Food service satisfaction survey - 4 point Likert scale

Good morning,

This is my first posting... I am a dietitian with a master degree. I am doing research in Clientele satisfaction (i.e. quality of Food service and Foods served) in an geriatric institution setting - long-term care.

For our latest survey, we had 30 questions which mainly used 4 point Likert scales (Very satisfied, Satisfied, Not very satisfied and Unsatisfied) to ‘force’ the results into either satisfactory or unsatisfactory category. Our sample size is small (n=20).

When calculating the average response and standard deviations, we allocated percentage values to each scale point: 100%-Very satisfied; 66% Satisfied; 33%-Not vey satisfied and 0%-Not Satisfied.

I am unclear as how to 'explain' the wide standard deviation associated with the averages resulting from such a scale…

1) Is there a way to explain or minimize the appearance of large STD DEV ? For example, we set our quality norm at 75%. Some results are 75%+/-15% and make it seem unsatisfactory (which is in fact only ½ point scale…) - in percentages, it looks huge…

For most of the questions, we also have a Ranking of Importance (Very important, Rather important, Of little importance, Not important) which we have used to obtained ‘weighted satisfaction rates’.

2) What type of other stat tests can I do with such a scale to complete my analysis?

3) Is there specific documentation I could consult to help with this type of surveys?

4) Would you know about Satisfaction surveys for demented patients or for their proxies?

Thank you,
Isabelle MSC RD


TS Contributor
Since you have such a small sample, I would also report, for each question, the % of responses that were on the "not satisfied" half of the scale. The questions with the highest % would be your areas for potential improvement, especially if they were "important" issues.
I think having good food should be our first priority in life. Without food Its not possible to live a happy life. I normally two times a week visit restaurants for junk food but I really love junk food. Even I asked my friend John to suggest me a list of restaurants and he helped me satisfy my junk food needs. I had a restaurant in past so I know how they prepare good food.


No cake for spunky
I think 20 cases is likely an issue. I expect if you have more your SE would be lower. I also think that its pretty much impossible to generalize to any large population with 20 cases.

The way to reduce it is likely gather more data. :p