Date of Award

Spring 5-1-1979

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Communication Disorders and Counseling, School, and Educational Psychology

First Advisor

Kenneth P. Walker

Second Advisor

Karl B. Zucker

Third Advisor

Walter L. Sullins

Abstract

This study was designed to investigate the effect of chi1dren 1s test behavior on examiners' scoring, questioning, and confidence in test resu1ts. The variables under investigation were leve1 of inte11igence and extraversion-introversion as operationally defined in terms of sociabi1ity, response speed, and persistence. E1even-year-old male accomplices read scripts containing responses to the Verba1 Scale of the WISC and operationally defined test behaviors. The test responses to the Verba1 Scale of the WISC were designed to yield superior (IQ= 128) or average (IQ= 97) test scores by varying the quality of test responses. Extravert-Introvert test behaviors independent of the scoring of the WISC protocol were introduced by operationa11y varying sociability, response speed and persistence. The degree of sociabi1ity was contro11ed by varying the number of comments which the chi1dren made which were not re1ated to test responses, with two comments indicating introverts and seven indicating extraverts. Response speed was operational1y defined as the lag between the end of the reading of the question and the start of the chi1d's response. Extraverts responded within five seconds while introverts took from 10 to 15 seconds to respond. Persistence was operationally controlled by varying the number of 11don't know11 responses and the number of responses which indicated that the child was pondering a response. Audio-recordings were played for thirty-two experienced and inexperienced school psychology graduate students at three Midwestern universities. Experienced examiners were those who had completed their final practicum course in school psychology, while inexperienced students were those who had completed only the course in individual intelligence testing. The criterion measures were the score given the twelve ambiguous responses, the number of questions per protocol, and the examiners' confidence in the accuracy of test results. Results showed that level of intelligence significantly (p:;:: .01) affected scoring, with 11 bright11 children receiving significantly more credit that average children for the same ambiguous responses. The number of questions per protocol did not differ significantly; however, there was a tendency (p = .10) for examiners to ask more questions of introverts than of extraverts. Experienced examiners were more confident (p < .03) in the test results than were inexperienced examiners. There was a significant interaction between levels of intelligence and extraversion-introversion. In the average group, examiners were more confident in the scores obtained by introverts. In the superior group, they were more confident in the scores obtained by extraverts. Behavioral expectancies for level of intelligence and extraversion-introversion test behavior were obtained by the examiners. An unexpected finding was that 11 bright11 children were perceived more favorably on all of the extravert-introvert behavioral expectancy checks. Since behavior was carefully controlled and equated across intellectual levels, this appeared to be a 11 halo11 effect operating in favor of high intelligence. The study clearly demonstrated that children's level of intelligence can bias the scoring of test responses, with bright children receiving more credit than average children. It further showed that even when test behavior is carefully controlled, a 11 halo 11 effect in rating can occur in favor of bright children.

Download

Share

COinS