Considerations on the Correlation Between Student Test Statistics in Cases of Simple and Multiple Linear Regressions

The paper brings arguments in favour of the idea that Student test statistics may not be considered only in comparison with the critical (tabelled) values in order to determine the significance of estimated parameters. Interesting information on the quality of estimation can be obtained by studying the correlation between the Student test statistics related to each explanatory variable in case of simple and multiple regressions, respectively. This way, it is possible to identify the factors which determine the change the Student test statistics related to considered explanatory variable as the number of the other explanatory variables increase in the linear regression equation. In this context, a factorial model for the analysis of the above-mentioned change is proposed. The proposed analysis methodology is practically applied to a private consumption function estimated for Romania during the period 1990-2009.