I have the following problem. I have conducted a survey about how people have consumed media in a certain way. The scale of responses was 1=never to 9=always. Also I have recorded exam scores for a specific exam for each of the respondents. These scores however are on a scale from 1.0 to 5.0; 1.0 being the best and they can take on the incremental values in between the two boundaries (increments of 0.1; i.e. 1.1,1.2,...,4.9,5.0).

I am now exploring whether there is a relationship between the survey results and the scores.

I am not sure which method best to use to establish if there's a relationship between the two. From other questions on here, I have seen that Spearman correlation and a chi squared test might work, but I am a bit concerned about the scale of the scores and I am not 100% sure whether one would even consider this truly ordinal data.



1 Answer 1


Spearman correlation will work with variables that are at least ordinal. That is, if you suspect your scores are really interval or quasi-continuous, there is no issue with using Spearman correlation for this kind of data.

Also, with an ordinal scale from 1 to 9, that's probably enough categories where respondents are treating it as interval in nature. That is, they are probably treating it as numeric, understanding, for example, that a response of 4 is half-way between a response of 3 and of 5. In this case, I would probably be okay with treating this variable as interval, and Pearson correlation, if other assumptions are met, would probably be okay. If this item was rated as (never, sometimes, often, always), I would advise against this approach.


Your Answer

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

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