Table 3:

Covariate Balance before Matching and Using Three Matching Methods

	1		2		3		4
	Unbalanced Covariates		Propensity Score Matching		Nearest Neighbor Matching		Inverse Probability Weighted Regression Adjusted
	Standardized Difference in Means	Variance Ratio	Standardized Difference in Means	Variance Ratio	Standardized Difference in Means	Variance Ratio	Standardized Difference in Means	Variance Ratio
Weekly Sweetened Beverage Consumption	−0.27	0.75	0.10	1.33	−0.18	0.76	0.09	1.38
Lives in Seattle	0.36	0.98	−0.02	1.00	0.07	1.00	0.00	1.00
High Income (≥ 260% FPL)	−0.16	1.02	0.02	1.00	−0.01	1.00	−0.01	1.00
Non-Hispanic White	0.05	0.95	0.08	0.91	0.05	0.95	0.06	0.94
Non-Hispanic Black	−0.05	0.86	−0.04	0.87	−0.02	0.93	−0.06	0.83
Non-Hispanic Asian	−0.30	0.37	−0.11	0.69	−0.05	0.85	−0.06	0.83
Hispanic	0.09	1.25	0.05	1.13	0.02	1.05	0.02	1.06
50 or Younger	−0.54	1.03	−0.04	1.00	−0.07	1.01	−0.01	1.00
Completed College or Above	−0.09	1.04	0.00	1.00	0.03	0.98	0.00	1.00
Married or Partnered	−0.07	1.00	−0.03	1.00	0.00	1.00	−0.02	1.00
Democrat	0.12	1.00	−0.04	0.99	0.00	1.00	−0.02	1.00
Independent	−0.02	0.98	0.02	1.02	−0.01	0.99	0.03	1.03

Note: Standardized difference in means is phone mean response minus web mean response, divided by the standard deviation of the outcomes. Variance ratio is the ratio of the variances in each group. For columns 2-4, the full set of matching variables includes all variables shown, interactions between all variables including squared terms, and interactions between all interactions, including quadrupled terms. We also employ matching on all variables except consumption for the consumption outcome, but these results are not shown. Results are not materially changed in that specification.