Original Article The Mental Health Implications of Emerging Adult Long-Term Cohabitation Emerging Adulthood 1-15 ª 2017 Society for the Study of Emerging Adulthood and SAGE Publishing Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/2167696817733913 journals.sagepub.com/home/eax Sara E. Mernitz1 Abstract Despite the growing prevalence of cohabitation, past attempts to identify mental health outcomes from cohabitation do not differentiate by cohabitation duration. The current study investigated the mental health implications from long-term cohabitation, defined as those lasting more than 3 years. Using the National Longitudinal Survey of Youth 1997, I compared the average individual mental health scores between time spent single, or time spent in a short-term cohabitation, and time spent in a longterm union. Results indicated that externalizing distress, defined as heavy episodic drinking, was lower during time spent in a longterm cohabitation than it was during time spent single. Unexpectedly, the average emotional distress rates were greater during time spent in a long-term cohabitation than they were during time spent single; men appeared to be driving that effect. Overall, long-term cohabitation did not provide an additional mental health benefit above and beyond short-term cohabitation. Keywords cohabitation, long-term cohabitation, mental health, NLSY97, quantitative methods Contemporary emerging adults are increasingly more likely to cohabit with a romantic partner prior to marriage (Cherlin, 2010; Manning, 2013), with few entering into marriage directly from a dating relationship (Furstenberg, 2011). Approximately 60% of emerging adults cohabit in the United States (Manning, 2013); these rates are higher than those found in the United Kingdom, Germany, Austria, and Belgium where between 30% and 50% of the adult population has ever cohabited (Kroeger & Smock, 2014). Given the high prevalence of emerging adult cohabitation, combining all cohabitors into a single category may mask important health differences within the cohabiting population. Cohabitation likely varies by duration, yet long-term cohabiting unions, and their associations with mental health, defined in the current study as both internalizing and externalizing distress, have not been explicitly examined. Past attempts to examine change in mental health from cohabiting unions have relied on comparing (1) current cohabitors and married individuals (i.e., Kim & McKenry, 2002; Fleming, White, & Catalano, 2010), (2) current cohabitors and dating or single individuals (i.e., Lamb, Lee, & DeMaris, 2003; Uecker, 2012), or (3) individuals’ mental health as they transition into and/or out of cohabitation (i.e., Mernitz & Dush, 2016; Musick & Bumpass, 2012). Although cohabitation is typically characterized by instability and short durations (Brown, 2000; Lichter, Qian, & Mellott, 2006), cohabitation has lengthened over time (Kennedy & Bumpass, 2008). These long-term cohabiting unions may have different mental health implications than short-term cohabitations for emerging adults. Long-term cohabitations may resemble marriage and provide similar reductions in distress to marriage. Cohabitation in the past has been characterized by lower commitment (Stanley, Rhoades, & Fincham, 2011) and lower relationship quality (Marcussen, 2005) than marriage. Yet, long-term cohabitations may provide similar benefits to marriage, such as greater emotional and social support, contributing to less emotional distress (Umberson, Thomeer, & Williams, 2013). Using the National Longitudinal Survey of Youth 1997 (NLSY97), I provide a first examination into the mental health implications of long-term cohabitation, defined in the current study as unions lasting longer than 3 years. This definition of long-term cohabitation is consistent with existing research on long-term cohabitation and union quality, where long-term cohabitation is defined as lasting longer than 3 (Willetts, 2006) or 4 years (Skinner, Bahr, Crane, & Call, 2002). I use modeling that carefully accounts for selection into long-term cohabitation to examine within-person change in mental health from time spent in a long-term cohabitation compared to time spent single, defined as not being in a union. Additionally, I examine change in mental health above 1 Department of Human Sciences, Institute for Population Research, The Ohio State University, Columbus, OH, USA Corresponding Author: Sara E. Mernitz, PhD, Population Research Center, University of Texas at Austin, 1885 Neil Avenue Mall, Columbus, OH 43210, USA. Email: email@example.com 2 and beyond any health benefits accrued in a short-term cohabitation, defined as those lasting 3 years or less. I also examine these associations for gender differences. The Importance of Long-Term Cohabitation for Emerging Adults The prevalence of emerging adult long-term cohabitation remains unknown, yet long-term cohabitation may be an alternative to marriage for many emerging adults. According to the theory of emerging adulthood and identity theory, a key developmental task during the transition to adulthood is establishing a romantic identity and forming intimacy with a romantic partner (Arnett 2000; Erikson, 1968). Premarital relationship experiences, including short-term cohabitation, offer ways to establish this identity through explorations in love and sex (Arnett, 2000), and these early experiences contribute to emerging adults’ ability to establish long-term relationships (Shulman & Connolly, 2013, 2015). Transitioning from these short- to long-term relationships can be difficult in emerging adulthood; emerging adults must learn to navigate their needs with those of their partner, establishing a balance between personal independence and closeness (Shulman & Connolly, 2013, 2015). This process does not occur solely in dating relationships; approximately 60% of emerging adults move into cohabitation with a partner in order to test the relationship for long-term potential (Stanley et al., 2011), suggesting that these abilities are developed in (and out of) many different relationships, including cohabitation. However, once emerging adults gain these abilities, they are able to establish emotionally intimate long-term relationships. Until recently, marriage was the primary way to meet this key developmental task of establishing intimacy with a romantic partner during the transition to adulthood (Arnett, 2000). However, emerging adults are getting married later (Cherlin, 2010), with the median age at first marriage occurring at age 29.5 for men and 27.4 for women (U.S. Census Bureau, 2016). Despite 76% of teens report wanting to marry in the future (Manning, Longmore, & Giordano, 2007), emerging adults often highlight many barriers to marriage like financial insecurity or unrealistic standards or expectations for marriage (e.g., Finkel, Hui, Carswell, & Larson, 2015; Smock, Manning, & Porter, 2005), suggesting that they may remain in long-term cohabitations. However, emerging adults might still be preparing for marriage in these long-term cohabitations. When emerging adults expected to marry earlier, or had an earlier “marital horizon,” they altered their behavior in preparation for marriage (Carroll et al., 2007; Willoughby & Carroll, 2015). Because developmental milestones met off-time, such as unions occurring much later in the life course, are theorized to be associated with health problems and risky behavior (Erikson, 1968), emerging adults’ expectations that they will soon marry may reduce distress. Trends in emerging adult union formation, such as the delay in first marriage and the growing prevalence of cohabitation (Cherlin, 2010), may be shaped by external barriers to Emerging Adulthood marriage—such as the disparity in access to marriage or difficulty reconciling competing developmental tasks. Today’s emerging adults face greater difficulty finding employment, even if they have a college degree (Settersten & Ray, 2010), and these economic barriers may make marriage but not cohabitation, less attainable for many emerging adults. Qualitative work finds evidence of “long-term engagements” among the economically disadvantaged, whereby cohabitors are engaged but have no formal plans to marry. Many highlight economic or social barriers, such as lack of desired employment or inability to afford a wedding, as the reasons they do not formally marry (Edin, 2000). Other scholars have suggested that competing needs for emerging adults, specifically employment and marriage/family needs, delay marriage (Shulman & Connolly, 2013, 2015; Willoughby & Carroll, 2015). These barriers to marriage contribute to a new transitional emerging adult romantic relationship stage whereby emerging adults reconcile their employment and marriage/family needs in the context of economic and family formation uncertainty (Shulman & Connolly, 2013, 2015). Even though emerging adults are capable of establishing highly intimate relationships, the difficulty reconciling their desire for a long-term relationship with their own, and their partner’s, life plans makes marriage in emerging adulthood unattainable for many (Shulman & Connolly, 2013, 2015). The postponement of marriage is a conscious response to these barriers. For these emerging adults, long-term cohabitation may become a viable way to reconcile these conflicting developmental tasks, especially for those facing greater social and economic barriers to marriage, leading to decreased distress. Cohabitation and Mental Health The mental health implications of entering into a cohabiting union are unclear. In the past, studies relying on data from the National Survey of Families and Households found that individuals in a cohabiting union reported greater depressive symptoms than those that married (Brown, 2000 [1987–1988 wave only]; Kim & McKenry, 2002 [1987–1988 and 1992–1993 waves]), but these associations have not been found among contemporary emerging adults. Using data from emerging adults in the National Longitudinal Study of Adolescent to Adult Health, cohabitors reported similar levels of depressive symptoms, but a higher frequency of drunkenness, than marrieds (Uecker, 2012). In the National Longitudinal Survey of Youth 1979 cohort, Duncan, Wilkerson, and England (2006) found that cohabiting emerging adults had less pronounced reductions in binge drinking than married emerging adults. When compared to single or unpartnered individuals, cohabitation in the past was not associated with emotional health, measured by depressive symptoms (Kim & McKenry, 2002; Lamb et al., 2003); more recent evidence from a community sample suggests that cohabitation was associated with decreased heavy drinking compared to being single (Fleming et al., 2010). Taken together, these findings provide mixed evidence for any observed emotional health benefits from cohabitation and Mernitz evidence of slight reductions in problem drinking from cohabitation compared to being single. Yet, much of this past research relied on comparing cohabiting individuals with either single or married emerging adults. Because cohabitation is overrepresented among the socioeconomically disadvantaged (Lichter & Qian, 2008) and those with a history of poor internalizing and externalizing distress (Sandberg-Thoma & Kamp Dush, 2014), failure to account for preexisting differences may bias results. When accounting for preexisting differences, transitioning from being single or unpartnered into cohabitation was associated with increased global happiness for participants in the National Survey of Families and Household 1987–1988 and 1992–1993 waves (Musick & Bumpass, 2012). Among emerging adults in the NLSY97 cohort, transitioning from being single into cohabitation was associated with less internalized distress, and transitioning from cohabitation into marriage was not associated with any additional improvement (Mernitz & Dush, 2016). Emerging adults who entered into cohabitation from being single reduced their heavy drinking, although these reductions were less pronounced than they were for those who entered directly into marriage (Fleming et al., 2010). Overall, after accounting for preexisting characteristics, these findings suggest that contemporary emerging adult cohabitation provides mental health benefits that may be comparable to marriage benefits. Scholars have used the marital resource model to explain observed mental health benefits and lack of distress from marriage (Umberson, Thomeer, & Williams, 2013). This model suggests that marriage provides many additional benefits that improve mental health from greater socioeconomic resources (Marcussen, 2005) to greater support networks (Frech & Williams, 2007). Long-term cohabitation might mirror marriage in providing many of these benefits, contributing to less distress for cohabitors. Unlike short-term cohabitors, long-term emerging adult cohabitors may have received these benefits for several years. Emerging adults report moving in together to reduce the cost of living (Huang, Smock, Manning, & Bergstrom-Lynch, 2011; Sassler, 2004) and may pool other economic resources. Further, long-term cohabitors may have access to a broader network of family and/or friends and may become more enmeshed with this network over time. Because greater support networks have been linked to better subjective well-being, especially when these support networks consisted of highquality social relationships (Saphire-Bernstein & Taylor, 2013), long-term cohabitors might experience less distress the longer they are in their union. Long-term cohabitors might also experience less distress because their romantic partner is able to provide support for a longer duration. Emerging adults in high-quality relationships, where support provision is likely to be high, had higher levels of happiness, regardless of whether they were experiencing conflict in a close friendship (Demir, 2010). These findings suggest that romantic partners become more critical for support provision in emerging adulthood than they were in adolescence. Given the associations 3 between support provision and romantic relationships, I expect that time spent in a long-term cohabitation compared to time spent single, and remaining in a cohabitation long-term, will be associated with reduced distress. Gender Differences in Associations Between Cohabitation and Mental Health Although gender differences in associations between cohabitation and mental health have received some attention, findings are mixed. Indeed, no significant gender differences were observed in associations between emotional health and cohabitation (Blekesaune, 2008; Uecker, 2012). However, using a contemporary sample of emerging adults, women received emotional health benefits from transitioning into a first cohabitation from being single, whereas men did not receive any emotional health benefit (Mernitz & Dush, 2016). Women also have greater networks of close relationships, which are linked to increased subjective well-being (Saphire-Bernstein & Taylor, 2013). Drawing from evolutionary theory, Taylor and colleagues (2000) proposed that women have learned to develop and maintain these close relationships over time in response to stress, termed the tend-and-befriend hypothesis. The hypothesis states that, unlike the fight or flight response to stress, women who were primarily responsible for child-rearing sought out social networks that helped keep them (and their children) safe. Thus, women are more likely to provide and receive support, both of which are associated with greater mental and physical health benefits (Taylor, 2011). Many of these discrepancies in the literature may be due to the way researchers conceptualize and operationalize mental health. Internalizing indicators of distress, such as emotional distress, were more accurate at evaluating women’s mental health, whereas externalizing indicators of distress, such as alcohol misuse/abuse, were more accurate at evaluating men’s mental health (Simon, 2002). Men were also more likely to underreport indicators of emotional distress on self-reported scales compared to women (Sigmon et al., 2005) and the prevalence rates for internalized and externalized distress differ by gender. Using data from the World Health Organization World Mental Health Surveys, Seedat et al. (2009) found that women were significantly more likely to develop mood and anxiety disorders (odds ratios ranged from 1.3 to 2.6), and men were significantly more likely to develop substance use disorders (odds ratios ranged from 0.2 to 0.4) worldwide. In studies where both internalizing and externalizing sources of distress were examined, cohabitation was unrelated to depressive symptoms for both genders but associated with more pronounced alcohol misuse among men (Horwitz & White, 1998; Marcussen, 2005). Yet, among emerging adults, both cohabiting men and women reported greater alcohol use than married men and women, although these associations were more pronounced for women (Uecker, 2012). Given these mixed findings, I expect to see no significant gender differences in associations between long-term cohabitation and mental health. 4 Emerging Adulthood Confounding Variables Unaccounted heterogeneity associated with both mental health and cohabitation likely poses a threat to the current study’s validity; thus, I control for several potential sources of bias. Because I examined within-person change in mental health, only time-varying sources are the primary threats to validity. Cohabiting unions were more common among those who are disadvantaged—for instance, emerging adults with lower educational attainment or those who were unemployed (Lichter, Turner, & Sassler, 2010). Further, cohabiting couples with children report poorer mental health than those without children (Brown, 2000). As these factors change over time, I control for these sources of bias. Current Study I used longitudinal data from the NLSY97 to (1) examine within-person change in mental health from entrance into a first long-term cohabitation compared to time spent single (i.e., never reporting any union), (2) examine within-person change in mental health from entrance into a first long-term cohabitation above and beyond any benefits gained from a short-term cohabitation, and (3) examine and test for possible gender differences in all associations. Based on the aims of this study, I propose the following hypotheses: Hypothesis 1: Compared to time spent single, time spent in a first long-term cohabitation will be associated with less internalizing and externalizing distress, measured as emotional distress and heavy episodic drinking. Hypothesis 2: Transitioning from time spent in a short-term first cohabitation to a long-term first cohabitation will be associated with less internalizing and externalizing distress. Hypothesis 3: There will be no observed gender differences in the association between transitioning into a long-term cohabitation and mental health. Method Sample I used data from the NLSY97 (n ¼ 8,984), designed to examine the family formation patterns, employment and educational experiences, and family backgrounds of youth born in the United States between 1980 and 1984. Data were collected annually from 1997 to 2011 and in 2013. These data are nationally representative; youth were selected from over 90,000 housing units within the continental United States and the District of Columbia. All civilian, noninstitutionalized youth were eligible to participant as long as they were between ages 12 and 16 before December 31, 1996, and living in the selected households. More detailed descriptions of available measures, procedures, and instruments are available online (https:// www.nlsinfo.org/content/cohorts/nlsy97). The current study used data from the 2000 to 2010 survey years as emotional distress, the main indicator of internalized distress, was only available biennially during those years; respondents were between ages 15 and 19 at the 2000 survey year and ages 25 to 29 at the 2010 survey year. Out of an initial sample of n ¼ 8,984 participants, I restricted the sample to those that reported a first cohabitation on, or after, the 2000 survey year (n ¼ 4,796) because emotional distress was first measured in 2000. One hundred and sixty-eight respondents reported a first cohabitation prior to the 2000 survey year and were not included in any analyses. The sample was further restricted to those that cohabited after age 18 and reported a long-term cohabitation lasting longer than 3 years (n ¼ 2,131); respondents who reported a short-term cohabitation only (n ¼ 2,262) were not included.1 Missing data were imputed using multiple imputation using chained equations (MICE; 20% data missing). In MICE, each variable acted as the dependent variable, and all other variables were regressed onto it (Johnson & Young, 2011); for instance, all dichotomous variables were imputed using logistic regression. I employed the multiple imputation and then deletion technique where the dependent variable was imputed in order to inform other values, but was not included in the final analyses (von Hippel, 2007). Thus, the sample size was still reduced for those missing information on their dependent variable (see Online Supplementary File 1 for sample restrictions for each analytic model). Measures Long-term cohabitation. Long-term cohabitations were measured using data from individual cohabitation histories, which contained changes in cohabitation every survey year (Center for Human Resources, 2013). Respondents reported the start and end dates for each cohabiting union.Out of all cohabitations, first cohabitations were most common (73%); thus, only these unions were included in the analyses. Duration was captured from the date the first cohabitation began until respondents transitioned out of their cohabitation (via marriage or dissolution) or until the 2010 interview date if no transition occurred. Long-term cohabitations were classified by using the average duration; any cohabitations lasting more than 3 years were considered long-term. Out of all emerging adults in the NLSY97, 25% were classified as long-term cohabitors. For the models examining within-person change in mental health from transitions from single (nonunion state) to longterm cohabitation, individuals were given a “0” during their single years (before a first cohabitation) and “1” during the years of long-term cohabitation. Periods of short-term cohabitation, after first cohabitation dissolution or transition into marriage, and any higher order cohabitation, defined as two or more cohabitation periods, were coded as missing. Thus, models examined the average mental health scores from when a respondent was single and compared these scores to the average mental health scores once they entered into a long-term cohabitation. For the models examining within-person change in mental health from transitions from a short-term to long-term Mernitz cohabitation, individuals were given a “0” each year they were in a short-term cohabitation (defined as a cohabiting union lasting 3 years or less). Respondents were given a “1” each year they were in a long-term cohabitation. Each year a respondent was single, postdissolution, or cohabiting with higher order partners was coded missing. Models compared the average mental health scores from when respondents were in a shortterm cohabitation to the average mental health scores from when respondents were in a long-term cohabitation. Mental health problems. Internalizing and externalizing distress were used to indicate mental health problems. The indicator of internalizing distress, emotional distress, was measured by the 5-item Mental Health Inventory 5 (MHI-5;Veit & Ware, 1983) at the 2000, 2002, 2004, 2006, 2008, and 2010 survey years. The MHI-5 is a valid indicator of depression and anxiety for both adolescents and young adults (Berwick et al., 1991; Ostroff, Woolverton, Berry, & Lesko, 1996) and is measured based on the occurrence of the following symptoms on a scale of 1–4, where 1 ¼ all of the time and 4 ¼ none of the time. “How much of the time during the last month have you” (1) “been a very nervous person?” (2) “felt calm and peaceful?” (3) “felt downhearted and blue?” (4) “been a happy person?” and (5) “felt so down in the dumps that nothing could cheer you up?” Responses to Questions 1, 3, and 5 were reverse coded and totaled, with higher values indicating greater emotional distress. Scale reliabilities ranged from a ¼ .77 to .82 each survey year. Heavy episodic drinking was used as an indicator of externalizing distress and was measured at each survey year (2000– 2010). Participants answered the question “On how many days did you have five or more drinks on the same occasion during the past 30 days?” The same occasion was defined as drinking 5þ drinks at the same time or within hours of each other. Given that many respondents reported 0 days where they drank at least 5þ drinks in the same occasion, this measure was dichotomized where 0 ¼ never and 1 ¼ heavy episodic drinking occurred. In primary care screening tests, this measure was shown to accurately identify harmful drinking patterns and alcohol abuse/dependence (Bush, Kivlahan, McDonell, Fihn, & Bradley, 1998). Thus, this measure likely differentiated between experimental and problem drinking among youth in the existing study. Gender. Gender was measured as a dichotomous indicator where 1 ¼ female and 0 ¼ male. Controls. I controlled for time-varying, dichotomous indicators of education, employment status, biological children, pregnancy, and current enrollment. Education was coded as an indicator of highest educational attainment and coded into four response categories: less than high school degree, high school degree (reference category), some college, or college degree and more than a college degree. Employment was coded from the employment status history file and was coded into three response categories: full-time employment (more than 35 hr 5 worked a week for at least 50 weeks), part-time employment (working less than 35 hr a week for less than 50 weeks), and not employed (reported working no hours). Biological children were coded as whether or not the respondent (or respondent’s partner) reported having given birth during the survey year. Pregnancy was coded as whether or not a respondent (or respondent’s partner) reported becoming pregnant during the survey year. Current enrollment was coded as whether or not the respondent was enrolled in any schooling. Additionally, I controlled for temporal variation in mental health by including yearly time dummy variables. These timing effects are represented by interactions between each year and each mental health outcome (either emotional distress or heavy episodic drinking). Analytic Plan To test all hypotheses, I used the Stata 14 statistical package to run pooled fixed-effects regression and logistic regression models (Allison, 1990; Johnson, 2005). Fixed-effects regression and logistic models account for within-individual change over time, while controlling for all time-invariant sources of bias (Allison, 1990). To illustrate how I examine change in a change score framework, I first use the following equation for emotional distress at Time 1: Emotional distressi1 ¼ b0 þ b2 Mi þ b3 Ui þ b3 Ti1 þ ei1 ; ð1Þ where b0 was a fixed constant, bj were the regression coefficients, Mi was a vector of measured time-invariant control variables, Ui was a vector of unmeasured time-invariant control variables, Ti was a vector of measured time-variant control variables, and ei was the error term. The equation for Time 2 looks similar, except this equation would include the regression coefficient for the event Xi (i.e., long-term cohabitation). Because any observed cohabitation does not occur prior to Time 1, it does not appear in the first equation. Emotional distressi2 ¼ b0 þ b1 Xi þ b2 Mi þ b3 Ui þ b3 Ti2 þ ei1 : ð2Þ The two equations were differenced in the following model to produce the change score model for the continuous variable emotional distress: ðEmotional distressi2 Emotional distressi1 Þ ¼ ðb0 b0 Þ þ b1 Xi þ ðb2 Mi b2 Mi Þ þ ðb3 Ui b3 Ui Þ þ ðb4 Ti2 b4 Ti1 Þ þ ðei2 ei1 Þ: ð3Þ All time-invariant coefficients (Mi, Ui, and b0) were differenced out of the equation, which reduced the equation to the following: ðEmotional distressi2 Emotional distressi1 Þ ¼ b1 Xi þ b4 T 0i þ e0i : ð4Þ For logistic regression models predicting heavy episodic drinking, the dependent variable is measured as the probability that heavy episodic drinking occurred for an individual at each 6 Emerging Adulthood time point. In these instances, the final equation would be: LogðPðHeavy episodic drinkingit Þ=1 LogðPðHeavy episodic drinkingit Þ ¼ b1 Xi þ b4 T 0i þ e0i ; ð5Þ where P is the probability of heavy episodic drinking, i represents the individual, and t represents the time point. The remaining equation is interpreted similarly to the regression Equation 4. Thus, in all of my fixed-effects regression models, I controlled for time-invariant sources of heterogeneity, regardless of whether they were observed or unobserved. In these models, certain variables that account for selection bias, such as family background characteristics, are controlled by the model design. Because I had observations from more than two time points, I use a pooled fixed-effects regression model and a conditional logistic regression model, which interprets an individual’s outcome as a deviation away from the mean at each point in time. Emerging adults who remain in a long-term cohabitation might differ from other emerging adults who enter into other unions (i.e., direct marriage or short-term cohabitation). Comparisons between other unions and long-term cohabitors may find a significant mental health effect that is due to selection. For instance, if married emerging adults reported less distress than long-term cohabitors, it might be that those who entered into direct marriage have less distress overall and not due to marriage. By using modeling that compares long-term cohabitors with themselves, the effect of long-term cohabitation on mental health is likely to be due to the union itself. For Hypothesis 1, that long-term cohabitation will be associated with better mental health compared to time spent single, I conducted several pooled fixed-effects regression and logistic regression models. I examined change in an individual’s emotional distress from models comparing an individual’s emotional health when single (defined as never reporting a union) with his or her emotional health when in a long-term cohabitation (defined as a cohabiting union with an above average duration). I repeated these analyses using fixed-effects logistic regression with the externalizing mental health outcome, heavy episodic drinking. For Hypothesis 2, that time spent in a long-term cohabitation will be associated with better mental health compared to time spent in a short-term cohabitation, I conducted pooled fixed-effects regression and logistic regression models. I analyzed within-person change in emotional distress from models comparing an individual’s emotional distress during their time in a short-term cohabitation and emotional distress during time spent in a long-term cohabitation. Again, I repeated these analyses using fixed-effects logistic regression when predicting the externalizing indicator of distress, heavy episodic drinking. For Hypothesis 3, that there will be no observed gender differences in mental health between time spent single or in a short-term cohabitation and time spent in a long-term cohabitation, I analyzed all models discussed for Hypotheses 1 and 2 separately by gender. Additionally, I calculated Wald’s test statistics based on the procedures outlined in Clogg, Petkova, and Haritou (1995) to compare group differences between men and women. I used the following equation: z ¼ ðbx by Þ=½s2 ðbx Þ þ s2 ðby Þ1=2 ; ð6Þ where bx is the coefficient for Group 1 (i.e., women), by is the coefficient for Group 2 (i.e., men), and s is the standard error for each coefficient. This equation produces a z statistic, which can be used to find the p value for each test. Results Descriptive Statistics Weighted descriptive statistics are presented in Table 1; weighting descriptive statistics ensures that the results are representative of the population. Because the NLSY97 oversamples certain respondents (i.e., Hispanic respondents), failure to weight the data results in the decreased ability to generalize to the population. Descriptive statistics for all long-term cohabitations (averaged around 7 years) indicated that individuals were predominately White, employed full-time, and received a high school degree. Most were not currently enrolled in school. The average age was around 24, and both genders were represented about equally. Few respondents had a child or a pregnancy, high levels of emotional distress, or reported heavy episodic drinking. Descriptive statistics were comparable by gender, yet there were statistically significant gender differences. Women reported more emotional distress than men (means of 9.87 and 9.37, respectively) whereas men reported more episodic drinking than women (48% vs. 34%). Women were more likely than men to be employed part-time (30% vs. 18%) and less likely to be employed full-time (59% vs. 70%); women were also significantly more educated (15% received at least a college education, 65% received a high school degree, and 16% reported less than a high school degree) and more likely to be enrolled in school (22% enrolled) than men (only 8% of men had a least a college education, 68% had a high school degree, and 20% had less than a high school degree; only 12% were currently enrolled). Lastly, women were younger (23.83 vs. 24.20), less likely to be Hispanic (14% vs. 16%), and more likely to have a child (9% vs. 8%) or report a pregnancy (9% vs. 6%). There were no significant differences in the length of long-term cohabitation (women’s cohabitations lasted on average 7.04 years whereas men’s cohabitations lasted on average 7.10 years). Fixed-Effects Regression Models Pooled-fixed effects regression models partially supported my first hypothesis that, compared to time spent single, time spent in a first long-term cohabitation will be associated with better mental health (see Tables 2 and 3, Model 1). Emotional distress was greater during time spent in a long-term cohabitation than during time spent single (not supporting Hypothesis 1). However, heavy episodic drinking rates were lower during time spent in a long-term cohabitation than they were during time Mernitz 7 Table 1. Weighted Descriptive Statistics for All Long-Term Cohabitors and by Gender. Full Sample Variables % Cohabitation duration (Years) Mental health Emotional distress Heavy episodic drinking 41 Controls Employment status Not employed 11 Part-time employment 24 Full-time employment 64 Education Less than high school 18 High school 67 Some college 4 College or more 12 Current enrollment 17 Age Female 51 Race White 62 Black 19 Hispanic 15 Had a child 9 Pregnancy 8 n 2,131 Women M (SD) Range 7.07 (2.30) 4–11 9.63 (2.52) 5–20 24.01 (2.81) % Men M (SD) Range 7.04 (2.31) 4–11 9.87 (2.46)a 5–19 % a 34 48 11 30a 59a 12 18 70 16a 65a 4 15a 22a 20 68 4 8 12 23.83 (2.84)a 18–30 18–30 — — 62 20a 14a 9a 9a 1,095 62 18 16 8 6 1,036 M (SD) Range 7.10 (2.29) 4–11 9.37 (2.55) 5–20 24.20 (2.77) 18–30 Note. M ¼ mean. SD ¼ standard deviation. a Significant gender difference. spent single (supporting Hypothesis 1). Inconsistent with my second hypothesis that, compared to time spent in a shortterm cohabitation, time spent in a long-term cohabitation will be associated with better mental health, there were no significant changes in mental health between time spent in a shortterm and time spent in a long-term cohabitation (see Tables 2 and 3, Model 2). Fixed-Effects Regression Models by Gender I hypothesized that there would be no observed gender differences in associations between long-term cohabitations and mental health; pooled fixed-effects regression models did not support this conclusion overall (see Tables 4 and 5). For men, emotional distress was greater during time spent in a long-term cohabitation than it was for time spent single (b ¼ .42, p < .001); for women, emotional distress during time spent single was not significantly different from emotional distress during time spent in a long-term cohabitation. A Wald’s test confirmed this gender difference (z ¼ 5.37; p < .01). Average emotional distress scores during time spent in a long-term cohabitation did not significantly differ from average emotional health scores during time spent in a short-term cohabitation for either gender. A Wald’s test confirmed that there were no gender differences (z ¼ 0.03; p > .10). When predicting change in heavy episodic drinking, women had less average heavy episodic drinking during their time spent in a long-term cohabitation than they had during their time spent single (b ¼ .54, p < .01); for men, this association was only marginally significant (b ¼ .34, p ¼ .07). A Wald’s test suggested that there were no significant gender differences (z ¼ 0.53; p > .10). Average heavy episodic drinking scores during time spent in a short-term cohabitation and time spent in a long-term cohabitation were not significantly different from each other for either gender. A Wald’s test was also nonsignificant, suggesting that there were no gender differences (z ¼ 0.11; p > .10). Long-Term Cohabitation Defined as a Standard Deviation Above the Mean Prior studies on long-term cohabitation defined these unions as cohabitations lasting longer than 3 years (Willetts, 2006) or 4 years (Skinner et al., 2002); these definitions were used because cohabitations needed to occur at both waves of available National Survey of Families and Households data. Although the length of those long-term cohabitations is comparable to the average length in the current study, I also measured long-term cohabitation defined as a standard deviation above the mean. In these instances, long-term cohabitors were classified as cohabitations lasting for at least 6 years (13% of all cohabitations were classified as long-term;n ¼ 1,205). Many of the findings were not replicated with this new definition of long-term cohabitation. For the full model, emotional distress 8 Emerging Adulthood Table 2. Pooled Fixed-Effects Regression Predicting Change in Emotional Distress From Long-Term Cohabitation. Variables Long-term cohabitation Long-term cohabitation versus single Long-term cohabitation versus short-term Controls Education (ref: high school) Less than high school Some college College Employment (ref: part-time) Not employed Full-time Had a child Pregnancy Current enrollment Year Emotional Distressc 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Person-yearsd n Table 3. Pooled Fixed-Effects Logistic Regression Predicting Change in Heavy Episodic Drinking From Long-Term Cohabitation. Model 1a Model 2b B SE B B SE B .20* .09 — — — — .02 .06 .11 .08 .02 .07 .13 .10 .22* .13 .08 .10 .15 .12 .06 .01 .04 .10 .05 .06 .05 .07 .07 .05 .01 .02 .01 .01 .17 .08 .05 .06 .06 .06 .07 .07 .13 .08 .38*** .08 .47*** .09 .62*** .10 .66*** .11 .45*** .12 .44*** .12 .59*** .12 .61*** .12 12,423 2,098 .04 .13 .01 .12 .45** .12 .43** .13 .60*** .13 .59*** .13 .38* .14 .38* .14 .54*** .15 .53*** .15 11,918 2,090 Note. aModel 1 compared the emotional distress from an individual’s time spent single (not in a union) to time spent in a first long-term cohabitation. bMode1 2 compared the emotional distress from an individual’s time spent in a shortterm cohabitation to time spent in a long-term cohabitation. cInteractions between each year and emotional distress control for temporal variation in distress. dPerson-years estimated how much time each participant contributed to the study. *p < .05. **p < .01. ***p < .001. became significantly lower during the time spent in a long-term cohabitation compared to emotional distress when single (b ¼ .33, p < .001); heavy episodic drinking was no longer significant when time spent in a long-term cohabitation was compared to time spent single (b ¼ .29, p ¼ .11). By gender, emotional distress scores were no longer significant for men in models comparing time spent in long-term cohabitation versus time-spent single (b ¼ .16, p ¼ .28); for women, emotional distress scores became significant in that time spent in long-term cohabitation was associated with less emotional distress than time spent single (b ¼ .53, p < .001). Wald’s tests indicated that there were gender differences in these associations (z ¼ 3.04, p < .01). For women, time spent in long-term cohabitation was no longer significantly associated with heavy episodic drinking compared to time spent single (b ¼ .27, p ¼ .27); findings for men for this association were also not Variables Long-term cohabitations Long-term cohabitation versus single Long-term cohabitation versus short-term Controls Education (ref: high school) Less than high school Some college College Employment (ref: part-time) Not employed Full-time Had a child Pregnancy Current enrollment Year Heavy Episodic Drinkingc 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Person-yearsd n Model 1a Model 2b B SE B B SE B 0.46*** .14 — — — — .02 .09 0.36** 0.15 0.22 .11 .22 .16 .19 .11 .06 .20 .28 .23 0.02 0.21** 0.46*** 0.56*** 0.12 .11 .07 .13 .12 .09 .23 .24** .32*** .65*** .17 .13 .08 .10 .11 .10 0.34*** 0.50*** 0.82*** 0.77*** 0.99*** 0.99*** 1.12*** 0.90*** 0.91*** 0.79*** 7,436 1,198 .10 .12 .13 .14 .15 .17 .19 .19 .19 .20 .14 .10 .11 .24 .31 .37 .44 .26 .29 .13 6,469 1,107 .20 .20 .20 .21 .21 .22 .23 .24 .25 .26 Note. aModel 1 compared heavy episodic drinking from an individual’s time spent single (not in a union) to time spent in a first long-term cohabitation. b Mode1 2 compared heavy episodic drinking from an individual’s time spent in a short-term cohabitation to time spent in a long-term cohabitation. cInteractions between each year and heavy episodic drinking control for temporal variation in distress. dPerson-years estimated how much time each participant contributed to the study. *p < .05. **p < .01. ***p < .001. significant, but the coefficient became positive (b ¼ .31, p ¼ .24). Wald’s tests confirmed there were no gender differences (z ¼ 0.02, p > .01). By gender, all associations between mental health and long-term cohabitation (compared to short-term cohabitation) were replicated. Discussion Emerging adult cohabitation has changed rapidly in prevalence and stability over time (Cherlin, 2010). While serial cohabitation, or cohabitation with more than two partners over time, has received scholarly attention (e.g., Lichter & Qian, 2008), the other end of the spectrum—long-term cohabitation—has received little attention. The prevalence of long-term cohabitation in emerging adulthood remains unknown, yet these unions Mernitz 9 Table 4. Pooled Fixed-Effects Regression Predicting Change in Emotional Distress From Long-Term Cohabitation by Gender. Model 1a Variables Long-term cohabitations Single versus long-term cohabitation Short versus long-term cohabitation Education (ref: high school) Less than high school Some college College Employment (ref: part-time) Not employed Full-time Had a child Pregnancy Current enrollment Year Emotional Distressc 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Person-yearsd n Model 2b Female Male Female Male B (SE) B (SE) B (SE) B(SE) .01 (.12) — .42*** (.13) — — .03 (.08) — .01 (.08) .13 (.10) .20 (.17) .03 (.12) .08 (.10) .15 (.21) .04 (.17) .01 (.15) .23 (.18) .12 (.14) .39** (.14) .06 (.26) .02 (.23) .09 (.09) .05 (.06) .05 (.10) .03 (.09) .02 (.07) .02 (.10) .03 (.07) .01 (.12) .19 (.11) .13 (.08) .06 (.09) .05 (.06) .09 (.07) .02 (.08) .15* (.08) .09 (.11) .11 (.08) .14 (.09) .03 (.10) .19 (.11) .08 (.10) .03 (.11) .48*** (.12) .56*** (.13) .72*** (.13) .73*** (.15) .49** (.16) .46** (.17) .71*** (.16) .76*** (.17) 6,389 1,081 .22* (.10) .29** (.11) .29* (.12) .38** (.13) .54*** (.14) .62*** (.15) .45** (.17) .47** (.17) .53** (.17) .52** (.18) 6,034 1,017 .05 (.16) .11 (.16) .15 (.16) .18 (.16) .41* (.17) .38* (.17) .11 (.18) .09 (.19) .35 (.19) .35 (.20) 6,330 1,082 .04 (.20) .17 (.19) .83*** (.19) .77*** (.19) .86*** (.20) .85*** (.20) .71*** (.21) .73*** (.22) .80*** (.23) .77*** (.23) 5,588 1,008 Note. aModel 1 compared heavy episodic drinking from time spent single (not in a union) to time spent in a first long-term cohabitation. bMode1 2 compared heavy episodic drinking from time spent in a short-term cohabitation to time spent in a long-term cohabitation. cInteractions between each year and emotional distress control for temporal variation in distress. dPerson-years estimated how much time each participant contributed to the study. *p < .05. **p < .01. ***p < .001. may be an alternative to marriage for many youth, especially youth facing economic and social barriers to marriage. These long-term unions may represent an attainable way to develop intimacy in romantic relationships, meeting a key developmental task (Arnett, 2000), at a time when marriage may be unattainable (Finkel et al., 2015). Consistent with my first hypothesis, average heavy episodic drinking rates were lower during a first long-term cohabitation than they were during time spent single. However, inconsistent with my first hypothesis, the average emotional distress scores were higher during time spent in a long-term cohabitation than they were during time spent single. According to rational-choice theorists, coresidence, such as cohabitation, increases an individual’s ability to monitor a partner’s behavior and vice versa (Friedman, 1995; Willis, 2006). Thus, long-term cohabitors may decrease their heavy episodic drinking because their partner is monitoring their behavior and may end the relationship if they do not decrease their drinking. As emotional distress may be less obvious to a partner—for instance, an individual can more easily identify when their partner is drunk rather than anxious, emotional distress may not be as noticeable to romantic partners. The higher rates of emotional distress during periods of long-term cohabitation may be due to testing the relationship. Approximately 60% of emerging adults report entering into cohabitation as a way to test their relationship (Stanley et al., 2011) and assess partner compatibility (Huang et al., 2011). Thus, these emerging adults may not be sure about the future of their relationship, and this long-term uncertainty may contribute to emotional distress over time. Contrarily, emotional health and well-being decline with union duration (Musick & Bumpass, 2012) and the observed emotional health decline may be due to the length of the union rather than cohabitation itself. Mernitz and Dush (2016) found that emerging adult first cohabitation was associated with emotional health benefits and that transitioning into marriage from cohabitation with the same partner was not associated with additional benefits. Thus, there is some evidence from emerging adult samples that emotional health may decline over time, regardless of union type. An alternative explanation for higher rates of emotional distress may be due to investment in these long-term cohabitations. The investment model (Rusbult, 1980) applied to 10 Emerging Adulthood Table 5. Pooled Fixed-Effects Logistic Regression Predicting Change in Heavy Episodic Drinking From Long-Term Cohabitation by Gender. Model 1a Variables Long-term cohabitations Single versus long-term cohabitation Short versus long-term cohabitation Education (ref: high school) Less than high school Some college College Employment (ref: part-time) Not employed Full-time Had a child Pregnancy Current enrollment Year Heavy Episodic Drinkingc 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Person-yearsd n Model 2b Female Male Female Male B (SE) B (SE) B (SE) B (SE) 0.54** (0.20) — 0.34 (.19) — — 0.02 (.13) — .04 (.13) 0.59*** (0.17) 0.14 (0.30) 0.02 (0.20) 0.15 (.16) 0.19 (.34) 0.64* (.29) 0.53 (.32) 0.02 (.34) 0.02 (.28) .80** (.30) .45 (.51) .14 (.40) (.15) (.11) (.18) (.15) (.12) 0.12 (.19) 0.32** (.11) 0.56*** (.15) 1.26*** (.17) 0.21 (.14) .35 (.19) .11 (.12) .12 (.14) .14 (.15) .13 (.17) 0.35* (.14) 0.54*** (.16) 0.93*** (.18) 0.74*** (.19) 1.04*** (.21) 0.98*** (.23) 0.89*** (.26) 0.70** (.26) 0.72** (.27) 0.56* (.27) 3,823 618 0.07 (.29) 0.12 (.28) 0.08 (.28) 0.08 (.30) 0.10 (.30) 0.20 (.32) 0.31 (.33) 0.19 (.35) 0.18 (.36) 0.06 (.38) 3,425 571 .35 (.30) .33 (.29) .09 (.29) .36 (.30) .50 (.31) .48 (.32) .52 (.33) .28 (.34) .34 (.36) .16 (.38) 3,044 536 0.01 0.11 0.61*** 1.06*** 0.09 (0.16) (0.10) (0.19) (0.19) (0.12) 0.29 (0.16) 0.38* (0.18) 0.62*** (0.19) 0.77*** (0.20) 0.87*** (0.22) 0.90*** (0.25) 1.26*** (0.27) 1.02*** (0.28) 1.06*** (0.28) 0.97*** (0.29) 3,613 580 0.03 0.28** 0.36* 0.18 0.35** Note. aModel 1 compared heavy episodic drinking from time spent single (not in a union) to time spent in a first long-term cohabitation. bMode1 2 compared heavy episodic drinking from time spent in a short-term cohabitation to time spent in a long-term cohabitation. cInteractions between each year and heavy episodic drinking control for temporal variation in distress. dPerson-years estimated how much time each participant contributed to the study. *p < .05. **p < .01. ***p < .001. romantic relationships suggests that commitment to a relationship is derived from relationship satisfaction, lack of alternative partners, and relationship investment (Rusbult, Agnew, & Arriaga, 2011). Although relationship satisfaction is most strongly associated with commitment, each component has an additive effect (Le & Agnew, 2003), and satisfaction alone is not the strongest predictor of relationship dissolution (Le, Dove, Agnew, Korn, & Mutso, 2010). Relationship investments can be intangible (i.e., relationship effort) or tangible (i.e., material possessions like shared furniture) and consist of both investments in the past and expectations of future investments; all intangible investments and future tangible investments were predictive of commitment (Goodfriend & Agnew, 2008). Long-term cohabitors have likely accrued many investments in their relationships, including plans for future investments (i.e., buying a home together), which may encourage them to remain committed to the relationship and continue cohabiting. However, they may also be experiencing a decline in relationship satisfaction, common among long-term relationships, including cohabitation (Brown, 2003; Skinner et al., 2002), which may contribute to increased emotional distress. My results for the mental health implications from transitioning into a long-term cohabitation from short-term cohabitation do not suggest that mental health declines over time. I found that there were no additional reductions in distress from remaining in cohabitation long-term, yet mental health did not decline over time either. Identity and intimacy development are key developmental tasks during the transition to adulthood (Arnett, 2000; Erikson, 1968), and failure to meet these tasks is associated with health problems and risky behavior (Erikson, 1968). Cohabitation provides an avenue to establish a romantic identity and develop intimacy with a romantic partner; thus, distress may be reduced for all cohabitors. Emerging adults who enter into a short-term cohabitation are successfully able to negotiate their personal needs with those of their partner; however, transitioning into a long-term cohabitation or marriage requires the ability to manage these needs in the context of economic and social uncertainty (Shulman & Connolly, 2013, 2015). Because the transition from short- to long-term cohabitation did not reduce distress, it may be that emerging adults are facing difficulty in other areas in life, specifically employment. Indeed, employment and romantic identities Mernitz become critically important in emerging adulthood (Arnett, 2000), and contemporary emerging adults have faced recent difficulty finding full-time employment, even among the college educated (Settersten & Ray, 2010). Failure in one domain (i.e., employment) can lead to negative experiences, which influence the other domain (i.e., relationships; Edwards & Rothbard, 2000). Thus, difficulty establishing an employment and romantic identity, a common occurrence during the emerging adult developmental period (Shulman & Connolly, 2015), might offset any decreased distress from successfully establishing a romantic identity in the context of long-term cohabitation. Gender Differences There were also significant gender differences in the mental health implications of long-term cohabitation (inconsistent with Hypothesis 3). For men, average emotional distress was greater during time in a long-term cohabitation than it was when men were single. Past research has suggested that men were more likely to report cohabiting as a way to test a relationship prior to marrying a partner than women (Stanley et al., 2011). Thus, the long-term length of these cohabitations might suggest that men are unsure about the future of their union and less committed to the relationship. The lack of commitment may stem from men perceiving better alternatives to their relationship (compared to women who are more invested and satisfied with their relationships; Le & Agnew, 2003). Because an individual’s level of commitment predicts his or her behavior in the relationship (Rusbult et al., 2011) and relationship behaviors that help maintain the relationship (i.e., forgiving a transgression; Rusbult, Hannon, Stocker, & Finkel, 2005) are associated with reduced emotional distress (Williamson & Gonzales, 2007), emerging adults in less committed relationships may report greater emotional distress. Men may be especially susceptible to these associations; for instance, less committed men viewed their partner’s qualities and virtues more negatively than their partners viewed themselves. Women, regardless of commitment level, viewed their partners more positively than their partners viewed themselves (Gagné & Lydon, 2003). Taken together, the combination of less commitment and negative relationship behaviors over time may contribute to increased distress for men. Even men highly committed to their cohabiting partner may exhibit increased distress in their long-term cohabitation because they might desire to get married to their partner, but be unable to do so. Indeed, studies have found that financial security is an important precursor to marriage (Smock et al., 2005), especially among men (Sassler & Goldscheider, 2004; Schneider, 2011), and the recent Great Recession experienced by this sample may have contributed to couples remaining in their cohabitation out of necessity, rather than desire. When mental health was measured by heavy episodic drinking, for both genders, heavy episodic drinking rates were lower during time spent in a long-term cohabitation than they were during time spent single (although this association was only 11 marginally significant for men, there were no gender differences). Inconsistent with past work, which found that cohabiting women experienced more pronounced declines in problem drinking (Uecker, 2012), I found evidence that there were no gender differences in heavy drinking. Scholars have suggested that emerging adults who plan to marry sooner rather than later decrease their substance use (Carroll et al., 2007), and it may be that many emerging adults in these long-term cohabitations expect to marry soon—or have an earlier marital horizon. Consistent with past research (i.e., Saphire-Bernstein & Taylor, 2013), these gender findings broadly provide some support for the notion that women benefit more from their relationships than men. However, these findings also highlight the importance of examining mental health from multiple domains. Specifically, long-term cohabitation appears less beneficial for men’s mental health compared to women’s mental health when considering internalized distress, but comparable when considering externalized distress. Further, as other studies have observed gender differences in relationship quality, commitment, and relationship maintenance behaviors (Gagné & Lydon, 2003; Le & Agnew, 2003; Skinner et al., 2002), understanding how long-term cohabitation is beneficial for mental health for both genders should consider relationship processes that likely influence these associations. Limitations and Future Research I acknowledge some limitations in the existing study. Emotional health was only measured biennially beginning in the 2000 survey year until the 2010 survey year. Thus, changes in emotional distress that might occur annually are masked in the current analyses. Because the data contain monthly cohabitation arrays, a more frequent measure of emotional distress and heavy episodic drinking would allow me to capture nuanced changes in health that may be more important for identifying a “honeymoon” effect, whereby unions only provide short-term health benefits, for long-term cohabitors. Further, for both emotional health and heavy episodic drinking, participants were asked to indicate their levels of distress in the past month, which likely varies. Future research containing more frequent measures of mental health where general levels of distress are assessed may help researchers better understand the implications of long-term cohabitation for health. An additional limitation is that the data do not contain an indicator for cohabitation union quality during the survey years (quality was only measured until 2007). Because quality is associated with mental health (i.e., Proulx, Helms, & Buehler, 2007), the relationship quality of these long-term unions is likely important for mental health. For instance, transitioning from short-term to a long-term cohabitation may provide mental health benefits if the union is of high quality and contribute to worse mental health if the union is of low quality. Thus, only the highquality long-term cohabitations may be akin to marriage and provide the most health benefits. Although quality was assessed in the NLSY97 during the short-term cohabitation time points, I do not have the measure available for the full 12 sample of long-term cohabitors during the time spent in longterm cohabitation. Future research should examine these associations accounting for relationship quality. Lastly, I am unable to distinguish between emerging adults who are not in any romantic relationship and those that are in a romantic relationship but not a marriage or cohabitation. The NLSY97 collected detailed information about cohabitation and marital status but did not collect romantic relationship entrance and exit dates for dating relationships. Thus, my category of “single” refers to those not in a union, which likely masks differences between emerging adults not in any relationship and those in a dating relationship. Future research could compare an individual’s mental health when they were in a long-term dating relationship to when they were in a long-term cohabitation. Relatedly, future research could also examine mental health differences between first- and higher order long-term cohabitors. There is evidence that emerging adults who enter into second cohabitations report less distress than when they were involved in a first cohabitation (Mernitz & Dush, 2016), suggesting that higher order relationships may reduce distress above and beyond a first relationship. Future work might also want to focus on the characteristics of emerging adults that are entering into these long-term cohabitations. Although the measurement of long-term cohabitation in the current study is consistent with existing research (Skinner et al, 2002; Willetts, 2006), those that cohabited for longer durations (a standard deviation above the average) exhibited different mental health outcomes. Thus, paying specific attention to who cohabits at varying durations, and possibly the relationship processes within these unions, is an important avenue for future research on long-term cohabitation. Future work should also examine associations between mental health and long-term cohabitation in a more recent sample of emerging adults. Although the NLSY97 provides highquality longitudinal data on cohabiting relationships, youth in the study entered into emerging adulthood around a decade ago. The recent Great Recession and the growing societal acceptance and prevalence of cohabitation might influence the generalizability of these results. For example, the Great Recession in the late 2000s might have lengthened the duration of cohabiting unions for contemporary youth because youth highlight financial security as an important precursor to marriage (Smock et al., 2005). In these instances, long-term cohabitation may be associated with increased distress if emerging adults want to, but are unable to, marry. However, the growing acceptance of cohabitation may reduce the likelihood that youth report wanting to marry in emerging adulthood, and longterm cohabitation could be linked to no change in, or even less, distress, regardless of the Great Recession. Emerging Adulthood identify an individual’s transitions into long-term cohabitation from time spent single or in a short-term cohabitation, accounting for preexisting individual characteristics to better isolate the effect of long-term cohabitations on health. Because the meaning of cohabitation has shifted over time (Furstenberg, 2011) and the age at first marriage has risen (Cherlin, 2010), examining different types of cohabitation, and their associations with health, is becoming increasingly important. Using a contemporary sample of emerging adults, I found that both internalizing (emotional distress) and externalizing (heavy episodic drinking) symptoms of distress were associated with long-term cohabitation. Long-term cohabitation was associated with increased emotional distress and decreased heavy episodic drinking compared to time spent single. Further, men reported more emotional distress from long-term cohabitation than women; there were no gender differences in heavy episodic drinking. There was no change in mental health from transitioning from short- to long-term cohabitation. Overall, my findings highlight the importance of considering multiple indicators of mental health status and suggest that the longterm cohabitations are an understudied union with important implications for health. Author Contribution Sara E. Mernitz contributed to conception, design, acquisition, analysis, and interpretation; drafted the manuscript; critically revised the manuscript; gave final approval; and agreed to be accountable for all aspects of work ensuring integrity and accuracy. Declaration of Conflicting Interests The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author received no financial support for the research, authorship, and/or publication of this article. Supplemental Material Supplementary material for this article is available online. Note 1. From 1997 to 2004, cohabitation was defined as a “sexual relationship in which partners of the opposite sex live together.” The phrase “opposite sex” was removed from 2005 to 2013, but the data still included references to living in a “marriage-like relationship” when discussing cohabitation (Center for Human Resource Research, 2013). One percent (n ¼ 113) of the full National Longitudinal Survey of Youth 1997 sample reported being in a same-sex cohabitation (n ¼ 60 cohabitors would be considered long-term; thus, I cannot distinguish between same- and different-sex longterm cohabitors). Conclusion This study provided an initial examination into the implications of long-term cohabitation for mental health. By using stringent change score analyses (i.e., Johnson, 2005), I was able to References Allison, P. D. (1990). Change scores as dependent variables in regression analysis. Sociological Methodology, 20, 93–114. Mernitz Arnett, J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469–480. Berwick, D., Murphy, J., Goldman, P., Ware, J., Barsky, A., & Weinstein, M. (1991). Performance of a five-item mental healthscreening test. Medical Care, 29, 169–176. Blekesaune, M. (2008). Partnership transitions and mental distress: Investigating temporal order. Journal of Marriage and Family, 70, 879–890. Brown, S. L. (2000). The effect of union type on psychological wellbeing: Depression among cohabitors versus married. Journal of Health and Social Behavior, 41, 241–255. Brown, S. L. (2003). Relationship quality dynamics of cohabiting unions. Journal of Family Issues, 24, 583–601. Bush, K., Kivlahan, D. R., McDonell, M. B., Fihn, S. D., & Bradley, K. A. (1998). The AUDIT alcohol consumption questions (AUDIT-C): An effective brief screening test for problem drinking. Archives of Internal Medicine, 158, 1789–1795. Carroll, J. S., Willoughby, B., Badger, S., Nelson, L. J., McNamara Barry, C., & Madsen, S. D. (2007). So close, yet so far away: The impact of varying marital horizons on emerging adulthood. Journal of Adolescent Research, 22, 219–247. Center for Human Resource Research. (2013). NLSY handbook. Columbus: The Ohio State University. Cherlin, A. (2010). Demographic trends in the United States: A review of research in the 2000s. Journal of Marriage and Family, 72, 403–419. Clogg, C. C., Petkova, E., & Haritou, A. (1995). Statistical methods for comparing regression coefficients between models. American Journal of Sociology, 100, 1261–1293. Demir, M. (2010). Close relationships and happiness among emerging adults. Journal of Happiness Studies, 11, 293–313. Duncan, G. J., Wilkerson, B., & England, P. (2006). Cleaning up their act: The effects of marriage and cohabitation on licit and illicit drug use. Demography, 43, 691–710. Edin, K. (2000). What do low-income single mothers say about marriage?Social Problems, 47, 112–133. Edwards, J., & Rothbard, N. (2000). Mechanisms linking work and family: Clarifying the relationship between work and family constructs. Academy of Management Review, 25, 178–199. Erikson, E. H. (1968). Identity, youth, and crisis. New York, NY: W. W. Norton. Finkel, E. J., Hui, C. M., Carswell, K. L., & Larson, G. (2015). The suffocation of marriage: Climbing mount Maslow without enough oxygen. Psychological Inquiry, 25, 1–41. Fleming, C. B., White, H. R., & Catalano, R. F. (2010). Romantic relationships and substance use in early adulthood: An examination of the influences of relationship type, partner substance use, and relationship quality. Journal of Health and Social Behavior, 51, 153–167. Frech, A., & Williams, K. (2007). Depression and the psychological benefits of entering marriage. Journal of Health and Social Behavior, 48, 149–163. Friedman, D. (1995). Towards a structure of indifference: The social origins of maternal custody. New York, NY: Aldine de Gruyter. Furstenberg, F. F. (2011). The recent transformation of the American family: Witnessing and exploring social change. In M. J. Carlson & 13 P. England (Eds.), Social class and changing families in unequal America (pp. 192–220). Palo Alto, CA: Stanford University Press. Gagné, F., & Lydon, J. (2003). Identification and the commitment shift: Accounting for gender differences in relationship illusions. Personality and Social Psychology Bulletin, 29, 907–919. Goodfriend, W., & Agnew, C. R. (2008). Sunken costs and desired plans: Examining different types of investments in close relationships. Personality and Social Psychology Bulletin, 34, 1639–1652. Horwitz, A. V., & White, H. R. (1998). The relationship of cohabitation and mental health: A study of a young adult cohort. Journal of Marriage and the Family, 60, 505–514. Huang, P. M., Smock, P. J., Manning, W. D., & Bergstrom-Lynch, C. (2011). He says, she says: Gender and cohabitation. Journal of Family Issues, 32, 876–905. Johnson, D. R. (2005). Two-wave panel analysis: Comparing statistical methods for studying the effects of transitions. Journal of Marriage and Family, 67, 1061–1075. Johnson, D. R., & Young, R. (2011). Toward best practices in analyzing datasets with missing data: Comparisons and recommendations. Journal of Marriage and Family, 73, 926–945. Kennedy, S., & Bumpass, L. (2008). Cohabitation and children’s living arrangements: New estimates from the United States. Demographic Research, 19, 1663–1692. Kim, H. K., & McKenry, P. C. (2002). The relationship between marriage and psychological well-being: A longitudinal analysis. Journal of Family Issues, 23, 885–911. Kroeger, R. A., & Smock, P. J. (2014). Cohabitation: Recent research and implications. In J. K. Treas, J. Scott, & M. Richards (Eds.), The Wiley-Blackwell companion to the sociology of families (pp. 217–235). New York, NY: Wiley-Blackwell. Lamb, K. A., Lee, G. R., & DeMaris, A. (2003). Union formation and depression: Selection and relationship effects. Journal of Marriage and Family, 65, 953–962. Le, B., & Agnew, C. R. (2003). Commitment and its theorized determinants: A meta-analysis of the investment model. Personal Relationships, 10, 37–57. Le, B., Dove, N. L., Agnew, C. R., Korn, M. S., & Mutso, A. A. (2010). Predicting nonmarital romantic relationship dissolution: A meta-analytic synthesis. Personal Relationships, 17, 377–390. Lichter, D. T., & Qian, Z. (2008). Serial cohabitation and the marital life course. Journal of Marriage and Family, 70, 861–878. Lichter, D. T., Qian, Z., & Mellott, L. M. (2006). Marriage or dissolution? Union transitions among poor cohabiting women. Demography, 43, 223–240. Lichter, D. T., Turner, R. N., & Sassler, S. (2010). National estimates of the rise of serial cohabitation. Social Science Research, 39, 754–765. Manning, W. D. (2013). Trends in cohabitation: Over twenty years of change, 1987–2010 (Family Profile FP-13-12). Bowling Green, OH: Center for Family & Marriage Research, Bowling Green State University. Manning, W. D., Longmore, M. A., & Giordano, P. C. (2007). The changing institution of marriage: Adolescents’ expectations to cohabit and to marry. Journal of Marriage and Family, 69, 559–575. 14 Marcussen, K. (2005). Explaining differences in mental health between married and cohabiting individuals. Social Psychology Quarterly, 68, 239–257. Mernitz, S. E., & Dush, C. K. (2016). Emotional health across the transition to first and second unions among emerging adults. Journal of Family Psychology, 30, 233–244. Musick, K., & Bumpass, L. (2012). Reexamining the case for marriage: Union formation and changes in well-being. Journal of Marriage and Family, 74, 1–18. Ostroff, J., Woolverton, K., Berry, C., & Lesko, L. (1996). Use of the mental health inventory with adolescents: A secondary analysis of the RAND health insurance study. Psychological Assessment, 8, 105–107. Proulx, C. M., Helms, H. M., & Buehler, C. (2007). Marital quality and personal well-being: A meta-analysis. Journal of Marriage and Family, 69, 576–593. Rusbult, C. E. (1980). Commitment and satisfaction in romantic associations: A test of the investment model. Journal of Experimental Social Psychology, 16, 172–186. Rusbult, C. E., Agnew, C., & Arriaga, X. (2011). The investment model of commitment processes. In P. A. M. Van Lange, A. W. Kruglanski, & E. T. Higgins (Eds.), Handbook of theories of social psychology (Vol. 2., pp. 218–231). Thousand Oaks, CA: Sage. Rusbult, C. E., Hannon, P. A., Stocker, S. L., & Finkel, E. J. (2005). Forgiveness and relational repair. In E. L. Worthington Jr. (Ed.), Handbook of forgiveness (pp. 185–205). New York, NY: Routledge. Sandberg-Thoma, S. E., & Kamp Dush, C. M. (2014). Indicators of adolescent mental health and relationship progression in emerging adulthood. Journal of Marriage and Family, 76, 191–206. Saphire-Bernstein, S., & Taylor, S. E. (2013). Close relationships and happiness. In S. A. David, I. Boniwell, & A. C. Ayers (Eds.), The Oxford handbook of happiness (pp.821–833). Oxford, England: Oxford University Press. Sassler, S. (2004). The process of entering into cohabiting unions. Journal of Marriage and Family, 66, 491–505. Sassler, S., & Goldscheider, F. (2004). Revisiting Jane Austen’s theory of marriage timing changes in union formation among American men in the late 20th century. Journal of Family Issues, 25, 139–166. Schneider, D. (2011). Wealth and the marital divide. American Journal of Sociology, 117, 627–667. Seedat, S., Scott, K., Angermeyer, M., Berglund, P., Bromet, E., Brugha, T., . . . Kessler, R. (2009). Cross-national associations between gender and mental disorders in the World Health Organization World Mental Health Surveys. Archives of General Psychiatry, 66, 785–795. Settersten, R. A., & Ray, B. (2010). What’s going on with young people today? The long and twisting path to adulthood. The Future of Children, 20, 19–41. Shulman, S., & Connolly, J. (2013). The challenge of romantic relationships in emerging adulthood: Reconceptualization of the field. Emerging Adulthood, 1, 27–39. Emerging Adulthood Shulman, S., & Connolly, J. (2015). The challenge of romantic relationships in emerging adulthood: Reconceptualization of the field. In J. Arnett (Ed.), The Oxford handbook of emerging adulthood (pp. 230–244). Oxford, England: Oxford University Press. Skinner, K., Bahr, S., Crane, D., & Call, V. (2002). Cohabitation, marriage, and remarriage: A comparison of relationship quality over time. Journal of Family Issues, 23, 74–90. Sigmon, S. T., Pells, J. J., Boulard, N. E., Whitcomb-Smith, S., Edenfield, T. M., Hermann, B. A., & Kubik, E. (2005). Gender differences in self-reports of depression: The response bias hypothesis revisited. Sex Roles, 53, 401–411. Simon, R. W. (2002). Revisiting the relationships among gender, marital status, and mental health. American Journal of Sociology, 107, 1065–1096. Smock, P. J., Manning, W. D., & Porter, M. (2005). “Everything’s there except money”: How money shapes decisions to marry among cohabitors. Journal of Marriage and Family, 67, 680–696. Stanley, S. M., Rhoades, G. K., & Fincham, F. D. (2011). Understanding romantic relationships among emerging adults: The significant roles of cohabitation and ambiguity. In F. Fincham & M. Cui (Eds.), Romantic relationships in emerging adulthood (pp. 234–251). New York, NY: Cambridge University Press. Taylor, S. E. (2011). Social support: A review. In H. S. Friedman (Ed.), The handbook of health psychology (pp. 189–214). New York, NY: Oxford University Press. Taylor, S., Klein, L., Lewis, B., Gruenewald, T., Gurung, R., & Updegraff, J. (2000). Biobehavioral responses to stress in females: Tend-and-befriend, not fight-or flight. Psychological Review, 107, 411–429. Uecker, J. E. (2012). Marriage and mental health among young adults. Journal of Health and Social Behavior, 53, 67–83. Umberson, D., Thomeer, M. B., & Williams, K. (2013). Family status and mental health: Recent advances and future directions. In C. S. Aneshensel, J. C. Phelan, & A. Bierman (Eds.), Handbook of the sociology of mental health (2nd ed., pp. 405–431). New York, NY: Springer. U.S. Census Bureau. (2016). Table MS-2. Estimated median age at first marriage, by sex: 1890 to present. Families and living arrangements. Retrieved February 28, 2017, from https://www. census.gov/hhes/families/files/ms2.csv Veit, C. T., & Ware, J. E. (1983). The structure of psychological distress and well-being in general populations. Journal of Consulting and Clinical Psychology, 51, 730–742. von Hippel, P. T. (2007). Regression with missing ys: An improved strategy for analyzing multiply imputed data. Sociological Methodology, 37, 83–117. Willetts, M. C. (2006). Union quality comparisons between long-term heterosexual cohabitation and legal marriage. Journal of Family Issues, 27, 110–127. Williamson, I., & Gonzales, M. H. (2007). The subjective experience of forgiveness: Positive construals of the forgiveness experience. Journal of Social and Clinical Psychology, 26, 407–446. Mernitz Willis, R. J. (2006). The economics of fatherhood. American Economic Review Papers and Proceedings, 90, 378–382. Willoughby, B., & Carroll, J. (2015). On the horizon: Marriage, timing, beliefs, and consequences in emerging adulthood. In J. Arnett (Ed.), The Oxford handbook of emerging adulthood (pp. 280–295). Oxford, England: Oxford University Press. 15 Author Biography Sara Mernitz is a postdoctoral fellow at the Population Reseach Center at the University of Texas at Austin. Her research broadly focuses on romantic relationships and their longitudinal associations with health. The current manuscript was completed at the Institute for Population Research at The Ohio State University.