Guy Bégin

Temporal Reliability and Convergent Validity of the Rating-Scale and the Peer-Nomination Sociometric Measures among Preschool and Kindergarten Children

An improved peer-nomination procedure (training plus more votes) and the rating-scale were compared to assess the test-retest reliability, and the convergent validity of both methodologies for 23 preschoolers (4- to 5-yr.-old) and 41 kindergartners (5- to 6-yr.-old). Five testing periods with both procedures were scattered over the entire school year. The results indicate that the peer-nomination procedure and the rating-scale reach acceptable and comparable levels of reliability among kindergartners. However, among preschoolers the rating-scale works a little better, and reliability is moderate. Data on convergent validity show both methodologies assess quite similar aspects of the group structure in the two groups studied. While the use of the rating-scale is more appropriate among preschoolers, the improved peer-nomination technique is recommended among older children in terms of temporal reliability, validity, and flexibility of descriptive possibilities.

COMPARISON OF DATA GATHERED ON SENSITIVE QUESTIONS VIA DIRECT QUESTIONNAIRE, RANDOMIZED RESPONSE TECHNIQUE, AND A PROJECTIVE METHOD

1266 students participated in an interview on sensitive issues. Three very sensitive hypothetical situations were assessed by means of three methods, the direct questionnaire, the randomized-response technique, and a “most people” projective-form questionnaire. The data indicate that the first two methods yielded similar proportions of endorsement. However, the projective method yielded much higher proportions of endorsement. These data indicate that the former two methods seem to assess the same concept which is quite different from the one measured through the projective questionnaire. Since the randomized-response method assures the subjects of the confidentiality of their responses, the authors believe that the direct questionnaire and randomized-response data are closer to reality, that is, more valid.

SEX MAKES A DIFFERENCE: 1. EVIDENCE FROM A MODELING STUDY CONDUCTED IN A NATURAL SETTING

The present experiment examined modeling effects on petition signing as an index of helping behavior. 263 naive subjects participated. It was predicted that, when the model agreed to sign, more subjects would agree to sign than in the control condition and, when the model refused to sign, fewer subjects would sign than in the control condition. The predictions were confirmed when the requester and subjects were of the same sex. When they were of the opposite sex, however, no model was needed. Finally, a cross-sex effect was found between the subjects and the requester.

Sociometric status and social interaction: are the neglected children socially less active?

Previous studies have clearly established that isolated children display less social behaviors than popular ones and that poorly accepted as well as isolated children are at risk for future adjustment problems. These findings prompt the author to predict that children of the various sociometric statuses should display differential rates of social interaction with peers. Predictions were tested by administering during the second term to 99 kindergarten children a sociometric measure and by videotaping each child for 1 hr. during free play. Whether or not the child was in social interaction with peers was subsequently coded at 5-sec. intervals by two observers. Sociometric scores and inter- and intraobserver agreements were high. Popular children were the most socially active, followed by the average, the rejected, and then the neglected children, which confirmed all hypotheses. These results give strong support for the validity of the difference between the neglect and rejected vs the popular and the average children and suggested that such work must include the childs rate of social interaction with peers as a covariable in the analyses to identify differences between statuses.

COMPARISON OF INTERJUDGE RELIABILITIES IN SCORING TAT PROTOCOLS AS A MEASURE OF NACH

Several indices of interjudge reliability are discussed with respect to the TAT-scoring methodology. Some improvements are suggested for the indices proposed by McClelland, et al. (15). Two indices of proportion of agreement are proposed. A comparison of McClelland, et al.s indices and those of proportion of agreement shows that the proportion of agreement in scoring subcategories is more severe and suggests that it should be added to evaluate more tightly interjudge reliability in scoring the TAT.

Sensitive Data Gathering through the Random Response Technique: A Validity Study

To test the validity of the data generated by the random response technique, 204 subjects participated in an electoral survey. Three simulations completed through the technique replicated accurately the results of the survey.