Interpreting regression coefficients for linear regression

noetsi

Fortran must die
#1
I was going to title this, dumbest question ever.

I am so used to saying for a given slope a value of .07 for a predictor means for a unit increase in X y would go up .07 units. But then, maybe getting old here, I wondered what that meant in a specific case.

My dependent variable is rehab rate which is a percent (so .01 to 1). My predictor is turnover which is also a percent. So the way I interpret this is if X goes up one percent rehab rate goes up .07 percent or 7 one hundreds of a percent. Which is obviously very small.

Is that right? What I am really confused here is what the regression considers a unit for a percent. Is a one unit increase for a percent, one percent?
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
It does 1.0 unit, so it depends on how you inputted your data. With the percentage, the thing you would look out for is that your means aren't close to 100%, since then you could could get predictions over a hundred percent, which may or may not be non-sensical given your context