Binary Logistirc Regression (SPSS)

基本算式: Ln (Odd) = Ln (Y'/1-Y') = a + bX
Y' represents the predicted probability of the event which is coded with 1 (one).

For dichomotous variables, the calculated odds ration (OR) of a logistic model is

• (the odd in favor of the event coded with 1) / (the odd in favor the event coded with 0)

筆記：

• Continuous variables in a logistic regression model - the importance of the "scale" - description 對應的某單位變動的勝算比，使用的單位不同，勝算比也不同。如：早產兒的體重使用 1 gram vs. 1 kilogram，odds ratio 有不同的大小。

參考資料：

1. 生物谷的教學
2. Multivariate Analysis with SPSS - 一個教學資訊豐富的網頁，包括如何使用 spss 做 binary logistic regression analysis。