driven a car	3.9	3.0	.685	<5% Total respondents
*7. Been joy‐riding in a stolen car	3.9	3.8	.963	<5% Total respondents
*8. Been involved (i.e., driving) in a hit‐and‐run accident	2.2	1.5	.653	<5% Total respondents
*9. Raced with other vehicles while driving a car or motor bike on the road	13.3	12.1	.752	Driving/Vehicle
*10. Engaged in any other illegal behaviour in a car (e.g., burnouts, doughies)	17.2	15.9	.758	Driving/Vehicle
*11. Taken and driven a car or a motor bike that belonged to someone else without the owner's consent	7.2	3.8	.199	Driving/Vehicle
*12. Stolen things or parts out of a car, motor bike, or bicycle	7.2	4.5	.329	Driving/Vehicle
13. Watched an R‐rated movie or played an R‐rated video game	74.4	54.5	.000	Media
L1. Failed to keep a promise (lie item)	85.0	84.1	.826
14. Bought or illegally obtained beer, wine, spirits, or other kinds of alcohol	28.9	37.1	.125	Alcohol
15. Drunk alcohol in a public venue where the age limit is 18 and over (e.g., a nightclub or pub)	12.8	18.2	.187	Alcohol
16. Drunk alcohol in a public location (e.g., park, school function, music festivals)	25.0	27.3	.651	Alcohol
*17. Got into a cinema, pool or other facility without paying the proper fee	32.8	22.7	.052	Cheat
*18. Pulled a runner without paying the proper fee (e.g., from a taxi, bus or restaurant, etc.)	12.8	7.6	.140	Cheat
19. Not attended classes or wagged school	33.3	47.0	.015	Cheat
20. Run away from home (at least overnight)	10.1	14.4	.243	N/A
*21. Shoplifted	18.9	18.9	.991	Theft
*22. Stolen money of less than $US20 (in one go) from shops, school, locker rooms, parents	30.0	22.0	.113	Theft
*23. Stolen money of more than $US20 in one go	16.7	9.8	.084	Theft
L2. Been late for school, work, an appointment, etc (lie item)	80.6	87.1	.124
EX. Been suspended or expelled from school (validity item)	15.0	12.1	.466
*24. Broken into a house/building with the intention of stealing something	5.0	3.8	.609	<5% Total respondents
*25. Broken into a house/building for other reasons	21.7	18.2	.449	Disturb
*26. Taken part in a robbery, using a weapon or force	3.3	1.5	.315	<5% Total respondents
*27. Cheated or stolen food, drinks, or other goods from dispenser machines, e.g., by tilting or banging the machines, or using the ‘wrong’ coins	25.0	12.9	.008	Cheat
*28. Stolen a mobile phone or any other electronic device that belonged to someone else	11.1	4.5	.038	Theft
*29. Purposely messed up other people's property, e.g., egging houses, throwing rocks	23.9	14.4	.038	Disturb
*30. Purposely damaged property by starting a fire	4.4	4.5	.966	<5% Total respondents
*31. Purposely started a bushfire	1.7	1.5	.916	<5% Total respondents
*32. Purposely damaged property in public places, e.g., street signs, road lamps	17.8	12.1	.171	Disturb
*33. Purposely put graffiti on walls, toilet doors, bus panels, or other public places	23.3	25.8	.622	Disturb
34. Purposely damaged school desks, windows, or other school property, e.g., kicking holes in the wall	23.3	15.9	.107	Disturb
L3. Done something that your parents did not want you to (lie item)	87.8	82.6	.196
*35. Purposely bashed someone up or got involved in a fist fight	27.8	10.6	.000	Fight
36. Possessed a weapon (e.g., knife) at school	13.9	3.8	.003	Fight
*37. Used or threatened to use a weapon of some sort, e.g., knife, bat, bottle, in a fight	13.4	2.3	.001	Fight
*38. Used or threatened to use force to get money or things from another person	8.4	3.8	.103	Fight
39. Illegally obtained or consumed cigarettes	15.6	15.9	.949	Drugs
*40. Bought, possessed, or consumed cannabis (also called marijuana, pot, weed)	19.0	22.0	.519	Drugs
*41. Bought, possessed, or consumed LSD (also called acid)	6.7	5.3	.610	Drugs
*42. Bought, possessed, or consumed ecstasy or other forms of amphetamines (e.g., speed, meth, ice)	7.3	7.6	.917	Drugs
*43. Sold or dealt any illegal drugs	8.9	6.8	.497	Drugs
*44. Illegally obtained or abused prescription or non‐prescription medication (e.g., Valium, Panadol, Nurofen, etc.)	6.1	15.2	.009	Drugs
*45. Downloaded movies or music illegally (piracy)	76.0	70.5	.275	Media
*46. Designed, obtained, sold, or used a fake ID	10.1	8.3	.606	Media
*47. Engaged in online identify theft (e.g., hacked into someone else's mobile or online networking account)	9.5	17.4	.039	Media
48. Gotten a piercing or tattoo without parental permission	5.6	15.2	.005	N/A
*49. Forced someone to do sexual things with you when that person did not consent or was underage	2.8	2.3	.774	<5% Total respondents
50. Sent a sexually explicit photo or text of yourself to others on your mobile phone or posted such materials online	8.4	11.4	.378	Media
51. Sent a sexually explicit photo or text of someone else to others on your mobile or posted such materials online	5.0	6.8	.504	Media
52. Made prank phone calls	55.9	57.6	.764	Media
*53. Threatened, harassed, or intimidated someone online (e.g., leaving nasty comments or spreading rumours on social media such as Facebook or Twitter)	18.4	11.4	.088	Media
*54. Threatened, harassed, or intimidated someone via mobile phone (e.g., making abusive phone calls)	11.7	3.8	.012	Media
*55. Intentionally hurt or been cruel to animals	7.3	1.5	.019	<5% Total respondents
56. Accessed pornographic materials	62.6	17.4	.000	Media
PW. Had contact with the police (e.g., informal or formal caution) without being charged for something that you did (validity item)	17.3	10.6	.096
CC. Appeared in the Children's Court for something that you did (validity item)	6.1	3.0	.205
L4. Told a lie to someone (lie item)	93.9	97.0	.205

Note. The 56 items of the checklist are interspersed with four lie items (labelled L1–L4), an item on suspension or expulsion from school (labelled EX), an item on police warning (labelled PW), and an item on Children's Court appearance (labelled CC). Item‐level χ² tests highlighted in bold indicate a significant (p < .05) gender difference in participation rates. This 56‐item and the 30‐item scale, complete with introductory instructions, are available from the authors.

Items marked with an * denote items that also hold relevance for adult samples.

Items 20 and 48 did not appear to fit neatly into any proposed factor categories and were subsequently removed from the analysis.

Appendix C

The Australian Self‐reported Delinquency Scale‐Revised: 30‐item scale.

Have you in the past 6 months.

Table

		Factor
1. Driven a car or a bike when drunk or over the legal alcohol limit	No Yes	Driving/Vehicle
2. Raced with other vehicles while driving a car or motor bike on the road	No Yes	Driving/Vehicle
3. Taken and driven a car or a motor bike that belonged to someone else without the owner's consent	No Yes	Driving/Vehicle
4. Stolen things or parts out of a car, motor bike, or bicycle	No Yes	Driving/Vehicle
*5. Engaged in any other illegal behaviour in a car (e.g., burnouts, doughies)	No Yes	Driving/Vehicle
*6. Bought or illegally obtained beer, wine, spirits, or other kinds of alcohol	No Yes	Alcohol
L1. Failed to keep a promise	No Yes
*7. Drunk alcohol in a public venue where the age limit is 18 years and over (e.g., a nightclub or pub)	No Yes	Alcohol
*8. Drunk alcohol in a public location (e.g., park, school function, music festivals)	No Yes	Alcohol
*9. Shoplifted	No Yes	Theft
*10. Stolen money of less than $US20 (in one go) from shops, school, locker rooms, parents	No Yes	Theft
*11. Stolen money of more than $US20 in one go	No Yes	Theft
12. Cheated or stolen food, drinks, or other goods from dispenser machines, e.g., by tilting or banging the machines, or using the ‘wrong’ coins	No Yes	Cheat
13. Got into a cinema, pool or other facility without paying the proper fee	No…Yes	Cheat
*14. Pulled a runner without paying the proper fee (e.g., from a taxi, bus, or restaurant, etc.)	No…Yes	Cheat
*15. Not attended classes or wagged school	No…Yes	Cheat
*16. Purposely messed up other people's property, e.g., egging houses, throwing rocks	No Yes	Disturb
17. Purposely damaged property in public places, e.g., street signs, road lamps	No Yes	Disturb
18. Purposely put graffiti on walls, toilet doors, bus panels, or other public places	No Yes	Disturb
19. Purposely damaged school desks, windows, or other school property, e.g., kicking holes in the wall	No…Yes	Disturb
L2. Done something that your parents did not want you to (lie item)	No Yes
*20. Used or threatened to use a weapon of some sort, e.g., knife, bat, bottle, in a fight	No Yes	Fight
21.Used or threatened to use force to get money or things from another person	No Yes	Fight
*22. Purposely bashed someone up or got involved in a fist fight	No Yes	Fight
*23. Illegally obtained or consumed cigarettes	No Yes	Drugs
*24. Bought, possessed, or consumed cannabis (also called marijuana, pot, weed)	No Yes	Drugs
*25. Bought, possessed, or consumed LSD (also called acid)	No Yes	Drugs
*26. Bought, possessed, or consumed ecstasy or other forms of amphetamines (e.g., speed, meth, ice)	No Yes	Drugs
*27. Sold or dealt any illegal drugs	No Yes	Drugs
*28. Designed, obtained, sold, or used a fake ID	No…Yes	Media
*29. Threatened, harassed, or intimidated someone online (e.g., leaving nasty comments or spreading rumours on social media such as Facebook or Twitter)	No Yes	Media
*30. Threatened, harassed, or intimidated someone via mobile phone (e.g., making abusive phone calls)	No Yes	Media
PW. Had contact with the police (e.g., informal or formal caution) without being charged for something that you did (validity item)	No Yes
L3. Told a lie to someone (lie item)	No Yes

Note. Items marked by * indicate new items or items with revised wording. The 30 delinquency items are interspersed with three lie items (labelled L1–L3), and an item on police warnings (labelled PW). The scale, complete with introductory and scoring instructions, is available from the authors.

Appendix D

Measurement invariance for males and females in 30‐item confirmatory factor analysis.

Table

Model description	Comparative model	χ^2	df	Δ χ^2	Δdf	p
1. Hypothesised 30‐item model (Model 1)		2,074.61	754	–	–	–
2. Factor loadings, variances and co‐variances constrained equal	Model 1	2,295.82	812	221.21	58	<.05
3. Factor loadings constrained	Model 1	2,165.24	776	90.81	22	<.05
4. Factor loadings constrained on Driving/Vehicle only (Model 2)	Model 1	2,079.03	758	4.42	4	Ns
5. Factor loadings constrained on Model 2 (Driving/Vehicle) and Alcohol (Model 3)	Model 2	2,079.29	760	0.26	2	Ns
6. Factor loadings constrained on Model 3 and Theft	Model 3	2,110.13	762	30.84	2	<.05
6b. Factor loadings constrained on Model 3 and Theft (Item10)	Model 3	2,100.86	761	21.57	1	<.05
6c. Factor loadings constrained on Model 3 and Theft (Item11)	Model 3	2,105.89	761	26.6	1	<.05
7. Factor loadings constrained on Model 3 and Media	Model 3	2,102.78	762	23.49	1	<.05
7b. Factor loadings constrained on Model 3 and Media (Item29)	Model 3	2,083.66	761	4.37	1	<.05
7c. Factor loadings constrained on Model 3 and Media (Item30)	Model 3	2,102.67	761	23.38	1	<.05
8. Factor loadings constrained on Model 3 and Cheat (Model 4)	Model 3	2,082.30	763	3.01	3	Ns
9. Factor loadings constrained on Model 4 and Disturb (Model 5)	Model 4	2,082.80	766	3.51	6	ns
10. Factor loadings constrained on Model 5 and Fight	Model 5	2,102.60	768	23.31	8	<.05
10b. Factor loadings constrained on Model 5 and Fight (Item20; Model 6)	Model 5	2,090.80	767	11.51	7	ns
11. Factor loadings constrained on Model 6 and Drugs	Model 6	2,101.80	771	11	4	<.05
11b. Factor loadings constrained on Model 6 and Drugs (Item24; Model 7)	Model 6	2,091.90	768	1.10	1	ns
11c. Factor loadings constrained on Model 7 and Drugs Item25	Model 7	2,099.90	769	8	1	<.05
11d. Factor loadings constrained on Model 7 and Drugs Item26 (Model 8)	Model 7	2,095.10	769	3.20	1	ns
11e. Factor loadings constrained on Model 8 and Drugs Item27 (Model 9)	Model 8	2,098.20	770	3.10	1	ns

Note. Genders differed on items within the Theft and Media factors, and on Items 21 (Fight) and 25 (Drugs). Items 1, 6, 9, 13, 16, 22, 23, and 28 were fixed parameters.