The Correlates of Self-Reported Delinquency: An Analysis of the National Longitudinal Survey of Children and Youth

2. Method

The National Longitudinal Survey of Children and Youth provided a useful data source for the examination of self-reported delinquency in Canadian youth. The NLSCY is a joint project between Statistics Canada and Human Resources Development Canada. The goal of the survey is, "to monitor the development and well being of Canada 's children as they grow from infancy to adulthood" (Statistics Canada, 2002, p.1). Initiated in 1994, the NLSCY was designed to follow a representative sample of Canadian children (newborns to 11 year olds) into adulthood. New interview and questionnaire data are collected every two years.

The data used in this study were drawn from Cycle III, which was collected in 1998. During Cycle III, individuals between 12 and 15 years of age filled out self-administered questionnaires designed to examine a number of different aspects of their lives (e.g., friends, family, school, feelings and behaviors , delinquent behavior , health, and work). Data on each child, including items such as his or her family's economic, labour force and health status, birth information, behavior , social relationships, school performance, and learning environment were collected from the person most knowledgeable (PMK) about the child, the PMK's spouse (if applicable), and the child's principal and teachers through questionnaires and interviews. In total, Cycle III provides data on 31,194 children from across the ten provinces.

2.1 Weighting procedures

The principle behind estimation in a probability sample, such as the NLSCY, is that each respondent represents several other individuals in the population. Applying a weight to a respondent allows for the calculation of the number of children that respondent represents. Conceptually, the basic weight of each child in the NLSCY is roughly equal to the inverse of the child's probability of selection. In general, one child represents approximately 300 children in the population. Given our secondary purpose of reporting on the prevalence of self-reported delinquency among youth 12 to 15 years of age in Canada , this study employs the cross-sectional weighting variable created by Statistics Canada.

2.2 Missing data

In order to effectively conduct statistical analysis, it was necessary to recode certain responses within the NLSCY. If a particular individual provided responses such as 'do not know' or 'not applicable', or if the individual refused to answer the question, the response was coded as missing. A portion of the respondents also failed to answer entire sections of the survey, which, along with our recoding procedures, resulted in a substantial reduction in the sample size used in the regression analysis.

2.3 Dependent variable calculation: Self-Reported Delinquency Score

In order to operationalise self-reported delinquency, 14 questions from the NLSCY were selected as dependent variables and matched with comparable Criminal Code offences. [1] Respondents were asked to indicate the frequency of each of the criminal behaviors in the 12 months preceding the survey using the following choices: never; once or twice; three or four times; or, five or more times.

Respondents were also asked about drug use (e.g., "Have you every tried marijuana and cannabis products?"), but the possible responses were dichotomous (i.e., "yes" or "no") and the timeframe was not specified as with the 14 selected questions. Therefore, drug use was not considered within the dependent delinquency score. As it is a Criminal Code offence to possess drugs in Canada , a decision was made to also exclude drug use from the list of independent variables.

In order to develop a meaningful measure of delinquency, which considered both the severity and the frequency of the behavior , weights were attached to each variable based upon the seriousness of the offence and subsequently multiplied by its frequency. Severity was defined using the mean custodial sentence length (in months) of each Criminal Code offence. [2] This information was obtained from the 1998 Adult Criminal Court Survey (ACCS) managed by the Canadian Centre for Justice Statistics (CCJS). Adult court data were selected rather than youth court data in order to create more variability and to accurately rank the seriousness of the crime based upon societal norms as maximum sentences in Canadian youth court are three years for most offences.

For example, the mean sentence length for sexual assault, using ACCS data, was 16 months. If a respondent in the NLSCY indicated that they had, i n the past 12 months, forced someone into having sex once or twice, the score would be 16, while three or four infractions would create a score of 32 and five or more infractions would total 48 points. Once each offence was weighted according to severity and frequency, a total overall self-reported delinquency score (SRDS) was calculated by summing each individual's score across all the offences. The SRDS can range from 0 (indicating no delinquent behavior ) to 312.

2.4 Independent variables

Previous studies on the correlates of delinquency were used as a guide in selecting the appropriate independent variables. The following scores [3] and individual variables were included:

  • Pro-social Score
  • Parental Nurturance Score
  • Parental Monitoring Score
  • Hyperactivity/Inattention Score
  • Emotional Disorder Score
  • Indirect Aggression Score
  • General Self Score
  • Friends Score
  • School Attachment Score
  • Extra-curricular Involvement Score
  • Volunteer Score
  • Victimization Score
  • Drug-use Score
  • Negative Peer Score
  • Negative School Behavior Score
  • Inconsistent Parenting Score
  • Peer Drug-use
  • Age of child
  • Gender of child
  • Household Income
  • Child witnesses abuse within the home
  • Social Economic Status
  • Child takes Ritalin
  • Child destroys own belongings
  • Child gives up easily
  • Parents hit child or threaten to
  • Parents get angry and yell at child
  • Child's time spent with friends
  • Parents encourage child to do well
  • Child repeated grade
  • Number of close boyfriends
  • Number of close girlfriends
  • Child's school aspirations

2.5 Multiple regression

Logistic regression was used to assess the combined and individual influence of each of the independent variables on the SRDS. In addition to the general regression analysis, partial regression analyses were conducted for male youth, female youth, youth engaged in violent offences, youth engaged in property offences, youth engaged in sexual offences, and youth engaged in drug trafficking.


Date modified: