To balance your findings, you will also create a "null" hypothesis, which claims that the relationship that is to be proven in the research hypothesis does not exist.
According to Alan Agresti and Barbara Finlay in “Statistical Methods for the Social Sciences,” the null is directly tested and predicts no effect, and the alternative contradicts the null and predicts an effect. As indicated in Research Methods Knowledge Base, a "one-tailed" hypothesis specifies a direction, either an increase or a decrease, while a "two-tailed" hypothesis does not specify a direction, only a change.
If your analysis produces some paradox, or some puzzle, or you are aware of some situation that might negate your results, it is probably helpful to address them in a chapter like this. If your operationalizations are controversial or especially novel, you may want an appendix to explain further. Remainder of dissertation written This is probably the most common model.
An appendix is a good place to present the results of alternate tests. Weaknesses: (a) no spin-off article ready until the last stages are completed, (b) the model may bomb and then you are left w/o any interesting results!!!!
(Note: Those using some type of qualitative research design, such as case studies, present results here as well. Show how your results might be generalized to a wide range of alternate areas of interest.
If you are doing a comparative case study dissertation, this chapter may be a series of short chapters about each case.) - - This is an optional chapter. - - If your data are self-collected, an appendix presenting and explaining the data is useful, if not suggested.Stay in touch with your committee and especially your advisor.If you have consulted your advisor and committee along the way, the defense will be easy. Submission deadlines come quickly and there is generally a lot of minor editing and polishing to do before it is in an acceptable format for the university.For example, taking media consumption of violence as the independent variable and aggression as the dependent variable, the null hypothesis could state, “Media consumption of violence has no effect on aggression,” while the alternative hypothesis would state, “Media consumption of violence has an effect on aggression.” Similarly, if you wanted to create a one-tailed hypothesis, you would indicate a direction, such as, “Media consumption of violence increases aggression." Make sure that your statements are brief and straight-to-the-point and keep in mind the results you will measure in your study.Research Methods Knowledge Base states that hypothesis testing assumes that both mutually exclusive hypotheses (research and null) exhaust every possible outcome and in the end, one is accepted and the other is rejected. There is no lit review in this chapter, but the theory should flow directly from the weaknesses outlined in the previous chapter (so it is appropriate to reference this work as you develop the theory).The hypotheses you test are introduced, elucidated and listed in this chapter. Every table should be relevant to one of the hypotheses being tested.If you don’t, I guarantee you will forget important things that you have done (filenames, variable names and definitions, sources, dropping/modifying variables) 2. 200 pages is probably sufficient in most cases for a quantitative dissertation (140 60).In any case, don’t try to set the departmental record for length. Skim other dissertations recommended by your advisor.Don’t put it off – it will keep you off the job market and will hurt your standing in the department. The Prospectus defense – should be a formality if your committee has commented on several drafts and you have incorporated their comments (or at least come up with an acceptable defense that doesn’t make their criticism sound trivial or stupid).It is best to include this defense in the manuscript. During data collection and analysis – Take notes on everything you do.