Pre-registrations for remote research through Discoveries Online
Our research team pre-registers all new projects through the Open Science Framework (for recent examples, see here and here). Below, we’ve provided a template for the language we use in pre-registrations to describe our various protocols for online data collection.
Sample Pre-Registration Text
Data Collection Procedures
Our sample will complete the study on their home or personal computers via an online platform (the Princeton & NYU Discoveries in Action Lab, or “PANDA”) developed specifically for unmoderated, remote developmental research (Rhodes et al., 2020). This platform captures both video data and survey data from participating families. Families on the platform are recruited in a multitude of ways, including but not limited to: community outreach events, social media advertising, posts on Prolific, and volunteer sign-ups at in-person data collection sites.
For this study, we will only recruit participants with children between the ages of 5.0-10.99 years. Parents of eligible children will be contacted via email through their PANDA account and offered the opportunity to log on and participate. Children and parents will then follow an IRB-approved consent and assent procedure. The study will be presented as a series of short videos and verbally narrated questions, which children will respond to with button presses. Following the child portion of the study, parents will be asked to fill out a brief survey on their own attitudes and beliefs. At the end of the session, families will be compensated with a $10 Amazon gift card for their participation.
We plan to stop data collection in the week when we have obtained usable data from at least 220 participants in each sample (or once we have obtained usable data from 50 participants per condition; whichever comes first). We will intentionally over-sample in this way to account for drops, which occur at a rate of approximately 10% of the full sample for online data collection based on previous work (Leshin, Leslie, & Rhodes, 2021).
All videos of participating children will be coded to make sure a visible child is participating in order to be included in the sample. Previous work on the PANDA platform has identified extremely low rates of parent or sibling interference (less than 1% of trials), and excluding such trials had no consequence for the overall pattern of findings (
Leshin et al., 2021). Therefore, we will randomly select 20% of videos to be coded by a trained researcher in order to estimate the interference rate. A second independent coder will code 50% of that subset (10% of the total). If the first coder's reliability is within the bounds of the interference level found in previous work, and if the first and second coders' reliability is high, we will conclude that the interference level is negligible and highly unlikely to influence the study patterns and therefore retain all trials for analyses. If the interference rate detected by the first coder is more than the rate estimated from previous work, however, then we will instead code every video trial-by-trial for instances of parental interference (with 20% also checked by a reliability coder), and we will exclude any trials where the parent or a sibling interferes such that there is a question of whether the answer reflects the child’s response. If more than 25% of a child’s responses are excluded for this reason, the participant’s data will be dropped from analyses. Our guide to interference coding is
We do not plan to exclude based on children's success on manipulation checks and comprehension questions; however, if more than 25% of our sample fails these, we will conduct planned exploratory analyses including just the subsample of participants who passed the respective measures. We plan to report the percentage of participants who passed and did not pass these comprehension checks.
Partial data from participants will be included if at least 50% of items are complete; participants with less than 50% completion will be excluded. We also plan to exclude any participants who are not within our age-range of 4.5-7.99 years old. For partial data that is included, we plan to just look at scores out of total responses completed.