Presentation on theme: "Experimental Plan and Study Design Janice Weinberg ScD Professor of Biostatistics Boston University School of Public Health."— Presentation transcript:
Experimental Plan and Study Design Janice Weinberg ScD Professor of Biostatistics Boston University School of Public Health
What to do: 1)Connect the experimental plan / study design to the study question(s) / aims 2)Justify your choice of study design. Why is your study the best way to address your study question? 3)Use correct study design terminology. Observational, cohort, case-control, pilot, prospective, retrospective, matched, etc… Clinical trial ? (parallel, cross-over, factorial, randomized, blinded, controlled, etc…)
4)Justify your sample size. This may be a power analysis or based on practical considerations (depends on the study) 5)Make sure your “unit of observation” / sample size is clear. How many samples, animals, subjects, experiments? Number of measurements per animal or subject? 6) Explain how (if) you have access to all relevant data needed to answer your study question(s). Explain what information will be collected. If any important data isn’t available explain why and discuss how your results could still be valid.
7)Consider sources of bias. Explain how you will control for these sources or explain why you will still get useful results in spite of them 8)Consider possible “effect modification”. Could results differ for subgroups of individuals? Possibly make this part of your plan 9)Address loss to follow-up in a prospective study. Explain how you will minimize loss to follow-up and missing data
10) Make a picture! – Study design illustration – Study flow chart – Study evaluation chart
Pet Peeves 1)Obviously trying to sound “smart”. We know you are smart. Focus on being clear. 2)A list of study procedures over time in the text, instead of using a chart. “At baseline we will collect A,B,C and D. At Week 1 we will collect A,B and C. At Week 2 we will collect A,B and C. At week 3 we will collect A,B,C and D.” (and so on…)
Pet Peeves 3)No study design illustration 4)A sample size calculation for an outcome that is not primary (or secondary) or that has no connection to the study question(s) 5)Unclear connection between the study design, study question(s) and analytic plan
Deal Breakers 1)An inability to determine the study design 2)A study design that cannot address the study question(s) 3)A study design that cannot be completed in the allotted time 4)Ethical issues. Is your study design ethical? 5)No sample size justification (for certain studies this is a deal breaker)
Your consent to our cookies if you continue to use this website.