METHOD
Participants and Procedures
The participants who took part in the Block Longitudinal Project (Block & Block, 1980; Gjerde, 1995) were recruited in preschool while attending either a parent cooperative or a university-run nursery school. The participants in this particular study were selected from 116 children who were recruited at age 3 and an additional group of 41 children who were added at age 4. Observer ratings in the Q-sort format were available for 102 of these 157 participants both in preschool and in young adulthood. This core longitudinal sample was later assessed at ages 5, 7, 11, 14, 18, 23, and 32, with observer ratings obtained at all ages with the exception of ages 5 and 32. At each age, the participants were seen at multiple occasions and completed a variety of tasks.1 Both as preschoolers and adolescents, most participants lived in urban settings. About two thirds of the participants are European American, one quarter are African American, and one twelfth are Asian American. Although this sample is heterogeneous with respect to social class and parental education, it is skewed toward middle-class socioeconomic status. For the core longitudinal sample included in this study, the average Duncan (1961) socioeconomic status score for fathers was 642.88 (SD = 215.95) when the participants were 3 years old. To ensure that the sample used in the present study did not differ substantially from the age 3 sample of 116 children, we compared the participants in this study with those excluded from the larger sample on ego resiliency, ego undercontrol, and parental socioeconomic status. Identical analyses were conducted for the age-4 sample that contained 41 additional children. None of these comparisons revealed significant differences. Nursery school teachers described the participants’ personality using the California Child Q-sort (CCQ) at pages 3 and 4 (Block & Block, 1980). Examiners used the California Adult Q-sort (CAQ) to describe the participants’ personality at age 14 (Block, 1961/1978). The participants also completed a variety of experimental procedures assessing four themes: ego resiliency, ego undercontrol, cognitive maturity, and socioemotional functioning (see Block & Block, 1980, for an overview of the measures administered and the CCQ). Each child was seen 10 times for a period between 15 and 25 min at the Harold Jones Child Study Center associated with the University of California at Berkeley. The procedures, conducted in rooms specially equipped for the testing of preschool children, were introduced as games or fun activities. The examiners were paid research assistants and Ph.D. psychologists, all of whom had received careful training in administering the tasks. The experimental data reported in this study were collected when the children were 3 years old.
Measures
Measuring child personality: The CCQ. Three nursery school teachers used the CCQ (Block & Block, 1980) to evaluate each participant at age 3. Three different nursery school teachers evaluated the children at age 4. Each teacher had known the child for at least 6 months. The CCQ is widely used in developmental and personality psychology (e.g., Gjerde, 1995; Hart et al., 1997; Robins et al., 1996). The assessors were provided with 100 statements about the characteristics of children, each printed on a separate card. The assessors described each child by arranging theCCQitems into a forced distribution using nine categories, ranging from highly uncharacteristic to highly characteristic of the child being described. The CCQ descriptions were aggregated for each child separately at each of the two ages. The two aggregates were then aggregated and this final age-3/age-4 aggregate was used in subsequent analyses. The alpha of theCCQitem, based on correlations among nursery school teachers, averaged .65 both at age 3 and age 4. The average CCQ item reliability was .72. TheCCQis an ipsative procedure; there is a “scale for every individual and a population of an individual trait score is distributed about that individual’s mean” (Guilford, 1952, p. 30). A Q-item reflects the salience relative to other Q-items with reference to a particular person. Thus, the Q-sort can be said to provide personcentered rather than variable-centered information (Block, 1971).
All assessors sorted the CCQ items into a forced distribution that stipulates the number of cards included in each of the nine piles. The use of this identical scaling method provides all participants with identical means and standard deviations across the 100 items. Thus, in inverse factor analysis, standardizing the columns in the transposed matrix is unnecessary. This is important because correlations among persons are influenced by differences among variable means (e.g.,Waller&Meehl, 1998). In sum, the forced distribution Q-sort method is uniquely suited for inverse factor analysis (Asendorpf & van Aken, 1999). Measuring preschool ego resiliency and ego undercontrol using the CCQ. Three psychologists used the CCQ to describe a prototypical ego resilient child (α = .91) and a prototypical ego undercontrolled child (α = .90). Each participant’s CCQ profile was then correlated separately with each of the two personality prototype definitions. If this correlation was high, the participant could be described in terms of this construct; if this correlation was low or negative, the participant was dissimilar from this construct (see Onishi, Gjerde, & Block, 2001, for a detailed description of this method). For the participants included in this study, the average ego resiliency score was .45 (SD = .32) and the average ego undercontrol score was .08 (SD = .32). These scores—based on the average correlation of the two prototype definitions with each participant’s overall Q-sort profile across the sample—are comparable to those reported by Robins et al. (1996) and Hart et al. (1997): .40 and .36 for ego resiliency and .06 and .09 for ego undercontrol, respectively.
Measuring adolescent personality: The CAQ. The CAQ shares the same psychometric principles as the CCQ, but the 100 items are written to reflect characteristics of adolescents and adults. At age 14, five advanced graduate students used the CAQ (Block, 1961/1978) to describe the personality characteristics of each participant. Prior to the assessment, the graduate students received thorough training in the Q-sort methodology and the content of each item was discussed. The assessors described the participants after having observed them performing in a variety of experimental procedures on four occasions, each lasting between 2 and 3 hr, in a laboratory setting. An interview also was conducted by one of the assessors. These five independent CAQ descriptions were averaged for each participant and this aggregate was used in subsequent analyses. Developing adolescent personality constructs using the CAQ. Knowledgeable psychologists used the CAQ to provide definitions of the following personality constructs: ego resiliency, ego undercontrol, shyness, and activity level. Shyness and activity level were included because they relate directly to ego undercontrol. Expert descriptions also were obtained for the Big Five trait dimensions: Agreeableness, Extroversion, Emotional Stability, Conscientiousness, and Openness to Experience. No psychologist defined more than one construct. Scores for each participant on each of these personality constructs were developed using the method described above for the CCQ-based measures of ego resiliency and undercontrol. The alpha reliabilities exceeded .85 for all constructs.
Measuring Preschool Ego Resiliency Using Laboratory Measures The motor inhibition task. This task (Maccoby, Dowley, Hagen, & Degerman, 1965) measured the child’s ability to change from a modal tempo to a slower tempo according to instructions. The child was first asked to draw a line, walk, and wind a toy without instructions, followed by instructions to perform these activities “very, very slowly, just as slowly as you can.” The final inhibition score is a composite regression-adjusted score that reflects how long it took the child to complete the task under the slow-speed instruction conditions, controlling for the child’s speed under the regular (or noninstruction) conditions. Estimated time was approximately 12 min.
The egocentrism task. This task (Flavell, Botkin, Fry, Wright,&Jarvis, 1968) assessed a child’s ability to predict which objects other persons can or cannot see. The examiner covered different parts of a double-sided picture board and asked the child what the examiner could see. An egocentric response was defined as one in which the child chose a picture that could be seen from his or her side but not from the experimenter’s side. Estimated time was approximately 5 min. The Lowenfeld mosaics test. This task (Lowenfeld, 1954) assessed the aesthetic quality of mosaic constructions. The child was given a tray and 132 geometric shapes and asked to “do something with the pieces on this tray.” The mosaics were evaluated both for imaginativeness and structure. In this study, we only used the imaginativeness ratings thought to reflect ego resiliency. Four judges, whose ratings were subsequently composited, scaled the mosaic photographs for imaginativeness using a sevencategory Q-sort. Estimated time was approximately 12 min.
Measuring Preschool Ego Undercontrol Using Laboratory
Measures
The actometer task. This task (Schulman & Reisman, 1959) was designed to measure activity level. An actometer is a self-winding watch activated by a child’s
movements. The participants wore the actometer on the wrist of the nonfavored hand for 2 hr. Four actometer measurements, each separated by 1 week, were conducted and then composited. The delay-of-gratification task. In this task (Funder, Block, & Block, 1983), the child received an enticingly wrapped gift. He or she was then given a puzzle task and told that she or he could have the gift later. The delay-ofgratification score is a composite based on (a) the number of verbal behaviors toward the gift, (b) the number of physical behaviors toward the gift, (c) the time before the child took the present, and (d) the swiftness of opening the present. Estimated time was approximately 5 min. Wechsler intelligence tests. The Wechsler Preliminary and Preschool Scale of Intelligence Test (WPPSI) was administered at age 4, the Wechsler Intelligence Scale for Children (WISC) was administered at age 11, and the Wechsler Adult Intelligence Scale-R (WAIS-R) was administered at age 18 (Wechsler, 1981). Drug usage. The videotaped age-14 interview included questions about a wide variety of issues (e.g., family relations, peers, and schoolwork). The participants also were asked about their usage of illegal drugs, including both marijuana and “harder drugs,” such as heroin and cocaine. Marijuana use was measured on a 6-point scale: never used marijuana (0) to used more than once a week (6).
“Harder drugs” was measured on a 2-point scale: never used (1) to used once or more (2). Two raters who showed near exact agreement separately coded drug use. (Block, Block, and Keyes [1988] provide a complete description of the development of the drug measures and relevant descriptive statistics.) We initially conducted separate analyses of marijuana and harder drugs use. Because the relations between the personality prototypes and the two drug measures were virtually identical, we composited the two drug measures (α = .91).