SAMPLE SIZE AND POWER ASSIGNMENT - Due by Thursday before 5 PM in the third week (email to small group leader)
Completion requirements
View
Start discussions: 1
Post replies: 1
Make forum posts: 1
SAMPLE SIZE AND POWER (1/2 page) ASSIGNMENT
(Note that you will turn in this assignment and the “STATISTICS” assignment at the same time. Incorporate these assignments into the same document as your earlier assignments. If necessary, revise the earlier sections of this document to reflect any changes you have made to your study.)
First complete the assigned readings and watch the video lecture for this topic.
Complete 1 or 2 (depending on your study design)
1. If your research question is primarily analytic:
- Specify your research hypothesis or hypotheses.
- Develop a sample size estimate using one of the examples in Chapter 6 of the Designing Clinical Research textbook as a model. This involves first deciding which statistical test you will use at the end of the study, then setting out the assumptions, and finally using one of the Chapter 6 appendices to estimate the sample size. For example, if your predictor variable is dichotomous and the outcome is continuous:
- State the correct statistical test (in this case, the t test).
- Develop your null and alternative hypotheses.
- Specify the effect size; standardized effect size; alpha and number of tails; and beta.
- Turn to Appendix 6A and estimate sample size.
- Comment on and justify the following:
- How you determined your estimated effect size and standard deviation. One of the best sources is prior publications of related work, so take this opportunity to become even more familiar with the relevant literature.
- Decisions about number of tails, size of alpha, amount of power, multiple hypotheses, etc. Imagine that you are writing this section for a grant application, and that you must convince a skeptical reviewer of the appropriateness of your plan.
- If you used the chapter tables, verify your calculation using the UCSF CTSI tool developed by our own Michael Kohn accessible via http://www.sample-size.net
2. If your research question is primarily descriptive, you won’t have a hypothesis but you still need to decide on a sample size:
- Follow the descriptive models set forth in Examples 6.4 and 6.5 in the Designing Clinical Research book.
- Comment on and justify your choice of expected standard deviation or proportion, desired precision (width of confidence interval), and confidence level.
- Verify your calculation use with Michael Kohn's handy sample size spreadsheet at http://www.sample-size.net
Adding discussions to this forum requires group membership.
There are no discussion topics yet in this forum.