Abstract
Because considerable prior research on open education has produced inconclusive or conflicting results, the present investigation was designed to overcome common weaknesses in past studies. A sample of 2175 students in 116 science classes provided their perceptions of actual and preferred classroom openness along five continuous dimensions (personalisation, participation, independence, investigation, differentiation) and responded to a cognitive and an affective outcome measure. Results replicated prior findings in that actual openness accounted for a significant amount of the variance in affective but not cognitive achievement. But actual‐preferred interactions also accounted for a substantial and significant amount of variance in both cognitive and affective outcomes. Findings supported the value of employing a person‐environment interactional framework in open education research, and suggested that actual‐preferred interaction could be at least as important as openness per se in predicting student outcomes in science classrooms.