[R] comparing two regression models with different dependent variable

Joris Meys jorismeys at gmail.com
Thu Jun 10 14:28:39 CEST 2010


On Thu, Jun 10, 2010 at 2:04 PM, Gabor Grothendieck
<ggrothendieck at gmail.com> wrote:
> On Thu, Jun 10, 2010 at 5:33 AM, Joris Meys <jorismeys at gmail.com> wrote:
>> This is only valid in case your X matrix is exactly the same, thus
>
> The poster did not give a full explanation so the best that can be
> done without getting into an extended question and answer is to make
> some assumptions, show the result and hope that the assumptions are
> sufficiently close that the poster can tweak it.

Don't get me wrong, I do appreciate your willingness to help people on
the list. Your solution is also mathematically sound. Yet, it is my
experience that proposing a method without a full understanding of the
setup and the data of the experiment, often leads to erroneous
results. Most often, the tested hypotheses turn out to be not the ones
of interest to the researcher, and they're unlikely to grasp the full
importance of all assumptions made and all subtle differences between
the tested hypotheses. It was clear from the OP that the TS has only
limited knowledge about statistics. Hence my further explanation.

>
> That the X matrices are assumed to be the same was clearly stated
> already and seems not to be some new revelation.  Ditto for the other
> comments below.

It is not because the same variables are used, that they contain the
same data. I could not conclude that from the little information we
have been given. It could have been the same persons tested twice, but
it could easily have been two seperate groups as well. Same predictor
variables, but other observations, and the X matrices are not the same
any more.

Cheers
Joris



More information about the R-help mailing list