3 Groups of Participants & Likert data

#1
Hi,

I'm new to the forum - though I have browsed and found the resources helpful for a long time. Thank you to all who contribute here.

I have an issue with some data that I can't see covered elsewhere. I have conducted a study on whether 3 different groups of students found our mobile website useful towards learning. The survey covers a range of areas, from content of the site, to whether mobile learning has improved their ability to study.

In total I have 16 Likert type questions - all with a scale of Strongly Disagree, Disagree, Neutral, Agree and Strongly Agree.

Immediately, I can observe from the whole sample (all 3 groups - n=601) that the results will be skewed to the right (Agree/Strongly Agree) for nearly all 16 Likert questions.

I'm unsure what type of tests I can use to illustrate whether there was a significant difference between the 3 groups of participants (biology/geography/maths students). Kruskal-Wallis seems like an option, but I don't know how to apply this to Likert type data.

Of the 601 sample, the group sizes are not equal. 281 are biology students, 267 are geography students and 53 are maths students. Does this affect what test I can use?

I have already created charts for each question, comparing the 3 groups % response to the question.

However, I'd like to take this further, and not quite sure how. I'd like to treat my data as ordinal rather than assuming it is interval.

From the extensive reading I have done, this would have been easier with only 2 groups of students, but I have what I have.

Any ideas would be greatly appreciated!
 
#2
Anyone? I'd really appreciate some help with this. Specifically, how to use Kruskal Wallis for my research. There doesn't appear to be any information about this on the internet - at least none that I can understand.
 

Karabiner

TS Contributor
#3
First of all, you should perhaps take into account that waiting for
more than just 3 hours for an answer would not be considered a
long time by most regulars here.

Second, it is not clear why you don't want to use Kruskal-Wallis H test or
what you mean by "don't know how to use". Ususally a statistical
software is used for carrying out the calculations.

With kind regards

K.
 
#4
Thanks for your reply. I said I was unsure whether I could use Kruskal-Wallis, not that I don't want to use it.

I don't know how to apply this test to my type of data, and if this can be done in Excel (or do I need SPSS)?

Can anyone give me an example of how to apply this test to Likert-type data? My scale is: Strongly Disagree (0), Disagree (1), Neutral (2), Agree (3), Strongly Agree (4) and for each of the questions, I have the three groups I mentioned above providing responses (Bio, Geo, Math students).

Thanks.
 

Karabiner

TS Contributor
#7
if this can be done in Excel (or do I need SPSS)?
The Excel help function will probably give you the information whether
the H-test is implemented in Excel or not. But I suppose it's not. So
you'd need another software (SPSS, or something else). How the test
is carried out will then dependent on which software you use.

With kind regards

K.
 
#8
Hi,

Thanks for the reply but I don't think you understand my problem.

I was looking for instructions to find out how to do this with my data. I can't find anything about this.

Specifically: Can anyone give me an example of how to apply this test to Likert-type data? My scale is: Strongly Disagree (0), Disagree (1), Neutral (2), Agree (3), Strongly Agree (4) and for each of the questions, I have the three groups I mentioned above providing responses (Bio, Geo, Math students).

Thanks.

I have SPSS, so instructions as to how to structure my data and then how to apply and interpret the analysis would be most appreciated.