1
$\begingroup$

So my data is on feedback students receive during their 4 years at university. I sent out a questionnaire and there are 23 questions in total, each question is looking at the perception and value of feedback. For each question the students select once answer: either strongly agree, agree, neutral, disagree or strongly disagree.

So there are 4 years in terms of students: year 1, year 2, year 3 and year 4. I'm trying to find whether there is a difference in their responses for each question. I was going to use chi square but since I have 4 groups to analyze I want to know what statistics would be best suited for this data.

So questions include: e.g. q5 the feedback is given in a timely manner: select one of the following strongly agree, agree, neutral, disagree or strongly disagree.

e.g. q8 feedback I receive is relevant to my goals as an undergrad: select one of the following strongly agree, agree, neutral, disagree or strongly disagree.

$\endgroup$
2
  • $\begingroup$ Are all 23 of the questions assumed to be measuring the same construct? $\endgroup$ Apr 22, 2015 at 18:00
  • $\begingroup$ Yes all 23 questions are measuring the same construct. $\endgroup$
    – user74279
    Apr 22, 2015 at 18:05

1 Answer 1

0
$\begingroup$

Your response scale

strongly agree, agree, neutral, disagree or strongly disagree

is ordinal, so you should look into ordinal regression. A chi-squared test, in comparison, will not use the ordering in the scale, so will give a less powerful test. There are many posts here about ordinal regression, so search the site. Some starting points:

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.