Hi, I need some information. After collecting data from various experiments, I am proceeding with a statistical analysis. Since I work with frequencies, I was told to use the chi-square test. After performing it, it tells me that my data is statistically significant.. but it doesn't tell me which conditions are, I should go and analyse them individually. Is there a post hoc test that tells me in which categories (which I am considering) there is this significant difference?
Thank you all!
As you wrote you are comparing frequency distributions. It's not that your data are significant, but deviations of your data from model assuming that all your data is a sample from one distribution is statistically significant (as measured by the chi-squared statistic). I wonder why you tested this hypothesis when this is not interesting for you? Just compare the frequencies you are actually interested in. If this is a kind of a screening to identify "some comparisons" of which the observed deviations are statistically significant, you need to consider an adjustment for multiple testing (e.g, Bonferroni, Holm, Hochberg, etc.).
Statistical analysis depends on your study objectives. Every statistical tool is design to solve some peculiar problems. Example, you can't use regression analysis if you are not interested in knowing relationship that exists between variables.
Hi Giulia,
could you please add more details about your "various experiments" and data collections?
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Physical Phenomena
- Luminescence
- Fluorescence