3

Consider the frequentist argument: "I will believe A since not-A makes the data improbable". Consider the Bayesian argument: "I will believe A since given my beliefs on A and the data generating process, it is more probable than not-A".

Are there any other argument types and how would you construct an argument? (Fiducial? Fuzzy logic? Dempster-Shafer? Possibility theory? Bounds? ...)

JohnRos
  • 5,336
  • 26
  • 56

1 Answers1

-1

Sure, you can make one that doesn't use hypothesis tests (or any statements about A and NOT A) at all. Something like:

We had evidence of a relationship between A and B because of (cite literature) but we thought that this needed further exploration because (give reasons). When we did (cite analysis done) we found (cite relationship found) which (strengthens/weakens/alters) our notion of the relationship. Reasons for this are (give more reasons)

Peter Flom
  • 94,055
  • 35
  • 143
  • 276
  • 1
    Thank you Peter. I agree this does qualify as a data driven argument. I guess I had a more quantitative approach in mind. – JohnRos Jan 11 '13 at 21:48