The purpose of the discussion is to evaluate and interpret the results of the study as well as to draw implications from them.

The discussion section should be the mirror image of your introduction: it should start out narrow, discussing your specific study, then become broader, discussing the wider implications of your findings. Open with a clear statement of the results of your study. Do not include statistics here. Do not simply state that the results were significant - tell the reader what the results mean. Were your hypotheses supported, or not? If not, what did you find instead? Be brief here -- don't spend a lot of time going over detailed results you covered in the results section. In the discussion, refer to your hypotheses in more theoretical terms (e.g., time's effects on test performance, gender differences in attitudes, mental processes inferred from reaction time studies).

Now you are free to discuss what you think the results mean and what their practical and theoretical implications are. You do this by using the similarities and differences between your research and the previous studies outlined in your introduction to help confirm and clarify your conclusions. Tie your results in with previous literature, and with the larger issues with which you started your introduction. Do not just restate what you have said in the introduction and only rarely refer to different studies in the discussion than were mentioned in the introduction. You must do these tasks whether your hypotheses were supported or not. If they were, what have we learned? Why are the results important and interesting? If your hypotheses were not supported, you have to show some thinking skills. Were the results in the right direction, but failed to achieve statistical significance? If so, discuss how you might increase the power of your manipulation in future research. Did you find significant results that were not predicted, or were even counter to your hypotheses? If so, how might you explain these findings? In retrospect, can you think of reasons why your results might have come out this way? Do they suggest new directions in which to take future research to explore these surprising findings?

You should also discuss any limitations of your study. Limitations should refer more to the inability to generalize information than to flaws in the study's design. Be careful that if you found significant results supporting your hypotheses that you do not explain them away with criticisms about your experiment. You must also be careful not to fill your discussion with excess or long lists of potential confounds in order to hold on to your original hypothesis when your results are not supported. Usually, negative findings should be accepted as such and should not be explained away.

Also, in this section, you should make suggestions for future research. Any suggestions should be improvements or elaborations to the experiment (not to degrade your study). The suggestions may be an attempt to answer further questions arising from your study, or to improve a weakness you could not avoid or did not see originally. Tie limitations into the need for future research.

It sometimes seems that every single paper students turn in makes the following points in the discussion section: (1) a limitation of this study was that there were not enough participants. If we'd only had more participants, everything would have been perfect. (People use this explanation even when their results were significantly reversed from what was predicted. Note that in such cases, more respondents would presumably only lead to a stronger reversal); (2) a limitation of the study is that it was run with university students. If it had been run with a random sample of the population, everything would be fine. (Unfortunately, they don't explain what it is about university students that might lead to results different from those found in the general population). In their suggestions for future research, they then say that the exact study should be rerun with a larger sample or with a different population. They don't make it clear exactly why that would be helpful.

It's not that these statements are necessarily wrong. If your results were in the right direction and very close to significant, it's appropriate to point out that a larger sample might have increased your power sufficiently to achieve significance. University samples are different from those in the general population, and in ways which might potentially influence results in a variety of fields. However, if these points are the only ones you make in the discussion section, you'll be showing a lack of imagination and potentially a lack of insight into your study. Use the discussion section to show you've understood what you've done, you know why it's important, you've carefully thought through its limitations, and you have worthwhile suggestions as to how the research could be improved or extended.

In summary, the key questions you are trying to answer in your discussion section are (1) did you find what you were looking for; (2) if so, what does it mean; (3) if not, why not; (4) what conclusions and theoretical implications can you draw from your results; (5) what are the limitations of what you did; and (6) where should things go from here?