Question

Given the following multiple linear model summary to predict health levels, in r studio:

                                       estimate     std.error     t value   pr(>t)

(Intercept)               -2.1234   2.3456 -2.431   0.1234
weight                     0.001234   0.001234   1.321   0.0123
log(height)               1.124338   0.1234   1.123   0.0123
log(age)                   0.1234   0.01234 10.123   <2e-16
weight:log(height) -0.002326   *******    *****    *****

What is the term (weight:log(height)) with estimate -0.002326 describing and what is the general statistical term for this method? Is it trying to combine the height and weight variates into one variate to make a better predictor?

Answer 1

The term (weight:log(height) is known as the interaction of the explanatory variables weight and log(height)  in the general statistical term for this method. The estimated value -0.002326 describes that decreased on the mean of health levels is 0.002326 when simultaneously increased a unit on weight and a unit on the log(height) (not on height) explanatory variables. Yes, it is trying to combine the height and weight variates into one variate to make a better predictor.