Causal inferences and the positivity bias in children: The role of the covariation principle

It is argued that developmental studies regarding Kelley's covariation prin ciple had provided children with incomplete covariation information, and th at the assessment of attributions differed conceptually from Kelley's model and pertinent adult studies. It is further proposed that younger children' s tendency for positive self-evaluations ("positivity bias") is due to thei r inferences not yet being based on the mechanism described by the covariat ion principle. In 2 studies, conducted in the United States and Germany, pa rticipants of different ages received complete covariation information and were asked for attributions, ability estimations, and expectations of succe ss. Children from grade 2 on selected the correct cause(s) when the effect covaried only with one cause, but only at a later age when covariation with 2 causes was presented. Ability estimations and expectation of success wer e more positive in younger than in older participants, replicating the "pos itivity bias." However, children did not simply become more negative ("pess imistic") with age but rather more realistic: Estimations of ability decrea sed when failure covaried with the person but not when there was covariatio n with the entity; expectation of success decreased when failure covaried w ith a stable cause (person or entity) but not when no covariation informati on was provided, and even increased when failure covaried with the circumst ances (a variable cause).