統計顯著性在班級團體輔導研究中之意義

The Effects of the Appliance of Statistical Significance on Classroom Guidance

推論統計的顯著水準概念，一般都訂在.05 或.01，而這種限定是否真能應用在班級團體輔導或一般諮商的研究上？研究者得考慮採用某一種顯著水準可能導致研究結果決定之錯誤程度，更需深思統計檢定的考驗力高低程度。統計考驗力是考驗假設的一個重要指標，同時它亦有助於決定適當的樣本數目。本文針對第一類型錯誤、第二類型錯誤、統計考驗力、P 值，及樣本大小有關之誤用與迷思提出說明；其次，探究統計考驗力的重要性及其影響因素，接著說明透過 Cohen 氏查表法與統計考驗力分析軟體 G＊Power 進行樣本規劃並實際分析 30 篇碩士論文研究假設的統計考驗力；最後，提供一項實際可行的假設考驗的步驟。

The idea about significance level in Inferential Statistic is generally at 0.05 or 0.01. However， whether this definition would be applicable to the research of class and group guidance or gen-eral counseling? The researcher should consider that the application of one kind of significance level might result in a mistake， which would cause the failure of the study. Therefore， the re-searcher should not be too considerate of the level of statistical check. The power of test is an important index to test the hypothesis， and that can help to decide an adequate size. This paper addresses several misuses and misconceptions related to Type-I error， Type-II error， power， p-values， and sample size . Then， makes a study on the power issue. Next， computing power for any specific study via Cohen，s power table and a computer program called G ŎPower. Finally， A modest hypothesis testing model is also proposed.