Which measurement would a researcher use to test for reliability when the data are in dichotomous format?

The KR-20 coefficient is specifically designed for assessing the reliability of tests that involve dichotomous data, meaning data that can have only two possible outcomes (such as “yes/no” or “true/false”). This coefficient is a type of reliability measure that evaluates internal consistency, which is essential when determining how well different items in a test measure the same underlying construct.

In cases where the data consists solely of dichotomous responses, the KR-20 coefficient provides a more appropriate and valid measure of reliability compared to other coefficients. While Cronbach's alpha is commonly used for continuous and ordinal data, it is not tailored specifically for dichotomous outcomes, even though it can be applied in those scenarios. Test-retest reliability focuses on the stability of scores over time and is not specific to dichotomous data. Interrater reliability measures the level of agreement between different raters or observers, which can be utilized in various data formats but does not specifically address the reliability of a dichotomous measurement in the same way the KR-20 does.

Thus, for dichotomous data, the KR-20 coefficient is the most relevant choice for testing reliability.