Will I Stay Married? Exploring Predictors of Expectations to Divorce in Unmarried Young Adults

Abstract

Expectations that one may eventually divorce may predict behavior in young adulthood and beyond, but studies that have looked at individuals’ assessments of their divorce likelihood have been limited. Guided by the expectancy-value theory of achievement motivation, we tested five categories of potential predictors of divorce expectations in a sample of 1,610 unmarried young adults from the Panel Study of Income Dynamics Transition to Adulthood study. Predictors were tested separately by gender and partnership status. Results suggested that some predictors mattered more for some groups than others, such as employment for single men, or certainty of marriage for partnered women. Consistent with prior research, caregiver divorce was significantly associated with expectations to divorce, but was only one of many factors found to predict these expectations. Socioeconomic factors and experiences and expectations of other relationships consistently predicted expectations. Expectations to divorce are multifaceted and complex.

The transition to adulthood is a prime time for forming identity and attitudes including desires and expectations for romance (Arnett, 2000; Willoughby, Hall, & Luczak, 2015). Expectations for intimate relationships can predict behavior (Arocho & Kamp Dush, 2017; Willoughby, 2014), thus, understanding the correlates of attitudes towards future relationships can help identify those most likely to hold such expectations and experience such potential outcomes. Further, as concerns regarding divorce might be a barrier to marriage in the first place (Miller, Sassler, & Kusi-Appouh, 2011; Waller & Peters, 2008), expectations of divorce specifically might help explain some of the current marital delay among emerging adults (Willoughby & James, 2017). We used the 2011 wave of the Panel Study of Income Dynamics (PSID) Transition to Adulthood (TAS) supplemental study to investigate how unmarried young adults’ characteristics predicted explicit expectations to divorce.

Attitudes towards Divorce

Many studies have examined general attitudes towards divorce (e.g., “Is divorce acceptable?”) across the life course and over time (e.g. Hatemi, McDermott, & Eaves, 2015; Kapinus, 2005; Thornton & Young-DeMarco, 2001). In addition, a few studies have examined “fears” of divorce in cohabiting couples (Gibson-Davis, Edin, & McLanahan, 2005; Miller et al., 2011; Waller & Peters, 2008). Both attitudes and fears are associated with marital outcomes and behavior, at least in adults (Amato & Rogers, 1999; Miller et al., 2011; Perelli‐Harris, Berrington, Sánchez Gassen, Galezewska, & Holland, 2017; Waller & Peters, 2008), but beyond general attitudes, individuals’ specific expectations can be especially salient for intentions and outcomes. Individuals internalize expectations and behave in ways that facilitate expected outcomes (Merton, 1948). Expectations of family life indeed predict both family life outcomes and other behaviors, even in adolescence and young adulthood (Arocho & Kamp Dush, 2016, 2017; Willoughby, 2014; Willoughby & Dworkin, 2009).

Young adults often hold marriage in high regard but may be hesitant to marry because they fear divorce (Willoughby & James, 2017). Feeling fearful of divorce (Miller et al., 2011; Perelli‐Harris et al., 2017), or even simply being exposed to it in one’s close environment (Waller & Peters, 2008), is associated with delaying marriage in adult samples. One’s self-assessment of their risk of divorce may play a role in behavior and outcomes of intimate relationships even earlier. Considering the current delay of marriage by many young adults (Payne, 2015) and the potential implications of marriage timing for relationship success and individual wellbeing (Glenn, Uecker, & Love Jr, 2010; McLanahan & Jacobsen, 2015), understanding the development of these expectations may help social scientists better understand relationship formation in emerging adulthood and implications for further development. Nonetheless, studies of explicit expectations of divorce have been limited. A few studies have examined divorce expectations in newlyweds (Baker & Emery, 1993; Campbell, Wright, & Flores, 2012), but newlyweds may be overly-optimistic and not be forthcoming with expectations of divorce. For example, Baker and Emery (1993) found that the median rated personal likelihood of divorce was 0%, even after respondents had accurately assessed the overall likelihood of divorce in the United States.

Studies of unmarried individuals have sampled mostly college students and focused on parental divorce as the main predictor (Black & Sprenkle, 1991; Boyer-Pennington, Pennington, & Spink, 2001; Helweg-Larsen, Harding, & Klein, 2011; Kirk, 2003). All found that offspring of divorced parents rated their chances of divorce higher than other students. These studies may not be generalizable to a wider population of emerging adults, including those who may work or form families instead of (or in addition to) attending college. It is important to study these issues more broadly, as the transition to adulthood is when most individuals form their first serious intimate relationships (Manning, Brown, & Payne, 2014) and develop their identity and behaviors concerning the relationships they have or desire to have (Arnett, 2000; Carroll et al., 2007).

Expectancy-Value Theory Applied to Relationship Expectations

In the expectancy-value theory of motivation (Wigfield & Eccles, 2000), experiences and perceptions inform expectations and values, which then lead to behaviors aimed at achieving goals. The theory identifies two motivators that predict the choices youth make as they pursue goals: value of the goal and expectations for success. These then work together to predict actions aimed at achievement. If an individual holds a goal in high regard but feels they have no hope of achieving it, they will not strive towards the goal, and similarly, one who knows they could achieve a goal that holds no value to them will also not be motivated to act. Therefore, understanding the balance of these two constructs should predict behavior. Applying this model to relationship expectations, expectations of divorce might hinder entrance into marriage or predict other behaviors even in unmarried individuals if they represent the individuals’ perceived likelihood of future success or failure in relationships. The theory does not only predict the outcome of these expectations, however, but also suggests a variety of experiences and personal characteristics that might predict the expectations that one may hold. These range from broad and overarching features (such as culture) to more personal characteristics that may be more immediately obvious to particular feelings and cognitions (e.g. previous experiences with the goal, socializer’s beliefs and behaviors, or one’s self-concept).

Two of the most fundamental predictors of values and expectations in this theory are cultural milieu, including gender role stereotypes, and previous achievement-related experiences. (Wigfield & Eccles, 2000). Males and females report different marital beliefs and associations between experiences and marital expectations (Carroll et al., 2009; Crissey, 2005). We believe gender will also be important in the association between other predictors and expectations and expectations for divorce. For instance, enduring desires for men to be breadwinners (Edin, 2000; Gibson-Davis, Edin, & McLanahan, 2005) may mean that men would be more sensitive to economic or achievement related factors when considering their chances of staying married, or holding stronger requirements to feel “ready” for marriage (Carroll et al., 2009) may mean that women are more sensitive to markers of maturity and stability when considering expectations for divorce. Further, previous research on attitudes towards marriage and divorce in relation to parental divorce has found gender differences (e.g. Black & Sprenkle, 1991). Given its role in culture and norms, gender likely interacts with predictors of divorce expectations. Thus, to address the fundamental component of cultural milieu identified in expectancy-value theory (Wigfield & Eccles, 2000), we examine the hypothesized predictors separately by gender.

One of the other fundamental predictors identified in the theory is previous achievement-related experiences. In our case, this is represented by relationship experience. Indeed, previous research has found significant differences in marriage and cohabitation expectations between partnered and single young adults (Crissey, 2005; Gassanov, Nicholson, & Koch-Turner, 2008; Manning, Longmore, & Giordano, 2007). Experiencing relationships may help youth be more aware of their own abilities to meet the demands of a partnership and may give them a better sense of their own likelihood of staying married. In this way, we see relationship experience as an example of the “previous achievement related experiences” identified in expectancy-value theory (Wigfield & Eccles, 2000), and believe one’s relationship status will play a fundamental role in the development of youths’ divorce expectations alongside gender.

In expectancy-value theory, cultural milieu and previous achievement-related experiences are two of the most basic drivers of both expectations and values (Wigfield & Eccles, 2000), and we believe this applies to the current study as well. Both gender and relationship status likely shape associations between these expectations and other characteristics and experiences. Rather than create many interaction terms for each model to see differences between genders and relationship statuses, we elected to separate the sample and analyze each combination discretely, thereby allowing all associations to vary between groups and displaying the full range of variation in associations simultaneously.

What Informs Expectations of Divorce?

Expectancy-value theory also details personal factors hypothesized to predict an individual’s values and expectations for success. Most proximal to expectations for success and values are factors including one’s self-concept and understanding of one’s own abilities (Wigfield & Eccles, 2000). This self-concept is a characteristic we highlight heavily in our hypothesized predictors of expectations for success in this study. If one already fears that they cannot make a relationship work or be “marriageable,” one may worry about experiencing divorce if they do marry. For this reason, we tested various characteristics of one’s current relationship and relationship experience. Both having children and being in relationships are predictive of marital attitudes and expectations (Crissey, 2005; Lichter, Batson, & Brown, 2004; Tasker & Richards, 1994), though the potential direction of associations with divorce expectations is not clear. Additionally, youth who have had more sexual partners may be less marriage-oriented (Hamilton & Armstrong, 2009), which may translate to an association with greater divorce expectations if they do expect to marry (Halpern-Meekin, 2012).

Being confident in one’s ability to maintain a satisfying marriage and avoid divorce may stem from socioeconomic circumstances as another measure of efficacy and confidence in success. Economic factors such as poor employment or income are often barriers to marriage (Edin, 2000; Edin, Kefalas, & Reed, 2004; Gibson-Davis et al., 2005; Reed & Edin, 2005) and an ability to provide is an important measure of marital readiness (Carroll et al., 2009). Also, marriage is increasingly stratified by education and the divorce rate of the least-educated has continued to rise while that of the most highly-educated has declined (Cherlin, 2010). Thus, both employment and educational attainment may play a role in expectations to divorce, and we measured these and other related characteristics as socioeconomic circumstances.

Interpretations of one’s previous experiences and one’s self-concept, both constructs in expectancy-value theory (Wigfield & Eccles, 2000), may stem in part from mental and emotional wellbeing. Mental health concerns may lead one to feel less sure of their ability to maintain a relationship. Anxiety and other issues with social relations are associated with poorer relationship functioning (Kashdan, Volkmann, Breen, & Han, 2007), and poor emotional wellbeing or distress might indicate depressive symptomology or other conditions, which are associated with poorer relationship functioning and less success in maintaining long-term relationships (Sandberg-Thoma & Kamp Dush, 2014). General wellbeing is closely associated with intimate relationships and might also be associated with expectations of divorce (Kamp Dush & Amato, 2005; Kim & McKenry, 2002).

Expectations and attitudes about family life may also play a role in one’s likelihood of reporting expectations for divorce. Expectations for other experiences, mainly marriage and parenthood, likely function alongside expectations of marital success (Willoughby et al., 2015). Interviews have suggested that the most marriage-oriented youth may be less accepting of divorce (Halpern-Meekin, 2012; Willoughby & Hall, 2015) and respondents may feel that marriage and parenting are connected in their expectations for their future (Plotnick, 2007) and rate themselves accordingly. Religiousness is also important in developing attitudes towards marriage and divorce (Thornton, 1985) and was included here as well.

Socializers’ (in many cases, parents’) experiences and beliefs play an important role in the development of youths’ beliefs and expectations (Wigfield & Eccles, 2000). Because of strong evidence for the intergenerational transmission of marital behavior and attitudes (Amato & DeBoer, 2001; Arocho & Kamp Dush, 2017), parental marital status, particularly experience with divorce, are likely to be important for young adults’ expectations of divorce. Additionally, those from lower socioeconomic family backgrounds may be more aware of the restrictions of marriage and costs of divorce, as well as the stresses placed on lower-SES marriages (Edin, 2000; Miller et al., 2011; Neff & Karney, 2016), and thus be more concerned with their risk of divorce than other young adults. Thus, we consider these childhood experiences in our analyses of divorce expectations.

In summary, the choice of predictors was grounded in empirical evidence, but due to the scarcity of previous work on expectations of divorce our analyses were exploratory in nature and we did not hypothesize overall directions of association. Therefore, our research questions were:

• 1)How are past and present experiences and personal characteristics associated with young adults’ likelihood of reporting expectations to divorce?
• 2)Do the factors predicting expectations to divorce differ by gender and relationship status?

Method

The Panel Study of Income Dynamics (PSID) began in 1968 and is produced by the Institute for Social Research at the University of Michigan. It began with a U.S.-representative probability sample of 2,930 families and an oversample of 1,872 low-income families, and by 2011 was made up of almost 9,000 families. When an individual in a sample household started their own household, their new household was added back into the PSID sample, and this process was repeated when that family’s children grew up and formed their own households. This design has created a rich dataset spanning multiple generations (PSID Transition into Adulthood Study 2011 User Guide, n.d.).

Transition to Adulthood Supplement.

The Child Development Study (CDS) was first conducted in 1997 on the children ages 12 and younger of the PSID families. The children were observed and interviewed and their caregivers gave assessments of the children. As children aged they moved into the Transition to Adulthood Supplement (TAS) sample starting in 2005. To qualify for the TAS, youth had to be at least 18 years old, out of high school, and their families (headed by either themselves or a family member) had to have completed a main PSID interview that year. The TAS has been repeated every other year, with additional youth added as they aged into the sampling frame. The TAS sample is ideal for this study because of attention placed on attitudes and expectations as well as detailed measurement of personal and contextual characteristics. We used data from 2011, when the most TAS respondents were interviewed (N = 1,907). We limited the sample to those respondents who had never married, reducing the sample by 253 youth. Additionally, 43 respondents indicated that they believed they would not marry and were not asked to state expectations of divorce, and 1 person was missing partner status and could not be imputed, thus N = 1,610 respondents were included in our analytic sample. For imputation and analyses, we divided the sample between males and females and those who were partnered (either dating or cohabiting) and single, making four groups: partnered males, n = 362; partnered females, n = 424; single males, n = 432; and single females, n = 392. It should be noted that the sample was primarily African American (17%) and White (66%) youth; the PSID was nationally-representative of the U.S. in 1968 but lacks current the racial and ethnic diversity.

Missing data.

Most individuals were missing data on at least one variable, although no single variable was missing more than 12%. We estimated missing data using the Multiple Imputation with Chained Equations procedure in Stata14 (White, Royston, & Wood, 2010). In this method, each variable was regressed on all other variables in the model to estimate missing values. This was repeated over 50 datasets. Analyses were conducted on all datasets separately and results were combined to generate an overall estimate. Missing data were estimated separately by gender and partnered status to preserve the differences between groups (Von Hippel, 2007).

Weighting.

About two-thirds of CDS 1997 respondents were eligible for the TA 2011, but only about 82% of the eligible youth were interviewed. The TA 2011 weight was provided in the dataset and built upon the CDS weight, and we applied it to both imputation and analyses. As all respondents were descendants or family members of descendants of the original PSID families, we adjusted standard errors for family nonindependence using the “vce cluster” option in Stata14.

Variables

See Table 1 for more detailed description of each variable and its coding for analyses.

Table 1:

Description and coding of analytic variables.

Name	Description	Coding
Expectations for Divorce	“What do you think are the chances that you will get divorced?”	1 = no chance 2 = some chance 3 = about 50–50, pretty likely, it will happen
Characteristics of the current relationship and family formation
Cohabitation	“Are you living with a partner in a marriage-like relationship?”	1 = cohabiting 0 = dating
Relationship Satisfaction	“…how satisfied are you with your (current) relationship in general?”	1 = Not at all satisfied to 7 = Very satisfied
Children	“How many (biological), adopted, or step-children do you have?”	1 = 1+ children 0 = 0 children
Total Sexual Partners	Combination of three questions: “Have you ever had sexual intercourse?” If yes, “Have you had sexual intercourse more than once?” If yes, “How many partners have you EVER had intercourse with?	0 – 97
Socioeconomic Circumstances
College Enrollment	“Are you going to school full-time or part-time?”	1 = part or full-time student 0 = not student
Education	Variable constructed from responses regarding education level, high school and GED, college attendance, and degrees	1 = less than high school 2 = high school or GED 3 = some college 4 = Associate’s degree or higher
Employment	Constructed from average hours worked each week in past year at up to five jobs.	1 = full-time (35+ hours) 2 = part-time (1–34 hours) 0 = unemployed (0 hours)
Income	“How much did you earn altogether from work in 2010, that is, before anything was deducted for taxes or other things, including any income from bonuses, overtime, tips, commissions, military pay or any other source?”	0–80 (Values divided by 1,000)
Financial Responsibility	Mean of responses to 4 items. Example: “How much responsibility do you currently take for earning your own living?”	1 = “Somebody else does this for me all of the time” to 5 = “I am completely responsible for this all of the time” α = .79
Financial Worry	Mean of responses to 3 items. Example: “How often do you worry that you may not have enough money to pay for things?”	1 = “never” to 7 = “daily” α = .72
Mental and Emotional Wellbeing
Social anxiety	Mean of responses to 4 items. Example: “How often do you feel nervous when meeting new people?”	1 = “never” to 7 = “daily” α = .82
Social wellbeing	Mean of responses to 5 items. Example: “In the last month, how often did you feel that you had something important to contribute to society?”	1 = “never” to 6 = “every day” α = .74
Emotional wellbeing	Mean of responses to 3 items. Example: “In the last month, how often did you feel happy?”	1 = “never” to 6 = “every day” α = .80
Psychological wellbeing	Mean of responses to 6 items. Example: “In the last month, how often did you feel good at managing the responsibilities of your daily life?”	1 = “never” to 6 = “every day” α = .82
Psychological distress	Summed score in response to 6 items. Example: “During the past 30 days, how often did you feel nervous?”	0 = “None of the time” to 4 = “All of the time” Score ranged 0–24, α = .75
Other Expectations and Attitudes about Family Life
Likelihood of marriage	“What do you think are the chances that you will get married?”	1 = pretty likely or it will happen 0 = some chance, about 50–50
Expecting marriage soon	“At about what age do you think you will get married?”	1 = within five years of current age 0 = more than five years from current age
Would have children	(A) For those who already have children: “Would you like to have more kids?” (B) For those who do not have children: “What do you think are the chances that you will have children? … This includes biological and adopted children.”	0 = no (A) or about 50–50, some chance, no chance (B) 1 = yes (A) or pretty likely, it will happen (B)
Divorce Hurts Kids	“How much do you agree with the following statement? ‘Divorce usually has a very negative effect on the children.’”	1 = “strongly disagree”to 7 = “strongly agree”
Single Parenting Hurt Kids	“How much do you agree with the following statement? ‘ Children suffer if they are raised by a single parent.’”	1 = “strongly disagree”to 7 = “strongly agree”
Family Form Limits Success	Mean of agreement to “Children from divorced families can be just as well-adjusted and successful as children from twoparent families.” and “Children raised by single-parent families can do just as well as children raised by two-parent ones.”	1 = “strongly disagree” to 7 = “strongly agree” r = .62
Religious importance	“How important is religion to you?”	1 = very important 0 = not at all important, not very important, somewhat important, no religion, atheist, agnostic
Childhood Experiences
Maternal Education	Retrieved from adult individual files, years of completed education.	0–17
Parent was Caregiver	Created from Child Development Study files and caregiver map.	1 = parent 0 = other relation
Primary Caregiver Changed	Created from Child Development Study files and caregiver map.	1 = change 0 = no change
Caregiver Divorce before 18	Created from adult marital history file in combination with respondent birthdate.	1 = yes 0 = no
Mother was Married	Created from marriage and birth history files.	1 = married 0 = never married, widowed, divorced/annulled, separated
Covariates
Race	“What is your race? Are you white, black, American Indian, Alaska Native, Asian, Native Hawaiian or Other Pacific Islander?”	1 = white 2 = black 3 = other/multiple
Hispanic Identity	“Are you Spanish, Hispanic, or Latino? That is, Mexican, Mexican American, Chicano, Puerto Rican, Cuban, or other Spanish?”	1 = Any Spanish, Hispanic, or Latino identity 0 = Not Spanish, Hispanic, or Latino
Age	Year and month of interview minus year and month of birth, rounded down	19–29
Year Reported Sexual Partners	Year of interview respondents reported number of sexual partners.	1 = 2007 2 = 2009 3 = 2011

Dependent Variable

Respondents were asked, “What do you think are the chances that you will get divorced? Would you say: No chance, some chance, about 50–50, pretty likely, or it will happen? We recoded this variable to three categories: “No Chance” of divorce, “Some Chance,” and “50% or Greater Chance” (“50–50,” “pretty likely,” and “it will happen” combined). Prior to collapsing, both “pretty likely” and “it will happen” held less than 2% of the sample each. We retained “some chance” and “50% or greater chance” of divorce as separate responses and compared the risk of reporting either to reporting “no chance” of divorce in all analyses.

Independent Variables

Characteristics of the current relationship and family formation.

For partnered youth, an indicator variable denoted cohabitation versus dating, and relationship satisfaction was assessed by responses 1–7 to the question, “how satisfied are you with your (current) relationship in general?” An indicator variable recorded whether or not any respondents had children (biological, adopted, or step). At each TAS wave respondents were asked if they had had sexual intercourse before, and how many total sexual partners they had had up to that point.

Socioeconomic circumstances.

We included an indicator of current college enrollment and a measure of the highest level of education respondents had completed each wave in four categories: less than high school, high school graduate or GED, some college, and college degree or higher. Young adults reported the average hours worked per week in the previous year at up to five jobs. We averaged these hours and then categorized youth as working full-time (35 hours or more per week), part-time (1–34 hours), and unemployed. Respondents reported their income from the previous year (divided by 1,000). Financial responsibility was an average of questions regarding the responsibility the respondents held for their own living expenses, rent, bills, and managing money; higher scores indicated more responsibility. Respondents were asked how much they worried about money, employment, and the future. These variables were averaged for financial worry, with higher scores indicating more worry.

Mental and Emotional Wellbeing.

Youth’s emotional and mental wellbeing responses were averaged (save for psychological distress, which was summed) into five scales according to their original design (PSID Transition into Adulthood Study 2011 User Guide, n.d.; Survey Research Center, 2008). Social anxiety dealt with issues of feeling nervous, shy, or self-conscious. Social wellbeing measured how much one felt like they contributed to society, belonged to the community, and had positive feelings towards society. Emotional wellbeing measured how often one felt happiness, interest, and satisfaction with life. Psychological wellbeing measured feelings of being able to manage responsibility, have trusting relationships, be challenged to grow, feel confident in one’s own ideas, feel like life has direction, and like one’s personality. Psychological distress measured feeling nervous, hopeless, restless, giving excessive effort, sad, and worthless. The social anxiety scale (and the related worry and financial responsibility scales, which we classify under socioeconomic characteristics), were developed for the Michigan Study of Adolescent and Adult Life Transitions (Survey Research Center, 2008). The three wellbeing scales were drawn from a variety of sources and first used in the Midlife in the United States study (Survey Research Center, 2008). The psychological distress questions are from the scale developed by Kessler et al., 2003 for the National Health Interview Survey.

Other expectations and attitudes about family life.

Respondents were asked to rate their likelihood of marriage; responses were dichotomized, with answers of greater than 50% chance of marriage coded as 1 and 50% or less as reference; those who indicated no chance of marriage were not asked about divorce expectations and thus not in the sample. Those who expected marriage within 5 years were coded as expecting marriage soon versus later. Respondents were asked if they believed they would have children in the future (or more children if they were already a parent) and this was also dichotomized.

Youth were asked to rate on a scale from 1–7 how strongly they agreed with the following statements: “Divorce usually has a very negative effect on the children,” “Children suffer if they are raised by a single parent,” “Children from divorced families can be just as well-adjusted as children from two-parent families,” and “Children raised by single-parent families can do just as well as children raised by two-parent ones.” The final two questions, both dealing with the ability of children to be successful in non-traditional family forms, were averaged due to the high correlation between them. The three resulting questions were coded so that higher scores indicated a more negative view of divorce or single parenting and included as separate predictors to test if some attitudes towards divorce and family structure were better predictors of one’s own expectations than others. Respondents also indicated religious importance, coded dichotomously as very important versus less because of the relatively few respondents reporting low religious importance.

Childhood experiences.

We used maternal education in 2011 as an indicator of childhood socioeconomic status. We used the PSID’s adult marital history to determine if the adult reported as the respondents’ primary caregiver in the earlier Child Development Study (CDS) had divorced after the respondent’s birth but before the respondent turned 18. Finally, we indicated if the respondent’s mother was married at the time of respondent’s birth.

Covariates.

Race was coded as White, Black, and Other, with ethnicity coded separately as Hispanic or not. Age was calculated using birth and interview date. As additional controls, we recorded the child’s relationship to their primary caregiver (biological parent versus other, such as grandparent or adoptive parent) from the primary caregiver reported during the respondent’s time in the CDS. An indicator denoted whether primary caregiver changed over the course of CDS participation; in cases of change, the last primary caregiver was used to measure marital history. Finally, because respondents were not asked to update their number of sexual partners each wave, when necessary we retrieved responses from previous waves and included the year of response as a control.

Results

Analytic Plan

We used multinomial logistic regression to model probabilities of rating oneself as having no chance, some chance, or a 50% or greater chance of divorce. We tested bivariate correlations between independent variables to assess the risk of multicollinearity. Of all correlations, less than 1% were greater than |0.50|, the largest being 0.66. Even among mental and emotional wellbeing scales, which might be expected to have the greatest collinearity, the largest correlation was 0.57. The largest variance inflation factor (VIF) was less than 4 (VIF of 4, 10, and 20 have been suggested as indicators of concern; see O’brien, 2007). Thus, we used models including all predictors. Differences between subsamples were assessed using the formula put forth by Clogg, Petkova, and Haritou (1995). Because of the large number of tests conducted, we consider differences in all analyses and comparisons significant at p < 0.01 to reduce the risk of Type I error, or false-positive results.

Descriptive statistics for the four groups are displayed in Table 2. Descriptive differences between groups were tested by appropriate bivariate regressions and significant differences are denoted in Table 2. In all models, the reference category was “no chance” of divorce. In Table 3, we report relative risk ratios (RRR), coefficients, and standard errors for the model for the full sample. RRR values can be interpreted similarly to odds ratios (positive associations above 1, negative association below 1). Tables 4 and 5 display the results of the model in each subgroup; Table 4 displays the values for the outcome “some chance” of divorce versus “no chance,” and Table 5 displayed the values for the outcome “greater chance” versus “no chance.”

Table 2:

Descriptive Statistics and Comparisons between Groups

	Partnered Males			Partnered Females		Single Males			Single Females
Variable	Mean / %	SE		Mean / %	SE	Mean / %	SE		Mean / %	SE
Outcome
Chance of Divorce (Reference: No Chance)
Some Chance	0.44			0.46		0.49			0.45
50% or Greater Chance	0.15			0.10d		0.20d,f			0.12 f
Current Relationship and Family Formation
Cohabiting	0.28 a			0.42 a
Relationship Satisfaction	5.97	0.07		6.10	0.06
Has Children	0.13 b			0.21 d		0.05 b,d,f			0.15 f
Number of Sexual Partners	5.43 c	0.59		4.02 e	0.22	3.85 f	0.41		2.60 c,e,f	0.22
Socioeconomic Circumstances
Student	0.38 c			0.37 e		0.44			0.52 c,e
Education (Reference: High School or GED)
Less than High School	0.08			0.05		0.06			0.04
Some College	0.26			0.18		0.22			0.14
AS or Higher	0.50 c			0.51		0.58			0.65 c
Employment (Reference: Full Time)
Unemployed	0.09			0.11		0.10			0.10
Part Time	0.41			0.47		0.42			0.51
Income (Thousands)	0.51			0.42		0.47			0.39
Financial Responsibility	5.45 a	0.72		3.16 a,d	0.48	6.37 d,f	0.64		3.70 f	0.47
Worry about Future	4.19 c	0.07		4.07	0.07	3.97	0.07		3.83 c	0.08
Mental and Emotional Wellbeing
Social Anxiety	3.11 a,c	0.10		3.50 a	0.09	3.43	0.10		3.68 c	0.09
Emotional Wellbeing	5.12 b	0.05		5.17 d	0.05	4.91b,d	0.05		5.02	0.06
Social Wellbeing	3.57	0.09		3.66	0.07	3.56	0.08		3.74	0.07
Psychological Wellbeing	5.14 b	0.06		5.20 d	0.05	4.91b,d	0.06		5.05	0.06
Psychological Distress	4.47	0.23		5.02	0.24	4.86	0.23		4.86	0.22
Other Expectations and Attitudes
Will Marry	0.74			0.75		0.65			0.69
Marry Within 5 Years	0.76 b,c			0.79 d,e		0.49 b,d,f			0.64 c,e,f
Will Have (More) Children	0.79 c			0.75		0.68			0.66c
Divorce Hurts Kids	5.98	0.11		5.73	0.10	5.66	0.11		5.71	0.10
Single Parenting Hurt Kids	4.37 ab,c	0.12		3.31 a,d	0.10	3.75 b,d	0.12		3.32 c	0.13
Family Form Limits Success	2.04 a	0.09		1.74 a	0.06	1.87	0.07		1.74	0.07
Religion is Very Important	0.27 c			0.32		0.26 f			0.41 c,f
Childhood Experiences
Mother Married at Birth	0.77			0.76		0.76			0.73
Caregiver Divorce	0.23			0.27		0.23			0.24
Maternal Education	13.37	0.25		13.11	0.22	13.40	0.23		12.96	0.39
Covariates
Age	23.84	0.18		24.39 d,e	0.18	23.51 d	0.17		23.35 e	0.20
Race (Reference: White)
Black	0.65			0.70		0.66			0.62
Other Race	0.19			0.17		0.16			0.18
Hispanic	0.16			0.13		0.18			0.20
Controls
Caregiver is Parent	0.95			0.97		0.95			0.97
Caregiver changed	0.12			0.06		0.10			0.08
Year reported sexual partners (reference: 2011)
2007	0.44 b,c			0.47 d,e		0.29 b,d			0.27 c,e
2009	0.22 a			0.30 d,e		0.22 d			0.20 e
N	362			424		432		392

Note: Females more likely to be partnered, p < 0.01. Letter superscripts denote differences at p < 0.01 between:

^a)partnered males and partnered females

^b)partnered males and single males

^c)partnered males and single females

^d)partnered females and single males

^e)partnered females and single females

^f)single males and single females.

Table 3:

Results of Multinomial Logistic Regression Predicting Expectations of Some and 50% and Greater Chance of Divorce of Full Model on Full Sample, N = 1,610

	Some Chance			50% or Greater Chance
Variable	Coefficient	SE	RRR	Coefficient	SE	RRR
Current Relationship and Family
Partnered	−0.12	0.15	0.89	−0.49	0.20	0.61
Has Children	−0.24	0.25	0.79	−0.04	0.30	0.97
Number of Sexual Partners	0.02	0.02	1.02	0.06 *	0.02	1.06
Socioeconomic Circumstances
Student	−0.01	0.21	0.99	−0.04	0.31	0.96
Education (Reference: High School or GED)
Less than High School	0.26	0.42	1.30	0.44	0.46	1.56
Some College	0.48	0.22	1.61	−0.19	0.28	0.83
College	0.34	0.31	1.41	−0.28	0.41	0.76
Employment (Reference: Full Time)
Unemployed	−0.09	0.34	0.91	0.39	0.40	1.48
Part Time	0.27	0.19	1.31	0.42	0.27	1.53
Income (Thousands)	0.01	0.01	1.01	0.01	0.01	1.01
Financial Responsibility Scale	−0.14	0.08	0.87	0.06	0.12	1.06
Worry about Future Scale	−0.07	0.06	0.93	0.13	0.08	1.14
Mental and Emotional Wellbeing
Social Anxiety	0.16	0.06	1.17	0.09	0.08	1.10
Emotional Wellbeing	−0.09	0.13	0.91	−0.28	0.16	0.76
Social Wellbeing	0.03	0.08	1.03	0.15	0.12	1.16
Psychological Wellbeing	−0.31 *	0.11	0.73	−0.10	0.15	0.91
Psychological Distress	0.00	0.03	1.00	0.04	0.04	1.04
Other Expectations and Attitudes
Greater than 50% of Marriage	−0.60 *	0.20	0.55	−0.81 *	0.27	0.45
Will Marry within 5 Years	−0.32	0.19	0.73	−0.11	0.25	0.89
Will Have (More) Children	−0.08	0.18	0.92	−0.15	0.29	0.86
Divorce Hurts Kids	−0.08	0.06	0.92	−0.05	0.07	0.95
Single Parenting Hurts Kids	0.06	0.05	1.07	0.04	0.06	1.05
Family Form Limits Success	−0.06	0.07	0.94	−0.05	0.11	0.95
Religion is Very Important	−0.45 *	0.17	0.64	−0.43	0.24	0.65
Childhood Experiences
Mom Married at Birth	0.16	0.21	1.17	−0.51	0.28	0.60
Caregiver Divorce	0.28	0.17	1.32	0.59 *	0.22	1.81
Maternal Education	−0.01	0.04	0.99	−0.02	0.05	0.98
Covariates
Male	0.14	0.17	1.15	0.59	0.24	1.80
Race (Reference: White)
Black	−0.65 *	0.20	0.52	−0.54	0.31	0.58
Other Race	−0.05	0.25	0.95	−0.39	0.39	0.68
Hispanic	−0.12	0.31	0.88	0.11	0.37	1.12
Age	−0.02	0.05	0.98	0.00	0.06	1.00
Controls
Caregiver is Parent	−0.28	0.44	0.75	−0.13	0.47	0.87
Caregiver changed	−0.25	0.29	0.78	0.39	0.31	1.48
Year reported sexual partners (reference: 2011)
2007	−0.26	0.22	0.77	−0.15	0.31	0.86
2009	−0.72 *	0.26	0.49	−0.42	0.33	0.65

Note:

^*p < 0.01.

Table 4:

Results of Multinomial Logistic Regression Model, by Groups, Predicting Some Chance of Divorce.

	Partnered Males			Partnered Females			Single Males			Single Females
Variable	b	SE	RRR	b	SE	RRR	b	SE	RRR	b	SE	RRR
Current Relationship and Family
Cohabiting	−0.49	0.41	0.61	0.10	0.34	1.10
Satisfied with Relationship	−0.40	0.20	0.67	−0.05	0.18	0.95
Has Children	0.28	0.50	1.33	−0.15	0.47	0.86	1.17	0.80	3.21	−0.41	0.51	0.66
Number of Sexual Partner	0.02	0.02	1.02	0.02	0.05	1.02	0.14	0.05	1.16 *	0.05	0.05	1.05
Socioeconomic Circumstances
Student	0.30	0.49	1.35	−0.66	0.40	0.52	−0.67	0.43	0.51	0.48	0.43	1.62
Education (Reference: High School or GED)
Less than High School	0.40	0.82	1.49	0.04	0.82	1.04	0.64	0.91	1.89	1.23	0.90	3.41
Some College	−0.18	0.51	0.84	1.22	0.54	3.39	0.50	0.49	1.65	0.74	0.52	2.10
College	0.89	0.67	2.43	0.43	0.75	1.54	0.16	0.67	1.17	1.00	0.64	2.71
Employment (Reference: Full Time)
Unemployed	−1.01	0.97	0.37	−0.57	0.55	0.57	1.94	0.76	6.93	−0.08	0.78	0.93
Part Time	0.12	0.46	1.13	−0.22	0.38	0.80	0.91	0.45	2.49	0.64	0.40	1.89
Income (Thousands)	−0.01	0.02	0.99	0.01	0.02	1.01	0.02	0.02	1.02	0.04	0.03	1.04
Financial Responsibility	−0.67	0.25	0.51 *	0.01	0.17	1.01	0.32	0.18	1.37	−0.50	0.20	0.61
Worry about Future	−0.02	0.14	0.98	−0.08	0.11	0.93	−0.07	0.13	0.93	0.03	0.12	1.03
Mental and Emotional Wellbeing
Social Anxiety	0.22	0.15	1.25	0.16	0.12	1.18	0.13	0.12	1.14	0.04	0.15	1.04
Emotional Wellbeing	−0.02	0.26	0.98	−0.33	0.29	0.72	−0.06	0.23	0.94	0.03	0.27	1.03
Social Wellbeing	0.07	0.18	1.08	0.32	0.16	1.38	−0.01	0.18	0.99	−0.10	0.19	0.90
Psychological Wellbeing	−0.63	0.25	0.53	−0.64	0.24	0.53 *	−0.56	0.23	0.57	0.17	0.28	1.19
Psychological Distress	0.05	0.06	1.06	−0.06	0.05	0.94	−0.04	0.07	0.96	0.06	0.07	1.07
Other Expectations and Attitudes
Greater than 50% of Marriage	−1.29	0.49	0.27 *	−1.40	0.42	0.25 *	0.02	0.41	1.02	−0.23	0.39	0.79
Will Marry within 5 Years	−0.43	0.49	0.65	−0.33	0.46	0.72	0.63	0.35	1.87	−0.91	0.37	0.40
Will Have (More) Children	−0.51	0.51	0.60	−0.52	0.41	0.60	−0.66	0.42	0.51	−0.03	0.36	0.97
Divorce Hurts Kids	−0.19	0.13	0.83	−0.24	0.12	0.78	0.02	0.12	1.02	−0.08	0.12	0.92
Single Parenting Hurts Kids	0.00	0.09	1.00	0.09	0.09	1.09	0.10	0.10	1.10	0.11	0.10	1.12
Family Form Limits Success	−0.32	0.15	0.72 c	0.03	0.14	1.03	−0.17	0.14	0.84	0.36	0.16	1.44 c
Religion is Very Important	0.26	0.44	1.30	−0.24	0.33	0.78	−1.12	0.36	0.33 *	−0.83	0.34	0.44
Childhood Experiences
Mom Married at Birth	0.78	0.53	2.18	0.45	0.39	1.56	0.45	0.44	1.56	0.33	0.40	1.39
Caregiver Divorce	0.06	0.45	1.06	0.46	0.35	1.59	−0.03	0.39	0.97	0.68	0.40	1.98
Maternal Education	−0.09	0.08	0.92	−0.07	0.07	0.93	0.07	0.06	1.08	0.00	0.07	1.00
Covariates
Race (Reference: White)  Black	−0.40	0.50	0.67	−0.41	0.50	0.66	−1.78	0.50	0.17 **	−0.47	0.44	0.63
Other Race	−0.34	0.52	0.71	0.27	0.51	1.31	0.10	0.53	1.11	−0.16	0.49	0.85
Hispanic	−0.52	0.56	0.60	0.15	0.48	1.16	−0.34	0.56	0.71	−0.33	0.63	0.72
Age	−0.09	0.12	0.92	−0.16	0.10	0.85 e	−0.23	0.10	0.79	0.20	0.09	1.23 e
Controls
Caregiver is Parent	2.65	1.09	14.11 a	−1.85	1.14	0.16 a	−1.74	0.94	0.17	−0.25	0.68	0.78
Caregiver change	−0.63	0.56	0.53	1.29	0.67	3.63	−0.32	0.55	0.73	−1.07	0.63	0.34
Year reported sexual partners (reference: 2011)
2007	−0.12	0.46	0.89	−0.38	0.41	0.68	−0.86	0.50	0.42	−0.58	0.49	0.56
2009	−0.73	0.57	0.48	−0.38	0.52	0.68	−1.64	0.55	0.19 *	−0.67	0.51	0.51
n		362			424			432			392

Note:

^*p < 0.01;

^**p < 0.001.

Letters denote differences at p < 0.01 between:

^a)partnered males and partnered females

^b)partnered males and single males

^c)partnered males and single females

^d)partnered females and single males

^e)partnered females and single females

^f)single males and single females

Table 5:

Results of Multinomial Logistic Regression Model, by Groups, Predicting Greater Chance of Divorce.

	Partnered Males			Partnered Females			Single Males			Single Females
Variable	b	SE	RRR	b	SE	RRR	b	SE	RRR	b	SE		RRR
Current Relationship and Family
Cohabiting	0.58	0.58	1.79	−0.98	0.50	0.38
Satisfied with Relationship	−1.45	0.29	0.23 ** a	−0.03	0.29	0.97 a
Has Children	−1.21	0.94	0.30	0.84	0.70	2.32	1.60	0.94	4.96	−0.17	0.78		0.84
Number of Sexual Partner	0.04	0.01	1.04 *	0.04	0.06	1.04	0.15	0.05	1.16 *	0.06	0.06		1.07
Socioeconomic Circumstances
Student	−0.70	0.94	0.50	0.12	0.60	1.13	−0.07	0.65	0.93	0.70	0.64		2.01
Education (Reference: High School or GED)
Less than High School	3.02	1.09	20.43 * b	0.44	1.22	1.55	−0.81	0.95	0.45 b	0.78	1.03		2.18
Some College	−0.74	0.76	0.47	−0.27	0.76	0.77	−0.74	0.57	0.48	0.27	0.71		1.31
College	−1.15	1.33	0.32	−0.57	0.97	0.57	−0.76	0.84	0.47	0.86	0.94		2.37
Employment (Reference: Full Time)
Unemployed	−2.25	1.42	0.11 b	−2.55	1.40	0.08 d	2.30	0.82	9.99 * b,d	1.50	0.98		4.48
Part Time	−0.43	0.71	0.65	−0.02	0.56	0.98	0.94	0.52	2.55	0.23	0.58		1.26
Income (Thousands)	−0.06	0.03	0.95 b,c	−0.02	0.03	0.98	0.05	0.02	1.05 b	0.08	0.04		1.08 c
Financial Responsibility	0.69	0.44	1.98 c	0.27	0.28	1.30 e	0.25	0.22	1.28 f	−0.78	0.29		0.46 * c,e,f
Worry about Future	0.79	0.20	2.20 ** b	0.26	0.15	1.30	−0.10	0.16	0.91 b	0.25	0.21		1.29
Mental and Emotional Wellbeing
Social Anxiety	−0.31	0.22	0.74	0.31	0.20	1.37	0.19	0.14	1.21	−0.35	0.22		0.71
Emotional Wellbeing	0.65	0.41	1.92 c	0.02	0.38	1.02	−0.38	0.26	0.68	−1.26	0.39		0.28 * c
Social Wellbeing	0.34	0.29	1.41	0.65	0.25	1.91 *	−0.19	0.27	0.83	0.15	0.30		1.17
Psychological Wellbeing	−1.07	0.35	0.34 * b,c	−0.83	0.41	0.44 e	0.25	0.30	1.29 b,e	0.87	0.39		2.39 c
Psychological Distress	0.01	0.08	1.01	0.05	0.09	1.05	0.13	0.07	1.13	0.12	0.10		1.12
Other Expectations and Attitudes
Greater than 50% of Marriage	−0.48	0.75	0.62	−1.69	0.56	0.18 *	−0.28	0.56	0.75	−0.91	0.59		0.40
Will Marry within 5 Years	−0.28	0.82	0.76	−0.11	0.63	0.89	0.61	0.45	1.84 f	−1.56	0.51		0.21 * f
Will Have (More) Children	−1.23	0.74	0.29	0.51	0.58	1.67	−1.11	0.63	0.33	0.18	0.63		1.20
Divorce Hurts Kids	−0.14	0.16	0.87	−0.10	0.17	0.90	0.16	0.14	1.17	0.05	0.15		1.05
Single Parenting Hurts Kids	0.16	0.16	1.17	0.14	0.13	1.15	0.05	0.12	1.05	−0.40	0.18		0.67
Family Form Limits Success	−0.52	0.20	0.60	−0.25	0.32	0.78	−0.27	0.20	0.76	0.31	0.27		1.37
Religion is Very Important	0.33	0.70	1.38	−0.54	0.54	0.58	−0.72	0.49	0.49	−1.31	0.59		0.27
Childhood Experience
Mom Married at Birth	−0.63	0.77	0.53	−0.84	0.67	0.43	0.87	0.52	2.39	−0.94	0.62		0.39
Caregiver Divorce	0.04	0.66	1.04	0.46	0.44	1.58	0.88	0.49	2.40	0.24	0.57		1.27
Maternal Education	0.12	0.14	1.12	0.04	0.11	1.04	−0.12	0.08	0.89	−0.04	0.10		0.96
Covariates
Race (Reference: White)
Black	−0.41	0.69	0.66	0.10	0.68	1.10	−0.31	0.52	0.73	−1.30	0.83		0.27
Other Race	−4.77	1.24	0.01 ** a,b,c	0.44	0.92	1.55 a	0.51	0.71	1.66 b	0.29	0.79		1.34 c
Hispanic	1.71	0.75	5.52	−0.69	0.82	0.50	−0.91	0.73	0.40	−0.87	0.92		0.42
Age	−0.12	0.15	0.88	−0.14	0.15	0.87	−0.01	0.12	0.99	0.13	0.16		1.14
Controls
Caregiver is Parent	2.62	1.02	13.76 b	−1.22	2.41	0.30	−1.74	0.77	0.18 b	2.56	2.71		12.91
Caregiver changed	2.75	0.81	15.63 *	0.08	1.19	1.08	1.38	0.66	3.99	−0.26	1.02		0.77
Year reported sexual partners (reference: 2011)
2007	1.87	0.80	6.51 a	−1.02	0.66	0.36 a	−0.67	0.66	0.51	−0.89	0.99		0.41
2009	2.47	0.99	11.85 a,b,c	−1.67	0.87	0.19 a	−0.82	0.65	0.44 b	−2.53	0.93		0.08 * c
n			362		424			432				392

Note:

^*p < 0.01

^**p < 0.001.

Letters denote differences at p < 0.01 between:

^a)partnered males and partnered females

^b)partnered males and single males

^c)partnered males and single females

^d)partnered females and single males

^e)partnered females and single females

^f)single males and single females

Predictors of “Some Chance” of Divorce

In the full sample, no characteristic of the current relationship or family experiences was associated with “some chance” of divorce versus “no chance.” The only significant association within subgroups was that, for single males, each additional sex partner was associated with 16% greater relative risk of reporting “some chance” of divorce versus “no chance.” There were also no significant associations between socioeconomic circumstances and “some chance” in the full model, with the only significant association in the subgroups being that, for partnered males, one-point greater financial responsibility was associated with 49% lower relative risk of reporting “some chance” of divorce.

Only one measure of mental and emotional wellbeing was significantly associated with “some chance” in both the full sample and in subgroups: psychological wellbeing. A one-point increase in psychological wellbeing was associated with 27% lower risk of reporting “some chance” of divorce in the full sample. When broken out to subgroups, it was only significant for partnered females, and a one-point change was associated with 47% lower risk.

There was a significant association between certainty of marriage and decreased risk of reporting “some chance” of divorce for the full sample (with 40% lower risk) as well as both partnered subsamples (partnered males who expected to marry had 73% lower risk, partnered females 75%). No attitude toward the effect of divorce or single parenting was significantly associated with reporting “some chance” of divorce. In the full sample, those who reported religion to be “very important” had 36% lower risk of reporting “some chance” of divorce, though when examined separately this association was only significant in single males, in that saying religion was very important was associated with a 67% lower risk of reporting “some chance” of divorce.

There were no significant associations between the childhood experiences in question and “some chance” of divorce in the full sample or the subsamples. In the full sample, Black youth had 48% lower risk of reporting “some chance” of divorce, and in the subsample models, only Black single males retained this significant association and has 83% lower risk of reporting “some chance” of divorce compared to “no chance.”

Predictors of “50% or Greater Chance” of Divorce

In the full sample, reporting one additional sexual partners was associated with a 6% greater relative risk of reporting “greater chance” of divorce compared to “no chance.” In the subsample analyses, this association was significant for both partnered (4% greater risk) and single males (16% greater risk). Additionally, partnered males who were more satisfied with their relationships had 77% lower risk of rating themselves likely to divorce, and this association was significantly different than the non-significant, negative association for partnered females.

As with “some chance” of divorce, no socioeconomic circumstance variables were associated with a higher risk of divorce in the full sample, but there was interesting variation in the subsamples. Partnered males with less than a high school education had more than 20 times the risk of reporting themselves as having a greater chance of divorce than those with a high school education, and this association was significantly different from the negative, and non-significant, association observed for single males. On the other hand, unemployed single males had nearly 10 times the risk of rating themselves as having higher chances of divorce than those employed full-time, and this association was significantly different from the non-significant and negative association in both partnered subsamples. Partnered males who expressed greater worry about the future has more than twice the risk of rating their divorce chances higher than partnered men with lower worry. Single women who reported one point greater financial responsibility had 54% lower risk of reporting a high chance of divorce, and this value was statically different than the associations for both male groups.

Turning to mental and emotional health, significant associations were only found in the subgroup analyses. For single women, one point greater emotional wellbeing was associated with 72% lower risk of reporting themselves to have higher chances of divorce, which was significantly different from partnered males’ positive, nonsignificant association. One point greater psychological wellbeing was associated with 66% lower risk of reporting higher chance of divorce for partnered males; this was statistically different from the positive, non-significant associations for both single groups. Perhaps surprisingly, partnered females who reported higher social wellbeing had nearly twice the risk of reporting higher chances of divorce than their less socially-adjusted counterparts.

Again, those more certain of marriage in the full sample had a lower risk, 55% lower, of reporting greater chances of divorce, though this time the association appeared to be driven by partnered females only, as partnered women with the greatest expectations of marriage had 82% lower risk of reporting high chances of divorce. Single females who believed they would marry soon were 79% less likely to report “greater chance” of divorce than those who expected marriage more than five years out, and this was statistically different from the positive, nonsignificant association for single males.

The only childhood experience significantly associated with reporting “greater chance” of divorce in the full sample was caregiver divorce. Those whose caregivers had divorced had 81% greater risk of reporting greater chances of divorce. This was not significant in any subsample, though in each sample the association was positive. In the subsamples, partnered males of “Other” racial background had 99% lower risk of reporting “greater chance” of divorce than White young men, and this association was also significantly different from that of every other subgroup.

Discussion

Concerns about potential divorce may lead young adults to delay marriage (Edin, 2000; Miller et al., 2011; Perelli‐Harris et al., 2017; Waller & Peters, 2008), but few studies have examined personal expectations of divorce. Those that have examined such expectations have focused on parental divorce nearly exclusively (Boyer-Pennington et al., 2001; Kirk, 2003) or have studied samples of college students (Black & Sprenkle, 1991; Boyer-Pennington, Pennington, & Sprink, 2001; Helweg-Larsen et al., 2011; Kirk, 2003), or newlyweds (Baker & Emery, 1993; Campbell et al., 2012). In this study, we examined a wide range of predictors in a sample of emerging adults both in college and not and how these associations varied by gender and relationship status. As emerging adulthood is an important developmental time for forming attitudes and identities towards relationships (Arnett, 2000), studying expectations to divorce in emerging adults is important for understanding the context of later relationship behaviors and eventual outcomes.

A large portion of these young adults, between 31% of single males and 43% of both single and partnered women, believed there was “no chance” they would divorce, and nearly half of each group felt there was only “some chance.” Perhaps it is not just newlyweds who rate their chances of success to be high (Baker & Emery, 1993; Campbell et al., 2012). Despite some reporting “fears” of divorce (Willoughby & James, 2017), many young people seem to have an optimistic view of their future relationships. Not all painted such rosy pictures, however, and some characteristics stood out as being significantly associated with divorce expectations.

For instance, economic factors seemed to be particularly important for males’ expectations, with low education standing out for partnered men and unemployment for single men as especially strong predictors of high expectations to divorce. This may reflect overall marriage culture, as some women speak of needing male partners to have stable jobs (Edin, 2000; Smock, Manning, & Porter, 2005), and men who already struggle to find full-time work may worry about living up to such demands. Additionally, as education is increasingly becoming a gradient by which spouses are sorted and divorce rates are stratified, low-educated men may have already felt the decreased economic opportunities afforded by their level of education (Cherlin, 2010). Supporting this notion, the more partnered males worried about their future or the less financial responsibility they took for themselves, the greater they rated their chances of divorce. Altogether, the strength of socioeconomic characteristics in predicting divorce expectations, especially for men, was reminiscent of the way that economic struggles actually do predict marital stress and eventually divorce (Cherlin, 2010; Neff & Karney, 2016), or even preclude marriage for some (McLanahan, 2004; Smock et al., 2005), as well as how strongly and stably low socioeconomic status predicts other less positive expectations of the future, including education and occupation (Mello, 2009). Those most likely to divorce because of these issues may already be aware of their risk, though not necessarily feel able to change their situations.

Experiences with relationships, both their own and their parents’, and expectations for marriage were also significant predictors of expectations for divorce. Partnered women were the least likely to report any chance of divorce when most certain of marriage, and single women who expected to marry within five years reported lower expectations to divorce. Those most oriented towards marriage may be the least accepting of divorce (Halpern-Meekin, 2012) and thus less likely to entertain the idea that they might experience it one day. For partnered men, certainty of marriage and satisfaction with the current relationship predicted lower expectations of divorce. For both single and partnered men, however, a greater number of sexual partners predicted higher expectations of divorce, perhaps suggesting a lower marriage orientation or that having had more partners was indicative of less confidence in maintaining a long-term relationship. Consistent with prior research, caregiver divorce was significantly associated with increased expectations to divorce in the full sample (Boyer-Pennington et al., 2001; Kirk, 2003), but the association was not significant for any subsample. Perhaps the children of divorce were resilient in their functioning following their parents’ relationship dissolution, and at this stage in their lives other factors were more important in predicting their personal expectations to divorce (Kelly & Emery, 2003).

In the end, we found significant predictors in every category, though they varied over groups and in strength. Some hypothesized predictors were not associated with expectations for any group, such as having or expecting to have children, psychological distress, maternal marital status at respondent’s birth, and maternal education. The development of expectations of divorce is obviously a multi-faceted process. Although our list of potential predictors was not exhaustive, we believe it does provide insights into who is most and least likely to report these expectations, and it highlights the especially important role of both socioeconomic achievement and relationship experiences and expectations in the attitudes that young adults hold regarding their chances of future marital success.

Limitations and Future Directions

Although our study was able to examine multiple predictors of divorce expectations, future studies should seek to improve measurement in this area of research. For example, the outcome was measured by a single item, which is the norm for field at this point (Arocho & Kamp Dush, 2017; Plotnick, 2007; Willoughby & Dworkin, 2009). Additionally, we were unable to establish temporal ordering for many of the predictors as they were measured the same year as expectations of divorce, although some variables were drawn from the Child Development Study and earlier TAS waves. Lastly, our sample was diverse in education and economic experiences, allowing us to capture emerging adults often missed by samples of college students (Arnett, 2000), but was not nationally-representative and results may not be entirely generalizable.

It should be noted that the outcomes of expectations to divorce are relatively unknown. Expectations to marry are predictive of marriage (Willoughby, 2014), thus we might believe that expectations to divorce are associated with an increased risk of divorce. Previous research on general attitudes of divorce would support this assertion (Amato & Rogers, 1999). However, expecting to divorce might delay or preclude marriage altogether (Edin, 2000; Miller et al., 2011; Waller & Peters, 2008), and other behavior and attitudinal outcomes could be associated with increased pessimism towards marital success (Perelli‐Harris et al., 2017). As this sample ages, it will be important to determine outcomes associated with the expectations reported here.

This exploratory study provided a preliminary picture of the characteristics and experiences associated with expectations that young adults hold for their future chances of divorce. Some would argue that a fulfilling marriage is becoming harder to achieve (Finkel, Hui, Carswell, & Larson, 2014), and marriage entry and success are increasingly segregated by socioeconomic lines (Cherlin, 2010). Our results highlight the role that socioeconomic circumstances and relationship experiences likely play in young adults’ cognitions about marriage success even before they wed (Willoughby et al., 2015), and these beliefs and attitudes may have consequences for how they experience family life throughout their development. Finally, understanding correlates of attitudes and expectations is important to help social scientists understand the current trends in both beliefs and behaviors in the United States today.