Can these data be grouped? Whithin a certain population, I have selected three samples (in three different periods of time) and determined for each the frequency of individuals that present a specific characteristic. I want to determine if all these results can be grouped together or if there are differences between samples. Can I do it by performing a chi-square test for observed vs. expected frequencies and pooling the data if there are no significant differences?
If I do the sampling (three times also) in other populations, should I repeat the same procedure for each in order to determine if data from the same population can be pooled?
 A: Question 1
If your dataset is something like this
Sample 1           Yes=200             No=180
Sample 2           Yes=156             No=130
Sample 3           Yes=300             No=320
You can use chi-squared test.
Question 2:
I am not sure if I understood your second question. What do you mean with other populations?  Please, give a practical example
A: I suggest three samples in different periods will not be usefully susceptible to sophisticated analysis.
Whether they might be grouped should most obviously depend on their similarity… here, handicapped by there being only three samples.
When you want to determine "if all these results can be grouped together or if there are differences between samples", what differences have you in mind?
To the extent they were similar - ideally, identical - grouping samples over different times might work.
In the same time, how could three samples suffice for a chi-square test for observed v expected frequencies, or pooling of data?
You should repeat the same procedure for anything you hope to compare but whether that would allow data to be pooled would depend on the details.
