In: International Journal of Social Research |DOI|

Abstract

Non-probability samples have become increasingly popular despite some criticism. This paper examines social media sampling as a tool for accessing hard-to-survey populations. Our study targeted individuals in Germany using ads on Facebook and X, yielding 4,590 respondents. We compared these samples with high-quality probability samples (SOEP, ESS) through measurement equivalence analysis of shared measures between samples. Results show that our social media sampling strategy yielded effective sample sizes for our target population that exceeded those from SOEP and ESS by ratios between 2:1 and 5:1. Our findings suggest that non-probability sampling can be a viable method for researchers examining relational patterns among variables in hard-to-survey populations. Because we observe varying levels of measurement equivalence, rigorous methodological strategies for post-hoc analyses are recommended. We propose measurement equivalence analysis as a post-hoc assessment strategy to quantify the analytical effectiveness of the employed sampling strategy.