Poster #201 - Evaluating the Parallel-Form Reliability of a Task Measuring Children’s Selective Learning From More Confident Individuals

• In Event: 1-115 - Poster Session 03
  In Poster Session: PS 03 Section - Social Cognition

Integrative Statement

From a very young age, children use different cues to decide whom to learn from. For instance, several studies have demonstrated that preschoolers prefer to learn from specific informants by using knowledge cues such as accuracy, visual access and confidence (see Mills, 2013, for a review). However, despite years of research on these topics, there are no studies evaluating the reliability of selective learning tasks. Our study intended to test whether children are consistent in their preference to learn from confident individuals. In other words, if one child strongly favours information from one confident individual in one task, can we assume that they are likely to show the same strong preference with a different confident individual on another task? Can we also assume that a child who shows no such strong preference on one task – or even prefers to learn from a more hesitant individual – would continue to do so when presented with different informants? We thus evaluated the parallel-form reliability of a task measuring children’s preference to learn from a confident rather than a hesitant individual.
Children aged 3 to 5 years (N=41) were introduced to videos of a confident and a hesitant model providing conflicting novel facts. In the first version of the task (form A), two models provided four novel facts about animals. In the second version of the task (form B), two different models provided four novel facts about different types of food. In each form of the task, children were asked to choose which of the answers provided by the two models they believed were right over four trials. These parallel forms were administered on the same day one after the other (order counterbalanced). Our results showed that children performed above chance (i.e., siding most frequently with the most confident individual) both on form A (M=2.59, SD=1.05, t=3.58, p=.001) and form B (M=2.41, SD=1.1, t=2.43, p=.02) of the task. Thus, it suggests that children were able to discern the models’ level of confidence, replicating past research. However, we found a positive yet only small-to-moderate correlation between children’s performance on forms A and B of the task (r=.31, one-tailed p=.026 – note that a one-tailed test is used here because there is a strong a priori expectation that two tasks measuring the same construct would correlate positively, not negatively). These results, while suggesting that children are somewhat stable across tasks in their preference to learn from confident individuals, also contradict the assumption that children would give highly consistent answers across different confidence-based selective learning tasks. This leads us to think that these tasks are not necessarily very reliable for the evaluation of individual differences and/or that performance on these tasks does not reflect a stable, situation-general attribute.

Authors

• Aimie-Lee Juteau, University of Ottawa
  Presenting Author

• Isabelle Cossette, University of Ottawa
  Non-Presenting Author

• Joana Ntunga Mukunzi, University of Ottawa
  Non-Presenting Author

• Patricia Brosseau-Liard, University of Ottawa
  Non-Presenting Author