I have little knowledge of statistics and was wondering what tests to use for an evidence-based project (not research, so I am not using a control group) that I am conducting.

  • What statistical test would I use for pre and post knowledge tests after a student educational intervention?

  • Also, are numeric test results considered interval data and would I being looking for the difference in means between the pre and post knowledge tests?

I also have pre- and post-confidence scores that have one negative value:

0: do not agree 

and two positive values:

1: somewhat agree and 
2: strongly agree; 

What statistical test would I use for the confidence survey? Would this be ordinal data and again would I be looking for the mean difference between the pre- and post-confidence surveys?

  • 2
    $\begingroup$ Last time I checked, "research" does not require a control group. $\endgroup$
    – Thomas
    Jun 16, 2013 at 18:21
  • $\begingroup$ I'm also not sure how this fails to count as research. $\endgroup$
    – Glen_b
    Jun 17, 2013 at 6:26

1 Answer 1


There is no need of a control group to define a project as a research. Your variables seem to be on a Likert scale, although they can also be considered as binomial (0: do not agree / 1: somewhat or strongly agree).

As a practical advice I would use Mann-Whitney U test for paired data if you use a 0/1/2 coding or a McNemar's test in case you use a binomial version which is for me more appropriate.

In general it is not appropriate to have one category for negative answers and two categories for positive answers.

Hope it helps.


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.