Logistic regression in SPSS: how to read the output

What each output table tells you, how to interpret the odds ratio Exp(B), and how to write the results in APA style.

Part of the series: Statistical Analysis for a Seminar Paper or Thesis: The Complete Guide

Logistic regression predicts an outcome with two categories: passed/failed, diagnosed/not diagnosed, dropped out/stayed. The SPSS output is long, but only a few tables really matter.

First: what SPSS is predicting

The Dependent Variable Encoding table shows which category was coded as 1. SPSS predicts the odds of being in that category. If the result looks the opposite of what you expected, this is the first place to check.

Block 0: you can skip it

The first part shows a model with no predictors at all, just for comparison. The important information is in Block 1.

Omnibus Tests of Model Coefficients

Tests whether the model with predictors is better than a model without them. Report the chi-square value, the degrees of freedom and the significance in the Model row. A significant result means the predictors together improve prediction.

Model Summary

• Nagelkerke R Square: an estimate of explained variance, between 0 and 1. It isn't the same as R² in linear regression, so describe it as a "pseudo R²".
• Cox & Snell is reported less often, because it can't reach 1.

Hosmer and Lemeshow Test

Tests how well the model fits the data. Here, unlike usual, a non-significant result is good: it means there's no meaningful gap between what the model predicts and what was observed.

Classification Table

The percentage of cases the model classified correctly. Compare it with the percentage in Block 0: a small improvement means the model doesn't add much to prediction.

Variables in the Equation: the main table

For each predictor:

• B: the coefficient, on the logit scale. Hard to interpret directly.
• Wald and Sig.: the test and significance of the predictor.
• Exp(B): the odds ratio. This is the value you interpret.

How to interpret an odds ratio

• Exp(B) = 1: no relationship.
• Exp(B) = 1.85: each one-unit increase in the predictor multiplies the odds of the outcome by 1.85, that is, raises them by 85%.
• Exp(B) = 0.60: each one-unit increase lowers the odds by 40%.

Odds are not probability. An odds ratio of 2 doesn't mean the probability doubled.

Ask for confidence intervals for Exp(B), under Options → CI for exp(B). A confidence interval that doesn't include 1 matches a significant result.

Reporting in APA style

The logistic regression model was significant, χ²(3) = 24.61, p < .001, and explained 21% of the variance in dropout (Nagelkerke R² = .21). Absences significantly predicted dropout, OR = 1.85, 95% CI [1.21, 2.83], p = .004.

Next steps

Logistic regression fits a dichotomous outcome; for a continuous one - linear regression, with APA reporting. For an association between two categorical variables, chi-square is enough. And how to build the coefficients table and the chapter: APA tables and how to write a results chapter.

Last updated: 2026-09-27