Model comparison table.

Due to the challenging nature of surveying exploited workers, scholars are often limited in their ability to standardize data collection efforts. Using two samples of survey data from workers in the Costa Rican fishing industry, we compare latent classes of exploitative labor between a subset of individuals surveyed through a proportional probability sample (PPS) and Vincent-link-tracing sampling (VLTS). Results suggest there are some differences between latent classes across the differently sampled groups, with those in the PPS sample being more defined by severe forms of exploitation. Despite these differences, latent classes largely looked similar, with each group having a more highly exploited class and a more minimally exploited class. This study suggests that sampling techniques may affect the broader understanding of the issue, but most of the variation of exploitative experiences may lie in the characteristics of specific contexts or occupations.