Can I change my hypothesis? - Dr Newman

"Can I change my hypothesis to: That there is a difference between a phone system that uses mnemonic commands and a phone system that uses arbitrary control codes"   "Or can I use the above and also check the original hypothesis: That a phone system using mnemonic commands will produce fewer errors than a system that uses arbitrary control codes?" in other words test for 2 hypothesise...? Reply from Jean Newman. "You should stick to the original hypothesis (i.e. mnemonic system will produce fewer errors...)."  "Just saying that there is a difference is too weak as a justification for doing an experiment.  Perhaps you're confusing it with the concept of a hypothesis that is taught in statistics, where you are trying to reject the null hypothesis (i.e., that there is no difference, although sometimes one actually has to make a prediction about the direction of the difference, as in the use of one-tailed t-tests).  Very few papers would be published if they just had to say that there was a difference somewhere, but not specify in advance how the conditions will differ.  The point is to use prior research to make predictions about the nature, and the direction of, that difference.  Statistics are still used, however, to "detect" a difference,  ...but the pattern/direction of differences must always be predicted in advance.   Note:  advanced students may realise that there are post-hoc tests that can be done to look for patterns "after the fact" of a difference being detected, but they're well beyond this course." Home: Staff/Stu report & notices. Home Staff-Student report: cont. Page 2   Lecture notes & lab report notices. Page 3 Lab: Slips of the Mind. Page 4