0
$\begingroup$

I have a treatment and a control group: both have pre- and post- values (I tested reading speed). I need to know if there are any significant differences between the groups (or if the treatment group has improved compared with the control group). Now I want to apply the Mann-Whitney test but I am not sure if Iy have to analyse

dependent variable: pre-test values
Group variable: treatment + control group variable

dependent variable: post-test values
Group variable: treatment + control group variable

or if I just can calculate the differences

dependent variable: post-test-pre-test values
Group variable: treatment + control group variable

Could anyone give me advice here?

$\endgroup$

2 Answers 2

1
$\begingroup$

If you have reading speed measured e.g. in pages per hour, such that you can calculate the differences, you can take the second option. By the way, you could also compare the ratios. It's up to you to decide what represents your idea best.

$\endgroup$
1
  • $\begingroup$ Could you please explain how to compare ratios? Is it analyzing (post/pre) ratio rather than (post - pre) difference? $\endgroup$
    – rnso
    Oct 19, 2014 at 13:24
4
$\begingroup$

If I am reading the results, I would be interested in knowing the whether there is a signficant difference between mean deltas of the two groups. It could be percent change as in following example. I could arrange the data as follows:

mydata
  id grp pre post  delta
1  1  tt  10   12  0.200
2  2  tt   8    9  0.125
3  3  cc  11   12  0.091
4  4  cc   9   10  0.111
...

wilcox.test(delta~grp, data=mydata, paired=F)
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.