I would like to add an item to the five suggestions that Beverly Garcia [20 Mar 95] posed.
6. Work directly with and for those who have the problem you are interested in researching. Ron Oliver developed an instructional approach that he thought was very successful, but he failed to address change management issues if the regular teachers weren't interested in adopting his solution. What if those teachers had been involved from the very beginning, asked if they thought there was a problem and, if so, how they would define it; how would they approach resolving the problem; is the solution compatible with their classroom/school cultures; is it effective; do they like it; do their students respond differently; etc. The researcher can still get publications as well as the satisfaction of knowing her work is actually being used. (Participatory research is not always feasible or acceptable to the client but, since we are an applied field, I think there is much more opportunity for its use in IT than in the more theoretical disciplines.)