The Pittsburgh Youth Study provides an excellent real-life laboratory for advancing and testing hypothesized developmental pathways. In 1986, OJJDP initiated support for three coordinated projects under the Program of Research on the Causes and Correlates of Delinquency, with study sites in Denver, Pittsburgh, and Rochester. A detailed overview of the three projects is provided in Urban Delinquency and Substance Abuse: Technical Report (Huizinga, Loeber, and Thornberry, 1993), which was collaboratively produced by the three research teams under the direction of the principal investigators: David Huizinga at the University of Colorado, Rolf Loeber at the University of Pittsburgh, and Terence P. Thornberry at the University at Albany, State University of New York.

The projects are longitudinal investigations that involve repeated contacts with the same juveniles and their primary caretakers over a substantial portion of their developmental years. This allows the researchers to pinpoint more accurately when a young person initiates certain disruptive behaviors and to examine potential causal factors that influence the onset, frequency, severity, and termination of problem behaviors, such as delinquency.

Initially, researchers at the Pittsburgh Youth Study randomly selected as subjects boys attending the first, fourth, and seventh grades in the Pittsburgh public school system. Of those subjects initially contacted, 84.7% of the boys and their caretakers agreed to participate. An initial screening of each boy, his primary caretaker (usually mother), and a teacher was used to generate retrospective data on the boys' disruptive and delinquent behavior. This information was used to develop a sample with an over representation of boys who had already demonstrated some disruptive behavior. For that purpose, the top 30% of the boys (approximately 250) with the highest rates of disruptive behavior and an equal number of the remaining 70% were selected from each of the three grade cohorts. Boys initially in the first, fourth, and seventh grades are referred to as the youngest, middle, and oldest samples, respectively. At the onset of the study in 1987/1988, Pittsburgh's sample of 1,517 boys ranged in age from 7 to 13.

Most of the study findings reported here are weighted to represent the general population of boys attending Pittsburgh's public schools. Across the three cohorts, slightly more than half (53 to 56%) of the boys are African-American; the remainder of the boys are Caucasian. About two-fifths (36 to 47%) of the boys come from families on welfare, and about two-fifths (36 to 44%) live with a single parent. This sample was drawn to capture urban youth considered at high risk for involvement in disruptive, delinquent, and drug-related behaviors. The researchers collected self-report and archival court and school data. (For more details on sample selection and methods, see Loeber et al., 1991.)

Measurement of Disruptive and Delinquent Juvenile Behaviors

The researchers for the Pittsburgh Youth Study questioned boys and their caretakers on a wide spectrum of factors during personal interviews occurring at 6-month intervals. The first interviews were conducted at the time of screening, and subsequent assessments were labeled alphabetically in sequence (A, B, C, etc.). The researchers also gathered archival data from sources such as school and court records.

Table 1 (below) displays the measurement instruments that were used to assess the progressive stages and component behaviors for each of the three developmental pathways in the middle and oldest samples (Loeber et al., 1993). Some of the instruments were administered to mothers, others to their sons. Generally, when both mothers and sons were asked about the same behavior during the same timeframe, the researchers required concurrence between the mothers and sons on behaviors considered commonplace and accepted single-source reports of behaviors considered more difficult to recall or less likely to occur.

The term "retrospective" generally refers to information gathered at either the screening interview or assessment A about life events that preceded the initial data collection. For those behaviors likely to have been initiated at a much younger age, the researchers relied heavily upon the reports of caretakers rather than those of the subjects. For example, caretakers were considered the more reliable source of information regarding the child's history of all authority conflict behaviors, the minor covert behavior of lying, and the less serious forms of aggression, all of which were assessed with the parent version of the Diagnostic Interview Schedule for Children. For those acts that qualified as delinquent, the researchers interviewed the boys directly utilizing the Youth Self-Reported Delinquency Scale.

Prospective data collection involved more overlapping of youth and maternal reports. For these data, recall deterioration was of much less concern since the reporting timeframe was limited to the past 6 months. Again, mothers were considered the more accurate source of information regarding defiant behavior and less serious types of aggression, such as annoying others and bullying. The most prospective data were collected in the youngest sample over nine half-yearly data waves. Boys were in first grade and averaged age 6.9 at screening, and most of them were in the fifth grade and averaged age 10.9 at assessment H. Therefore, when comparing the youngest boys to those 3 (middle sample) and 6 (oldest sample) years older, it is important to note that many of these boys may still be at risk for the onset of specific problem behaviors and may not have reached their peak years for delinquency. Beginning with assessment G, the youngest boys were administered the same instruments as the older boys (the Youth Self-Reported Delinquency Scale and the Youth Self-Report).

Sequences in the Age of Onset of Disruptive and Delinquent Juvenile
Behavior

A key issue in developmental pathways is establishing temporal progressions in which less serious behaviors in a given pathway occur before more serious behaviors emerge. A necessary step is sequencing the age of onset of specific behaviors. To do so, the researchers conducted analyses of a wide range of disruptive behaviors initially among the middle and oldest samples, including data collected from the screening and the next five data collection waves (S through E), and later on the youngest sample (waves S through H).

The oldest boys were selected for this presentation of findings because at age 16 (by assessment E) they were more likely than the younger cohorts to have experienced onset of the full range of behaviors. These results from the oldest sample of boys are shown in figure 4, in what are termed "box-and-whisker" plots. Here, the whisker lines extend over the full age range of reported onset, the interior line in the box depicts the median age of onset, and the left and right box sides indicate the ages by which 25% and 75% (respectively) of involved youth experienced onset of a specific behavior. The researchers preferred to focus on the median age, which reflects the point in time by which 50% initiated a specific behavior. The number provided in parentheses indicates how many of the sample of 506 oldest boys reportedly ever engaged in a behavior category (Loeber et al., 1993).

Based on retrospective and prospective data, the median ages of onset for the oldest sample show that stubborn behavior tended to occur earliest at median age 9, with a wide range of onset -- the 25th percentile at age 3 and the 75th percentile at age 13. This was followed by minor covert acts, such as lying and shoplifting, at median age 10. Defiance, which involves doing tasks in one's own way, refusing to follow directions, and disobeying, emerged next at median age 11. Aggressive behaviors, such as bullying and annoying others, followed at age 12, along with property damage, such as vandalism and firesetting. More seriously aggressive acts, such as physical fighting and violence, came last at a median age of 13. Also at that age, authority avoidance, such as truancy, running away, and staying out late at night, emerged.

It should be noted that many children in the oldest sample at age 16 had not yet gone through the full-risk period. This implies that the median ages of onset are restricted and are likely to change when the age range is extended to include children who experience a later onset of problem behavior.

It was important to validate whether the developmental sequence applied equally to African-American and Caucasian boys. Substantial agreement was found between the ages of onset of problem behaviors for African-American and Caucasian boys across the samples.

Because of the earlier cited work (Frick et al., 1993; Loeber and Schmaling, 1985) on behavioral dimensions, the researchers next distinguished between the onset of authority conflict behaviors and overt and covert problem behaviors. They found that in each of the three dimensions of problem behavior, the onset of some behaviors occurred before the onset of others (Loeber et al., 1993; Loeber et al., in press). Specifically, the researchers found that for authority conflict behaviors before age 12, the onset of stubborn behaviors tended to occur before the onset of defiance and disobedience, which in turn occurred before the onset of authority avoidance (truancy, running away, and staying out late at night).

Likewise, the researchers found developmental sequences in covert problem behaviors, in that the onset of minor covert acts (lying and shoplifting) tended to occur before the onset of property damage (vandalism and firesetting), which in turn tended to occur before the onset of moderate to serious forms of delinquency (fraud, burglary, and serious theft, such as stealing a car).

In the same manner, the researchers documented developmental sequences in overt problem behavior, in that the onset of minor aggression (bullying and annoying others) tended to precede the onset of physical fighting (including gang fighting), which in turn tended to precede the onset of violence (rape, attacking someone, and strong-arming). These three developmental sequences were replicated across the three grade samples, which strengthened the findings.

Because developmental sequences are based on analyses of variables (i.e., behaviors), they do not reveal to what extent individuals progress over time from one behavior to the other. This aspect can be best captured by developmental pathways.

Previous

Contents

Next