📸 Poisson Regression in SPSS via Generalized Linear Models
What Is Poisson Regression?
Poisson regression models count outcomes — non-negative integer values such as number of hospital visits, accidents, publications, or crimes. Standard linear regression is inappropriate for counts because it can predict negative values and assumes normally distributed errors.
The model uses a log link function: log(μ) = β₀ + β₁X₁ + ... The exponentiated coefficients give Incidence Rate Ratios (IRR).
Running Poisson Regression in SPSS
Step 1: Analyze → Generalized Linear Models → Generalized Linear Models.
Step 2: Type of Model tab: select Poisson loglinear.
Step 3: Response tab: add the count DV. Predictors tab: add independent variables.
Step 4: Statistics: select Exponentiated parameter estimates (IRR) → OK.
Overdispersion check: If Variance/Mean ratio >> 1, use Negative Binomial regression instead.
📸 Poisson regression output — Incidence Rate Ratios
APA Reporting
Poisson regression indicated that smoking significantly increased the event rate (IRR=1.51, p<.001) and the new treatment significantly reduced it (IRR=0.73, p=.001), after adjusting for age.