Which statistical test should I use for my Likert-scale surveys?

#1
I have 2 surveys, both using a Likert-scale with only 4 points (strongly agree, agree, disagree, strongly disagree).

Both surveys have 10 questions.

The first survey has 25 responses and the second survey has 6 responses.

There is only 1 group for each survey.
 
Last edited:

Karabiner

TS Contributor
#2
What do you want to know from your analysis? You just told us that you have a variable with 25 observations, and another variable with 6 observations. It is even unknown wether these are 31 different respndents.

With kind regards

Karabiner
 
#3
What do you want to know from your analysis? You just told us that you have a variable with 25 observations, and another variable with 6 observations. It is even unknown wether these are 31 different respndents.

With kind regards

Karabiner
Thank you for your reply. They are 31 different respondents in total. There are two surveys, each with different questions relating to the same product. The surveys were allocated to 2 groups of people and the questions are targeted to that specific group. So there are 25 different respondents from one and 6 from another.

I just want to know whether a product is effective. All of the questions relate to satisfaction/ effectiveness of the product and they have been asked to strongly agree, agree, disagree, strongly disagree.

I can see this by looking at the numbers on bar charts but want to know if it's statistically significant.
 

Karabiner

TS Contributor
#4
So you want to compare what group A said in terms of questions a, b, c about product X
with what group B said in terms of questions d, e, f about product X ? Since you have different
measures, I don't know what sense this should make.

With kind regards

Karabiner
 
#5
So you want to compare what group A said in terms of questions a, b, c about product X
with what group B said in terms of questions d, e, f about product X ? Since you have different
measures, I don't know what sense this should make.

With kind regards

Karabiner
No I don't want to compare the groups with each other. I would like to see if each response in itself is statistically significant compared with the number of other responses within each individual question.
Thank you
 

Karabiner

TS Contributor
#6
No I don't want to compare the groups with each other. I would like to see if each response in itself is statistically significant compared with the number of other responses within each individual question.
Thank you
This is still not quite clear to me. Could you explain this a bit further, or give some example?

With kind regards

Karabiner
 
#7
This is still not quite clear to me. Could you explain this a bit further, or give some example?

With kind regards

Karabiner
Ok so question 1 asks if the product easy to use.

Responses are:
Strongly agree - 13
Agree - 11
Disagree - 1
Strongly Disagree - 0

Can I statistically say that the product is easy to use i.e. the agree answers are significant?

It looks obvious that the majority says yes but some questions are not so clear cut.

Thanks
 

Karabiner

TS Contributor
#8
You have to determine which percentage of (for example) "agree+strongly agree"
answers you would generally consider as a favourable evaluation of a product.
Something like: "I consider it as a positive evaluation if more than 66% [or 80%,
or 50%...whatever] say it is easy to use". If such a benchmark is identified, then you
can make a test of significance. The test would indicate whether in your sample the
rate positive responses is high enough to assume that it is > 66% in the population.

Statistical tests of signficance do not tell you whether something is "signficant" in the
sense of relevant, important, large.

With kind regards

Karabiner