Using Mechanical Turk to Study Parents and Children: An Examination of Data Quality and Representativeness

Abstract

ABSTRACT Objective Mechanical Turk (MTurk) is a popular data collection tool that is increasingly used to study parents and children. We examined the quality of data obtained via MTurk, the representativeness of parents on MTurk, and whether the wording of MTurk recruitment materials is related to sample characteristics. Method Participants were 650 parents with children aged 4 to 17 recruited for two separate studies (64.6% female, 83.8% Caucasian, mean age = 37.6 years). The Mental Health Survey (N = 322) recruitment materials described a study about mental health and the Health Survey (N = 328) materials only referenced health; both surveys restricted participation to parents in the United States (U.S.). Parents completed measures about demographics, mental health service seeking history, and parent and youth psychopathology. Results Participants provided reliable responses on study measures, with low rates of missing data. Participants were more female, less racially/ethnically diverse, and more educated than the U.S. population, and also had more children per household. Over 40% of parents and 30% of children had previously received mental health services and many scored in the clinical range on measures of psychopathology. Contrary to hypotheses, mental health help seeking was higher in the Health Survey than the Mental Health survey. Conclusions MTurk is a feasible method for child and adolescent clinical psychology research, although participants may not be fully demographically representative of the general population and could possibly be better considered an at-risk or clinical sample than a community sample.