適用於認知心理學實驗室之變異數分析暨趨勢分析程式

ANOVA and Trend Analysis Statistical Program for Cognitive Experiment

本文介紹專為認知心理學實驗室資料所設計之電腦統計程式。大多數的認知實驗室資料均可循相似的統計程序進行分析，故只要將資料依特定的格式輸入，不需任何的指令，本軟體便可在一分鐘內完成如下的統計分析作業。首先，程式所提供的多種資料轉換公式，能將實驗室常見的反應時間、百分比或頻次資料進行轉換，以符合變異數分析之假設。接著，程式會執行最高16因子且涵蓋各式實驗設計（全受試者內、混合設計、全受試者間）之變異數分析，並自動對達到顯著水準的五階以下交互作用效果進行一系統的低階單純主要效果或單純交互作用效果分析。若主要效果或各階的單純主要效果達顯著，則程式會繼續執行杜凱氏（Tukey）事後多重比較，並且計算出不等距的趨勢分析正交對比係數，以及進行最高10 次方之趨勢分析。為檢核程式的精確度，作者輸入三本統計教科書中30多個範例，結果發現誤差均在小數點一位以下，此應可確保統計結果的正確性。

This article introduces a statistical software specially designed for cognitive experiment. This program provides several data transformation formula to guarantee that reaction time or proportional datum which frequently observed in laboratory could comply with the hypotheses of ANOVA. In addition, this program can conduct as many as sixteen factors’ ANOVA which combine complete between-subject, complete within-subject or mixed experiment design. If any interaction effect less than five orders is significant, its simple main effects or simple interaction effects would be computed automatically. Furthermore, This program would execute Tukey HSD posterior multiple comparisons and trend analysis consequently for any main effect or simple main effect which reaches statistical significant level.