«Evidence of the Effectiveness of Child Support and Visitation: Examining Food Insecurity among Children with Nonresident Fathers Steven Garasky Æ ...»
Separate categories are created (a) for children whose nonresident parent is deceased or whose whereabouts are unknown,3 and (b) for children whose MKA reported only extended summertime and/or holiday visits with the nonresident parent or could not classify the frequency of visits. Financial involvement is measured using MKA reports of the total amount of child support coming into the household for each family member.4 We calculate the average monthly amount of child support the family received in the last 12 months and control for the number of children in the family as a proxy of nonresident parents’ ﬁnancial involvement.
Unfortunately, the NSAF does not allow the analyst to distinguish deceased fathers from absent fathers whose whereabouts are unknown. Questions about nonresident father involvement are limited to children whose fathers are known to be alive and living elsewhere.
We use the MKA’s report of whether any household member received child support because the NSAF does not contain reliable child-level data on child support amounts (Adam Saﬁr, personal communication, 01/08/04). However, the household-level measure of child support is correlated with the MKA report of whether the child’s nonresident parent made any ﬁnancial contributions in order to support the focal child in the last 12 months (r = 0.520, p 0.001), suggesting that our use of the household-level measure is valid. About two-thirds (62%) of focal children in households that received child support in the previous year were reported as having a parent who paid child support in the previous year.
110 J Fam Econ Iss (2007) 28:105–121
Our multivariate analyses include characteristics of the child, the MKA and the household. Child characteristics include the child’s sex and age, whether he or she was born within a marriage, whether a child support order has been established, and whether he or she has any siblings living elsewhere. Characteristics of the MKA include his or her sex, age, level of education, employment status, health status, nativity, and the frequency in which he or she attends religious services. Householdlevel characteristics include household income excluding child support; household composition measured with indicators for having a single parent, a married stepparent, or a cohabiting stepparent; and the number of minor children and adults (other than parents and stepparents) in the household.
We estimate a probit model of the form:
where FOODINS is an indicator of whether the household experienced this aspect of food insecurity (= 1) or not (= 0). Experiencing an aspect of food insecurity is a function of father involvement (vector FI) as measured through child visitation and the payment of child support. For simplicity, we assume that the father’s decision to be involved is exogenous to the resident family’s ability to meet their food needs. X is a vector of the other explanatory variables that were discussed above. l is an error term.
We examine the effects of child support and visitation on aspects of food insecurity in separate estimations of Eq. 1 using our six indicators of food insecurity (responses to each of the three food insecurity questions separately and our three composite measures of food insecurity severity). Because the NSAF employs a complex cluster sampling design, special weighting procedures are employed so that standard errors are not underestimated (Flores-Cervantes, Brick, & DiGaetano, 1997).
Table 1 provides descriptive statistics for our analytic sample (n = 7,861). The characteristics of the sample reﬂect our focus on families with incomes below 200% of poverty. A majority of these families have problems meeting their food needs. For example, 57% reported that in the last 12 months they worried that their food would run out before getting money to buy more. Half of the families experienced an instance where their food did not last and they did not have money to get more, while one-in-three (32%) households reported that adults had to cut the size of their meals or skip meals because there was not enough food. Overall, nearly two-thirds (62%) of the families in our sample experienced at least one of these three food problems in the previous year, half (49%) experienced at least two of these food problems in the previous year, and one-quarter (26%) experienced all three of these problems.
J Fam Econ Iss (2007) 28:105–121 111 Table 1 Analysis sample characteristics
Severity of food insecurity in previous year Ever experienced at least one of the above 62.3 Ever experienced at least two of the above 48.9 Ever experienced all three of the above 26.1
All analyses were weighted using NSAF sample weights Focal children are split evenly between boys and girls, averaging 9 years in age.
Forty-one percent of the focal children are non-Hispanic white, 37% non-Hispanic black, with the remainder being Hispanic (22%). About half (54%) of these children were born to a married couple. Forty-two percent of them have awards for child support.
The MKA in this sample is almost always a woman (98%). The average age of these respondents is 33 years. Our MKAs have limited educations with one-third not graduating high school and another one-third having only a high-school diploma or GED certiﬁcate. Over half (55%) of the MKAs did not work in the previous year.
Twenty-one percent suffer from fair or poor health; 7% are foreign-born. About one-in-three MKAs attend religious services on a weekly basis.
The majority (75%) of the households in our sample are headed by a single mother. On average, these households have three minor children each. Few households had other adults in them beyond parents and step-parents. Family incomes without including child support average about $14,000 annually in 1996 dollars.
As can be seen in Table 2, father involvement in the previous year for these children varied considerably. Overall, over half (57%) of the focal children received an in-person visit from their father in the previous year. More speciﬁcally, about half (49%) of the children either did not receive a visit in the previous year from a father who was known to live elsewhere or received a visit from him less than once per month, while one-fourth were visited at least once a week. The fathers of 13% of the children was deceased or their whereabouts were unknown. Only one-third of the children resided in families that received any child support. Overall, the average amount of child support received by families receiving some support was $219 per month, or slightly over $2,600 annually (1996 dollars).
Consistent with Grall (2003) and Garasky, Peters, Argys, Cook, Nepomnyaschy, and Sorensen (2006), we ﬁnd that having a child support award is not fully correlated with receiving child support (see Table 3). About 59% of the children with a child support award had families who received child support payments. Interestingly, among families reporting not having a child support award for the focal child, 16% reported receiving some child support payments.
Table 4 investigates relationships between father involvement and our measures of food insecurity. We employ a dichotomous measure of visitation frequency (any versus no visits). Children whose nonresident father is deceased or whose whereabouts are unknown are classiﬁed as having no visits. In addition, we examine effects of receiving any child support versus none. We see from these bivariate explorations that father visitation and our indicators of food insecurity are linked. Considering each measure independently (Panel A), we ﬁnd only for the measure ever experiJ Fam Econ Iss (2007) 28:105–121 113 Table 2 Father involvement
All analyses were weighted using NSAF sample weights enced food not lasting is the difference statistically signiﬁcant between households with nonresident fathers who visited in the previous year and those with nonvisiting fathers (46% vs. 53%, p 0.01). Overall, families with visiting fathers were less likely to report that any one of the three food access problems occurred in the previous year (60% vs. 65%, p 0.05) as shown in Panel B. The receipt of child support is not statistically related to the indicators of food insecurity in these bivariate relationships, although for most measures households that received some child support are slightly less likely to report experiencing difﬁculty meeting their food needs.
Tables 5 and 6 report our estimated coefﬁcients for the independent variables of our logistic regressions. Table 5 reports results for the measures ever worried that food would run out, ever had food not last, and ever had adults cut the size of their meals or skip meals. Table 6 lists regression coefﬁcients for which the dependent variable measures the severity of food insecurity, in terms of whether the focal child’s family experienced at least one of the above food acquisition problems, at least two of these types of food problems, and all three types. In analyses not Table 3 Relationship between child support order and child support received
*p 0.05 **p 0.01 reported here, we ran four separate models for each outcome: Model 1 incorporated our set of core explanatory variables (vector X in Eq. 1). This core consisted of the characteristics of the focal child, the MKA, and the household as discussed earlier.
Model 2 added to the core variables categorical indicators of the frequency of inperson visits by the nonresident father. Model 3 added the average monthly amount of child support received by the resident family to the set of core explanatory variables. Finally, Model 4 included the core variables, the indicators of visitation and the amount of child support received. Because our substantive ﬁndings remain unchanged across models, Tables 5 and 6 show only the results for the full model (Model 4) for each outcome.5 Next, we discuss brieﬂy our results for the core explanatory variables as reported in Tables 5 and 6. We follow that with a discussion of the effects of father involvement on our indicators of food insecurity.
As one would expect, family income (excluding child support) is related to all of our indicators of food insecurity. The likelihood of being unable to meet food needs decreases as income increases. The number of other adults in the household increases the likelihood of meeting food needs. Households with additional adults are more likely to have someone with the time and ability to prepare more economical meals for the family using basic ingredients (e.g., less reliance on expensive packaged foods). Living in a married step-parent household reduces the likelihood of adults ever cutting the size of their meals or skipping meals relative to living in a single mother household. Clearly, the composition of the household affects the distribution of food across its members.
All analyses were weighted using NSAF sample weights. Reference categories in italics *p 0.10 **p 0.05 ***p 0.01 Regarding the characteristics of the MKA, households for which the MKA is female tend to experience more aspects of food insecurity. This result is consistent with previous research on divorce that indicates that single fathers have more sources of social support compared to single mothers (Arendell, 1995; Greif, 1985).
For instance, children whose primary caretaker (MKA) is their father may be more likely to get meals from friends and relatives than children living with their mother.
The level of education of the MKA also is related to experiencing food insecurity, but in an inconsistent and unexpected way. One would expect having difﬁculty meeting food needs to be negatively related to education if we assume that more educated adults are able to make better use of their resources as they provide food for the members of their households. Our results support this expectation as we ﬁnd MKAs with less than a high school education to be more likely to report experiencing aspects of food insecurity compared to those MKAs with a high school degree or a GED certiﬁcate, our base group. However, unexpectedly we ﬁnd that MKAs with some college education also are more likely to report problems acquiring enough food compared to the base group. Perhaps this population of low-income college-educated women is unique in some way. MKAs in fair or poor health also are more likely to be in households that experience indicators of food insecurity.
Without data regarding family medical expenses, we can only speculate that an MKA in poor health probably has higher medical expenses (e.g., prescription drug costs), which reduce the family’s budget for food. The effects of being a foreign-born MKA on the household’s ability to meet food needs are mixed, which could be the result of cultural differences in perceptions of food insecurity risks (i.e., they are less likely to worry but are more likely to run out of food). Experiencing indicators of food insecurity is not related to the age of the MKA, the employment experiences of the MKA in the previous year after we control for household income, nor the frequency with which the MKA attends religious services. The lack of effect of religious services is surprising given the involvement of churches in soup kitchens and the like. In general, the characteristics of the focal child do not affect our indicators of food insecurity for the child’s household.
As can be seen in Tables 5 and 6, our indicators of visitation show that frequent—more than once a week—visits by the father reduce the likelihood that the focal child’s resident family will experience episodes of food insecurity. This result is robust in that it is found for each of the three separate food insecurity measures and J Fam Econ Iss (2007) 28:105–121 117 Table 6 Logistic regression estimates of father involvement on severity of food insecurity (ever worried food would run out, ever had food not last, ever cut/skipped meals)