I am looking at sense of belonging for 100 undergraduate students who are part of a program for first year students.
Students who participated in the program will rate their sense of belonging on a 50 item likert scale questionnaire (1=strongly disagree, 5=strongly agree). Students who did not participate in the program will complete the same questionnaire.
Which statistical test to I use to compare the results of both groups?
You do not say how this 50 item scale is structured. Are there sub-scales, what are the reliabilities of those sub-scales? Certainly 50 items are not needed for one dimension. Likewise, I assume that you do not intend to treat each item separately.
My best on your research,
John
Have a look at the link
https://towardsdatascience.com/statistical-tests-when-to-use-which-704557554740
Jasmine,
I see no answer to my questions regarding the structuring of this scale, so I still have little to go on.
However, if I can assume that the Likert-type items are summed to generate constructs from the 50 items, then a usual analysis might be to consider the reliability of such constructs via calculating coefficient alpha, and if satisfactory, contrasts between those who did and did not participate, given that assumptions are met, could be conducted on the means of those composites with a t-test, either Gosset or Welch depending on whether the HOV assumption is met. Again, with 50 items, I am guessing that there is more than one composite captured with the summations, so you should pay attention to protecting the Type I error rate with some appropriate adjustment.
This does not attend to validity, but, without more knowledge, there is little that I can say about that.
My best,
John
- Badges
- Science topic
- Similar topics
- Medical Science Nursing
- Dissertations