0

I have one variable (x), I am simply trying to understand the relationship between (x,y).

I run two regression models, first model is predicted by removing the intercept and the second including the intercept. What I find when I simply graph the Beta Coefficient: movement in Model 1 appears to be on a larger scale (-0.5,0,0.5) than in Model 2 (-0.005,0,0.005).

What meaning does this have for the relationship between X and Y?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Terry McGrenaghan
  • 1
  • 1

1 Answers1

1

If you had first standardized your $x$ and $y$ variable, the two models would be the same. If you haven't done that, and apperently you didn't, then you should include the constant. There are other exceptions where it might be useful to exclude the constant, but I don't think those apply.

Maarten Buis
  • 19,189
  • 29
  • 59
  • Hi Maarten, thanks for the reply. I understand they need to be standardized (which they aren’t in this case), although this is the initial part of an argument. I standardise the factors later on. I just want to describe the effects on the coefficients in this scenario. – Terry McGrenaghan Sep 24 '14 at 13:58
  • By omiting the constant you are forcing the line to go through the point (0,0). This is fine if you first standardized your variables, as the line will then by definition pass trough that point. If you did not standardize your variable, this is almost always an unreasonable restriction. – Maarten Buis Sep 24 '14 at 14:07