The Implications of In-Person vs Remote Recruitment on the Baseline and Daily Characteristics of Samples of Individuals in Recovery from Substance Use

Google map

Statistical generalizability frequently relies on the assumption that participants are sampled from a population via a simple random sample. That is, it is assumed that each member of the population of interest has an equal probability of being included in the sample. In practice, however, when studying hard-to-reach populations, it is often not practical, or even possible, to draw such samples. Thus, research on hard-to-reach populations often relies on convenience sampling methods. However, decisions made when recruiting and onboarding convenience samples such as where to recruit from and what modality to use (e.g., social media, fliers, word of mouth), may unintentionally oversample certain members of a population. In turn, oversampling may result in samples that systematically differ from the population, thus creating discrepancies between convenience samples. People in recovery from substance use disorder (SUD) represent one example of a hard-to-reach population. We aimed to assess if convenience sampling strategies introduced selection bias as reflected by the baseline and daily characteristics of the resulting samples. This research aim was assessed by comparing two samples of people in recovery from substance use disorder drawn from two separate studies. The two study protocols used mostly identical measures for the baseline and daily assessments, allowing for direct comparisons to be made between the two samples.