About the Course:
This course, "Designing Valid Statistical Studies" covers the issues that need to be addressed in order for a study to produce statistically valid conclusions. We mainly deal with statistical study designs that are not experiments, and are used with observational data. We use examples mainly from the literature, which is where the methodology of observational studies has been used and developed, but the principles apply to any study based on observational data. The first session covers different study designs, with an emphasis on designs that can be used with observational data (clinical trials are reviewed briefly, but those seeking a more in depth treatment should take the company's sequence of courses in clinical trial design and analysis). Participants will learn what types of bias and extraneous factors can endanger studies, and how to avoid or adjust for them.
Course Content:
MODULE 1: Review Types of Studies, Validity
Methodologic issues Clinical trial (brief overview) Observational Case-control Cross-sectional Validity vrs. precision Internal vrs. external validity
MODULE 2: Bias and Confounding
Selection bias (brief overview) Information bias (brief overview) Confounding
MODULE 3: Controlling Extraneous factors
At design stage At analysis stage Randomization Matching
MODULE 4: Matching and Stratification