Stepwise regression model

A method in multiple regression studies aimed to find the best model. This method seeks a model that balances a relatively small number of variables with a good fit to the data by seeking a model with high
(the most parsimonious model with the highest percentage accounted for). The stepwise regression can be started from a null or a full model and can go forward or backward, respectively. At any step in the procedure, the statistically most important variable will be the one that produces the greatest change in the log-likelihood relative to a model lacking the variable. This would be the variable which would result in the largest likelihood ratio statistics, G. (A high percentage accounted for gives an indication that the model fits well. See also multiple regression correlation coefficient - .)