2

I've looked at this post for reference. And suppose my model is

$\hat{score} = 12 + 0.4*Age_i + 0.5*Height_i + 0.8*Age_i*Height_i$, then is the following interpretation of my coefficients correct?

For a one unit increase in age, the average score changes by $0.4 + 0.8*Height_i$, holding height constant.

For a one unit increase in height, the average score changes by $0.5 + 0.8*Age_i$, holding age constant.

Community
  • 1
Adrian
  • 1,665
  • 3
  • 22
  • 42
  • There are some other explanations given in http://stats.stackexchange.com/questions/56784/how-to-interpret-the-interaction-term-in-lm-formula-in-r?noredirect=1&lq=1 if you want more detail than @YvalSp provided. – mdewey Nov 22 '16 at 12:25

1 Answers1

1

Pretty much. But:

For a one unit increase in age, the average score changes by $0.4+0.8∗Height_i$, holding height constant.

Holding $height$ constant as long as $height\not=0$. If it equals $0$ than the interaction term is also $0$. Also, if $height>0$ and held constant, than for every increased unit of $age$ , the average predicted score will increase by $(0.4+0.8)$, not by $0.4+0.8*height$.

$\beta_1$ (0.4) will be used alone only when $height=0$, or else the interaction will be also coming into play.

The same principle goes, naturally, also for both covariates.

Yuval Spiegler
  • 1,821
  • 1
  • 15
  • 31