「兒童五大人格特質量表」之信效度及測量不變性：探索性結構方程模式取向

Reliability, Validity, and Measurement Invariance of Children's Five-Factor Personality Scale: An Exploratory Structural Equation Modeling Approach

本研究旨在編製適用於國小五年級學童之「兒童五大人格特質量表」，比較驗證性因素分析（CFA）與探索性結構方程模式（ESEM）的模式適配度、檢視量表之因素結構，並檢驗量表的跨性別測量不變性。本研究依據人格五因素理論編製「兒童五大人格特質量表」，在預試方面，以290名國小五年級學童為對象，經項目分析篩選題目；在正式樣本方面，採分層隨機叢集抽樣，抽取全臺灣28所學校共2,056位國小五年級學童為受試對象，隨機平分為2組樣本，以樣本一比較ESEM與CFA模式的適配度，檢視量表的因素結構，據此修正模式與篩選試題，再以樣本二交叉驗證此修正模式的適配性，並進行跨性別的測量不變性檢定。研究結果顯示，「兒童五大人格特質量表」56題修正模式的因素結構符合理論假設，ESEM模式具有良好的信效度，且多群組ESEM分析結果亦顯示量表具跨性別測量不變性。

The purpose of this study was to develop the Children's Five-Factor Personality Scale and to test its reliability, validity and measurement invariance across gender. Due to the overly restrictive assumptions, confirmatory factor analyses (CFA) typically fail to support the five-factor structure of personality. We show that an approach of the exploratory structural equation modeling (ESEM), an integration of CFA and exploratory factor analysis (EFA), can overcome these problems. The items were originally developed based on the five-factor theory of personality and related literature, and were revised according to experts' comments. A pilot study with item analyses was conducted to evaluate psychometric quality of items and to select items. Later, using stratified cluster sampling method, a sample of 2,056 5th graders was collected from 28 elementary schools in Taiwan. An ESEM was conducted to exam the factor structures and compared with the CFA model. The final version of the scale includes 56 items. The construct validity, convergent validity and measurement invariance across gender were examined and discussed in the study. The results show the scale exert good reliability and validity with measurement invariance across gender.