潛在成長曲線模式常應用於縱貫資料，用以分析重複測量變數在觀測個體的起始狀態與觀測期間的成長率或衰退率的變化情形。本文旨在比較潛在成長曲線模式於不同的研究條件下，選模指標的統計檢定力及探討影響選模指標表現的因子。本文探究專為小樣本設計的統計量及訊息指標。由模擬結果發現，影響選模檢定力的主要因子為樣本數、截距與斜率的共變異數、及重覆測量的次數。就檢定力而言，整體上BIC選模的表現最好，其次為T□，接著依序為T□、T□、T₂*及AIC，而adjusted-BIC的表現最差。

Latent Growth Curve Modeling (LGCM) is often used to analyze the change or trend of repeated measures in longitudinal data. The objective of this study is to investigate the statistical power of model selection indices in Latent Growth Curve Modeling under different research settings. We studied several model selection indices specially developed for small sample sizes and compared their statistical power with information criteria. The results of the simulation indicated that the most influential factors are sample sizes, covariance of intercepts and slopes, and number of repeated measures of the observed variables. In terms of statistical power in model selection, overall, BIC has the best power, followed by T□, T□、T□、T₂* and AIC, while adjusted-BIC has the worst performance.