0
$\begingroup$

I've got several dummy variables that I'm combining into two factors, so

FACT1 = DUM1 + DUM2 + DUM3
FACT2 = DUM4 + DUM5

Because there's a different number of dummies going into each factor, I thought I'd better normalize them so they're comparable on the same range.

FACT1_n = FACT1/3
FACT2_n = FACT2/2

Now I want to fit a model

y = a + b1(FACT1_n) + b2(FACT2_n)

and I'm not sure how to interpret the coefficients.

Is this even the right way to go about this?

EDIT: I should note that I'm not asking how to interpret coefficients on dummy variables. I'm asking about how to interpret a coefficient on a sum of dummy variables that has been normalized.

$\endgroup$
  • 1
    $\begingroup$ Expressions like "DUM1 + DUM2 + DUM3" are not defined until you state how you are numerically encoding the dummies. How are you encoding them? (Once you tell us that, this question ought to answer itself...) $\endgroup$ – whuber Aug 28 '13 at 19:08
  • 1
    $\begingroup$ @whuber the dummies are 0/1 for no/yes. They come from a survey that has no/yes questions, which ask about a few concepts (the factors) in various ways. Sorry if I'm being dense, but it hasn't answered itself. $\endgroup$ – Trevor Aug 28 '13 at 22:08
  • $\begingroup$ @Trevor: In my view adding dummy variable change the influence of each dummy variable and hence normalising. I think you should use Custom Model as combine effect of dummy variables. Combine effect of DUM1,DUM2,DUM3 and DUM4,DUM5 yields you the effect of FACT1 and FACT2 respectively. $\endgroup$ – SAAN Aug 29 '13 at 6:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.