Public concern about high-risk behaviors has increased pressure on researchers to assess the attitudes and behaviors of adolescents. The method most frequently used for this task is the self-report survey, which raises the issue of whether this source of information is reliable. Two questions were considered in the present study: What is the relationship between self-report surveys, parental interviews, and school and police records in their assessment of adolescent behaviors and attitudes? and What types of questions elicit responses that best reflect police and school data? These questions were posed during a recent analysis of data collected as part of a study evaluating an intervention program for high-risk youth.
As common as the self-report survey method is for the assessment of adolescent risk behaviors and attitudes, there is a paucity of empirical research comparing it with other, more concrete methods. From a clinical perspective, many studies have considered the reliability of different data sources (Kazdin, 1989; Knight, Hensley, & Waters, 1988; Reynolds, Anderson, & Bartell, 1985; Kotsopoulos, Walker, Copping, Cote, & Stavrakaki, 1994; Achenbach, McConaughy, & Howell, 1987). Achenbach and colleagues (1987) meta-analyzed studies that measured youth emotional/behavioral problems by means of different data sources (i.e., self, peers, teachers, parents, and clinicians). The interinformant Pearson's r was considerably higher in samples that collected data from informants other than the youths themselves. In addition, the mean interinformant r was significantly lower for adolescents as compared with children (Achenbach et al., 1987). It thus appeared that self-report as a measure of conduct disorder, particularly for adolescents, was less strongly associated with other data sources.
Self-report surveys are also commonly used in the study of personality. Moskowitz (1986) concluded that, despite potential methodological pitfalls, self-report can be used reliably to assess personality characteristics defined as having stability, coherence, and generality across situations. More important for the present study, Moskowitz asserted that self-report surveys are reliable in assessing personality characteristics that are defined situationally and fall within time parameters. Criterion validity, however, was not examined.
Assessments of substance use also rely heavily on self-report data. However, researchers assessing adolescent substance use have found inconsistencies in interitem responses (Barnes & Welte, 1986) and fallacious reporting of fictitious drug use (Swisher, Shute, & Bibeau, 1984). In a survey of 2,470 low-income middle school students, Farrell, Danish, and Howard (1991) found that 43% had inconsistencies in their responses regarding substance use. They concluded that this was due to random responses, and maintained that their method by and large produced accurate data. Other researchers also have concluded that use of a self-report survey is an accurate way to assess drug and alcohol consumption, but have suggested using multiple methods to improve its validity (Johnson, 1985). Despite all these assertions, the accuracy of self-report surveys of drug and alcohol use seems controvertible.
The literature suggests that adolescents' self-report data should be examined in greater detail. This study evaluated the self-report method when assessing high-risk adolescents.
The data for this study was compiled as part of an evaluation of a Northern California intervention project aimed at high-risk youth. One component of the intervention is a youth center that was designed by and targets gang-exposed youth. The center endeavors to be a safe and supportive environment for youth who are surrounded by, and involved in, gang culture. The youth who participated in this study were selected on the basis of their regular attendance at the center. …