Medical-statistic

Week 4 Worksheet
Students: Answer each of the following questions when reading the assigned articles for Week 4
1. What is the research question the paper sought to address? [or, what were the hypotheses?]
2. What were the main findings of this paper?
3. Assume the findings in the paper are valid. What is the clinical significance of the findings? How should it affect the way dentists practice?
4. What was (were) the main outcome variable(s)? (dependent variables) How were they measured, and do those measures seem appropriate?
5. What was (were) the main grouping variable(s)? (independent variables) How were they measured, and do those measures seem appropriate?
6. Were statistical differences found between or among the groups, if statistical comparisons were made? [Statistical differences means the groups were found to be different at the p<.05 level. This means a chance of a Type 1 error was approximately 5%. A Type 1 error is concluding from the statistical analysis that there are group differences, when in fact there are no group differences. The p value just tells you how likely a Type 1 error is if the rest of the study is perfect, but it doesn’t tell you how important those differences are. ] 7. As a result of the statistical analysis, what conclusions are valid, regarding the study question? 8. Do you see any obvious ways in which a Type 1 error might have occurred? Type 1 error happens when statistical differences are obtained between groups that aren’t real. The statistics may have found differences, but the analysis wasn’t valid for some reason. For example, if the two groups being compared were different on characteristics other than the outcome variable, then the statistical difference wouldn’t be due to the treatment, but to some other variable. 9. Do you see any obvious ways in which a Type 2 error might have occurred? Type 2 error happens when statistical differences are mistakenly NOT obtained between groups. For example, the treatment may have been effective, but the sample size wasn’t big enough to capture it. This is called “Power.” If the sample size wasn’t big enough, the study didn’t have enough power, and a type 2 error occurred. Type 2 error can also occur if the groups are different on some important characteristic that masks the treatment effect. 10. Does the study present effect sizes, or odds ratios? If so, deduce how important the group differences are, if they exist. If not, what does that mean about this study? 11. Was the study design a randomized controlled trial? 12. Do the authors draw causal conclusions? Does the study design warrant causal conclusions? Why or why not? 13. What are the limitations of the study discussed by the authors? 14. What other limitations, if any, do you see? [ok to say none] 15. Do you think the findings are valid (generally true)? Why or why not? 16. If valid, do you think the findings are important (“clinically significant”)? Why or why not?