Hi,

Why do we need to check if the sample data collected for hypothesis testing is normal or not before carrying out the statistic computations?

I am doing a project to identify critical factors behind consumer perception of hostel food and Dhaba food. I have gathered 44 responses and the responses have been on nominal scale. Which test statistic should be used in this case and what are the conceptual steps in doing so?

Can the nominal data be assumed to be continuous and statistic like two sample t test can be used??

- Anirban

Hi Anirban,

Normality test is done for continuous data so that we choose the right tests using mean or median.

As you have nominal data, you may like to do tests for equality or inequality of proportions. If you imagine your nominal data in a bar chart, you are not likely to see a bell shape. For example, a head or tail outcome for toss of a coin shall have only two bars in a bar chart based on occurrences.

Hope this helps.