Section 1: Chapter 4:
USMLE Biostatistics: Study Samples, Populations, & Randomization
Understanding the implications of study samples, study populations & randomization are fundamental for success in the USMLE as well as in successfully producing, utilizing, and applying medical research.
One of the key elements to consider when reading a research manuscript or USMLE biostats question is the study population, as well as whether or not that population was randomized and if it was, how was it randomized? The sample selected for a study has huge implications on the validity of the study and very often, the main source of error or bias in a study has to do with the population selected. For this reason, researchers dedicate a lot of time towards ensuring that their study population is adequate and unlikely to create bias. Randomization of the study population is an extremely important concept in research. Randomization is performed in order to ensure that the study groups (control group and experimental groups) are identical in every regard except for the variable being tested. When this is the case, any differences in outcomes observed between groups can only be explained by the variable being tested (the exposure), because everything else was the same between groups. This process of assuming that the exposure group and control group are identical in every regard with the exception of the exposure variable is fundamental in research, and it is when this assumption is incorrect that errors (bias) occurs.
In this chapter we will discuss and name the different ways that study samples are obtained in research as well as their implications on validity and risk of bias. We will also introduce the concept of randomization and explain the rationale for using it and what can happen when randomization is not adequate. We will discuss the following topics as they relate to USMLE and medical school level biostatistics.
- Definition of study samples and study population
- Definition of probability samples; what are probability samples and why are they important in research?
- Definition of non-probability or convenience samples; what are convenience samples and why are they bad for research? (they increase the risk of selection bias)
- Methods for obtaining probability samples; random sampling, systematic sampling, stratified sampling, and cluster sampling.
- Methods for obtaining non-probability samples; quota sampling and convenience sampling.
- Study sample representatives and the selection bias.
- Randomization; what is randomization and why is it important in research?
- Assessing randomization quality; what are baseline characteristic tables?
- Video length: 22 min
- Practice exercises: 10 USMLE style multiple choice questions with in-depth explanation
- Estimated time to complete video lecture and practice exercises: 1 hour
- Next chapter: Chapter 5 – Random Error, Systematic Error, Selection Bias, and Methods to Control for the Selection Bias