15 Data validation
The MB5 data dictionary and validation tool are under development. Thank you for your patience as we finalize these resources!
In order to confirm that all labs have implemented the experiment correctly and in a way that will produce analyzable data, we ask each lab to run pilot data for one or more “participants”1 through the MB5 Data Validator. The validator checks data files against our data templates to ensure that all required variables are present and that data is in the expected format (e.g., integer, numeric, string) and response range. Examples:
1 This data can be generated in any way you’d like: the “participant(s)” can be a real (consented) baby, a research assistant, the lab PI, a local celebrity, etc. It just can’t be data that will be included in the final analysis!
participant_age_daysshould contain an integer between 90 and 455lab_idshould be a string that matches an entry in the MB5 Project Trackerpretermshould be one of the options in the set {‘preterm’, ‘not_preterm’}test_looking_time_novelshould be an integer between 0 and 10000 (looking time in ms)
15.1 ALL LABS
All labs, regardless of data collection apparatus or procedure, will need to validate a participant data file (1 row per participant, with columns for each participant-level variable (e.g., participant_id, participant_age_days, participant_sex, parent1_education) and a trial data file (3 rows per trial/36 rows per participant, with columns for trial and trial-phase level data; e.g., trial condition, looking time, presentation time). See Chapter 21 for more details and instructions.
15.2 Labs NOT using Eye Trackers
Labs not using eye trackers are asked to submit raw output files from PsychoPy or whatever stimulus-presentation software your lab uses. These file should include stimulus and timing info for trials and key presses.
15.3 Labs using Eye Trackers
Labs using eye trackers are asked to provide raw gaze data for all participants (either as a single file or 1 file/participant). This file (or files) should consist of 1 row per sampling period, with each row marked by participant_id, trial_num, and sample-level variables including x_gaze, y_gaze, and pupil_diameter.
15.4 Please refer to Chapter 21 for more information about about data formatting and to view/download the MB5 Data Dictionary and Templates
15.5 Validate your pilot data
Once you have your participant and trial data files ready to go, run them through the ManyBabies Data Validator. (Refer to the manual linked on the validator page for complete instructions).
15.6 Uploading validated pilot data
Please submit your lab’s validated pilot data using the MB5 Green-Lighting Form.
Please ensure that all uploaded materials are clearly labeled in the filename with your LabID (e.g. mb5_babylabPrinceton_pilot-participant-data.csv; mb5_babylabPrinceton_pilot-trial-data.csv). Check the list of LabIDs to confirm your lab’s unique LabID before uploading.