Skip to main content
. 2011 Jun 3;47(2):794–818. doi: 10.1111/j.1475-6773.2011.01278.x

Table 4.

Decomposition Results for Health Care Expenditure among Whites and Latino Citizenship/Nativity Category by Time of Residence in the United States

Latinos Noncitizen Latinos ≤10 Years Noncitizen Latinos >10 Years Naturalized Latinos ≤10 Years Naturalized Latinos >10 Years U.S.-Born Latinos
Model 1: probability of any health expenditures
Whites (reference group): predicted value=0.88
Predicted value 0.68 0.53 0.61 0.64 0.77 0.73
Total difference −0.20 −0.35 −0.27 −0.24 −0.11 −0.15
Total explained −0.15 −0.26 −0.17 −0.13 −0.05 −0.11
Significant individual factors (% of total difference)
 Age 5 5 −12 9
 Education attainment 11 9 15 8 14
 Uninsured 21 24 18 53 21 15
 Usual source of care 24 25 28 40 22 21
 English response 10 12 10 7
Model 2: ln(total health expenditures)
Whites (reference group): predicted value=7.27
Predicted value 6.70 6.31 6.47 6.76 6.95 6.79
Total difference −0.57 −0.95 −0.80 −0.51 −0.32 −0.48
Total explained −0.27 −0.73 −0.39 −0.32 −0.01 (NS) −0.19
Significant individual factors (% of total difference)
 Age 16 20 6 25 29
 Education attainment 11 9 14 12 9
 Worse health status −16 −21 −19 −12
 Uninsured 17 25 21 19 8
 Usual source of care 10 14 10 14 8
Model 3: OOP share
Whites (reference group): predicted OOP share=0.39
Predicted value 0.45 0.58 0.51 0.43 0.41 0.41
Total difference 0.06 0.19 0.12 0.04 0.02 0.02
Total explained 0.06 0.16 0.09 0.04 0.01 0.02
Significant individual factors (% of total difference)
 Education attainment −6 −16 −16 −10
 Worse health status −12 −10 −40 −49 −22
 Uninsured 112 91 100 340 189 140
 Medicaid −18 −6 −10 −101 −50 −48
 Usual source of care 15 11 11 69 21 32

Notes. Data sources: Medical Expenditure Panel Survey and National Health Interview Survey (2000–2007). Factors that explain 5% or higher of the total differences are reported. Nonsignificant results for each decomposition model are excluded for brevity. All regression models include year fixed effects. Among individual factors, positive/negative coefficients indicate the share of explanatory variables positively/negatively associated with health spending disparities. The share of observed characteristics explained by the model is the ratio of total explained over total differences. For example, the share of observed characteristics between Latinos and whites for outcome 1 is −0.15/−0.20=75%. The remaining share (25%) corresponds to unobserved heterogeneity.

NS, nonsignificant difference with the reference category; OOP, out-of-pocket expenditures.