Heaven (1996): Personality, Self-Esteem and Delinquency

Why investigate personality over time?

Heaven’s (1996) study, Personality and Self-Reported Delinquency: A Longitudinal Analysis, investigated whether personality could help predict young people’s later delinquency. Delinquency means antisocial or law-breaking behaviour, particularly among young people. Here, it was measured through students’ own reports of interpersonal violence and vandalism/theft, rather than through official criminal records.

Eysenck’s criminal personality theory linked criminal behaviour to extraversion, neuroticism and psychoticism. Extraversion involves outgoing, stimulation-seeking behaviour; neuroticism involves emotional instability; psychoticism includes traits such as aggression and low empathy. Earlier research had supported some links between personality and delinquency, but findings were not consistent.

An association at one moment does not necessarily tell us whether a personality measure can predict later behaviour. Heaven therefore followed the same young people over time. This placed the study within the individual differences approach: it investigated whether differences between people were related to differences in their behaviour.

The aim and the role of self-esteem

Heaven aimed to test whether psychoticism, extraversion and self-esteem measured at Time 1 were significant predictors of self-reported delinquency about two years later. The hypothesis was that these measures would significantly predict delinquency over time.

Self-esteem is a person’s overall sense of their own worth. Heaven included it instead of neuroticism for two reasons. First, Furnham and Thompson (1991) and Farrington (1992) had questioned how useful neuroticism was in predicting self-reported delinquency. Second, low self-esteem was often regarded as an aspect of neuroticism. Researchers had suggested that delinquent behaviour might compensate for feelings of low self-worth. This was a proposed explanation, not something the study had already established.

Who took part?

At Time 1, in August 1992, the sample consisted of 282 adolescents: 146 females and 136 males. They attended two Catholic high schools in New South Wales, Australia. Their ages ranged from 13 to 15, and the modal age—the most common age—was 14.

The schools were in a large provincial centre and were classified as independent rather than private schools. Their modest fees attracted students from a relatively wide range of socio-economic backgrounds.

At Time 2, in June 1994, 123 females and 112 males took part. The female follow-up rate was 84%, and the male follow-up rate was 82.3%. The modal age was now 16. This loss of participants between testing occasions is called attrition.

How personality and delinquency were measured

The research was a longitudinal study using self-report questionnaires. Longitudinal research collects information from the same participants on more than one occasion. It was also correlational: Heaven measured variables and analysed relationships between them, rather than manipulating personality in an experiment.

At both testing occasions, students completed a questionnaire booklet containing:

For the delinquency measures, students used four response options ranging from Never, scored 1, to Often, scored 4. Higher scores therefore indicated more frequently reported delinquent behaviour. These numerical scores allowed relationships between the measures to be analysed through correlations.

What the students did

All students in the relevant classes were tested during class time at Time 1 and again at Time 2. The questionnaires took about 30–40 minutes to complete. Students had the option of withdrawing, although none exercised that option during testing.

Questionnaires were completed anonymously, with printed numbers allowing responses to be matched across the two occasions. Students were assured that their responses were confidential and that individual answers would not be made available to school authorities.

These arrangements mattered because questions about violence and theft were sensitive. A student who feared punishment might conceal their behaviour. Confidentiality could encourage more honest answers, although it could not guarantee accuracy.

What Heaven found

Males reported more delinquency than females at both Time 1 and Time 2. This was a difference within the sample, not a claim that every male was more delinquent than every female.

Psychoticism measured at both times correlated with delinquency measured at both times. Extraversion at Time 2 also correlated with delinquency at Time 2. That same-time association alone does not establish that earlier extraversion successfully predicted later delinquency.

When the Time 1 measures were used to predict delinquency at Times 1 and 2, psychoticism was the best predictor. This gave some support to the association between psychoticism and delinquency proposed by Eysenck’s theory.

However, the three measures explained only a modest percentage of the variation in delinquency. In other words, they accounted for only a limited amount of the differences in reported delinquency between students. Being the best predictor measured did not make psychoticism a powerful predictor overall.

What the results mean for the theory

The findings did not provide significant support for the overall hypothesis that the three measures would predict delinquency over time. Their usefulness as longitudinal predictors was limited, despite the associations found.

Correlation also does not establish causation. Personality and delinquency might be associated without personality causing delinquency. Other psycho-social and environmental influences might contribute to behaviour or predict it more effectively.

The study therefore qualifies a purely personality-based explanation of crime. It offers some support for Eysenck’s emphasis on psychoticism, but not for treating personality as a sufficient explanation of delinquency. This connects the individual differences approach with social influences and cautions against a reductionist explanation that overlooks a young person’s wider circumstances.

Evaluating the evidence

Following the same students was a strength. Heaven could examine whether earlier measures predicted later behaviour, rather than relying only on a snapshot of different people. Nevertheless, establishing that a measure came first does not remove other possible explanations for a correlation.

Self-report made unrecorded behaviour accessible, but accuracy was uncertain. Students could report acts that had never reached police or school authorities. On the other hand, social desirability might lead them to under-report delinquency or present themselves more positively. Confidentiality reduced a reason for concealment, but memory errors or inaccurate self-perceptions could still affect validity.

The sample restricted generalisation. Including males and females from relatively varied socio-economic backgrounds was useful. However, all participants were adolescents attending just two Catholic schools in one Australian region. Findings may not apply equally to adults, young people outside school or other cultural and educational settings.

Attrition could introduce bias. Fewer students took part at Time 2. If those who were not followed up differed systematically in personality or delinquency, the remaining sample could give a misleading picture of the relationships. This is a possibility, not evidence that the missing students had particular characteristics.

Standardised scales made comparison easier, but limited depth. The same measures and scoring systems were used at both times, supporting consistent comparison. Yet brief questionnaires and fixed response options may not capture the full complexity or context of personality and delinquent behaviour.

Overall, Heaven provides useful evidence of associations, especially involving psychoticism, but the method and relatively weak prediction do not justify claiming that personality causes later delinquency.

Aim and design

• Heaven (1996): could Time 1 psychoticism, extraversion and self-esteem predict self-reported delinquency about two years later?
• Self-esteem replaced neuroticism because its predictive value had been questioned and low self-worth might be linked to compensatory delinquency.
• Longitudinal, correlational study using self-report questionnaires.

Sample and procedure

• August 1992: 282 students; 146 females, 136 males; ages 13–15, modal age 14.
• Two Catholic independent high schools in New South Wales, Australia; relatively varied socio-economic backgrounds.
• June 1994: 123 females and 112 males; follow-up rates 84% and 82.3%; modal age 16.
• Number-coded, anonymous questionnaires in class; 30–40 minutes; withdrawal allowed; individual responses withheld from school authorities.

Measures

• Revised short forms: psychoticism and extraversion.
• 10 items: global self-esteem.
• Six items: interpersonal violence; eight items: vandalism/theft.
• Delinquency responses: Never = 1 to Often = 4, with four options.

Findings and conclusion

• Males reported more delinquency at both times.
• Psychoticism at both times correlated with delinquency at both times.
• Extraversion correlated with delinquency at Time 2.
• Psychoticism was the best predictor, but the measures explained only modest variation in delinquency.
• Some support for Eysenck’s psychoticism link, but no significant support for the overall hypothesis; limited longitudinal predictive usefulness.
• Correlation does not establish causation; psycho-social and environmental factors may predict delinquency better.

Evaluation anchors

• Same participants over time → tests prediction beyond a snapshot.
• Self-report → accesses unrecorded acts, but honesty and recall affect validity.
• Two schools in one region → limited generalisability.
• Attrition → possible bias if missing students differed systematically.
• Standardised scales → consistent comparison, but limited detail.