Comparing time-based regressions


New Member

I have a ton of temperature data over time for 38 different subjects (over 100,000 lines of data) that I ran regressions on for each subject. I want to compare how much people's temperature varies with time and use that as a variable (ex: is there a correlation between people with a stronger time-temperature link and age, etc).

I'm wondering what's the best measure to use: correlation coefficient (r), or the regression slope? Because this is a small change in temperature over a long period of time, the slopes are showing up like 0.000 or 0.001 (though looking at the graphs it's definitely positive slopes). However, the correlation coefficients go as high as .7 for the same data.

Also, any quick & easy way to get these outputs re-routed into values in cases in SPSS, or have to some of it by hand? (Like write down the r / slope for each subject, recode it into a variable, etc)