Table 1.
Initial Covariate Balance between Treated (Labeled) and Control (Unlabeled) Restaurant Customers.
Table 2.
Comparison of Absolute Standardized Biasesa of Covariates between Treated (Labeled) and Control (Unlabeled) Subjects after Propensity Score Matching and Weightingb.
Fig 1.
Boxplots of Absolute Standardized Biases for the Covariates in the Propensity Score Model.
Methods. 0-original sample, no matching or weighting; 1- greedy 1:1 nearest neighbor matching, 2- greedy 1:1 nearest neighbor matching with exact match on race and caliper of 0.2, 3- full matching (constrained), 4- inverse probability weighting
Table 3.
Effect of Menu Labeling on Nutritional Outcomesa Before and After Propensity Score Matching/Weightingb.