On the Assessment of the Adequacy of the Fitted Regression Model Using the Confidence Interval

In this study, an attempt was to critically evaluate the application of confidence interval approach in selecting most suitable predictor variables in a regression model. Data were sourced secondarily on some economic factors with probable effects on crime rate. The multiple regression model was fitted with the F-test statistic and confidence interval estimates as measuring criteria for assessing the model aptness. The result from the study revealed that the number of unemployed male citizens is evident as an increasing factor of crime rate in the society. This result was notable from the evaluation of parameter estimates using the confidence interval at α = 0.05, since only its predictor coefficient interval estimate excluded 0. The result also reported that the narrower the interval of an estimate the more efficient it becomes. In general, the study found that confidence interval estimation was more efficient in assessing a regression model than the F-global test (ANOVA) which ascertained that the model was significant at α = 0.05. Also, it is obvious that the confidence interval criterion gives a better understanding and insight of on the predictive power of the model.