Reduction of demand characteristics in the measurement of certainty during modeled conservation

The purpose of the present experiment was to document the cognitive processes mediating the observational learning of conservation. Two formats for measuring certainty were contrasted among nonconservers assigned to one of give groups. Under a public format, two groups used a certainty rating scale and showed the experimenter how sure they were about the correctness of a peer model's statements. Under a private format, two groups employed the same scale but were assured that their rating was not known by the experimenter. Within each format, one group was presented with the model's conservation answers and one group with the model's opinions on arbitrary matters. Conservation was demonstrated to the fifth group which did not perform the certainty assessment. All conservation groups were post-tested immediately after modeling and 6 weeks later and all groups displayed equivalent stable learning. Whereas public certainty was constantly high, private certainty increased as the model proceeded through conservation exemplification, in accordance with a hypothesis of sequential imitation and comprehension processes. Private certainty was also higher for conservation than for opinion, although not on the first two items. Public certainty was high regardless of modeled content, and public rating latency was always found to be shorter. These results strongly suggested the presence of demand characteristics for high undifferentiated ratings under the public format of certainty appraisal. This contamination prevented valid monitoring of the course of cognitive rule processing.