Well it depends on what you want to know. The old adage holds for research: Proper Prior Planning Prevents Poor Performance. Good Luck, D. Booth

When in doubt, you can always use a non-parametric test (i.c., Kruskal-Wallis or (Extension)of the Median test), but first check the distributions for normality with histograms (and QQ-plots), and the Shapiro test. The Fisher F- or t-tests are known to be quite robust and to have more power.

Dear Abdullah Al Zabir ,

You have made a classic mistake. You analysis should always be driven by your research question. You questionnaire design should have been informed by your literature review and should have enabled you to answer your research question.

Havng said this, there are a few practical issues:

• It is useful to have some demographic questions so you can split your responses into groups
• Likert questions are often organised into groups. This can simplify the analysis when a dimension reduction technique is employed to turn the group of questions into a single scale variable.
• It is important to ensure your sample size is adequate.