Infant temperament measured by multiple observations and mother report.

Observers and mothers rated infant behavior (n = 50) in the home on dimensions of temperament once a week for 8 weeks. Although week-to-week correlations were modest (intraclass correlations of .14-.36), aggregates of the 8 observations had high reliability for both observers and mothers. Mother reports were tied to our observation sessions by having mothers (a) rate their infants' behavior during the period when our observations were made and (b) use a questionnaire that mirrored the scoring system used for scoring the videotaped observation sessions. When direct observations were compared with mother reports (on the aggregated weekly reports and on 4 widely used questionnaires), little evidence of mother-observer correspondence was found. The interpretation of the large literature that has used maternal report is discussed, as well as the importance of direct observation of infant behavior when temperament is assessed.