1
$\begingroup$

Potentially simple question here but I can't work out the answer, so...

I have created a single question questionnaire with five possible yes/no answers and then calculated the percentages for each answer (N=34) with the distribution being normal as per a Kolmogorov-Smirnov test.

I then split this sample into two subsamples, male and female, with the aim of comparing these data with the population but whilst the female data is still normally distributed, for one of the five answers for the males the K-S statistic is now significant (0.003).

Do I still calculate the means (parametric) for males and females and compare these with the population mean values or do I calculate medians for the (otherwise normally distributed) population and compare them with the medians for the males and females? Or is there an entirely different approach?

All advice most appreciated!

$\endgroup$
1
$\begingroup$

In the normal distribution the median = the mean. If you have reason to suspect that your data is not normally distributed then you should use a non-parametric test to compare them (i.e., the Mann-Whitney U-test).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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