No abstract

A wide-spread stereotype that influences women’s paths into STEM (or non-STEM) fields is the implicit association of science and mathematics with “male” and with requiring high levels of male-associated “brilliance.” Recent research on such “field-specific ability beliefs” has shown that a high emphasis on brilliance in a specific field goes along with a low share of female students among its graduates. A possible mediating mechanisms between cultural expectations and stereotypes on the one hand, and women’s underrepresentation in math-intensive STEM fields on the other hand, is that women may be more likely than men to feel that they do not belong in these fields. In the present study, we investigated field-specific ability beliefs as well as belonging uncertainty in a sample of n = 1294 male and female university students from five STEM fields (Mathematics, Physics, Computer Science, Electrical Engineering, and Mechanical Engineering) at a prestigious technical university in Switzerland. Field-specific ability beliefs of both men and women emphasized brilliance more in more math-intensive fields (Mathematics, Physics) than in less math-intensive fields (Engineering). Women showed higher beliefs in brilliance than men did, and also reported higher levels of belonging uncertainty. For both genders, there was a small, positive correlation ( r = 0.19) of belief in brilliance and belonging uncertainty. A relatively small, but significant portion of the effect of gender on belonging uncertainty was mediated by women’s higher belief in brilliance.

In a quasi-experimental classroom study, we longitudinally investigated whether inquiry-based, content-focused physics instruction improves students' ability to apply the control-of-variables strategy, a domain-general experimentation skill. Twelve third grade elementary school classes (Mdnage = 9 years, N = 189) were randomly assigned to receive either four different physics curriculum units (intervention) or traditional instruction (control). Experiments were frequent elements in the physics units; however, there was no explicit instruction of the control-ofvariables strategy or other experimentation skills. As intended, students in the intervention classes strongly increased their conceptual physics knowledge. More importantly, students in the intervention classes also showed stronger gains in their ability to apply the control-of-variables strategy correctly in novel situations compared to students in the control classes. Thus, a high dose of experimentation had the collateral benefit of improving the transfer of the control-ofvariables strategy. The study complements lab-based studies with convergent findings obtained in real classrooms.

President of the German Society of Psychology. His main fields of research include technology supported individual and collaborative problem-solving and learning, computer-mediated communication, cognitive psychology, and environmental psychology. Inferences from distributed information in group problem-solving COLLABORATIVE PROBLEM-SOLVING WITH DISTRIBUTED INFORMATION: THE ROLE OF INFERENCES FROM INTERDEPENDENT INFORMATIONWe study inferences from distributed, interdependent information in group problemsolving. Three inference types (collaborative, individual, and shared) are distinguished based on information sharedness and distribution, with a special focus on collaborative inferences that generate new information no individual group member could have inferred. In an experiment, n=27 dyads solved a specifically designed inference task. Inferences from shared information were the most likely, individual inferences from unshared information less likely, and collaborative inferences from unshared, distributed information the least likely to be drawn. An analysis of inference patterns in dyads' discussions points towards the individual-and grouplevel processes involved in drawing collaborative inferences, and explains why first support measures explored in this study were not optimally designed.KEYWORDS group problem-solving; group decision-making; distributed information; inferences Inferences from distributed information in group problem-solving In our increasingly specialized world, complex problems are usually tackled by groups of persons from diverse knowledge backgrounds. In such groups, the information, ideas, perspectives, and expertise relevant for solving a joint problem are thus distributed across group members. When groups pool and integrate their members' knowledge efficiently, substantial synergies may be reached (e.g. Brodbeck, Kerschreiter, Mojzisch, & Schulz-Hardt, 2007;Stasser & Birchmeier, 2003). However, starting with the seminal work by Stasser & Titus (1985), a large body of research on decision making in small groups has demonstrated that groups focus on shared information, i.e. knowledge that is known to all members from the start, and neglect unshared information, i.e. knowledge that is known only to individual group members (for reviews see e.g. Mojzisch & Schulz-Hardt, 2006;Stasser & Birchmeier, 2003;Tindale & Kameda, 2000;Wittenbaum, Hollingshead, & Botero, 2004). As a consequence of this information pooling bias, groups consistently fail to detect the optimal solution in exactly those situations where they would have the greatest advantage over individual decision makers: hidden profiles. In a hidden profile, solution-relevant information is distributed in such a way that individual group members are led to prefer a suboptimal solution alternative, while only the combined information reveals the optimal solution (e.g. Stasser & Titus, 1985).In a hidden profile, each information item is known by at least one group member, and the main challenge faced by the group is to overcome members ...