中華心理學刊 民 91，44 卷，2 期，227-238
Chinese Journal of Psychology 2002, Vol.44, No.2, 227-238
陳欣進（空軍官校心理諮商科）；簡伯丞（空軍官校心理諮商科）；鄭秋瑾（空軍官校心理諮商科）；吳瑞屯（國立台灣大學心理學系）
摘要
相對於以因素分析方法為基礎發展的測驗，訊息處理理論比較能夠透過實驗法的因果推論幫助瞭解各訊息處理階段的特性與運作方式。Sternberg（1977）以訊息處理理論為基礎，發展了能分析訊息處理歷程的的成分分析（componential analysis）方法。本研究以飛訓學生為對象進行心像旋轉能力測量，並且對於測量結果進行成分分析。研究結果顯示，本研究所採用的心像旋轉作業具有相當高的內效度，並且能夠有效區分不同階段的訊息處理歷程。在飛訓甄選的運用上，通過初階飛行訓練的學生，在表徵刺激複雜度的轉折點參數估計上未達顯著，但是沒有通過的學生則有顯著的參數估計。而當心像旋轉作業各成分分析參數皆放入空軍官校現行飛訓甄選模式的計算時，轉折點參數也是第三個進入模式的變項。上述訊息顯示，本研究發展的心像旋轉能力測量方式可以是一項甄選飛訓學生的有效測驗，同時複雜圖形的處理效率可能會是一項具有高鑑別力的甄選指標。
關鍵詞：成分分析、心像旋轉、飛訓甄選、測驗設計
A COGNITIVE APPROACH TO MENTAL ROTATION TESTING ON PILOT SELECTION
Hsin-Chin Chen（Counseling and Psychology Research Center, Chinese Air Force Academy）；Po-Cheng Chien（Counseling and Psychology Research Center, Chinese Air Force Academy）；Chiu-Chen Cheng（Counseling and Psychology Research Center, Chinese Air Force Academy）；Jei-Tun Wu（Department of Psychology, National Taiwan University）
Abstract
Many studies had demonstrated that it is necessary to include spatial ability tests in pilot selection test battery, but the process underlying traditional mental rotation tests used for distinguishing appropriate pilot trainees may not easily be obtained, because scores of traditional tests designed by the factor analysis paradigm combined with many processes other than that we are concerned, thus faded the measurement. To overcome this problem in pilot selection, instead of the factor analysis approach, an experimental procedure was called for to extract the crucial process that could discriminate suitable pilot trainees. The present study adopted a method similar to 'componential analysis' by Sternberg (1985), an information processing paradigm, to design a mental rotation task to use in pilot selection of trainees at the Chinese Air Force Academy. Subjects were told to make a judgment whether two blocks showed on the computer screen were the same when rotated. Three within-subject variables and one between-subjects variable were manipulated. The only between-subjects variable was the kind of pilot trainees (pass or fail), three within-subject variables were rotation angles (60, 120, 180, 240, or 300 degree), the way stimulus being displayed (2 blocks presented at the same time or one after the other), and the number of block reflections (two or three). Four model parameters were estimated in the present componential analysis, labeled 'display', 'reflection', 'rotation', and 'constant' respectively. Among these parameters, 'display' which represents the difference of processing time between two ways of stimulus display was related to the capability of encoding spatial stimulus, 'reflection' which represents various processing time between two and three reflections was related to the processing efficiency in complex stimulus, 'rotation' which represents the processing time increased when added rotation angles was related to the processing efficiency in rotating stimulus, and 'constant' was related to all other processing not included in any of the other parameters. The results showed that the task adopted has a good internal validity. Compared to those trainees having passed the basic flying training, the data showed that those who did not pass obtained significant estimations on the parameter regarding 'reflection' which indicates complexity of stimulus. In case of including parameters estimated by componential analysis into the parameter pool of a quantitative stepwise selection model established by the Chinese Air Force Academy, 'reflection' was the third variable chosen for the selection model. As the data showed in this research, cognitive approach test of mental rotation could be a useful tool among others on pilot selection, with the processing efficiency in rotating complex stimulus as a discriminative index on pilot selection.
Keywords：Componential analysis, Mental rotation, Pilot selection, Test design
