CHILD STUDY JOURNAL/Volume

20/No. 3/1990

Page 139

CAUSAL ATTRIBUTIONS IN YOUNG CHILDREN Robert D. Friedberg Mesa Vista Hospital Constance J. Dalenberg California School of Professional Psychology, San Diego

Previous research has not fully examined preschoolers' attributions. Accordingly, the purpose of the study was the investigation of the causal explanations children use to account for common occurrences. In this study, 60 preschool children were presented with videotaped puppet shows designed to elicit causal attributions. Children viewed 12 vignettes across three domains (cognitive, physical, social) and two outcomes (success, failure) and then answered a question tapping their causal beliefs. Overall, as hypothesized, the results revealed that most children predominantly used internal, unstable, and specific attributions. Further, as predicted, the presence of a self-enhancing bias emerged. Children became more external and unstable in their attributions in response to failure. Implications for further research and clinical practice are discussed.

As any visit to the psychology, self-improvement, or even occult section of a local bookstore will attest, people seem compelled to search for the answers to life's fundamental "Why?" questions. Causal attributions are inferences which serve people's need to understand, organize, and form rules regarding everyday events (Fincham, 1983; Harvey & Geary, 1981). Children construct causal schemes beginning in infancy and increasing in complexity during their schoo Iyears (Abramo vitch & Freedman, 1981).It is important to explore young children's early understandings of their world, particularly because these understandings may predict social and personal well-being. Research examining attributional processes in preschool and kindergarten children has been equivocal and controversial. Some authors (Rholes, Blackwell, Jordan, & Walters, 1980; Rholes &

Page 140

Friedberg

&

Dalen berg

Ruble, 1984; Ruble & Rholes, 1981) have contended that due to cognitive immaturity children's attributions are not important mediators of behavior whereas others (Dalenberg, Bierman, & Furman,1984;Karniol & Ross, 1976;Kassin, 1981;Miller, 1985)have argued that children have a surprising degree of sophistication if asked the right questions. Accordingly, preschoolers' attributions are a function of their level of cognitive sophistication and their social experience (Dix & Herzberger, 1983). Therefore, their socialcognitive skills differ as a result of the kind of socialization agents, activities, and tasks involved (Higgins & Wells, 1986).The presence of the halo schema (Kun, 1977), additive effect (Karniol & Ross, 1976),discounting (Dalenberg et al., 1984),actor-observer differences (Abramovitch & Freedman, 1981), and sex differences (Lochel, 1983) in attributions points to a level of causal sophistication that merits further attention. The equi vocal results of previous studies examining children's causal ascriptions can be traced to a lack of adequate ways of measuring young children's attributions. Berndt and Heller (1985) noted that much of the research confounds age-related verbal fluency with social-cognitive capaci ties. Further, overly demanding social-cognitive tasks typically lead to egocentric errors (Flavell, 1985). Although measures for children's internal-external orientations exist (Mischel, Zeiss, & Zeiss, 1974; Nowicki & Duke, 1974), they are flawed representations of children's causal beliefs. These instruments confuse the effects oflocus and stability (Weiner, 1979). Further, these measures fail to reflect children's familiar experiences by not including task- and domain-specific experiences and success and failure outcomes for each situation. The attributional dimensions of internal/ external, unstable/ stable, and global/specific are well-documented in the literature examining causal ascriptions in elementary school children and adults (Abramson, Seligman, & Teasdale, 1978; Kaslow, Rehm, & Siegel, 1984;Nolen-Hoeksema, Seligman, & Girgus, 1986;Seligman, Peterson, Kaslow, Tannenbaum, Alloy, & Abramson, 1984). The objective of this study was the extrapolation and downward extension of Seligman's et al. attributional dimensions to preschool and kindergarten children's causal reasoning. This project also

CHILD STUDY JOURNAL/Volume

20/No. 3/1990

Page 141

investigated whether the self-enhancing bias, a fundamental attributional phenomenon demonstrated in older populations (Miller, 1976;Zuckerman, 1979),emergesinyoungchildren'scausal ascri ptions. It was hypothesized that overall, children would be (a) more internal than external, (b) more unstable than stable, and (c) more specific than global in attributional style. Further, it was expected that children would be more external, unstable, and specific in response to failure outcomes.

Method Subjects

.

Sixty preschool and kindergarten children participated in the study (Mean Age = 57.9 mos., SD=5.45). There were 32 boys (Mean Age =5825, SD=4.49) and 28girls (Mean Age = 57.57, SD=6.44). The children were predominantly Caucasian and drawn from seven area preschools .

'

Instruments The Preschool Attributional Puppet Show (PAPS), designed for this study, assessed preschoolers' causal ascriptions. It consisted of 24 puppet skits composed of four to five sentences and was presented via videotape. Following each skit, the subjects' task was to attribute a cause to an outcome by answering an open-ended "Why" question. The vignettes used in the show were culled from the responses of three experienced preschool educators (Mean yrs. of experience = 13.3 yrs) who completed a survey on the ten most frequent cognitive, physical, and social experiences a preschooler encounters. All subjects were presented with 12 vignettes. The cognitive vignettes included (a) completing a puzzle, (2) remembering words to a song, (3) sorting bu ttons, and (d) building a tower. The physical vignettes included (a) running a race, (b) climbing, (c) catching a ball, and (d) kicking a ball. The social vignettes included (a) sharing, (b) recei ving help, (c) waiting a turn, and (d) social acceptance. Each vignette had a successful and unsuccessful outcome yielding two

Page 142

Friedberg & Dalenberg

sets of 12 vignettes. Thus, each child responded to four outcomes (success, failure) across each domain (cognitive, physical, social). Sample vignettes are presented in the Appendix. To facilitate identification and simplify the task, the puppet and the child were the same gender. The boy and girl puppets looked different and had distinctive visual cues associated with them. Administration time for the PAPS was approximately 25-28 minutes.

Procedure During the first testing session, each subject viewed the PAPS. An informal warm-up period, which served as a rapport builder and introduction to the task, preceded the formal assessment of causal beliefs. Before each videotaped vignette began, the child was encouraged to pretend he/she was the key puppet in the story. After each story, recall was tested. Children were allowed to hold the puppet presented during the show to increase their attention span, heighten their identification, and make the situation more real for them (5.E. Denham, personal communication, April 6, 1986). When children responded with "1don't know" answers, the interviewer tested the limits by asking, "If you did know, why ..?" or "Tell me a reason."! During the second testing session, the children were administered the Peabody Picture Vocabulary TestRevised (PPVT-R) (Dunn & Dunn, 1981) as a vocabulary check.

Coding of Responses Children's responses were coded on a two point scale along three attributional dimensions by three trained raters blind to the hypotheses. The raters were three undergraduate students with experience in teaching roles with children. Reliability wascalculated as the total agreements divided by the total agreements plus disagreements. Reliability on the locus dimension between raters 1 and 2 was .97, between raters 1 and 3 was .96, and between raters 2 and 3 was .97. Consistency between rater 1 and rater 2 remained strong across the other two dimensions (Stability, .80; Globality, .92). Rater 3 had difficulty agreeing with the other two judges on the stability dimension (Rater 1, .45; Rater 2, .48), but did reach an

CHILD STUDY JOURNAL/Volume

20/No. 3/1990

Page 143

adequate level of consensus on the globality dimension (Rater I, .65; Rater 2, .68). Children were classified along the attributional dimensions based on the frequency of their causal ascriptions. In order to be classed inan attributional dimension (e.g., internal or external), the majority of a child's attributions (e.g., 60% or greater) must have fallen into that particular attributional category. Children who did not show a dominant attributional style were placed in a mixed category. Children's sample responses are presented in Table 1. Results Differences in Causal Attributions The average child produced codable responses to a majority of vignettes (Mean Response Percentage = 77.8%). This, in itself, is an important finding because in previous research children's uncodable and '1 don't know" responses accounted for a larger proportion of the responses (Frieze, 1981; Lochel, 1983). Children were not included in the statistical analyses if their overall response percentage (the number of cod able responses divided by the number of vignettes) was less than 25%. One-way, one-tailed chi square analyses revealed that the number of children who explained ou tcomes with an internal style (n=29) was significantly greater than the number of children who used either external (n=8) or mixed (n=21) styles to account for outcomes, X2(1,N=58)=5.84, p