假設檢定策略的學習與遷移效果

Abstract

自Wason (1960) 的研究發現人在假設檢定的過程中有證真偏好的傾向後，有相當多的研究致力於去除證真偏好，但成效不一。本研究的目的即在探討證偽式假設檢定策略 (falsificatioist strategy) 的學習效果，及其近程與遠程的遷移效果。共有66名大學生被隨機分派至三種策略教導（證真式、證偽式、及控制組）的實驗情境中。在以三題系統測試作業做為策略教導的練習題後，所有受試者均接受二題系統測試作業的測試，以衡量假設檢定策略的學習效果。接下來以一題與系統測試作業不具有表面相似性、但實際卻其有相同問題結構的錯誤診斷作業做為衡量近程遷移的效果。最後，以一題與系統測試作業在表面相似性、及結構相似性都偏低的10-20-30法則發現作業做為衡量遠程遷移的效果。結果顯示，證偽式假設檢定策略有顯著的學習效果；近程遷移的效果雖非百分之百，但仍顯著。可是遠程遷移的效果則不明顯。根據研究的結果，我們對假設檢定行為的認知歷程，及證真偏好的現象提出一些解釋。

The learning and transfer effects of an instructional model for falsificationist hypothesis-testing strategy was evaluated in the present study. Sixty-six college students were randomly assigned to three experimental conditions and received different instructions on hypothesis-testing strategy (falsificationist, verificationist, vs. control) with three system testing tasks. After instruction, learning effect was examined with two system testing tasks similar to those used at the instructional phase. A fault diagnosis task which was isomorphic to the learning tasks was used to examine the near transfer effect. A rule discovery task (10-20-30 task) similar to Wason's 2-4-6 task was used to examine the far transfer effect. Results showed that instruction for the falsificationist strategy was effective on learning and near transfer. However, no far transfer effect was observed. Implications in cognitive process of hypothesistesting behavior and confirmation bias were discussed.