What counts as a comparison? Say we have a test in which we are comparing many variables to a dependent variable, using different methods. Say ttest, chi2 and then finally a multiple regression on parts of the data.
Would we include every single ttest, chi2 and every variable in the multiple regression when doing our bonferroni correction?
What if we have 3 dependent variables that we are investigating in the same study. Would we could everything in the same way over the course of the 3 variables, or would each dependent variable be counted as one "unit"?
 A: You perform a Bonferoni correction to avoid an inflation of the Type 1 error rate (rejecting the Nullhypothesis although it is true). The Chance of making a Type 1 Error when you do a single test for example a t-test is your alpha level, usually 0.05.
However, when you perform a second test then the chance that you get at least one Type 1 Error increases to almost 10% instead of the 5% that you had when you only did one test.
In order to avoid this inflation we use the above mentioned Correction.
Technically, for each p-value that you/the computer calculates you need to correct for. But most researchers do not do that. In fact, staying true to the definition, you would need to correct for any test for which you calculated a p-value across your whole life.
As you can see that is almost impossible and here is also where the whole concept of p-values kind of breaks apart.
In a more practical way, correcting your p-value is something that researchers do not really agree on. SOme only do it for post-hoc tests of ANOVAS. People, with a more thorough understanding usually correct for more test. But it all depends on the situation, your supervisors and how strict you feel about Type 1 inflation.
The best idea is too just try avoiding todo multiple tests. 
For example you can use a MANOVA for your three dependent variables, instead of testing them separately ( If they are related). Also if you have to correct do not use the Bonforroni Corrections ( it has a low power) I recommend the Benjamini-Hochberg correction. 
