The Urban Institute
Support for this research was provided by the Ford Foundation. The views expressed in this paper are those of the author and do not necessarily reflect the views of the Institute or its funders.
Prepared for NICHD Workshop "Improving Data on Male Fertility and Family Formation" at the Urban Institute, Washington, D.C., January 16-17, 1997
For this first time, in 1980, the Current Population Survey (CPS) asked men who had a previous marriage whether they had children living elsewhere and whether they provided financial support to those children as part of its supplement on marital and fertility histories. Researchers found, however, that nonresident fathers with a prior marriage were seriously underrepresented in these data and that those who self-identified as nonresident fathers were significantly more likely to report that they provided financial support to their children living elsewhere than custodial mothers reported receiving it (Cherlin, Griffith and McCarthy 1983).
The 1987/88 National Survey of Families and Households also included detailed questions about fertility and marital histories of male and female adults and asked all parents whether they had children who lived elsewhere most of the time. Again, researchers found that nonresident fathers were seriously underrepresented in these data and that nonresident fathers who self-identified as such tended to report that they paid child support (Seltzer and Brandreth 1994).
Based on this experience, most researchers concluded that scarce resources for survey research should be spent on interviewing and analyzing custodial mothers. Since 1987, no survey of the entire adult population has asked men whether they have children living elsewhere. Instead, both the CPS and SIPP have continued to collect information about child support from custodial mothers. The SIPP asks men about their fertility, but it does not ask them where their children live. Research on nonresident fathers has certainly continued, but it has had to rely on subnational (or subgroup) data to shed light on this issue.
Since the early 1980s, child support has become a major policy issue. It is now viewed as a key element of our income security policy for low-income families. The federal government no longer guarantees cash assistance to poor single mother families; welfare is considered transitional support. These families are expected to eventually rely on their own earnings and child support. Yet we do not have a nationally representative survey that can identify nonresident fathers of poor children, which means we have no reliable estimates of their ability to pay child support. Without this information, policies will continue to be made on incomplete, and possibly misleading data. Thus, it is time to develop a methodology for large, national surveys that will produce accurate information about nonresident fathers.
The purpose of this paper is to describe what we do and do not know about nonresident fathers= ability to pay child support based on two national surveys that try to identify this population--the National Survey of Families and Households (NSFH) and the Survey of Income and Program Participation (SIPP). Both surveys are nationally representative of the civilian, noninstitutionalized adult population. Other national surveys may identify certain subgroups of nonresident fathers, such as young nonresident fathers, but I am unaware of any other national survey that attempts to canvas the entire (noninstitutionalized) population of nonresident fathers.
How Are Nonresident Fathers Identified in National Surveys?
To date, researchers have taken both direct and indirect approaches to identifying nonresident fathers in national surveys. By asking the following questions of all adult males, the NSFH permits a direct approach to identifying nonresident fathers: (1) How many children have you ever fathered? For those who say they have fathered a child, they ask: (2) Do you have any biological children age 18 or younger who do not live in this household at least half of the time? In contrast, the SIPP asks women and men about their fertility but it only asks mothers about the living situations of their children who live elsewhere. It does not ask these questions of fathers. Thus, I have developed an indirect approach to identifying nonresident fathers in the SIPP. For payers of child support, I use a question in the SIPP that asks respondents whether they provide financial support for children living elsewhere. For nonpayers, I determine which fathers report having fathered more children than currently live with them. Unfortunately, I cannot discern the ages of the children who live outside of the household. Thus, I impose a series of age and marital history restrictions on the definition of a nonresident father to more accurately capture this population.(1)
Two other important differences between the SIPP and NSFH should also be mentioned. First, the SIPP allows proxy respondents to answer the questions for an interviewee, but the NSFH does not. Second, the SIPP imputes answers to many questions if a respondent does not answer it, but the NSFH does not. However, the SIPP does not impute male fertility and includes flags that indicate if an answer is imputed. In my indirect method of identifying nonresident fathers, I do not use imputed values.
These two surveys yield similar fertility information about men (age 19 years or older) despite their different survey designs. The NSFH finds that 65 percent of men had fathered a child, while the SIPP finds that 67 percent of men had fathered a child (see Table 1). In addition, the number of births per adult male are quite similar in the NSFH and SIPP. In the NSFH, adult males report 1.78 births/male; in the SIPP, adult males report 1.7 births/male.
In contrast, the NSFH and SIPP yield significantly different percentages of men as nonresident fathers. In the SIPP, 8.6 percent of men were identified as nonresident fathers, or 7.3 million men, but only 6.9 percent of men in the NSFH were identified as such, or 5.6 million men.(2)
Table 1 also shows that both surveys underrepresent fathers. In the NSFH, 14 percent fewer men than women reported that they had been a (biological) parent, while 10 percent fewer men than women reported that they had been a (biological) parent in the SIPP.(3)
Men also report fewer births than women. In the NSFH, men report 84 percent as many births as women; in the SIPP, they report 89 percent as many births as women.
In both the NSFH and the SIPP, there are significantly smaller numbers of nonresident fathers than custodial mothers. In the NSFH, there are 61 percent as many nonresident fathers as custodial mothers; in the SIPP, there are 74 percent as many nonresident fathers as custodial mothers.(4)
What Do We Know About Nonresident Fathers Who are Identified in National Surveys?
Although the SIPP identifies more men as nonresident fathers than the NSFH, both surveys provide remarkably similar demographic profiles of nonresident fathers identified in national surveys. As Table 2 shows, nonresident fathers in these two surveys are predominantly white, ever-married, in their thirties, with at least a high school education.(5)
The only characteristic in which the NSFH and SIPP differ substantially is the extent to which nonresident fathers pay child support. In the NSFH, 78 percent of nonresident fathers report that they provided financial support for their children living elsewhere, but only 55 percent of nonresident fathers in the SIPP report that they paid child support. The questions about paying child support are different in the two surveys, which may lead to this discrepancy. The NSFH asks nonresident fathers with a child support order how much they are suppose to pay and then asks whether any payments were missed. For nonresident fathers without an order, the question in the NSFH is similar to the question asked in the SIPP. Both of these questions ask nonresident fathers whether they provide any financial support for their children living elsewhere.
Although nonresident fathers in the NSFH are more likely than nonresident fathers in the SIPP to report that they pay child support, the former report spending about the same proportion of their income on child support as do the latter. The average nonresident father in the NSFH spends 8.6 percent of his income on child support, while the average nonresident father in the SIPP spends 8.0 percent of his income on child support. Both figures are considerably less than the amount that state guidelines suggest nonresident fathers should pay in child support. Thus, both surveys indicate that the average nonresident father could pay more in child support.
On the other hand, both surveys show that a sizable minority of nonresident fathers have low incomes. In the NSFH and the SIPP, 22 percent of nonresident fathers have personal incomes after paying child support that fall below 150 percent of the poverty threshold for an individual. I have used this definition of low income rather than the official definition of poverty that relies on family income, because the NSFH does not collect family income for all respondents. It should also be noted that child support guidelines are based on nonresident parents' personal income rather than family income, which is another reason for focusing on personal income. I use the poverty threshold for a single person because it provides a measure of the amount of income an individual needs to meet his basic needs (not because I think all nonresident fathers live alone).
Table 2 also shows that 14 to 30 percent of nonresident fathers who report that they do not pay child support also report having high incomes (which I define as above 150 percent of the poverty threshold for a single person). Thus, I find a large minority of nonresident fathers who are deadbeat dads-- they can afford to pay child support but do not.
How Many Nonresident Fathers are Missing in National Surveys?
To ascertain the extent to which nonresident fathers are underrepresented in these surveys, I compared the number of children that nonresident fathers report living elsewhere to that reported by custodial mothers. Custodial mothers reports are used as a reference point because it is generally believed that their reports of children eligible for child support are more accurate than those of nonresident fathers (Cherlin et al. 1983). In addition, the NSFH and SIPP do not survey the institutionalized population. Because some nonresident fathers are institutionalized and custodial mothers are not, custodial mothers should provide a more accurate report of child support-eligible children.
Custodial mothers are identified in the NSFH and SIPP using questions that ask parents whether any of their children who live with them have a parent living elsewhere.(6)
The characteristics of custodial mothers in the 1987/88 NSFH and the 1990 SIPP are similar to those in the 1990 Current Population Survey-Child Support Supplement (CPS-CSS)--which provides the more commonly used data to describe custodial mothers (U.S. Bureau of the Census 1995).
In both surveys, nonresident fathers report having fathered fewer children who live elsewhere than those reported by custodial mothers. In the NSFH, nonresident fathers report a total of 8.6 million children living elsewhere, while custodial mothers report that they have 16.4 million children with a father living elsewhere. In the SIPP, nonresident fathers report a total of 13 million children living elsewhere and custodial mothers report 16.6 million children with a father living elsewhere.
To ascertain how many nonresident fathers are missing in these surveys, I divide the deficit of children reported by nonresident fathers when compared to custodial mothers by the average number of children reported by custodial mothers.(7)
This procedure yields 4.3 million nonresident fathers missing in the NSFH, or 44 percent of all nonresident fathers. In the SIPP, 2.1 million nonresident fathers are missing, representing 22 percent of the population of nonresident fathers.
Why are Nonresident Fathers Underrepresented in National Surveys?
There are three basic reasons why nonresident fathers are underrepresented in these surveys. First, both surveys are restricted to individuals who reside in households, meaning that individuals who live in group quarters, such as correctional institutions or military barracks, are not interviewed. About 1.1 million men between ages 19 to 54 were institutionalized in 1987 and 1990 and about 1.5 million men (ages 19 to 54) lived in other group quarters in 1987 and 1990 (U.S. Bureau of the Census 1992, 1988). In other words, about 2.6 million men between the ages of 19 and 54 were intentionally missed by the NSFH and the SIPP.(8)
The second reason nonresident fathers are underrepresented in these surveys is that they reflect the Census undercount of certain subpopulations, especially young black males. For example, it is estimated that the 1990 Census undercounted black males in their early thirties by 14% (Robinson et al. 1993). This undercount is incorporated into the NSFH and the SIPP because both surveys rely on the Census to develop their survey weights. About 2.2 million men between the ages of 19 and 54 were undercounted in 1987 and 1990 (U.S. Bureau of the Census 1988; Robinson et al. 1993).
The third reason that nonresident fathers are underrepresented in the NSFH and the SIPP is because men are significantly less likely to report that they have children living elsewhere than are women likely to report that they have children living with them with a father living elsewhere. As shown in Table 1, only 6.9 percent of adult men in the NSFH report that they have children living elsewhere (with the mother), while 10.3 percent of adult women report that they have children living with them who have a father living elsewhere. Similarly, in the SIPP, only 8.6 percent of adult men, but 10.6 percent of adult women say that they have child-support eligible children.
To estimate how many nonresident fathers are missed by the NSFH and the SIPP for these reasons, I first examined the Survey of Inmates in State Correctional Facilities, conducted by the U.S. Department of Justice in 1991. These data show that 42 percent of men in state prisons in 1991 had minor children and were not currently married. Since most institutionalized men (between 19 and 54) are in correctional institutions, I applied this figure (42 percent) to the adult male institutionalized population in 1987 and 1990 to estimate the number of nonresident fathers who were institutionalized at the time of the NSFH and SIPP surveys.(9)
Using this procedure, I estimate that about 500,000 nonresident fathers were institutionalized in 1987 and 1990 (see Table 3).
Since I have no information on the undercounted population, I used the same figure that I used for the institutionalized population (42 percent) to estimate the number of nonresident fathers who were undercounted by the NSFH and the SIPP. This yielded about 900,000 nonresident fathers who are undercounted in 1987 and 1990 (Table 3).
To estimate the number of nonresident fathers in 1987 and 1990 who lived in other group quarters (most of whom are in the military or at college), I assumed that men (19 to 54) who lived in other group quarters in 1987 and 1990 were just as likely to be nonresident fathers as men (19 to 54) in the NSFH and the SIPP. Based on this assumption, I estimate that 149,094 nonresident fathers were living in group quarters in 1987 and 183,675 were living in group quarters in 1990. Thus, a total of about 1.5 million nonresident fathers were not interviewed by the NSFH or the SIPP either because they were undercounted, institutionalized, or living in other group quarters.
The number of nonresident fathers who are underreporting their children living elsewhere is estimated as the residual category. I subtract the estimated number of nonresident fathers who were not interviewed by the NSFH or the SIPP because they were undercounted, institutionalized, or living in other group quarters from the total number of nonresident fathers who are estimated to be missing in these surveys. Using this procedure, I estimate that 2.8 million nonresident fathers are underreporting their children living elsewhere in the NSFH, and about 500,000 nonresident fathers are underreporting their children living elsewhere in the SIPP (Table 3). In other words, 65 percent of the underrepresentation of nonresident fathers in the NSFH is caused by underreporting, but only 25 percent of the underrepresentation of nonresident fathers in the SIPP is caused by underreporting.
Where Should Research Go From Here on Nonresident Fathers and Their Ability to Pay Child Support?
At this point, 22 to 44 percent of nonresident fathers are missing in national surveys. Given the magnitude of the problem, I make the following recommendations to improve our understanding of nonresident fathers and their ability to pay child support.
First, we need a description of nonresident fathers who are not interviewed in national surveys, most of whom are undercounted. One of the key reasons individuals are undercounted by household surveys is because these surveys are limited to individuals who are usual residents. The U.S. Bureau of the Census conducted a survey in 1993 of 999 households, called the Living Situations Survey, which asked a series of probing questions about who is associated with each household. They found that these additional probes resulted in a 38 percent increase in the number of persons per household and an 5 percent increase in the number of usual residents (Martin undated). This survey asked some demographic questions, but a more thorough analysis of these questions needs to be conducted. For example, how many households would have been typed single-mother households according to the simple usual residence question, but in fact had a father present at least some of the time?
A subset of these roster probes should be added to the Current Population Survey (CPS) and the SIPP, the two largest government surveys that are fielded on a regular basis and include detailed measures of income--a critical variable when analyzing nonresident fathers. Before these probes are added, however, further research needs to be conducted to ascertain whether these probes increase the number of usual residents, which probes are the most useful, and whether these probes affect the response rate to surveys such as the CPS and SIPP.
For individuals who are identified by these roster probes who are not usual residents, key information (e.g., sex and age) should be obtained at the time of the initial interview. In addition, follow up interviews with a subset of these individuals should be conducted to ascertain whether they are working, their other income sources, and whether they have children. Without this additional information, we will never have a reasonable profile of nonresident fathers and their ability to pay child support.
We also need to do a better job of identifying nonresident fathers who are interviewed in national surveys but who do not self identify themselves as nonresident fathers. At this point, little effort has gone into testing different approaches that may improve response rates among men to questions about fertility and children living elsewhere. Does the wording of these questions matter? Does the order of the questions matter? Does a context for these questions help improve response rates? Does it matter whether a proxy is used to answer these questions?
The U.S. Census Bureau should experiment with question design regarding nonresident fathers and the payment of child support. The SIPP already has a fertility supplement that asks fertility questions of both men and women. As I showed above, male fertility is not that different from female fertility in the SIPP. As currently designed, the SIPP goes on to ask mothers (but not fathers) about the living situations of their oldest and youngest child. These questions should be tested on a sample of fathers. The CPS already tested questions about child support payments in the 1996 CPS-Child Support Supplement. The results of these questions should be examined.
Although many researchers recommend that subnational studies of nonresident fathers be undertaken to learn more about their attitudes and behaviors, these studies will not provide a national profile of nonresident fathers and their ability to pay child support, which is critical to policy formation. Furthermore, administrative data on nonresident fathers is insufficient because they do not include the entire universe of nonresident fathers.
The only way to produce reasonably accurate estimates of nonresident fathers= ability to pay child support is to improve upon a large, on-going national survey. The SIPP and CPS are the most likely candidates because they already have questions that identify custodial mothers and are currently viewed as the most reliable source of information on child support.
Bachu, Amara. 1996. Fertility of American Men. U.S. Bureau of the Census, Population Division Working Paper No. 14.
Cherlin, Andrew, Jeanne Griffith, and James McCarthy. 1983. A Note on Maritally-Disrupted Mens Reports of Child Support in the June 1980 Current Population Survey. Demography 20 (3): 385-389.
Garfinkel, Irwin, Sara S. McLanahan, and Thomas L. Hanson. 1997. A Patchwork Portrait of Nonresident Fathers. Unpublished paper, March.
Martin, Betsy. Features of Living Situation Survey. (Undated).
Robinson, J. Gregory, Bashir Ahmed, Prithwis das Bupta, and Karen A. Woodrow. 1993.
"An Estimation of Population Coverage in the 1990 United States Census Based on Demographic Analysis." Journal of the American Statistical Association 88 (September): 1061-1079.
Seltzer, Judith and Yvonne Brandreth. 1994. What Fathers Say About Involvement with Children After Separation. Journal of Family Issues 15 (March): 49-77.
Sorensen, Elaine. 1997. A National Profile of Nonresident Fathers and their Ability to Pay Child Support. Urban Institute Working Paper, April.
U.S. Bureau of the Census. 1995. Child Support and Alimony: 1991. Current Population Reports, Series P-60, No. 187. Washington, D.C.: U.S. Government Printing Office.
U.S. Bureau of the Census. 1992. General Population Characteristics: United States. 1990 CP-1-1. Washington, DC: GPO.
U.S. Bureau of the Census. 1988. United States Population Estimates, by Age, Sex, and Race: 1980 to 1987. Series P-25, No. 1022. Washington, DC: GPO.
|Source: 1987/88 National Survey of Families and Households (NSFH), 1990 Survey of Income and Program Participation (SIPP)|
|Total Number of Men/Women||80,998,000||84,834,000||90,469,000||84,834,000|
|% Who Report Being a Father/Mother||65.2%||67.3%||75.9%||74.8%|
|Births per Adult||1.78||1.7||2.13||1.93|
|% Who Report Being a Nonresident|
|Nonresident Fathers||Nonresident Fathers|
|Mean (in years)||36.1||35.2|
|Not a High School Graduate||18.8%||21.4%|
|High School Graduate||38.8||42.1|
|Some Post-Secondary Education||42.4||36.5|
|Mean (in years)||12.6||12.3|
|Income and Payment Status|
|High Income Nonpayers||14.4||30.2|
|High Income Payers||64.1||47.6|
|Mean % of Personal Income|
|Paid Towards Child Support||8.6%||8.0%|
|Source: 1987/88 National Survey of Families and Households (NSFH), 1990|
|Survey of Income and Program Participation (SIPP)|
|Note: Low Income is defined as having personal income (after paying child support)|
|below 150% of the poverty threshold for a single person|
|Source: 1987/88 NSFH; 1990 SIPP; U.S. Census Bureau 1988, 1992; Robinson et al 1993|
|% Who are Nonresident Fathers||42%||42%|
|Nonresident Fathers Who are Undercounted||920,700||930,998|
|% Who are Nonresident Fathers||42%||42%|
|Nonresident Fathers Who are Institutionalized||462,197||473,464|
|Total Other Group Quarters*||1,538,245||1,550,508|
|% Who are Nonresident Fathers||10%||12%|
|Nonresident Fathers Who are in Other Group Quarters||149,094||183,675|
|Total Number of Men Who are Underreporting|
|That They are Nonresident Fathers||2,823,287||522,890|
*Total Undercount, Institutionalized, and Other Group Quarters are limited to men between the ages of 19 and 54 to eliminate older men who are probably not nonresident fathers
1. 1For further details regarding these restrictions, see Elaine Sorensen, A National Profile of Nonresident Fathers and their Ability to Pay Child Support. Urban Institute Working Paper, 1997.
2. 2Nonresident fathers in the NSFH are limited to those whose focal child is under 18 and lives with the mother.
3. 3The nonresponse rates to the fertility question are not that different by sex in the NSFH and SIPP. In the NSFH, only one person (a woman) did not answer the fertility question; in the SIPP, about 5 percent of women and men did not respond to the fertility question.
4. 4Custodial mothers in the NSFH and the SIPP are limited to those who indicate that at least one of their children living with them is under 18 years old.
5. 5Neither of these descriptive profiles has been altered to adjust for the under representation of nonresident fathers in these surveys. I have simply applied the population weight that is supplied by the NSFH and SIPP to the individual records.
6. 6All custodial mothers have physical custody of their children, but some of them share legal custody with the father.
7. 7I use the average number of children reported by custodial mothers because other research has shown that men underreport their fertility. See, for example, Bachu 1996.
8. 8I examine 19 to 54 year olds to limit the population in question to those who are most likely to be nonresident fathers. I impose the lower age limit because the NSFH only interviews adult men who are at least 19 years old.
9. 9See Garfinkel et al. 1997 for a similar approach to estimating the number of nonresident fathers in prison.