The question is to compare means of sales volume in different periods. So I got sales period 1 and sales period 2. Sample size is 40. The skewness, kurtosis, test of normality and histogram all show sales period 1 has a normal distribution but sales period 2 does not. Should I use paired samples t test or Wilcoxon? I did both test but the results are compeletely different.

  • 1
    $\begingroup$ If the data are paired the individual distributions wouldn't need to be normal. Which Wilcoxon test did you use? What does the distribution of pair differences look like (say in a normal Q-Q plot)? What do the individual values consist of? Would you expect the changes under the alternative to consist of noise around constant shifts or more like noisy percentage changes around a typical percentage change? $\endgroup$ – Glen_b Oct 25 '14 at 6:30
  • $\begingroup$ This link might help: stats.stackexchange.com/questions/2492/… $\endgroup$ – rnso Oct 25 '14 at 9:09

Your Answer

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

Browse other questions tagged or ask your own question.