This paper examines the factors that are most likely to contribute to healthy marriages among low-income couples. Using data from the Fragile Families and Child Wellbeing survey, we analyzed couples who were unmarried at the time of their child's birth, but who subsequently married within the first year after that birth.

The analysis revealed four factors that were sig­nificant predictors of subsequent marriage among couples who were unmarried at the time of their child's birth. These factors were:

• Parental marital attitudes and relationship skills;
• Mother's race;
• Mother's age (25 or older); and,
• Father's employment.

Neither the annual earnings nor education level of mothers or fathers were found to be significant predictors of post-birth marriage among unmar­ried parents.

The analysis also indicates that improving pater­nal employment alone would have, at best, a mod­est impact on marriage. Increasing fathers' employment, so that all fathers were currently employed and worked 52 weeks per year, would increase the marriage rate among unmarried cou­ples only slightly; from a base rate of 11.3 percent up to 13.2 percent.

The Fragile Families data indicate that the mari­tal attitudes and relationship skills of a couple play an important role in encouraging marriage. An 11-point scale for each parent was devised, measuring attitudes toward marriage, gender trust, support­iveness, and conflict in the relationship. An upward shift of one point for each parent on this scale doubled a couple's probability of marriage.

The analysis suggests that healthy marriage pro­grams should put their primary emphasis on improving couples' attitudes and relationship skills. Effective job training and employment ser­vices can also play a positive role in encouraging healthy marriage, but job training should play an ancillary and supportive-rather than a domi­nant-role in marriage promotion programs.

Background

Each year, one in three U.S. children is born out of wedlock. Children born and raised without married fathers in the home are more likely to suf­fer from a wide array of social maladies, such as increased poverty, welfare dependence, more emo­tional and behavioral problems, increased school failure, and expanded criminal activity. In recent years, a new consensus has emerged among both liberals and conservatives on the benefits of mar­riage to children, adults, and society, and on the need for government to develop policies to pro­mote healthy marriage.

A strong policy to promote healthy marriage would have overlapping components: enhancing the relationships of married couples; reducing divorce; reducing out-of-wedlock childbearing; and promoting healthy marriage among unmar­ried parents. Of these, promoting marriage among unmarried parents (generally termed "fragile fami­lies") at the "magic moment" of a child's birth (or shortly thereafter) has drawn, by far, the most attention.

Out-of-wedlock childbearing has increased dra­matically during the last four decades, rising from 7 percent of all births in the mid-1960s to 34 per­cent today.[1] There are four broad theories to explain the rise in out-of-wedlock childbearing (or the share of parents who are not married at the time of a child's birth). These theories focus on the role of:

Male Wages and Employment. It is often argued that the decline in the earnings and employment of low-skill fathers has made them less attractive and reliable as husbands and breadwinners. This, in turn, has led an increasing share of women to devalue marriage as a support to childrearing, and to opt for sin­gle parenthood.

Welfare Penalties Against Marriage. All means-tested benefit programs inherently penalize marriage among low-income parents by ensuring that a mother will receive higher benefits if she does not have an employed hus­band or male partner in the home. Many argue that the anti-marriage features of welfare have strongly contributed to non-marriage among low-skill parents.

Cultural Values and Norms. The last four decades have seen dramatic changes in cultural norms and values concerning non-marital sex, the importance of marriage, maternal employ­ment, co-habitation, and out-of-wedlock childbearing. In large sub-classes within the U.S., marriage is no longer seen as an impor­tant prerequisite to childbearing. For a large portion of the population, the father's expected role within the family has become severely attenuated, and any link between marriage and childrearing has become tenuous. It seems likely that changes in cultural values and norms have had a significant effect on individ­ual attitudes and behavior.

Individual Skills and Attitudes. Individual skills and attitudes can play an important role in marriage formation and stability. Critical attitudes and skills can include views on the importance of marriage, life-planning skills, the willingness to defer gratification, commu­nication skills, conflict resolution skills, fidel­ity, and the capacity to develop trust and commitment. Education and counseling pro­grams to alter attitudes and improve skills can, potentially, play an important role in increas­ing healthy marriage.

It is likely that each of these four factors has played a role in the current high level of non-mar­riage among parents. However, in shaping policies to promote healthy marriage, it is important to understand the relative weight of each factor in encouraging out-of-wedlock childbearing. Which of these factors, if adjusted, is likely to have the largest impact on encouraging non-married par­ents or "fragile families" to become and remain married?

Present Analysis

This paper uses data from the Fragile Families and Child Wellbeing survey to assess the factors contributing to marriage. The Fragile Families sur­vey is a nationally representative sample of couples in large cities at the point of a child's birth. Roughly 40 percent of these couples were unmar­ried at the time of their child's birth. The survey also provides follow-up data showing that among those couples who were unmarried at the time of their child's birth, roughly 11 percent had married one year later. This paper will examine the factors that contributed to marriage among unmarried couples during the one-year follow-up period, or the year subsequent to their child's birth.

Examination of "fragile families" (or couples who are not married at the time of their child's birth) is especially important because they are likely to be a primary intervention group of any healthy marriage initiative. Our paper will focus on comparing the relative role of: 1) male employment and earnings and 2) couples' marital attitudes and skills in foster­ing marriage among this group. What are the roles of economics and attitudes in promoting marriage among fragile families?

What Variables Affect Marriage?

In order to assess the factors that are important in the future marriage of unwed parents, this anal­ysis looks at the Fragile Families couples who were not married at the time of their child's birth to determine what factors help explain why some couples got married within one year of the birth and why other couples did not. A set of eight logistic regression models was used for this pur­pose. Many of the explanatory factors used in these models are borrowed from previous aca­demic research using the Fragile Families data.[2]

The independent variables in these models include:

• Father's annual earnings (expressed in thou­sands of dollars). Do men with higher earnings get married more often than those with lower earnings?
• Mother's annual earnings (expressed in thou­sands of dollars). Do women with higher earn­ings get married more often than those with lower earnings?
• Race. Is there a difference in marriage rates across racial lines?
• Parents of different races. Do marriage rates change if the parents are of different races?
• Father's education. Are fathers with more edu­cation more likely to marry?
• Mother's education. Are mothers with more edu­cation more likely to marry?
• Mother's and father's health. This variable is a self-evaluated, five-point health scale, rating health from low (1) to high (5). It might be reasonably theorized that healthier individuals may be more predisposed toward marriage.
• Mother's age. Older individuals may be more likely to marry.
• Couple has other children. Has the couple had other children together in the past? If the cou­ple has already had children together, it might make them more likely to marry.
• Mother has had a child with another man. If there are children in the home who are not the bio­logical offspring of the current father, it might make marriage less likely.
• Physical and drug/alcohol abuse. Does such dis­cord decrease the probability of future marriage?
• Religious observance. This scale variable reports the level of religious attendance and worship (without regard to the specific denomination), from never (1) to weekly (5). More religious par­ents may be more predisposed toward marriage.
• Mother's marriage attitudes and relationship skills. This variable measures the mother's attitudes about marriage and the quality of her interac­tions with the father on an 11-point scale. Do pro-marriage attitudes and relationship skills increase the probability of marriage?
• Father's marriage attitudes and relationship skills. This variable measures the father's attitudes about marriage and the quality of his interac­tions with the mother on an 11-point scale. Do pro-marriage attitudes and relationship skills increase the probability of marriage?

The dependent variable, again, is a binary (yes/ no) variable about whether or not the single par­ents became married within roughly a year of their baby's birth. This period around the time of a child's birth is typically known as the "magic moment" when it is most likely that single parents will get married to each other.[3]

We first examined the impact of these variables on marriage in the logistic regression analysis shown as Model I in Table 1 of the Appendix. Among the variables examined, only the following were found to be significant:

• Mother's race;
• Mother's age (25 or older);
• Mother's eleven-point Mari­tal Attitude and Skills Scale; and,
• Father's eleven-point Mari­tal Attitude and Skills Scale.

Critically, neither education nor annual earnings were signif­icant for mothers or fathers.

In Model II (shown in Table 1 of the Appendix), we removed most insignifi­cant variables, but retained the education and income variables for both parents. Again, race, age of mother (over 25), and parental attitudes were found to be significant. Parental educa­tion and incomes remained insignificant and low in power.

Because education and income are correlated, it is possible that the presence of both variables in the regression masks their signifi­cance. To test for this possibility, in Model III we omitted the mother's and father's income variables but retained the education variable. Little changed in the regres­sion-the education variables remained insignifi­cant. In Model IV, we reversed this process, retaining the income variables while omitting the education variables. Again, little changed in the regression-the variables race, age, and parental attitudes remained significant, while father's income and mother's income remained insignificant and nearly flat.

Fathers' Employment and Marriage

Although male annual income does not appear to contribute to marriage among fragile families, it is possible that male employment does. It may be that the stability of a father's employment, rather than his income per se creates, over time, a sense of confidence that facilitates marriage.

As Chart 1 shows, fathers in fragile families who maintained 52 weeks of employment during the year following a child's birth were more likely to marry. We tried several variables to test the impact of paternal employment on marriage among fragile families: Whether a father was employed at the time of a child's birth was found to be insignificant, as was unemployment during the year after birth.

However, father's employment at the time of the one-year follow-up interview was found to have a significant and robust link to marriage. This finding agrees with Mincy and Dupree.[4] At first glance, the fact that father's employment at the time of birth fails to predict subsequent marriage, while employ­ment one year after the birth is strongly linked to marriage, appears puzzling. It may be that a pattern of improvement in the father's employment instills confidence that fosters marriage.

Through trial and error, we developed a three-part set of dummy variables that best explained the role of paternal employment in post-birth mar­riage rates. These variables were:

Currently Employed: 52 Weeks. These fathers were employed at the one-year follow up and were employed for 52 weeks in the year after their child's birth.

Currently Employed: Less than 52 Weeks. These fathers were employed at the one-year follow up survey, but had worked less than 52 weeks in the year after their child's birth.

Not Currently Employed. These fathers were unem­ployed at the time of the one-year follow up survey.

The effects of these variables are shown in Table 2 of the Appendix, which reproduces the first four models in Table 1, except that the fathers' income variable has been replaced by the fathers' employment variable. The find­ings in Table 2 replicate those in the models in Table 1, except that the paternal employment variables remain sig­nificant as predictors in all four models. Overall, we found five variables to be significant in explain­ing marriage within one year after a child's birth. These were:

• Father's employment;
• Mother's race;
• Mother's age (25 and over );
• Father's marital attitudes and relationship skills; and
• Mother's marital attitudes and relationship skills.

The final regression using these variables is shown in Table 3 of the Appendix.

The Impact of Fathers' Employment

The effects of fathers' employment on marriage in the year after a child's birth are summarized in Chart 2. The chart shows that, holding race, moth­ers' age, and parental attitudes constant, a father who was employed at the follow-up survey and who had maintained employment during the 52 weeks after his child's birth was almost twice as likely to marry the child's mother as was a father who was unemployed at the one-year follow-up. Some 13 percent of employed fathers became mar­ried compared to 7 percent of those who were not employed.

Although doubling the marriage rate appears to be a strong effect, two caveats must be applied. First, because the marriage rate of unemployed fathers was only 7 percent, doubling that rate yields only modest gains. Second, as Chart 3 shows, roughly three-quarters of the fathers who were unmarried at the time of the birth were employed at the one-year follow-up, and nearly half had maintained employment for a full 52 weeks in the year after the birth. As a conse­quence, increases in employment would affect less than half of all the unmarried fathers, diminishing the potential impact of enhanced employment on the overall marriage rate.

This is shown in Chart 4, which simulates the effects of increased male employment. The chart shows the projected marriage rate if all the fathers who were unmarried at the time of the birth had maintained employment for a full 52 weeks in the year after the birth and were still employed at the follow-up survey. Under these conditions, the overall marriage rate at one year after the child's birth would rise by roughly two percentage points-from 11.3 percent to 13.2 percent. This suggests that increases in employment alone are not likely to have an appreciable effect in expanding marriage among fragile families.

Parental Marital Attitudes and Relationship Skills

All eight regression models (Tables 1 and 2 of the Appendix) show that parental attitudes and relationship skills are strong and significant predictors of marriage. The attitude and skill scores for the moth­ers and fathers are based on 14 ques­tions taken from the baseline survey. The questions measure four different factors:[5]

• Positive attitudes toward marriage. These ques­tions measure the extent to which the parent believes marriage is important and beneficial for children and adults.
• Gender trust. These questions measure the degree to which the individual believes the opposite gender is exploitative and unfaithful.
• Support in the relationship. These questions measure how much affection, support, and encouragement individuals receive from their partners.
• Conflict in the relationship. These questions measure the extent of disagreement and con­flict in a relationship.

The exact questions used for the scale are shown in Table 4 of the Appendix. To compose an overall attitude and skill score for each parent, an average score for the questions under each factor was determined. The average scores for the four factors were then summed. This procedure yielded a possible range of scores (for each parent) from 4 to 14. For pur­poses of simplicity, the summary scores were recal­ibrated to a scale of 1 to 11.

Thus, each parent has a potential attitude score ranging from a low of 1 to a high of 11. The median score of mothers who were not married at the time of their children's birth was 7.83 with a standard deviation of 1.152. The median score of fathers who were not married at the time of their child's birth was 8.16 with a standard deviation of 1.147. The scores for the mothers and fathers can be added to produce a joint couple score ranging from 2 to 22. The median joint couple score for parents who were not married at the time of their child's birth was 16.0 with a standard deviation of 1.81.

Each of the regressions in Tables 1, 2, and 3 shows that the mother's attitude and skill score has a greater impact on marriage than the father's. However, this is mis­leading because the mother's responses to the supportiveness questions, in fact, refer to the father's behavior-and vice versa. To understand marital behavior, it is best to look at the mothers' and fathers' scores in tandem.

Effect of Attitude and Skills on Marriage

A couple's attitude and skills score is highly correlated with marriage rates. Chart 5 shows data for all the couples in the Fragile Families survey, both those who were married at the time of their child's birth and those who were not. The chart shows the percent of couples at each joint-score level (on a scale of 2 to 22) who were married when their child was born.[6] The higher a couple's score on the attitude and skills scale, the greater the likelihood that they were married at the time of their child's birth. Virtually none of the couples with joint scores below 14 were married at the time of their child's birth. By contrast, over 80 percent of couples with scores above 18 were married at the time of their child's birth.

Chart 6 shows that this linkage also obtains for unmarried couples during the period after a child is born out of wedlock. Among couples who were not married at their child's birth, higher attitude and skill scores lead to a higher probability of mar­riage during the first year after birth. Among the one-quarter of couples with the lowest attitude and skills scores, the predicted rate of marriage is only 4.9 percent. Among the one-quarter of cou­ples with the strongest scores, 27.7 percent were likely to marry.[7]

The Impact of Improving Attitudes and Skills on Marriage

Chart 7 shows the potential effect of improving attitudes and skills on marriage. Under current cir­cumstances, 11.3 percent of couples that were unmarried at their child's birth do marry within the subsequent year. Raising each couple's attitude and skills score by two points (on a scale of 2 to 22) would nearly double the probability of mar­riage, raising it to 21.9 percent.[8]

Marital attitudes and skills will play an increas­ingly prominent role in debates about marriage policy. The healthy marriage programs contained in both the House and Senate welfare reform bills direct funds toward marriage skills education pro­grams. These programs, in turn, will be specifi­cally designed to address many of the relationship issues measured in the attitude and skill scales used in this paper.

Our current knowledge concerning the capacity of marriage education programs to greatly improve attitudes and skills in target populations is lim­ited.[9] It is unclear to what extent such attitudes and behaviors are rigid or malleable. However, on the surface, at least, shifting attitude scores does not appear difficult. In general, a one point shift in a parent's attitude score can be achieved by shifting the intensity of response to a few of the 14 ques­tions listed in Table 4. For example, a one-point rise in a parent's attitude score will occur if the parent states that he or she "strongly agrees" (rather than just "agrees") with the following two statements:

• "It is better for a couple to get married than to just live together," and,
• "It is better for children if their parents are married."

Comparing the Effects of Fathers' Employment and Parental Attitudes and Skills on Marriage

The foregoing discussion has indicated that changes in a couple's marriage attitudes and rela­tionship skills are likely to have a far greater effect on increasing marriage than are changes in pater­nal employment. This is demon­strated in Chart 8. Under status quo conditions, 11.3 percent of unmarried parents will marry dur­ing the first year after their child's birth. If all unmarried fathers had maintained employment for the first 52 weeks after their child's birth, the marriage rate one year after the birth would have risen to 13.2 percent.

By contrast, if paternal employ­ment remained unchanged, but each parent had a one-point increase in their attitudes and skills score, the marriage rate would rise to 21.9 percent. If improvements in attitudes and increases in employment were combined, the effect on marriage would be somewhat stronger. If all fathers maintained full employ­ment and each parent had a one point upward shift in his or her attitude score, the marriage rate in the group as a whole would be expected to rise to 25.1 percent.

Targeting Healthy Marriage Services

Because future healthy marriage funds will be quite limited, it is important that programs be targeted toward couples that have reasonable prospects for entering and sustaining marriage. Clearly, services provided to couples with poor attitude scores (below a joint score of 16) are unlikely to lead to many marriages, healthy or otherwise. By contrast, providing marriage skills training to unmarried couples with reason­ably strong initial attitudes (above a joint score of 16) may substantially increase marriage rates and improve relationship quality.

In some cases, effective employment services will also play a useful role with this group. If a couple has positive marital attitudes and decent relationship skills, but the father has difficulty sustaining work, increases in his employment may substantially improve the couple's marital prospects. However, provid­ing job training across the board to all unmarried parents is unlikely to have a discernable impact on mar­riage rates.

The Role of Job Training in Healthy Marriage Promotion

The preceding analysis shows that job-training programs are likely to have a modest effect in increasing marriage among non-married parents. Marriage programs that place a dominant focus on pro­viding job training to non-married fathers are unlikely to be successful. There are four reasons why a pri­mary focus on job training is inap­propriate as a strategy for promoting healthy marriage:

1. As noted, training and counsel­ing to improve attitudes toward marriage and relationship skills are likely to be more effective in promoting marriage. In part, this is due to the fact that increasing paternal employment will be important only for a minority of unmarried fathers, whereas improvements in relationship attitudes and skills would have a beneficial effect on all unmarried couples.
2. The federal government has conducted exten­sive job training programs for 40 years. The impact of these programs on employment and wages has been modest at best. Pinning hopes for promoting healthy marriage on old-style programs with mediocre track records seems to be an unwise strategy.
3. The federal government already spends around $6 billion per year on job training. Piling addi­tional funding on top of this sum is unlikely to accomplish much. To the extent that job train­ing would be a useful support to marriage pro­motion, the best strategy would be to re-target existing job training funds for this purpose rather than to divert limited funds available for marriage skills education into job training pro­grams that are already amply funded.
4. In general, marriage skills training per couple will be far less expensive than job training. Therefore, given limited funds, a marriage pro­gram with a focus on job training will reach far fewer couples than a program focused on mar­riage and relationship education.

Promoting healthy marriage is a new policy goal. Meeting this goal will require developing entirely new programs, rather than re-treading old job training programs, which were never devel­oped to promote marriage in the first place. The analysis presented here suggests that marriage pro­motion programs should focus on providing mar­riage education with the aim of improving couples' attitudes and relationship skills.

This does not mean that effective job training and related employment services cannot play a mean­ingful role in marriage promotion. In some situa­tions, helping a father maintain steady employment could greatly facilitate a couple's marriage. Overall, however, job training and employment services should play an ancillary and supportive-rather than a dominant-role in marriage promotion.

Is the Glass Half Empty
or Half Full?

In the preceding analysis we have outlined the changes in attitudes and employment that could potentially lead to a doubling of the marriage rate among currently unmarried fragile families. On the one hand, doubling marriage rates sounds impressive. On the other hand, because the base mar­riage rate among these couples is low (around 11 percent in the first year after the child's birth), even if the rate were doubled, most couples would be unaffected. If the upper boundary for subsequent marriage among parents who have non-marital births is really around 20 percent, this could be judged a cause for pessimism.

For a number of reasons, we feel this pessimistic interpretation is unwarranted. First, if the post-birth marriage rate among unmarried par­ents were raised from 11 percent to 22 percent, roughly 400,000 children would be affected over the course of a decade. This would be a notable pol­icy success. Second, without interven­tion, a number of unmarried couples will marry in the second and third years after their child's birth. Presum­ably, pro-marriage services could raise the marriage rates in these succeeding years as well, thereby raising the overall marriage rate well above the first-year figure. Third, we currently do not know how much marriage skills training can shift attitudes in the target popula­tion: The actual changes could be considerably greater than the one-point shifts described in this paper. Finally, an effective healthy marriage policy will target not merely unmarried parents at the time of their child's birth, but many other groups as well. These additional groups may, in fact, be better candidates for healthy marriage promotion.

A Broader View of Healthy Marriage Promotion

Policy discussions about how to promote healthy marriage are of very recent origin. Much of the discussion to date, has focused on "fragile families" at the "magic moment" of a child's birth (i.e., unmarried couples at the time of, or shortly after, their child's birth). This focus has been rea­sonable given the abundant, and often surprising, data about these couples provided by the Fragile Families survey. The child's birth also provides social service agencies easy access to nearly all low-income mothers. Thus, the maternity ward seems to be an excellent venue for beginning an intervention.

Although parents who are unmarried at the time of a child's birth should be an important target group in any healthy marriage initiative, there are reasons to believe that these parents are not neces­sarily the ideal candidates for marriage programs, and that the "magic moment" of birth, while important, may not be the optimum point for ini­tiating an intervention.

A serious pro-marriage initiative would target a broader array of groups in a variety of venues. It should include:

• Education about the value of mar­riage and life-skills planning for high school students who are at risk of out-of-wedlock childbearing;
• Marriage skills training for low-income married couples at the time of a child's birth. Childbirth places considerable strain on relationships and this can lead to divorce. It is possible that lower-income married couples could benefit from pro-mar­riage services as much or more than unmarried parents;
• Pre-marital counseling programs for engaged couples and marriage enrichment programs for married couples. These programs have poten­tial to reduce future divorce. While it would not be necessary for the gov­ernment to broadly subsidize mid­dle-class use of these programs, government funds should be used as a catalyst to promote awareness and make such programs more widely available; and
• Marriage and relationship skills training for young unmarried adults prior to a child's conception.

It is clear that many unmarried, new parents are not well prepared for either marriage or parenthood. There is wide­spread agreement that the best point of interven­tion with these young couples would have been prior to their child's conception, rather than after the child's birth. However, while the government has virtually guaranteed access to low-income mothers at the time of birth, contact with young, low-income adults at an earlier stage is com­monly thought to be difficult or impossible.

In fact, this perception may be erroneous. The federal government currently funds some 4,700 birth control clinics through the Title X program. These clinics provide birth control to 4.4 million low-income persons each year-most of which are young adult women. Many of the clientele of these clinics will become members of the "fragile fami­lies" of the future.

In addition to birth control, it should be rela­tively simple for these clinics to offer referrals to programs providing life-planning, marriage, and relationship training to those who are interested. The goal of such programs would be to encour­age young adult women to delay childbirth and to develop stable marital relationships before bringing children into the world. The potential for outreach through the Title X clinics may actually be greater than through maternity wards. Expanding healthy marriage services to cover points prior to a child's conception may considerably increase the effectiveness of future programs.

Conclusion

For 40 years, the attitude of the welfare system toward marriage has ranged from indifference to hostility. Fortunately, this is beginning to change. An increased recognition of the importance of marriage should lead to policies to help couples, who are interested, enter into and sustain healthy marriages. Success in this endeavor will entail new and different policies, rather than a continuation of old programs that had only a marginal relation­ship to marriage in the first place.

Robert Rector is Senior Research Fellow in Domestic Policy and Kirk A. Johnson, Ph.D., is Senior Policy Analyst in the Center for Data Analysis at The Heritage Foundation. This study is based on a paper presented at the Administration of Children and Fami­lies' Seventh Annual National Welfare Research and Evaluation Conference, May 26, 2004.