Focusing on the relevant probabilities in population assignment to identify and correct bias

Abstract Population assignment methods enable biodiversity managers to identify the population origins of individuals or samples, which can help to determine the population‐specific impacts of human activities. Although population assignment is a powerful tool, accuracy can be low when populations have unequal odds of being sampled, which may arise from factors such as differences in population sizes, migration patterns, or behavior. When the probability of being assigned to population 1 given the individual is from population 1, , is assumed to be equal to the probability of being from population 1 given that the individual is assigned to population 1 , assignment confidence will be overinflated if the probability of sampling differs between groups. One potential consequence of this overconfidence could be inaccurate delineation of protected areas. To assess the reliability of population assignment, the relevant probability must be used: the probability that an individual belongs to a population given that it was assigned to the population. Understanding and calculating this conditional probability is essential to avoid misinterpretation of results. We demonstrate how to calculate conditional probabilities for assignments, and to identify the conditions under which the ability to assign individuals accurately is poor. We also demonstrate how the probability of correct assignment can be low when dealing with unequally sampled populations. To mitigate these issues, we suggest using population assignment with multiple sources of data and in conjunction with other methods to improve population assignment.