[R] Help with slope comparisons

[David Winsemius]

On Nov 11, 2010, at 4:16 AM, Ravi Kulkarni wrote:

>
> Hello Forum,
>  I have data for two groups of subjects for a ratio variable (cortisol
> level) assessed three times during the day (at 0600, 0900 and 2100  
> hours).
> So I have three means for each group. I have been asked to do a "slope
> analysis" to compare the two groups.
>  According to my understanding, this means that I regress the three  
> means
> on the time for each group and compare the slopes using a t-test  
> (there is a
> post on this: Etienne Toffin etoffin at ulb.ac.be
> Tue Jan 27 20:08:36 CET 2009).
>  My question: (if what I said is correct) it seems strange to me  
> that we
> calculate a regression line with just three data points - is this  
> really
> what is meant by a "slope analysis"?

Doubtful. What was probably meant was to do a regression with an  
interaction model based on the original data. Sounds like either you  
or your supervisor need some consulting resources.

-- 
David.
>
>  Any help most appreciated...
>
>  Ravi
> -- 


David Winsemius, MD
West Hartford, CT