Frequentist inference is the only form of statistics taught in my department, and I feel like it has a strong hold over many students here. But when I read data science blogs, I get the feeling that frequentist methods, ANOVAs, t-tests etc. are really looked down upon, which is in stark contrast to all the other graduate students around me

I understand that a problem with frequentist methods is the strong assumptions that get made (which vary depending on the specific method). This is in essence saying you need clean data and that is very rarely the case in real life

Can anyone provide some real life examples of where/why frequentist methods would fail? I'm looking for some strong arguments that I could make against those who hold such a strong pro belief in my department

  • $\begingroup$ For some weaknesses of the frequentist approach, you may look for (many are available on CV) posts discussing differences between Bayesian and frequentist approaches $\endgroup$ – Christoph Hanck Aug 20 '15 at 6:27
  • $\begingroup$ are the Bayesian methods akin to methods/tools used in the data science/analysis fields in real world job markets? $\endgroup$ – Simon Aug 20 '15 at 6:37
  • $\begingroup$ There definitely is quite some overlap. Many people argue, as you will see, that the Bayesian approach much more naturally allows to answer the questions we should really be interested in. That said, the Bayesian approach has very solid philosophical underpinnings, and the discussion thereof of course tends to be quite far away from applications. $\endgroup$ – Christoph Hanck Aug 20 '15 at 6:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.