Do all dissertation samples sizes need to equal 95% confidence and 5% interval? Would 90% / 7% work? Is it up to the school or up to the dissertation committee? I am starting to get to the end of the road and really focusing to get this done.
I think the bottom line is that whatever your committee says is OK is OK. Much can depend on the size of sample(s) and, especially, on the tests being applied. While a small sample and a non-parametric test (e.g., chi square) might want a 99/1 or 98/2, while many of the t-tests used will be considered significant at the .1 level (i.e., 90/10). (A curious coincidence -- or is it significant -- is that the t-test was invented by Bill Gossett, but not the one who used to contribute here in the early days.) --John Bear, who decided not to go to Stanford in part because Quinn McNemar taught advanced statistics to doctoral students at 7 am, five days a week.
John's right - it is up to the committee. One of the things that the committee will probably examine is what the standard is in your field of research. Shawn
My Econometrics prof used to say that the 95/5 bit was because it's the default in most stats programs. Remember the words of Homer Simpson: "Oh, people can come up with statistics to prove anything, Kent. 14% of people know that."