Skip to main content
. 2018 Nov 19;179(1):63–70. doi: 10.1001/jamainternmed.2018.5011

Table 1. Sample Characteristics Before and After Propensity Score Matching, by Participation Status in SNAP.

Characteristic Percentagea P Valueb
SNAP No SNAP
Before Propensity Score Matching (n = 1385)
Older adults, No. (%) 503 (36.3) 882 (63.7) NA
Age, y
65-69 34.9 26.5 <.001
70-74 29.1 23.8 .04
>75 36.0 49.6 <.001
Male sex 26.7 37.2 <.001
Race/ethnicity
Non-Hispanic white 46.7 59.7 <.001
Hispanic 18.6 13.5 .01
Non-Hispanic other 34.7 26.8 .002
US citizen 96.1 97.4 .16
Single or living alone 90.5 78.6 <.001
Educational level, bachelor’s degree or higher 10.1 10.1 .99
Residing in the South 44.0 42.8 .68
Employed 2.7 4.6 .08
Natural logarithm of per capita income, mean (SD) 9.2 (0.5) 9.3 (0.5) <.001
Fair or poor health status 56.2 46.2 <.001
Any functional limitation 90.3 84.6 .003
Any prescription drug coverage 87.2 79.3 <.001
≥$500 in out-of-pocket medical costs 27.3 51.1 <.001
After Propensity Score Matching (n = 1302)
Older adults, No. (%) 480 (36.9) 822 (63.1) NA
Age, y
65-69 34.5 34.5 .97
70-74 29.4 30.0 .81
>75 36.0 35.4 .85
Male sex 26.7 26.6 .99
Race/ethnicity
Non-Hispanic white 46.9 45.7 .71
Hispanic 18.8 18.9 .93
Non-Hispanic other 34.4 35.3 .75
US citizen 96.3 96.3 .94
Single or living alone 90.4 90.7 .87
Educational level, bachelor’s degree or higher 10.2 10.7 .82
Residing in the South 43.8 42.3 .65
Employed 2.7 3.1 .70
Natural logarithm of per capita income, mean (SD) 9.2 (0.5) 9.2 (0.5) .82
Fair or poor health status 55.8 54.2 .62
Any functional limitation 90.2 89.9 .85
Any prescription drug coverage 87.3 86.2 .62
≥$500 in out-of-pocket medical costs 27.5 27.4 .98

Abbreviations: NA, not applicable; SNAP, Supplemental Nutrition Assistance Program.

a

Sample weight not shown.

b

P values are from t tests for equality of means before and after matching, generated during the first stage of the propensity score–matching process. Before matching, t tests are based on an unweighted regression of each variable on the SNAP indicator. After matching, the regression is weighted using the weight generated from the matching process. Because of weighting and potential differences in the sample sizes for each regression, the proportions shown may not always match a simple ratio of the subgroup sample size and total sample size.