Wilcoxon Mann–Whitney ranksum test

[Lovely Tolin]

Hi all,

I am currently working on analyzing the impact of a particular project. The data that I have is ordinal - Likert scale data with options ranging from 1 (Not at all) to 7 (very participative) - wherein extent of participation was asked "before" the start of the project and "after" the project. In this case, I would like to figure out if there is a significant difference between the before and after responses. One of the tests suggested is Wilcoxon Mann-Whitney rank sum test for ordinal data. The stata code for this is:

 ranksum varname [if] [in], by(groupvar)

I was quite confused since I do not really have a grouping variable. What I do have are variables for "before" and "after" responses. Hence, I would like to ask if there's anyway to get around this and still implement the test without the need of having a grouping variable? If not, can I simply use the paired t-test instead?

Thank you in advance.

Lovely

[Marcos Almeida]

Welcome to the Stata Forum/ Statalist,

You seem to have paired data, therefore you’d need Wilcoxon - signrank - test, instead of - ranksum.

[Lovely Tolin]

Welcome to the Stata Forum/ Statalist,

You seem to have paired data, therefore you’d need Wilcoxon - signrank - test, instead of - ranksum.

Hi Marcos!

Thanks so much for this, problem solved instantly.

Best regards,
Lovely

[Marcos Almeida]

Thank you for informing your query reached a satisfactory closure.