测量中的生成解释与选择解释──对美国加州学生和中国香港学生的研究结果

Generated-versus Selected-Explanations: Results from California and Hong Kong

解释可以用来检验学生的推理在多大程度上基于科学知识和证据, 本文研究的是学生在两类经典多选题中的解释是否可以测量他们的科学推理以及测量的有效性问题。本研究在传统多选题中附加了两类解释项目──成解释和选择解释,用以了解学生做出选择的原因。这项研究刚试了794 个美国加州学生和916 个中国香港学生, 检验了三类题目( 传统多项选择、成解释、选择解释) 的心理测量学特征, 并且比较了美国加州学生和中国香港学生在三类试题上的表现。为了检测不同国家的潜在偏见, 本文还进行了题目功能差异分析。研究结果显示美国加州学生的分数更高, 并且无论时美国加州学生还是对中国香港学生, 生成解释比另外两类题目难度都要大。本文最后从评估和课程的角度对表现上的差异进行了讨论。

Explanations can be used to examine the extent to which students' reasoning is based on scientific knowledge and evidence. This study investigates whether and how effectively explanations added to typical multiple choice (MC) items can measure student scientific reasoning. Two types of explanation items were used and compared: generated-explanation (GE) and selected-explanation (SE) items. Both explanation items were added to traditional MC items in eliciting student justification of their choices on the MC items. The assessment was administered to 794 Californian students and 916 Hong Kong students. We examined the psychometric properties of the three item types (MC, GE, and SE) and compared student performance on these three item types between Californian and Hong Kong students. A differential item functioning analysis was condueted to detect items with potential bias. Results show that GE items are more difficult than the other two item types for both Californian and Hong Kong students and tend to favor Californian students. We discuss the performance difference from assessment and curriculum perspectives.