There are two individual provinces to compare. Variable is difference between expeted level and actual level. Both expected and actual levels are measures as 1-5 Likert scale.
First suggestion that I have is: find mean values of both actual and expected levels and find the gap. Then doing t test for the variable gap.
Second suggestion that I have is: get the difference between actual and expected ranks. Then doing Mann Whitney U test for that difference.
I'll assume that by "1-5 Likert scale", you mean a single Likert-type items, with potential values like "Very Small, Small, Moderate...".
If so, you will probably want to treat the data as ordinal. In this case, t-test wouldn't be ideal, but a (Wilcoxon-)-Mann-Whitney test would work.
Thanks Salvatore.
But the matter here, it wants to compare 'difference of expected rank and perceptional rank' for two provinces. This analysis wants to compare the 'difference between two ordinal scales'.
Expected rank is first variable, perceptional rank is second variable and we derived difference between these two as third variable. And wants to compare third variable for two provinces. Although variable 01 and 02 are ordinal, can we consider 3rd as well?
Hm... So, are there items being ranked for each of the two provences by several respondents ? Or are there several provences being ranked by several respondents and you are just interested in two of them ? Or... ?
Thanks Salvatore for your support.
No there are 30 respondents from province A and another 30 from province B. (Independent samples). Questionnaire is same.
There ask expected rank and perception for same fact (ex: expected level of online material and actual level of online material. Then get the gap between these two ranks.
There are values for these gap for two provinces. I want to test that "are there a difference between these two provinces for the derived variable gap"
First variables (expectaion level and actual level) are in Likert scale. Derived variable is difference of those two. So could we consider this derived variable as ordinal too?
If you are subtracting one value from another, you are already considering them interval and not just ordinal.
When you say, "scale" are you summing up the responses for several questions ?
Mathematical operations cannot be legitimately be carried out with figures that are not of equal intervals.
So @ Mangiafico's comments are quite correct.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS