You need to perform ANOVA (which is still a type of multiple t test) since you have more than 3 treatments and are making multiple comparisons. Otherwise you cannot make a comparison of how treatment 1 ranks alongside 2 unless you include all data together. Strictly speaking, there is no such thing as a control, but rather it is also a treatment since you are comparing it with 2 others. You may have pseudoreplication since you performed all the tests under the same conditions (but it's okay). Now with regards to Bonferroni, it must be used with multiple comparisons to decrease Type 1 errors that are associated with the comparisons. But you only apply Bonferroni in ANOVAs, not individual t tests. I hope that helps :)

Andrew Paul McKenzie Pegman Thanks for the response! I have a comment, if sample size is more (n=50 or 100 of one group), Bonferroni correction must be used to reduce the error and if n=3 for one group, Bonferroni correction is not needed, is it true? I see most of the articles using one way Anova without Bonferroni correction.

TIA

Jochen Wilhelm

Now I encountered an another problem. I have two sets of data (1 to 3 column, I set and 4 to 6 column, II set). I wanted to determine significance differences for 2 and 3 vs 1 or 5 and 6 vs 4.

I used one way Anova followed by Dunnett test on Sigma plot. The first set is pretty good but with the second set, it shows "equal variance is failed (P=0.009), Do you to want to run one way anova on ranks" (attached image), can you please help me out how could I solve this issue?