haHA! That's pretty funny (under the guise of helpful) to post a question and then attempt to answer it based on one source of seemingly empirical 'evidence'...particularly from such a questionable source.
The reality is NEVER as cut and dry as any study (or studies)...particularly without knowing the statistical significance of each. Top that off with test structure, content, and environment (among others) and you (Caslon9000) are surely passing on largely erroneous information. WTF?
Meanwhile, as a student (as most others have mentioned), it really depends on the test, my familiarity with the information, what other information is presented on the test, the time I have, etc., and even then there is no exact science to slate the 'rightness' or 'wrongness' of methodology.