Table 2

Association of income and the EITC refund size with mental health and health behaviours

OutcomeModel 1:
association of income with health (OLS)		Model 2:
association of EITC with health (OLS)		Model 3:
effect of income on health (IV)
Psychological distress−0.026**
(−0.035 to –0.018)		−0.070*
(−0.11 to –0.031)		−0.13**
(−0.21 to –0.059)
Very good/excellent health0.0013**
(0.00085 to 0.0018)		0.0000046
(−0.0027 to 0.0027)		−0.0028
(−0.0096 to 0.0040)
Currently drink alcohol0.00098**
(0.00047 to 0.0015)		0.0025
(−0.00039 to 0.0055)		0.011
(−0.0023 to 0.025)
3+ drinks per day−0.00012
(−0.00053 to 0.00030)		−0.0016
(−0.0039 to 0.00078)		−0.0071
(−0.018 to –0.0038)
Currently smoke0.0018
(−0.00013 to 0.00050)		0.00086
(−0.0012 to 0.0029)		0.0044
(−0.0060 to 0.015)
Cigarettes per day0.012**
(0.0056 to 0.019)		0.0022
(−0.033 to 0.038)		0.011
(−0.17 to 0.19)

  • Study sample was drawn from the Panel Study of Income Dynamics for survey years 1985–2015. All models involved multivariable linear regressions, adjusting for age and age-squared, education, marital status, number of children, household inflation-adjusted pre-tax earned income and income-squared, year fixed effects, and individual fixed effects. Robust SEs were clustered at the individual level. Coefficients are presented per US$1000 of income or EITC.

  • Model 1 represents the OLS model, regressing each outcome on pre-tax inflation-adjusted household income.

  • Model 2 represents the reduced form of the IV analysis, regressing each outcome on EITC refund size.

  • Model 3 represents the IV analysis, with EITC refund size as the instrument for post-tax inflation-adjusted household income.

  • *p<0.05, **p<0.01.

  • EITC, earned income tax credit; IV, instrumental variable; OLS, ordinary least squares.