I'm not sure what you are correlating. I imagine you have four sets of measurements for A & B: pre & post by control & treatment. Which of the four correlations are you asking about?

Hi Martin,

I basically have control mice fed a normal diet and treatment mice fed a high fat diet.

I measured two variables, A and B.

After dietary treatment both were decreased. But I also observed a negative correlation between A and B.

So I'm just a bit confused re how to interpret this seeing as they decreased when I measured them.

Thanks

Irina

I'm still confused. I understand that A and B both fell pre-post in both groups. Which group had a negative correlation between A and B. Was the correlation positive before feeding? What is your research question(s)?

I inserted the measurements for protein A and B in two separate columns in SPSS (so each column contained both the control and high fat diet measurements by mouse number). I ran partial correlations controlling for dietary treatment. I wanted to know how protein A is influenced by protein B.

Sorry, more questions. How many times did you measure A and B in each mouse? How many rows does each mouse have in SPSS?

Hi Martin - A and B were only measured once due to time constraints.

The sample size was small, about 7 data points for each condition (control and high fat).

Sorry, I'm now very confused. If you only measured the variables once, how do you know they decreased?

Sorry Martin, I'm probably not explaining this very well. What I meant is that they only measured once in each mouse (i.e. I didn't get the chance to re-run experiments and take the average measurement for each mouse which I know would have improved the reliability of the measurements).

So both the control fat and the high fat groups had about 7 mice and I measured A and B only once in each of the mice (total of around 14 mice).

Hope I've explained it a bit better now.

I got confused because of the word "decreased". To me, it implies measurements before and after the interventions. I think it would have been clearer to say A and B were lower in the treatment group after the intervention. Also, controlling for the difference in diet, a partial correlation indicated an inverse relationship between A and B, i.e. the difference in diet is a moderating influence on the relationship between A and B. That's all your results indicate.

Additional moderating influences on the relationship between A and B are a matter for speculation, so tentatively speculate.