Multiple regression

#1
To arrive at the relative weights of four variables in their contribution to satisfaction with an appraisal system, I used multiple regression. The beta weights of the four variables are:0.66142, 0.19799, 0.26149 and 0.00340 respectively; the t-tests indicate that except the first one (0.66142) the other weights are not significant.

When only one variable is proving significant, is it necessary to use Multiple Regression? If not Multiple Regression, what is the other tool that can be used?
 

Masteras

TS Contributor
#2
At first i would suggest use stepwise regression to see whether it is the only significant. If yes, ususally this one is kept only. but this is based upon statistics, the researcher (not the statistician) may wish to keep non sign ificant variables inside the model. in this case the statistician does as told. so it is not a theorem that if non-significant throw it out, but rather a joint decision.