I am trying to perform a count regression models but I want to determine first the predictors that has a significant relationship to my response variable
Before you perform a count regression model, you can check assumption to determine normality, linearity, and equality of variance, then run ANOVA or Chi-Squared test methods to identify which predictors have a significant relationship with your response variable.
Determining the Relationship of Variables: If you have a dependent variable that is a ratio and an independent variable that is nominal with more than 2 levels, you might consider using Analysis of Variance (ANOVA) or Multivariate Analysis of Variance (MANOVA) if you have more than one dependent variable.
Chi-Squared Test: If both your variables are categorical (nominal), you can use a chi-squared test to determine if there’s an association between them.
These statistical methods can determine whether the means of your dependent variable(s) are significantly different in each category of your independent variable.
(OR) Alternatively, you can use a method called variable selection. This involves fitting your model with different combinations of predictors and comparing the fit of these models using a criterion such as the Akaike Information Criterion (AIC) or the Bayesian Information Criterion (BIC). The model with the lowest AIC or BIC is typically chosen as the best model.
Hope this helps!
You may conduct correlation analysis prior to regression to identify which variables may be significant predictors in your regression analysis.
Why do you want to determine the significant predictors before conducting the regression? Instead, you should just run the regression (with a set of dummy variables to capture the categories of your nominal IV). That will tell you which predictors are significant when you take the full set into account.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling