Confidence intervals allow interval estimates, which provide substantially more information than p-values that force researchers to dichotomize a result as “significant” or “non-significant”. With the move away from p-values, given their many many many foibles, journals and researchers are moving towards using confidence intervals to guide data interpretation. Unfortunately, confidence intervals, when interpreted incorrectly, are susceptible to many of the same shortcomings as p-values due to widespread misconceptions, particularly when dichotomized to include or not include the null. P-value functions offer a useful way to visualize confidence intervals and allow for more reliable interpretations that do not succumb to needless dichotomization.

These figures use data (updated twice weekly) from The COVID Tracking Project to present the state-level inequities between proportion of COVID-19 infections and deaths by race/ethnicity in each state as compared to the proportion of people by race/ethnicity living in each state (estimated from 2019 US Census Bureau Projections).