Premium
Predictive, concurrent, prospective and retrospective validity of self‐reported delinquency
Author(s) -
Jolliffe Darrick,
Farrington David P.,
Hawkins J. David,
Catalano Richard F.,
Hill Karl G.,
Kosterman Rick
Publication year - 2003
Publication title -
criminal behaviour and mental health
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.63
H-Index - 54
eISSN - 1471-2857
pISSN - 0957-9664
DOI - 10.1002/cbm.541
Subject(s) - predictive validity , juvenile delinquency , psychology , socioeconomic status , longitudinal study , demography , prospective cohort study , referral , clinical psychology , medicine , psychiatry , family medicine , population , environmental health , surgery , pathology , sociology
Background The self‐report method is widely used to measure offending. Previous studies suggest that it is generally valid, but that its validity may be lower for blacks than for whites. Aim To assess the validity of self‐reported offending in relation to court referrals, and to investigate how it varies with types of offences, sex and race. Method Annual court and self‐report data were collected between ages 11 and 17 for eight offences in the Seattle Social Development Project, which is a prospective longitudinal survey of 808 youths. Results Self‐reports predicted future court referrals. Predictive validity was highest for drug offences, for males and for whites, and lowest for females and Asians. The probability of youths with a court referral reporting offences and arrests was highest for drug offences, for males, for whites and for blacks. Retrospective ages of onset agreed best with prospective ages for drug offences, Asians and whites. More Asians than blacks or whites failed retrospectively to report offences that had been reported prospectively. Conclusions The validity of self‐reports of offending was high, especially for drug offences, for males and for whites. Contrary to prior research, validity was high for black males. It was lowest for Asian females. Sex and race differences in validity held up after controlling for socioeconomic status. Differential validity probability did not reflect police bias. Copyright © 2003 Whurr Publishers Ltd.