Section 1: Chapter 6:
USMLE Biostatistics: The Measurement Bias and Methods to Prevent It
The different kinds of research biases are a favorite of the USMLE. Learn everything you need to know about the measurement or information bias here.
The measurement bias, also known as the information bias, is an extremely high yield topic for the USMLE biostatistics section. As mentioned in the previous chapter, there are three main “families” of bias or systematic error; selection bias, measurement/information bias & confounding bias. There many subtypes of selection bias and measurement bias. In this chapter we will focus on the measurement bias and the different kinds of measurement bias that are frequently encountered in the USMLE.
So what is the measurement bias? The measurement bias or information bias is a type of bias which occurs due to errors in the way that the variables are measured in the study. In other words, the measurements made in the study are systematically different from the true values of the variables. For example, a blood pressure cuff which is too tight will tend to overestimate blood pressure values. Therefore, a study which measured blood pressures using a cuff which is too tight, will suffer from measurement bias because the measurements made are systematically different from the true values.
There are many possible ways that a study may incorrectly measure the variables it seeks to measure. For this reason, there many different types of measurement bias. Some of the most common types of measurement biases to know for the USMLE are listed below. While these biases all describe different mechanisms, they all lead to the same net result, namely, systematically incorrect measurements.
- Observer Bias: The observer bias is a type of measurement bias which occurs due to the preconceived notions of the observer regarding the study. For example, a pathologist may be more likely to diagnose mesothelioma if they know that the subject was exposed to asbestos.
- Recall Bias: The recall bias is a type of measurement bias which occurs when the study participants are asked to remember past events, but they tend to remember certain events more than others. For example, cancer patients may be more likely to recall an exposure to a potential carcinogen more than non-cancer patients.
- Reporting Bias: The reporting bias is a type of measurement bias which occurs due to selective withholding of certain information by the study subjects or researchers. For example, study participants may under report how many cigarettes they smoke per day.
- Surveillance Bias: The surveillance bias is a type of measurement bias which occurs when the exposure group and control groups are screened for the outcome variable differently. For example, researchers may scrutinize patients who they know received the exposure more intensely than patients who they know did not.
- Hawthorne Effect: The Hawthorne effect is a type of measurement bias occurs when study participants change their behavior because they know they are being observed.
- Lead-Time Bias: The lead time bias occurs when earlier detection of a lethal disease is confused for increased survival when in reality there is no increase in survival time. This is a type of measurement bias because the measurements made are incorrectly attributed to increased survival, when in reality, there is no increase in survival.
In this chapter we will discuss the measurement bias, the different kinds of measurement bias including the observer bias, recall bias, surveillance bias, Hawthorne effect, and Lead-Time bias and the methods to prevent the measurement bias. Specifically, the following topics will be discussed.
- Definition of the measurement/information bias
- Examples of the measurement bias
- Types of measurement bias
- Definition and examples of the observer bias
- Definition and examples of the recall bias
- Definition and examples of the reporting bias
- Definition and examples of the surveillance bias
- Definition and examples of the Hawthorne Effect
- Definition and examples of the Lead-Time Bias
- Methods to prevent the measurement bias in research; calibrating measurement devices, establishing measurement criteria, and blinding participants
- Video length: 15 min
- Practice exercises: 10 USMLE style multiple choice questions with in-depth explanations
- Estimated time to complete video lecture and practice exercises: 1 hour
- Next chapter: Chapter 7 – The Confounding Bias and Methods to Prevent It