Abstract
As a result of the COVID-19 pandemic, online methodologies for developmental research have become an essential norm. Already, there are numerous options for recruiting and testing developmental participants, and they differ from each other in a variety of ways. While recent research has discussed the potential benefits and practical trade-offs of these different platforms, the potential empirical consequences of choosing among them are still unknown. It is critical for the field to understand not only how children's performance in an online context compares to traditional settings, but also how it differs across online platforms. This study offers the first comparative look at the same developmental task across different online research methodologies, allowing for direct comparison and critical examination of each. We conducted three versions of a test of preschoolers' ability to generate and apply second-order inferences to predict novel outcomes. Experiment 1 is an in-person task conducted at public testing sites in the vicinity of the university. In Experiment 2, we conducted an online-moderated version of the same task, in which an experimenter presented a recording of the procedure during a live video call with families over Zoom. Finally, Experiment 3 is an online-unmoderated version of the task, in which the same videos were presented entirely asynchronously using the Lookit platform. Results suggest that online methodologies may introduce difficulties and age-related differences in young children's performance not observed in person. We consider these results in light of the previous online developmental replications, suggest possible interpretations, and offer initial recommendations to help future developmental scientists make informed choices about whether and how to conduct their research online.