Correlation between the dependent variables in an experiments (Urgent )


I run an experiment with one independent variable let say A (I manipulate this variable across two conditions for insistence under one condition participants use the A software and under another condition they did not). The experiment has three dependent variables let say B, C, D.It is a repeated measure design, so I compared each of the dependent across two conditions

The results of paired t test showed that when participants used the software A, the dependent variables B and C increase, while the dependent variable C decrees.

But when I run correlation between the dependent variables B, C, D I found no correlations. I was expecting positive correlation between B and C and negative correlation with C.

To me it looks like if A cause the increase in B and C then we would expect that B and have positive correlation.
So, the questions is do dependent variables correlates ?


TS Contributor
I run an experiment with one independent variable let say A
Could you please provide us with the details of the actual study? What is its topic,
what are the research questions, what are the variables and how were they measured
(precisely), how large is your sample size? What do you mean by correlation/no
correlation, is it the difference between r=0.00 (no correlation) and r=0.01 (correlation),
for example?

With kind regards

Last edited:
The independent variable is using a new interface, so participants under one condition used the new interface and under the other condition used an old version of the interface. The dependent variables are participants engagement, their levels of uncertainty and ease of use of the the interface .
The results of paired sample test show :
-An increase in participants levels of engagement
-A decrease in their levels of uncertainty
- They reported higher ease of use of the new interface
However when measure the correlations between the three dependent variables, they were no significant as following:
Engagement and uncertainty = r=-0.23, p= 0.39.

Engagement and ease of use r = -0.05, p =0.84.

Ease of use and uncertainty r=.044, p=.871.

If the new interface increase participants levels of engagement, decrease in their levels of uncertainty, and they reported higher ease of use of the new interface, why these variables do not correlate?

The sample size 16.
As children get older their vocabularies increase, as do their absolute muscle strengths in the bicep muscle. But I expect no correlation between those two variables at any age.

I al not quite clear what you are correlating, however. You must have measured each variable twice on each subject, no? This sounds like a crossover design. So when you did you correlations, was it using interface A or not A?
Thank you EdGr for your reply.
I did the correlation within each condition, when they used A and when they did not.

When they used the interface the correlations were as following:

Engagement and uncertainty r=-0.22, p= 0.20.

Engagement and ease of use r = 0.26, p =0.16.

Ease of use and uncertainty r=-0.22, p=0.20.
OK, I see what you did. These correlations seem logically very low (in other words, we would expect some correlation between, say, engagement and ease of use) but the fact that both increased doesn't mean that there should be one. Imagine having classes that trained on algebra and reading. Both go up in the "after" condition, but they need not be correlated.

By the way -- you never sorted or manipulated the Excel file without keeping the original, did you? A bad sort can screw up the data! Excel will happily sort one column and leave the rest alone.. Most regular stats packages won't. Bad calculations can ruin a file too. You should make sure there isn't an error somewhere when the results don't seem right.