As far as I know it is impossible to use mean on ordinal level. Unless you index all values into more complex variable.
To be more precise: of course, it is phisically possible to count means on ordinal level - software will do anything you wish; but there is no sense in it, because such values mean nothing on this level. E.g. you have from 1 (very good) to 5 (vary bad) and you received value of the mean 3,56. What does it stand for?
Based from my reading, if you give a score for each categories, you should sum up the total score and later ranked it based on other finding as support of the ranking was formed or your own rank. Let say, high, average and low. Something like that. I haven't go in depth about this though but I use it for my final year project after looking some of my friends ongoing research. :D
Likert scales are ordinal and therefore we use median and mode to calculate the central tendency. Don't use mean to calculate central tendency as this is done for interval data only. If your data is normally distributed, use parametric tests such as Pearson Correlation for correlation. If your data is not normal distributed, use spearman's correlation to calculate correlation.
Simply compute the average and use t-test to check the significance of the distance of the average from the mid-point of Likert scale .
- Badges
- Science topic
- Similar topics
- Statistics
- Survey