Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 0 down vote favorite

I have a set of answers to a number of 7 point Likert scaled questions that ranged from very bad to very good. I have about 200 answers for each question.

Q1(type1): Very bad, ..., ..., ..., ..., Very good

Q2(type2): ...

Q3(type1):

Q4(type1):

Q5(type2):

Q6(type1):

...

What I'd like to do now is check which type1 question matches with which type2 question in such a way that I will know that a question of type 1 is as bad or as good as a question of type 2.

I'm trying to figure out which test to use but I'm stuck. It can't be that I have to run lots of pretests but a variance analysis also doesn't seem right. I'm assuming my answers are interval scaled.

share|improve this question
What do you mean by "as bad or as good"? – Peter Flom Mar 1 '12 at 13:55
2  
I'd like to say, people find Q1 as bad as Q5 for example. Q5 and Q1 are types of things that people can find very bad to very good. – J. Maes Mar 1 '12 at 13:58

1 Answer

up vote 2 down vote

If you are assuming that the Likert scaled variables are interval scaled (which is questionable, but probably OK for this purpose) then you can look at the mean. You just need to define how close the means have to be to be "as bad" or "as good". Alternatively, if you are only interested in the extremes, you could match the proportion saying "very bad" or the proportion saying "very good".

If you want to get fancier (although I don't think you need to) you could look into "item difficulty" measures from item response theory, and equate "difficulty" with one end of the scale.

share|improve this answer
I was looking for a solution for which I don't have to compare the means one by one with the use of t-tests, but in one big test. Can this be done? I'm using SPSS – J. Maes Mar 1 '12 at 14:05
If you want to know which ones match with which, you have to compare them all. But I would not use a t-test or any other test, I would look at the means (all of them) and decide based on what you mean by close. A t-test (or ANOVA or whatever) would tell you if there are statistically significant differences among the tests. I don't see how this answers your question. – Peter Flom Mar 1 '12 at 18:30
That's what I'd like to to, compare the means and say which ones are (alpha 5%) the same. The problem is that I have about 7 variables in the first category and 7 in the second. I can't seem to find out which test I can use to compare all the means from the first group with the ones in the second. – J. Maes Mar 1 '12 at 18:34
I still don't really understand what you are trying to accomplish, but have you looked at correlations? – joint_p Mar 1 '12 at 23:00

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.