The Importance of Probability-Based Sampling Methods for Drawing Valid Inferences

Understanding how well a sample of respondents represents the larger population from which is was drawn is critical to being able to generate valid inferences about the population. Probability-based sampling approaches have been a theoretical and empirical cornerstone of high-quality research about populations. However, non-probability sampling methods, such as those applied with the vast majority of opt-in survey panels, increasingly are being used to draw inferences about populations. This chapter examines critical issues that researchers need to be aware of when comparing these different approaches to sampling in their own work and when evaluating the work of others, and highlights the need for particular scrutiny of work produced using non-probability samples.