A Comparison of Partial and Complete Paired Comparisons in Sociometric Measurement of Preschool Groups

Abstract

Low test-retest reliabilities over periods from ten days to five months have been obtained on a par tial-rank order sociometric, the PSI, of preschool- age children's peer preferences. These results have been interpreted to mean that preschool-age chil dren do not have stable and enduring friendships with their peers. An alternative possibility is that the reliabilities of partially ranked data are so low as to obscure the existence of stable individual friendships in this age group. A full-rank order so ciometric instrument, the PCST, utilizing color photographs of the children in a preschool group as aids in eliciting friendship choices from the chil dren, was developed and tested on a group of three- year-olds and a group of four-year-olds. The socio metric measurements from both the PSI and the PCST were most reliable for the four-year-old group. Correlations between the PSI and the PCST, when corrected for attenuation, revealed that the two measures were probably assessing the same peer choice behavior, although the PCST was markedly superior in reliability. Administration time for the PCST was higher but substantially less than for previous paired-comparisons procedures.