I work in a quantitative social science where it is not common to formally state hypotheses in papers or presentations. I am applying for access to public health data, and the applications ask for explicit hypothesis statements. The applications will be reviewed by boards consisting of administrators, MDs, and researchers with PhDs in fields like public health and social work. Is the norm to state a general "scientific hypothesis," or a formal null/alternative hypothesis that will be used in statistical tests?

To be concrete, I am interested in studying the effect of X on some health outcome Y, because I have a hunch that X has a negative effect on Y. In that sense, my "scientific hypothesis" might be "X has a negative effect on Y." However, I will statistically test the null hypothesis "X has no effect on Y," against a two-sided alternative, i.e. "X has a positive effect on Y" or "X has a negative effect on Y."

Which of these hypotheses should I include in my application? If it's ambiguous, what are some relevant considerations? My feeling is that the scientific hypothesis might better reflect my motivations for the study, the null hypothesis is more precise.

  • I'm not so much an hypothesis person (more of critical realism). In your case, since the application explicitly ask for one, is there anything against going with what you have ... statistically test the null hypothesis Sep 28 at 6:16
  • 1
    This is a good question, and I suggest that you un-accept the one answer that has appeared, as this might lead to more answers from people who have experience of this or useful information.
    – toby544
    Sep 28 at 18:57
  • Fair enough, I'll un-accept the answer, if and if there are not additional answers within the next week or so I'll re-accept Peter Flom's answer.
    – sdg
    Sep 28 at 19:49

2 Answers 2


I would state my research hypothesis, unless the place you are submitting to has explicit suggestions. E.g. I've helped write a bunch of applications for NIH and NIDA funding, and those places have fairly detailed instructions. (But my experience is from many years ago).

  • Do you mean the "scientific hypothesis", as OP calls it? It is not clear.
    – toby544
    Sep 28 at 18:55
  • Yes. Or "alternative" hypothesis.
    – Peter Flom
    Sep 30 at 11:09

I would first question your statement that hypotheses are unusual in quantitative social science research in public health, in my experience, they are extremely common and your research should always be driven by a theoretically derived hypothesis.

That being said, data providers often ask for hypothesis or even a short description of the project, however, usually noone follows up on that. Just pose your hypothesis as framed above (X is negatively associated with Y) and you should be finde.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .