Dear allstaters
Sorry for this potentially stupid question.
I am of the understanding that if I fit a model with both the independent and dependent variables being functions of the population then I will create a dependent association between the two variables in the model independent of the real association I am really interested in. For example, if the independent variable is deaths/population and the dependent variable is pubs/population, will or can the two things be correlated even when the no. of deaths is independent of the no. of hospitals?
I think its easy to show this with examples but there are lots of published articles where people do it and I'm having trouble finding a refence to support my claim.
Thanks for you comments,
Patrick
|