As shown in , over a third (35%) of the fathers reported that they were in suboptimal health, 15% were disabled, and 8% had experienced a depressive episode in the past 12 months. Only 60% of the fathers completed high school. Fewer than half (40%) had been offered health insurance by their baseline employer. The vast majority (85%) were Black or Hispanic. Over half (55%) had household incomes below the FPL (the remainder had incomes between 100–200% of the FPL). Three quarters of the fathers were employed at least part-time. Most (70%) were living with their children’s mothers (married or cohabiting).
CHARACTERISTICS OF LOW-INCOME FATHERS BY HEALTH INSURANCE STATUS
Of the 1,653 fathers, 474 (29%) had private insurance, 228 (14%) had public insurance, and 951 (58%) had no insurance. The last figure is slightly higher than the national figure for low-income young men cited earlier (54%). In contrast, 20% of the 1,653 mothers in this sample had private insurance (either through the father or themselves), 44% were publicly insured, and 37% were uninsured; 20% of the children had private insurance, 66% were publicly insured, and 14% were uninsured (most of these figures not shown in tables). Thus, fathers had much higher rates of uninsurance than the mothers and children. Notably, fathers’ uninsurance is a risk factor for children not having insurance. Of the 228 children who were uninsured at follow-up, 68% had fathers without insurance and the other 32% had fathers who had health insurance (not shown in tables).
includes the results from t-tests that indicate significant differences between pairs of the three insurance groups (fathers with private, public, and no insurance). While past research indicates that uninsured and insured men differ in many ways, it has not generally distinguished between privately and publicly insured men.31
As expected, we find that there are many differences between the private and uninsured groups and between the private and public groups. We also find that the public group more closely resembles the uninsured group than it does the group of privately insured fathers.
Compared with privately insured fathers, uninsured fathers tend to report poorer health, are more likely to be disabled, and have lower levels of human capital (they are younger, less educated, less likely to have served in the military, less likely to have been offered health insurance at baseline, more likely to have a history of incarceration, more likely to have income below 100% of the FPL, less likely to be employed, and less likely to be married to the child’s mother). Also, they are more likely to be Hispanic. The mothers are more likely to be on public insurance, less likely to be employed, and less likely to have their own private health insurance. Finally, they are more likely to have a child in suboptimal health. Thus, uninsured fathers have greater immediate health needs than privately insured fathers, and they also have much lower earnings capacity.
Compared with publicly insured fathers, uninsured fathers are younger, less educated, less likely to be disabled, more likely to be foreign-born, less likely to be Black, more likely to be Hispanic, and less likely to live with the child’s mother. The results for disability, nativity, Hispanic ethnicity, and living arrangements are consistent with lack of eligibility for Medicaid. Additionally, the mothers are more likely to have suboptimal physical or mental health and less likely to be on public insurance. This last finding highlights interdependency in parents’ insurance status.
Even in this low-income sample, fathers with private insurance differ significantly from those with public insurance in many ways. They tend to have greater levels of human capital (they are older, more educated, more likely to have had an offer of employer insurance at baseline, less likely to have an incarceration history, less likely to have income below the FPL, more likely to be employed, and more likely to be married). They are also less likely to be disabled or depressed. Finally, they are less likely to be Black and more likely to be foreign-born, and the mother is less likely to have public insurance and more likely to have her own private insurance.
In the cases where both privately and publicly insured fathers differ significantly from those who are uninsured, we find that uninsured fathers are younger and less educated than insured fathers; that uninsured fathers are less likely than publicly insured fathers, but more likely than privately insured fathers, to be disabled; and that uninsured fathers and publicly insured fathers are more likely to be associated with mothers who have public insurance, compared with fathers who are privately insured.
Additional analyses (results not shown) indicate that fathers with multiple risk factors for being uninsured (unmarried, unemployed, and less than a high school education) constitute a substantial proportion of the uninsured. Seventy-eight percent of the 167 fathers who were unmarried, unemployed, and had a low level of education were uninsured. Even among low-income fathers who are better positioned to have coverage, many do not. Of the 230 fathers who were married, employed, and graduates of high school, over one third (35%) were uninsured. For the uninsured fathers in the last group, there is evidence of considerable unmet need: 62% reported having suboptimal health, 57% were disabled, and 59% were depressed.
presents the results of a multinomial logistic regression model that assesses the association of each characteristic with the father’s insurance status, controlling for all of the other characteristics. The analysis indicates whether the bivariate associations persist taking all other variables into account, and indicates the magnitude of those effects. Results are presented for the likelihood of private insurance and for the likelihood of public insurance, with no insurance as the reference outcome in both cases. Because no insurance is the reference outcome, the RRRs represent the change in private (or public) insurance status relative to no insurance for a unit change in a given characteristic, holding the others constant. For example, being disabled more than the doubles the probability that a father will have public rather than no insurance coverage. Having suboptimal rather than excellent or very good health makes a father 35% less likely to have public insurance coverage than to be uninsured. In keeping with the bivariate associations from , many paternal and family characteristics differ significantly (using p<.05 unless indicated otherwise) between fathers with private rather than no insurance and between fathers with public rather than no insurance. However, education, race/ethnicity, military service, depression, and the child’s physical health, which were significant in some of the bivariate comparisons in , are not independently associated with the father’s insurance status in the multivariate context (although some of these factors may operate through the father’s employment). The characteristics that differentiate privately insured fathers from uninsured fathers are not necessarily the same characteristics that differentiate publicly insured fathers from uninsured fathers. These differences again highlight the importance of separately analyzing privately and public insured fathers rather than pooling them together to contrast them with those who are uninsured.
MULTINOMIAL LOGISTIC REGRESSION ESTIMATES OF ASSOCIATIONS BETWEEN FATHERS’ INDIVIDUAL AND FAMILY CHARACTERISTICS AND FATHERS’ INSURANCE STATUS (N = 1653)a
Age, employment (full- or part-time), having had (at baseline) an employer offer health insurance, being married to the focal child’s mother, and the mother having her own private insurance all increase the likelihood that the father had private (vs. no) insurance. Being foreign-born, being very poor, ever having been incarcerated, and having a child whose mother has public health insurance all decrease the likelihood that the father had private (vs. no) insurance. The strongest associations were for employment (RRR=4.61, p<.001 for full-time; RRR=3.12, p<.001 for part-time). The majority of these results are consistent with the corresponding bivariate associations from .
Being foreign-born and the mother having suboptimal physical and/or mental health decreased the likelihood that the father had public (vs. no) insurance. Notably, being in suboptimal health themselves also decreased the likelihood of public (vs. no) insurance for these fathers, again suggesting that those with greater need for coverage are falling through the public safety net. Being disabled and being married to or cohabiting with the focal child’s mother (vs. not living with her) increased the likelihood that the father had public insurance, which we would expect given the eligibility requirements for public insurance. In addition, the mother being employed and the mother having public health insurance also increased the likelihood of public (vs. no) coverage, although not at conventional levels for the former (p=.07). For public insurance, many of the state indicators were highly significant and had large magnitudes. Fathers living in one of the sampled states (Pennsylvania) were over three times as likely as those living in another state (Texas) to be publicly insured (results not shown). However, the small number of states represented in the data and the small samples in certain states preclude our making inferences about the effects of state health policies.