折半信度 (Split-Half Reliability)

Split-half reliability means to separate the test to 2 equivalent parts and to calculate the correlation the results of  these 2 parts. It is often used to estimate the stability and internal consistency of the test when the items of test are enough.

中文解釋：
折半信度係指將題目較多的問卷分成兩個內容相當的短版問卷之後，兩個短版問卷得分的相關。折半信度可用於估計施測結果的穩定性與問卷的內部一致性。

判斷標準：
相關係數 (Pearson's r或Spearman's ρ) ≧0.75為良好，0.40-0.74為中等，≦0.40為差。

臨床意義：
1.可確保被拆成兩半的問卷內容同質性高。
2.折半信度只使用半份問卷估計信度，若半份問卷的信度良好，則使用整份問卷施測信度更高。

研究設計：
先將問卷拆為「內容等量」的兩半，再邀請問卷的適用對象填寫整份問卷。最後計算適用對象的於兩半問卷的作答結果之相關係數。

參考資料：

Cronbach, Lee. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297-334.

Elena M, Andresen. (2000). Criteria for assessing the tools of disability outcomes research. Archives of Physical Medicine and Rehabilitation, 81, S15-S20.

Klein, C., & Fischer, B. (2005). Instrumental and test-retest reliability of saccadic measures. Biological Psychology, 68, 201-213.

Streiner, D. L. (2003). Starting at the beginning: An introduction to coefficient alpha and internal consistency. Journal of Personality Assessment, 80, 99-103.