Data used for a meta-analysis on the effectiveness of nudges, i.e. choice architecture interventions.
Format
A data frame with 447 effect size measurements and 25 columns:
publication_id, integer ID number for the publication. Note that two publications were erroneously assigned the same ID number, so this is not a unique publication identifier.
study_id, integer ID number for the study.
es_id, integer ID number for the effect size measured.
reference, publication citation in "Author(s) (year)" format. Due to two publications being assigned the same reference, this is also not a unique publication identifier.
title, title of the publication. Due to the error in assigning publication ID numbers, this is the unique publication identifier within the data set.
year, year of the publication.
location, geographical location of the intervention. This is given as a factor, rather than an integer, using the information provided in the codebook.
domain, factor giving the intervention's behavioural domain.
intervention_category, factor giving the intervention's category, based on the taxonomy in Münscher et al. (2016).
intervention_technique, factor giving the intervention's technique, based on the taxonomy in Münscher et al. (2016).
type_experiment, factor giving the type of experiment, as defined by Harrison and List (2004).
population, factor giving the intervention's target population. This is given as a factor, rather than an integer, using the information provided in the codebook.
n_study, sample size of the overall study.
n_comparison, combined sample size of the control and the intervention for the measured effect size.
n_control, sample size of the control condition for the measured effect size.
n_intervention, sample size of the intervention condition for the measured effect size.
binary_outcome, logical for whether the outcome scale is binary or continuous.
mean_control, mean of outcome for the control condition.
sd_control, SD of outcome for the control condition.
mean_intervention, mean of outcome for the intervention condition.
sd_intervention, SD of outcome for the intervention condition.
cohens_d, extracted effect size of intervention.
variance_d, variance of extracted effect size.
approximation, logical for whether effect size extraction involved approximation.
wansink, logical for whether the study was (co-)authored by Brian Wansink. This was added on revision, because, a few years before publication, Wansink had many papers retracted or corrected, due to various questionable practices, resulting in Wansink being determined to have committed scientific misconduct. This column was added to check whether the findings were robust to the exclusion of non-retracted studies by the Cornell Food and Brand Laboratory, of which Wansink was the director.