Survey Analysis

Hey Guy's I really need your help. I am not a statistician. I have to analyze surveys that have 6 observations (people) and 20 factors (food preference with answers; like, dislike, neither or no)
What method should i use to analyses this on SAS or Excel


Ninja say what!?!
6 isn't much. Any way you can get more? I doubt you'll be able to infer anything with a sample size of 6.


Ambassador to the humans
Yeah about the best you're probably going to be able to do with six observations is descriptive statistics.


TS Contributor
Hi zqkal,

The best analysis for you will depend on what you want to discover in your data, i.e., your objectives with the analysis. If you can tell us a bit more about your project (and if you can gather more data) we could contribute with something more meaningful.
I am doing a research on the effect of after school meal preparation on obesity. I have a study and control group. Both groups take the survey before and after the intervention.
The survey asks
1.Have you tried this item before? YES or NO
I asked each student to fill the survey before and after the intervention for 20 food items. i.e., each person has 20 response before and after.
i have 6 in the study and 9 in the control group
I want to know

1) Among those who haven't tried the food item (as reported on the pre-survey) before the intervention, what is the likelihood that they tried it after the intervention (as reported on the post-survey)
2)Among those who have tried a food item, how did their food preference for that food item change from pre-survey to post-survey


No cake for spunky
Each method you use (ANOVA, EFA, regression) will have a different minimum sample size. A very general rule of thumb is that you need a 100 cases plus several more for each variable you will analyze. I suggest Fidel and Tabachnick's "Multivariate Statistis" as they have a good general coverage of methods, required sample size etc. It is a great place to start if you are new to stats because it addresses very practical questions like this.


TS Contributor
I'd also suggest some multivariate technique but you sample size wouldn't be enough. If you could increase your sample it would be great but if not, then maybe using Mac Nemar's test only in the most important meals could give some information.