Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Sep 1.
Published in final edited form as: J Fam Psychol. 2016 Jun 23;30(6):698–707. doi: 10.1037/fam0000221

Is Marriage a Buzzkill? A Twin Study of Marital Status and Alcohol Consumption

Diana Dinescu 1, Eric Turkheimer 1, Christopher R Beam 2, Erin E Horn 1, Glen Duncan 3, Robert E Emery 1
PMCID: PMC5014643  NIHMSID: NIHMS789897  PMID: 27336180

Abstract

Married adults have consistently been found to drink less than their single or divorced counterparts. This correlation may not be causal, however, as people nonrandomly “select” into marriage and into alcohol use. The current study uses a sample of 2,425 same-sex twin pairs (1,703 MZ; 722 DZ) to control for genetic and shared environmental selection, thereby eliminating a great many third variable, alternative explanations to the hypothesis that marriage causes less drinking. Married twins were compared with their single, divorced, and cohabiting co-twins on drinking frequency and quantity. Married co-twins consumed fewer alcoholic beverages than their single or divorced co-twins, and drank less frequently than their single co-twins. Alcohol use patterns did not differ among married and cohabiting twins. These findings provide strong evidence that intimate relationships cause a decline in alcohol consumption.

Keywords: twin study, behavior genetics, marital status, alcohol consumption

Married adults are better off than their unmarried counterparts in a variety of domains (Waite & Gallagher, 2002) including greater life satisfaction (Uecker, 2012) and increased longevity (Schone & Weinick, 1998), an effect known as the “marriage benefit.” One putative benefit of marriage, and the focus of the present study, is decreased alcohol use and abuse (Liang & Chikritzhs, 2012; Prescott & Kendler, 2001). When compared with married adults, greater alcohol consumption is characteristic of the divorced (Heien & Pompelli, 1987; Linsky, Straus, Colby, & Jr., 1985) and the never married (Power, Rodgers, & Hope, 1999; Simon, 2002). Although reduced alcohol use among married individuals is robustly observed, methodological issues inherent in correlational research cast doubt on the mechanisms behind this so-called marriage benefit regarding its role in decreased alcohol use and abuse.

Social Causation Versus Nonrandom Selection

Marriage advocates assert that marrying is causally related to decreased alcohol use or abuse. Married partners, for example, may monitor each other's behavior, support positive coping, alter each other's values, and encourage delayed gratification more generally (Emery, Horn, & Beam, 2012; Leonard & Eiden, 2007). Alternatively, the apparent benefits of marriage may be explained by nonrandom selection into marriage and alcohol use. For example, it is possible that healthier, wealthier, and better-adjusted individuals make more attractive marital partners and consume less alcohol but for reasons unrelated to marital status (Carr & Springer, 2010). Marital status, like childhood poverty or family structure, cannot be experimentally manipulated, presenting a serious barrier to making causal inferences about its role on alcohol use patterns (Heckman & Smith, 1995; Kenny, 1975). Indeed, married and non-married adults may differ in many measurable and unmeasurable ways, and selection factors may account for some or all of the observed association between marital status and drinking behavior. Many of these confounds may be eliminated by controlling for covariates purportedly associated with both phenotypes, though two important sources of selection remain unaccounted for in cross-sectional data from unrelated participants: genetic and shared environmental selection.

The Utility of Behavior Genetics in the Study of Behavioral Data

Twin studies are a type of natural experiment offering a powerful tool for parsing selection from causation in the analysis of cross-sectional data and permitting more stringent tests of causality (Heath et al., 1993). By comparing twins reared in the same household, who by definition are matched for family and cultural background and genetic predisposition, genetically informed studies can help support causal arguments by controlling for family-level selection factors (i.e., genetic and shared environmental selection). Comparing identical (i.e., monozygotic or MZ) twins discordant for marital status controls for measured and unmeasured genetic and shared environmental selection, including aspects of family life that we can assess only imperfectly (e.g., family of origin dynamics) or not at all (e.g., the entirety of shared childhood experiences) (McGue, Osler, & Christensen, 2010).

It should be noted that associations within twin pairs do not conclusively prove causation, because without randomization it is impossible to control for all potential confounds. Other third variable factors, in addition to marital status, could account for differences in the discordant MZ twins’ drinking behavior. Nevertheless, twin studies do control for genetic and shared environmental confounds at the level of families, which brings them a step closer to causal interpretation than non-genetically informed research (Turkheimer & Harden, 2013).

Review of Genetically Informed Research on Marital Status and Alcohol Use

Twin studies have demonstrated that genetic factors account for variation in both marital status (Horn, Xu, Beam, Turkheimer, & Emery, 2013; Johnson, McGue, Krueger, & Bouchard, 2004; Trumbetta, Markowitz, & Gottesman, 2007) and alcohol use (Dick, Rose, Viken, Kaprio, & Koskenvuo, 2001; Heath, Jardine, & Martin, 1989; Kaprio et al., 1987; Prescott & Kendler, 1999; Prescott et al., 1994). Few twin studies, however, have examined the two together, which is a necessary step to determine whether marital status has an effect on alcohol consumption above and beyond family-level selection factors that might impact both phenotypes.

Studies comparing twins of different marital statuses have yielded discrepant findings. Some researchers have found lower frequency and quantity of alcohol consumption in recently married twins compared with their unmarried co-twins (Prescott & Kendler, 2001). Others have observed no differences between married and divorced co-twins (Osler, McGue, Lund, & Christensen, 2008). It is unclear if the conflicting results are due to factors such as age differences between samples, different ways of defining marital status, or error in measurement. The inconsistent results in prior research and the importance of the topic justify the need for further, genetically-informed research on marital status and alcohol use.

Cohabitation: An Important Consideration

Prior research has focused on the distinction between marital and “non-marital” states (e.g., divorce or singlehood). Throughout the developed world, and increasingly in the U.S., cohabitation is becoming more common as a step towards—and in some cases an alternative to—legal marriage (Cherlin, 2010). A recent twin study showed that having a partner reduced drinking, but marriage was no more protective than cohabitation (Horn et al., 2013). It is critical to replicate and extend such findings to build a knowledge base that adequately reflects societal shifts in union formation. The present study makes an important contribution to this literature by comparing drinking patterns among single, married, cohabiting and divorced individuals.

Gender Differences

Potential gender differences are also of interest, as a recent meta-analysis reported small but significant differences in marital satisfaction between the genders (Jackson, Miller, Oka, & Henry, 2014), and family scholars often consider “his and her marriage,” whereby partners’ experiences of the same union differ (Beam, Marcus, Turkheimer, & Emery, 2016; Jackson et al., 2014). In terms of alcohol use, men benefit more than women from being married (Monin & Clark, 2011), perhaps because wives are hypothesized to moderate their husbands’ drinking (Holmila & Raitasalo, 2005). Husbands’ premarital drinking behavior has been found to predict their wives’ drinking behavior after the first year of marriage (Leonard & Mudar, 2003), suggesting that men may act as “inciters” when it comes to women's drinking after marriage (Holmila & Raitasalo, 2005). Consequently, an association between marital status and alcohol consumption may differ by gender. Examining gender differences in a twin study context will help disentangle the genetic and nonrandom selection effects on alcohol consumption for men and women, controlling for the possibility that women seem to generally experience marriage differently from men, and offer less biased estimates of the effect of marital status on individuals’ drinking behavior.

Drinking Frequency Versus Drinking Quantity

The present study also makes a distinction between drinking frequency and drinking quantity. Prior genetically-informed research found no significant differences between married and cohabiting individuals on a latent alcohol use construct (Horn et al., 2013). However, differentiating between drinking frequency and quantity is important (Thompson, Stockwell, & MacDonald, 2012), as the two behaviors may have different motivations, may occur in different settings, and may transmit different social cues. Understanding these differences carries important clinical and public health implications, as findings could have meaningful treatment implications for different marital status groups. The current study is the first to use a genetically informed design to investigate the effect of marriage and marriage-like relationships on different types of drinking behavior.

The Present Study

We report a study that uses twin data uniquely positioned to help address the question of whether selection or causation mechanisms are at play in any significantly observed associations between marital status and drinking behavior. Twin research is a robust, quasi-experimental tool that allows us to account for the presence of possible selection factors and perform a more accurate analysis of correlation and causation than traditional correlational studies. This method has been used extensively in investigations of observed social and behavioral outcomes, from marriage and family research (Beam et al., 2011; Emery et al., 2012) to BMI and alcohol use (Dinescu, Horn, Duncan, & Turkheimer, 2015; Kaprio, 2015). The current study uses genetically informed phenotypic regression models (Turkheimer & Harden, 2013) to examine the effect of marital status on alcohol consumption in both men and women. Specifically, we examine whether the relationship between marital status (marriage versus divorce, singlehood, and cohabitation) and drinking behavior is accounted for by nonrandom genetic and environmental selection effects, causal processes, or both. We accomplish this by first examining the presence (or absence) of correlations in the full sample and then investigate whether the correlation remains significant after adjusting for genetic and shared environmental effects in twin pairs. Because we are interested in the concept of the “marriage benefit,” we only examine comparisons between married individuals on the one hand, and divorced, single, or cohabiting individuals on the other. Second, we study whether the observed effects of marital status on drinking behavior differ between men and women. We hypothesize that marriage has a protective effect on both types of drinking behavior when compared with various classifications of unmarried individuals and that this benefit is more pronounced for men than for women.

Methods

Sample

Data were drawn from the University of Washington Twin Registry (UWTR; currently known as the Washington State Twin Registry), a community-based sample of Washington State adult twins reared together. Twins were ascertained upon registering with the Washington State Department of Licensing, and were mailed necessary documentation, including a survey. Once one twin completed the survey, the co-twin was recruited and sent the same package. Zygosity was determined through self-report survey questions, which have been shown to have 95-98% accuracy compared with biological markers (Eisen, Neuman, Goldberg, Rice, & True, 1989). For the current study, questionnaires were sent to 83% of the overall registry, with a 64% response rate. More detailed construction methods and response rates are described in detail elsewhere (Afari et al., 2006; Strachan et al., 2013).

The variables used in the current study come from the Health and Wellbeing Questionnaire (year 2010), which assessed various indices of health and behavior such as general lifestyle habits and physical and mental health. The questionnaire includes measures of self-reported drinking, as well as demographic information including marital status. All data are cross-sectional. We used all same-sex twin pairs for which marital status data were available for data analyses (807 male pairs, MZ = 563, DZ = 244, age range 18–95, M = 40.4, SD = 17.6; 1,618 female pairs, MZ = 1,140, DZ = 478, age range 18–97, M = 39.45, SD = 16.6). Less than 1% of the data (n=20 same-sex pairs) were missing marital status information and were excluded from all multivariate analyses. Participants who did not report a marital status did not differ significantly from those who did report marital status information on drinking quantity (t(16)= 0.04, p = 0.96), but significantly differed in drinking frequency (t(17)= 4.73, p < .001), with the missing data group drinking less (M= 0.83) than the nonmissing data group (M=1.79).

The institutional review boards at all participating institutions approved the study, and informed consent was obtained from all participants prior to collection of the information.

Measures

Marital Status

Marital status was determined from a single self-report item: “Are you currently...” with response categories: single; never married; married; widowed; divorced; separated; and living with partner. Twin pairs in which at least one twin was widowed were excluded, since widowhood is not within the scope of the investigation. The divorced and separated groups were combined to test the effects of relationship dissolution on alcohol consumption; this group will henceforth be referred to as “divorced.” The final analysis included four marital status groups: married (N = 2,312), cohabiting (N = 355), divorced (N = 460), and single (N = 1,632).

Alcohol Use

We used two indices of alcohol consumption: self-reported drinking frequency (“How often do you have a drink containing alcohol?”) and drinking quantity (“How many drinks of alcohol do you have on a typical day when you are drinking?”). These items are part of the AUDIT-C, a brief alcohol screen that has been shown to be valid for detecting heavy drinking and/or alcohol abuse (Bush, Kivlahan, McDonnell, Fihn, & Bradley, 1998). Self-reported drinking underestimates actual drinking (Stockwell et al., 2004), so any significant results should be interpreted as conservative estimates.

Drinking frequency was measured using five levels: 0 = never, 1 = monthly or less, 2 = 2-4 times a month, 3 = 2-3 times a week, 4 = 4 or more times a week. Drinking quantity was similarly split into 6 levels: 1 = 1-2 drinks, 2 = 3-4 drinks, 3 = 5-6 drinks, 4 = 7-9 drinks, 5 = 10 or more drinks, 0 = I don't drink.

Data Analysis

We carried out twelve comparisons: married vs. single (never married) twins, married vs. cohabiting twins, and married vs. divorced twins, each conducted separately on drinking frequency and drinking quantity, and separately on males and females. For each comparison, we fit a series of structural equation models using robust weighted least squares (WLSM) estimation in the structural equation modeling software Mplus 7.11 program (Muthén & Muthén, 2013). When there are covariates in the model, WLSM treats data as missing as function of the covariates (Muthén & Muthén, 2010), implementing pairwise deletion of cases with any missing data on the predictor. There was less than 1% reduction in pairs of twins for the multivariate structural equation models presented in this report. We used root mean square error of approximation (RMSEA; Browne & Cudeck, 1992) as a measure of model fit, with values under .05 indicating good model fit and values greater than .05 but less than .08 indicating acceptable model fit. Wald chi-square tests were used when comparing nested models (e.g. a model in which the parameters were constrained to be equal between men and women and a model in which they were allowed to be free) (Wald, 1943).

We began by fitting univariate twin models, which partition the variance of an outcome into three components: additive genetic influences (A), shared environmental influences (C), and non-shared environmental influences (E). Genetic factors (A) represent variation in a phenotype that is associated with variation in genotype, and is correlated 1.0 for monozygotic (MZ) twins and 0.5 for dizygotic (DZ) twins. Shared environmental factors (C) are the cumulative effect of environmental factors that make individuals reared in the same family similar to one another, and by definition are correlated 1.0 for both MZ and DZ twins. Non-shared environmental factors (E) represent twins’ unique life experiences that contribute to their within-pair differences and include measurement error. These factors are uncorrelated between twins. Although we report the univariate variance decomposition of both marital status and alcohol consumption, classical univariate ACE decompositions were not the focus of the present study.

Next, we fit simple phenotypic regression models to the data (see Figure 1), in which we regressed drinking behavior on marital status to obtain an estimate of the observed association between marital status and drinking behavior (parameter bphen in Figure 1). This result represents the regression of marriage on drinking behavior without statistically adjusting for the mediating effects of genetic and shared environmental selection confounds, and is equivalent to a population-level association.

Figure 1.

Figure 1

Open in a new tab

Structural Equation Model Representation of Models Used in Data Analysis Phenotypic model: bphen only, dotted paths absent.

Biometric model: includes paths bA, bC, and b’phen.

Finally, we fit biometric models to the data (Turkheimer & Harden, 2013), in which the effect of marital status on drinking behavior was estimated holding constant any genetic and shared environmental effects common to both marital status and drinking behavior (see Figure 1). In this model, we simultaneously regressed drinking behavior on phenotypic marital status (parameter b'phen in Figure 1) and on the A and C variance components of marital status (parameters bA and bC in Figure 1). If the effect of marital status on drinking behavior remained significantly different from zero after adjusting for the A and C regressions in the model, it would be interpreted as what we refer to as a quasi-causal effect, meaning that in a pair of identical twins discordant for marital status (e.g., one twin is married and the other is divorced) the married twin drinks less or less often than the unmarried co-twin (Turkheimer & Harden, 2013). Conversely, if the regression parameters were no longer statistically significant and was substantially reduced in magnitude after controlling for the genetic and shared environmental effects, a selection hypothesis would be supported. In other words, a pair of identical twins discordant for marriage would demonstrate similar drinking behaviors.

To test for gender differences, we tested whether the bphen parameters could be constrained to be equal across gender without significant decrement in model fit. Age was included as a covariate to account for the possibility that drinking behavior and marital status change with age. Age is controlled by design in comparisons within pairs of twins.

Results

Descriptive Results

Frequencies of the four marital statuses, as well as means and standard deviations of the self-reported drinking variables are presented in Table 1 by gender. Of the four groups studied, single adults reported drinking least frequently and cohabiting individuals most frequently. In terms of quantity, cohabiting men and women reported the highest drinking levels, while married individuals reported drinking the least. Women reported drinking less than men in terms of both frequency and quantity.

Table 1.

Frequencies, means, and SDs of variables of interest, by gender

Variable
 Male
 Female
Frequency
    Married 703 1605
    Single 584 1040
    Cohabiting 92 263
    Divorced 105 336
Mean (SD)
Drinking frequencya
    Married 2.15 (1.42) 1.66 (1.35)
    Single 1.78 (1.37) 1.64 (1.18)
    Cohabiting 2.35 (1.41) 1.95 (1.24)
    Divorced 2.25 (1.44) 1.73 (1.33)
Drinking quantityb
    Married 1.03 (0.76) 0.86 ( 0.65)
    Single 1.37 (1.22) 1.18 (0.94)
    Cohabiting 1.49 (1.19) 1.24 (0.82)
    Divorced 1.38 (1.15) 0.96 (0.69)
Open in a new tab
a

categorical variable, scale of 0-4 (see Methods section for description)

b

categorical variable, scale of 0-5 (see Methods section for description)

Univariate ACE Decompositions

The univariate ACE decompositions for marital status and drinking behavior are presented in Table 2. Variance component estimates that were negative were subsequently set to zero. All marital status and alcohol phenotypes included some between-family variation, either genetic or shared environmental, in addition to significant nonshared environmental variance. There was substantial shared environmental variance in five out of the six combinations of three marital status comparisons and two genders. Additive genetic contributions to marital status were much smaller, and not significantly different from zero in most comparisons. The univariate ACE decomposition of alcohol use showed a combination of genetic and shared environmental variance. There was substantial nonshared environmental variance in all comparisons.

Table 2.

ACE Variance Components of variables of interest, by gender

Male
 Female
Phenotype
 A
 C
 E
 A
 C
 E
Marital Status
Married vs. Single 0.08 (0.10) 0.82 (0.09) 0.10 (0.02) 0.09 (0.09) 0.78 (0.09) 0.13 (0.02)
Married vs. Cohabiting - 0.70 (0.10) 0.30 (0.10) 0.09 (0.36) 0.41 (0.31) 0.49 (0.09)
Married vs. Divorced - 0.37 (0.10) 0.62 (0.10) 0.29 (0.08) - 0.71 (0.08)
Alcohol Use
Drinking frequency 0.27 (0.08) 0.14 (0.08) 0.59 (0.02) 0.40 (0.06) 0.07 (0.06) 0.52 (0.02)
Drinking quantity 0.19 (0.12) 0.33 (0.11) 0.49 (0.03) 0.21 (0.08) 0.42 (0.07) 0.36 (0.02)
Open in a new tab

Note. Bolded values are significant at <.05;

Phenotypic and Biometric Models

Drinking Frequency

Table 3 presents the results (estimates and standard errors) of the phenotypic and biometric models in which drinking frequency was regressed on marital status.

Table 3.

Unstandardized Parameter Estimates for Phenotypic and Biometric Models for Drinking Frequency

Drinking frequency
Married vs. Single
 Married vs. Cohabiting
 Married vs. Divorced
Estimate (se) Male Female Male Female Male Female
Regression coefficients
        Phenotypic model
    bphen 0.02 (0.03) 0.02 (0.03) −0.2 (0.05)* −0.2 (0.05)* −0.03 (0.04) −0.03 (0.04)
        Biometric model
    bA 1.58 (0.7)* 0.13 (1.82) 1.68 (7.6) −0.55 (3.4) - −0.05 (0.2)
    bC −0.2 (0.5) 0.3 (0.6) −0.48 (0.88) −0.18 (0.76) 0.23 (0.21) -
    b′phen −0.1 (0.03) −0.1 (0.03)* −0.07 (0.08) −0.07 (0.08) −0.04 (0.05) −0.04 (0.05)
Covariates
    Age 0.15 (0.02) 0.06 (0.02) 0.12 (0.2) 0.04 (0.02) 0.09 (0.01) 0.02 (0.02)
Goodness of Fit
    RMSEA (CFI/TLI) 0.02 (0.999/0.999) 0.02 (.982/.985) 0.01 (.997/.997)
Open in a new tab

Note. Bolded values are significant at <.05; bphen is the full phenotypic effect; b′phen is the genetically informed phenotypic effect; bA and bC are the indirect effects of marital status on the phenotype.

Married vs. Single.

In the phenotypic model, the drinking frequency of married individuals did not differ significantly from that of their single counterparts (bphen = 0.02, SE = 0.03). Moreover, there was no significant difference in bphen between genders (χ2= 1.73, df = 1, p = 0.19). In the biometric regression model, however, we found evidence of a suppression effect in males, whereby the protective influence of marriage was masked by a large genetic confound in the opposite direction from b'phen (bA = 1.58, SE = 0.70, b’phen = −0.10, SE = 0.03). In other words, marriage is significantly related to lower drinking frequency within pairs of identical twins, but this effect is not evident in population level studies because of genetic confounds in the opposite direction. This suppression effect is illustrated in the MZ males panel at the upper left of Figure 2. Concordant single pairs of MZ twins (i.e., both members of the pair are single) in the lighter, rightmost bar of the panel, drink significantly less frequently than concordant married pairs of married MZ twins, in the darker, leftmost bar. This comparison, however, includes between-pair genetic and environmental confounds of the relationship between marital status and drinking. In contrast, within pairs of MZ twins discordant for marital status (middle bars of the panel), with the between-pair confounds controlled, the married twin, represented by the darker shade, drinks less frequently than his single co-twin, represented by the lighter shade. Genetic confounds can also be observed by comparing the MZ and DZ panels (top row panels of Figure 2). In discordant pairs of twins, MZ married twins drink less than their single co-twins, while DZ married twins in discordant pairs drink more than their single co-twins. This difference is also explained by a genetic confound, because the within-pair difference between DZ twins includes the genetic difference between them. As was the case with the phenotypic model, no gender difference in b'phen was observed (χ2 = 0.07, df = 1, p = 0.79).

Figure 2.

Figure 2

Open in a new tab

Illustration of quasi-causation model results Married vs. Single male MZ and DZ twins Drinking Frequency and Drinking Quantity

Married vs. Cohabiting

In the phenotypic model, we observed a significant effect of marital status on drinking frequency, in which married twins drank significantly less often than cohabiting twins (bphen = −0.20, SE = 0.05). There was no significant difference in bphen between men and women (χ2 = 0.70, df = 1, p = 0.40). However, when between-family confounds were controlled, the effect became smaller and non-significant (b’phen = −0.07, SE = 0.08), suggesting that genetic and shared environmental factors confounded the effect of marriage on drinking frequency in this group comparison. Again, no gender differences in b’phen were found (χ2 = 0.25, df = 1, p = 0.61).

Married vs. Divorced

We found no evidence of an effect of marriage on drinking frequency when comparing married and divorced twins. This was the case in both the phenotypic model (bphen = −0.03, SE = 0.04) and the biometric model (b'phen = −0.04, SE = 0.05). Gender differences were similarly absent in the phenotypic (χ2 = 0.53, df = 1, p = 0.47) and biometric (χ2 = 2.09, df = 1, p = 0.15) models.

Drinking Quantity

Table 4 presents the results (estimates and standard errors) of the phenotypic and biometric models in which drinking quantity was regressed on marital status.

Table 4.

Unstandardized Parameter Estimates for Phenotypic and Biometric Models for Drinking Quantity

Drinking quantity
Married vs. Single
 Married vs. Cohabiting
 Married vs. Divorced
Estimate (se) Male Female Male Female Male Female
Regression coefficients
        Phenotypic model
    bphen −0.3 (0.03)* −0.08 (0.02)* −0.51 (0.03)* −0.17 (0.03)* −0.5 (0.03)* −0.1 (0.03)*
        Biometric model
    bA 2.79 (2.65) 0.64 (1.29) - - −1.58 (1.96) −0.08 (0.17)
    bC −0.43 (0.37) −0.34 (0.35) −1.2 (0.11)* −0.3 (0.13)* 0.33 (0.87) -
    b′phen −0.15 (0.04)* −0.06 (0.03)* 0.21 (0.09)* −0.06 (0.04) −0.1 (0.04)* −0.1 (0.04)*
Covariates
    Age 0.01 (0.02) −0.08 (0.02) −0.1 (0.02) −0.09 (0.01) −0.06 (0.01) −0.11 (0.01)
Goodness of Fit
    RMSEA (CFI/TLI) 0.03 (.995/.996) 0.04 (.964/.971) 0.03 (.972 /.977)
Open in a new tab

Note. Bolded values are significant at <.05; bphen is the full phenotypic effect; b′phen is the genetically informed phenotypic effect; bA and bC are the indirect effects of marital status on the phenotype.

Married vs. Single

In the phenotypic model, married men drank significantly less than their single counterparts (bphen = −0.30, SE = 0.03). When comparing the effect of marital status on drinking quantity in men versus women, we found a smaller—but still significant—effect in women (bphen = −0.08, SE = 0.02), and a significant difference in bphen between men and women (χ2 = 29.3, df = 1, p < 0.001). The effect remained significant after adjusting for genetic and shared environmental confounds in the biometric models for both men (b’phen = −0.15, SE = 0.04) and women (b’phen = −0.06, SE = 0.03). There appears to be a protective effect of marriage with respect to drinking quantity. This effect is apparent in Figure 2, MZ Males panel in the lower left. Concordant pairs of married MZ twins, in the darker, leftmost bar, drink significantly less than concordant pairs of single MZ twins, in the lighter, rightmost bar. The effect is even more pronounced within pairs of MZ twins discordant for marital status, which suggests that genetic and shared environmental confounds could not be responsible for it, as these are controlled in a comparison within pairs of MZ twins. Thus, this analysis offers robust evidence for a quasi-causal effect of marital status on drinking quantity. Moreover, there was a significant difference in bphen between genders in this model (χ2 = 3.28, df = 1, p = 0.07), suggesting that the effect of marriage on drinking quantity is greater in men than in women.

Married vs. Cohabiting

The phenotypic model for men suggested that married men drank significantly less than cohabiting men (bphen = −0.51, SE = 0.03). However, this association appeared to be due to the combination of a significant negative shared environmental confound (bC = −1.2, SE = 0.11) and a significant positive quasi-causal effect (b’phen = 0.21, SE = 0.09), suggesting that married men drank less than their cohabiting counterparts for shared environmental reasons, and not because of their marital status per se. Married MZ male twins drank significantly more per occasion than their cohabiting co-twins. For women, the effect of marital status on drinking quantity was also mediated by nonrandom selection, as evidenced by the substantially reduced, no longer statistically significant effect of marriage on drinking quantity after controlling for genetic and shared environmental confounds (bphen = −0.17, SE = 0.03; b’phen = −0.06, SE = 0.04); as was the case with men, the protective effect of marriage appeared to be due to an underlying shared environmental selection effect (parameter estimates?). Significant gender differences in regression parameters were observed both in the phenotypic (χ2 = 75.04, df = 1, p < 0.001) and in the biometric (χ2 = 7.88, df = 1, p = 0.005) models, suggesting that the effects of marriage versus cohabitation differ significantly between men and women, such that the effect is weaker or nonexistent in women relative to men.

Married vs. Divorced

In the phenotypic model, married men and women drank significantly less than their divorced counterparts, more so for men (bphen = −0.50, SE = 0.03) than for women (bphen = −0.10, SE = 0.03; gender test: χ2 = 82.25, df = 1, p < 0.001). When controlling for between-family confounds, there was a protective effect of marriage such that married twins drank less than their divorced counterparts (b’phen = −0.10, SE = 0.04). This quasi-causal effect was not significantly different between men and women (χ2 = 0.13, df = 1, p = 0.71).

Discussion

The results of the current study advance knowledge on the associations between a variety of marital status categories and two alcohol use measures, drinking frequency and drinking quantity. We used a genetically informed design to parse selection from potential causation in this relation, and predicted that marriage (compared with divorce, singlehood, or cohabitation) would be associated with drinking less frequently and consuming fewer drinks on each occasion. We also expected this protective effect to be more pronounced for men than for women.

Findings

At the full-sample level, married individuals drink less frequently than their cohabiting counterparts, but as often when compared to their single or divorced counterparts. However, when controlling for genetic and shared environmental confounds, marriage appears to have a protective effect in married compared to single twins. In other words, single and married twins differ in the frequency of their drinking, but this difference is not observable in the overall UWTR sample because of genetic selection confounds. This suggests that individuals who drink less frequently tend to marry, while individuals who drink more frequently are more likely to remain single. Within twin pairs, there is no difference between married and cohabiting or married and divorced twins with regard to drinking frequency.

Findings for drinking quantity revealed a more consistent protective effect of marriage. In comparison to both single and divorced adults, married adults drank significantly fewer alcoholic beverages per occasion before and after controlling for genetic and shared environmental selection. These results suggest that people do change their drinking quantity habits when they enter, and when they exit, marriage. They drink more when they are single, less when they are married, and again more following marital dissolution.

In contrast, shared environmental selection effects were the best explanation of differences between married and cohabiting twins in how much they drank. After controlling for genetic and environmental selection effects, the differences between marriage and cohabitation were non-significant for women and, interestingly, produced a contrast effect in men in which married men drank more than unmarried ones. In other words, although there was a significant difference in quantity of drinking among marital status groups when examined in the uncontrolled sample, with cohabiting individuals drinking more than their married counterparts, cohabiting male men actually drink less than their married co-twins and cohabiting women drank just as much as their married co-twins controlling for genetic and environmental selection confounds.

A more nuanced picture emerges combining the results from our drinking frequency and drinking quantity analyses. Our evidence suggests that individuals drink the most, in frequency and quantity, when they are single. Drinking goes down significantly, both in frequency and quantity, after marriage. Following marital dissolution, both men and women experience and increase in drinking quantity, but their drinking frequency does not change, possibly because their drinking frequency habits stabilize after having been married. When compared to marriage, cohabitation presents an interesting picture. Cohabiting men and women drink more frequently than married men and women, but quantity-wise cohabiting men drink less while women do not differ significantly.

Replicating one other study examining marriage and alcohol use (Horn et al., 2013), the similar findings for marriage and cohabitation suggest that the relationship itself, not the legal status of marriage, may account for the benefits of partnership on drinking quantity. Since cohabitation generally is a step toward rather than an alternative to marriage in the U.S. (Cherlin, 2010), future research should compare the relative protective effect of cohabitation among those who intend to marry versus those who do not. People's future relationship goals and plans may influence relationship dynamics. In fact, one study found that drinking declined in the year prior to marriage for both men and women (Miller-Tutzauer, Leonard, & Windle, 1991).

The comparison between married and cohabiting twins is also noteworthy for demonstrating the value of genetically informed designs. In the full sample without twin controls, we observed that married men and women drank less than their cohabiting counterparts. The effect, however, was greatly reduced after controlling for nonrandom genetic and environmental selection effects. Put differently, genetic and shared environmental selection effects accounted for the observed association. The individuals who drink less tend to be the same people who are more likely to marry rather than cohabit. While other nonshared experiences cannot be ruled out as possible explanations for these findings, this genetically-informed study provides evidence that marriage has a quasi-causal effect on reducing drinking quantity for both men and women. In other words, an individual-specific environmental experience has occurred to only one twin and not the other, and has impacted that twin's drinking behavior. In our design, twin pairs were chosen such that one twin is married and the other one has a different relationship status, which offers the best evidence to date that marriage was a factor that influenced the outcome of interest – in our case, drinking behavior.

Limitations

Despite the above strengths, the present study also has limitations. First, the data used were cross-sectional. Ideally, genetically informed research should use longitudinal data to make longitudinal predictions. Additionally, while the sample was community based, it was primarily white (92.3%), which makes it impossible to generalize the present findings to non-white populations. Moreover, there is no available information on how study responders differed from non-responders, so it is possible that the results are biased by self-selection. Finally, future studies might compare cohabiting, single, and divorced adults, which the current study did not attempt to do.

Significance

Notwithstanding these limitations, the present study is the first genetically informed study of marriage to examine drinking frequency and drinking quantity separately. The study makes important contributions to prior longitudinal observational work. Longitudinal studies suggest that marriage may impact drinking behavior through such mechanisms as changing one's values and beliefs (Leonard & Eiden, 2007). However, it is impossible to ascertain whether married participants were exposed to unmeasured influences, such as an upbringing that predisposed them to be more open-minded and thus accept changes in values that facilitated their selection into both marriage and positive drinking behaviors. Twin designs permit the statistical control for genetic and shared environmental selection, which provide wholesale elimination of two broad categories of possible confounds – genetic and shared environmental selection factors. Using a twin design, we found that marriage (and cohabitation) had a protective effect on drinking behavior, particularly drinking quantity. This strong evidence confirms and enhances prior findings illustrating that relationship status directly influences drinking, particularly among men.

In addition to providing support for protective effects of marriage, these results underscore the importance of distinguishing between drinking frequency and drinking quantity. Married and unmarried adults did not differ much in terms of drinking frequency, although we suspect that they differ in where they imbibe (e.g., at a bar versus at home). Yet, these groups clearly differ in drinking quantity, which also may be linked to the social circumstances of consumption (which we could not examine in the present research). This difference is key, because excessive quantities of alcohol consumption are widely known to have detrimental effects on physical health and relationships (Crane, Testa, Derrick, & Leonard, 2014; Rehm et al., 2003; Rodriguez, Neighbors, & Knee, 2014). In fact, the detrimental effect of excessive alcohol consumption on relationships may explain one benefit of marital and cohabiting relationships. Married couples may make an effort not to drink too much in order to avoid conflict, including conflict due to their partner's reminders about not drinking excessively. Increased health behavior monitoring may be one of the benefits of marriage (Kiecolt-Glaser & Newton, 2001; Schone & Weinick, 1998), with wives more often monitoring their husbands (Holmila & Raitasalo, 2005).

Implications

The present study has implications for public health initiatives and clinical interventions. Recent years have seen a rise in marriage promotion and advocacy: in 2005, the Marriage and Fatherhood Provisions of the Deficit Reduction Act included financial support for marriage education and support of married couples, and in 2013, civil rights advocacy culminated in the federal extension of marriage rights to the LGBT community. Our findings that marriage and marriage-like relationships may not simply correlate to healthier behaviors, but make a causal contribution to protecting against potentially harmful behaviors, highlight the critical public health implications of these efforts. Clinically, treatment interventions already consist of screening and customizing treatment based on people's known risk factors, such as blood pressure in the prevention and treatment of cardiovascular diseases (Kannel, 1996) or being an athlete in the treatment of eating disorders (Bratland-Sanda & Sundgot-Borgen, 2013). Our findings are a potential first step towards customizing interventions based on marital status and the availability of not only social support, but of a stable close relationship.

Acknowledgments

This work was supported by the National Institute of Child Health & Human Development (1R01HD056354-01) and the National Institute on Aging (F31AG044047-01A1, T32AG020500, T32AG000037-37). The National Institute of Child Health & Human Development and the National Institute of Aging played no role in the study design; the collection, analysis, or interpretation of data; the drafting of this manuscript; or the decision to submit it for publication.

References

  1. Afari N, Noonan C, Goldberg J, Edwards K, Gadepalli K, Osterman B, Buchwald D. University of Washington Twin Registry: construction and characteristics of a community-based twin registry. Twin Research and Human Genetics : The Official Journal of the International Society for Twin Studies. 2006;9(6):1023–1029. doi: 10.1375/183242706779462543. http://doi.org/http://dx.doi.org/10.1375/twin.9.6.1023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Beam CR, Horn EE, Hunt SK, Emery RE, Turkheimer E, Martin N. Revisiting the effect of marital support on depressive symptoms in mothers and fathers: a genetically informed study. Journal of Family Psychology. 2011;25(3):336–344. doi: 10.1037/a0023758. http://doi.org/10.1037/a0023758. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Beam CR, Marcus K, Turkheimer E, Emery RE. Gender Differences in Interpretations of Marital Quality: Application of the IFACE model. 2016 Under Review. [Google Scholar]
  4. Bratland-Sanda S, Sundgot-Borgen J. Eating disorders in athletes: overview of prevalence, risk factors and recommendations for prevention and treatment. European Journal of Sport Science. 2013;13(5):499–508. doi: 10.1080/17461391.2012.740504. http://doi.org/10.1080/17461391.2012.740504. [DOI] [PubMed] [Google Scholar]
  5. Browne MW, Cudeck R. Alternative Ways of Assessing Model Fit. Sociological Methods & Research. 1992;21(2):230–258. http://doi.org/10.1177/0049124192021002005. [Google Scholar]
  6. Bush K, Kivlahan D, McDonnell M, Fihn S, Bradley K. The AUDIT Alcohol Consumption Questions (AUDIT-C). An Effective Brief Screening Test for Problem Drinking. Archives of Internal Medicine. 1998;158(16):1789. doi: 10.1001/archinte.158.16.1789. http://doi.org/10.1001/archinte.158.16.1789. [DOI] [PubMed] [Google Scholar]
  7. Carr D, Springer KW. Advances in Families and Health Research in the 21st Century. Journal of Marriage and Family. 2010;72(3):743–761. http://doi.org/10.1111/j.1741-3737.2010.00728.x. [Google Scholar]
  8. Cherlin A. Demographic Trends in the United States: A Review of Research in the 2000s. Journal of Marriage and the Family. 2010;72(3):403–419. doi: 10.1111/j.1741-3737.2010.00710.x. http://doi.org/10.1111/j.1741-3737.2010.00710.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Crane CA, Testa M, Derrick JL, Leonard KE. Daily associations among self-control, heavy episodic drinking, and relationship functioning: An examination of actor and partner effects. Aggressive Behavior. 2014;40(5):440–50. doi: 10.1002/ab.21533. http://doi.org/10.1002/ab.21533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dick DM, Rose RJ, Viken RJ, Kaprio J, Koskenvuo M. Exploring gene-environment interactions: Socioregional moderation of alcohol use. Journal of Abnormal Psychology. 2001;110(4):625–632. doi: 10.1037//0021-843x.110.4.625. http://doi.org/10.1037//0021-843X.110.4.625. [DOI] [PubMed] [Google Scholar]
  11. Dinescu D, Horn EE, Duncan G, Turkheimer E. Socioeconomic Modifiers of Genetic and Environmental Influences on Body Mass Index in Adult Twins. Health Psychology. 2015 doi: 10.1037/hea0000255. http://doi.org/10.1037/hea0000255. [DOI] [PMC free article] [PubMed]
  12. Eisen S, Neuman R, Goldberg J, Rice J, True W. Determining zygosity in the Vietnam Era Twin Registry: an approach using questionnaires. Clinical Genetics. 1989;35(6):423–432. doi: 10.1111/j.1399-0004.1989.tb02967.x. http://doi.org/10.1111/j.1399-0004.1989.tb02967.x. [DOI] [PubMed] [Google Scholar]
  13. Emery RE, Horn EE, Beam CR. Marriage and Improved Well-Being: Using Twins to Parse the Correlation, Asking How Marriage Helps, and Wondering Why More People Don' t Buy a Bargain. Marriage at the Crossroads: Law, Policy, and the Brave New World of Twenty-First-Century Families. 2012:126–141. [Google Scholar]
  14. Heath AC, Jardine R, Martin NG. Interactive effects of genotype and social environment on alcohol consumption in female twins. Journal of Studies on Alcohol. 1989;50(1):38–48. doi: 10.15288/jsa.1989.50.38. http://doi.org/http://dx.doi.org/10.15288/jsa.1989.50.38. [DOI] [PubMed] [Google Scholar]
  15. Heath AC, Kessler RC, Neale MC, Hewitt JK, Eaves LJ, Kendler KS. Testing hypotheses about direction of causation using cross-sectional family data. Behavior Genetics. 1993;23(1):29–50. doi: 10.1007/BF01067552. http://doi.org/10.1007/BF01067552. [DOI] [PubMed] [Google Scholar]
  16. Heckman JJ, Smith JA. Assessing the case for social experiments. The Journal of Economic Perspectives. 1995;9(2):85–110. http://doi.org/10.1257/jep.9.2.85. [Google Scholar]
  17. Heien D, Pompelli G. Stress, Ethnic and Distribution Factors in a Dichotomous Response Model of Alcohol Abuse. Journal of Studies on Alcohol and Drugs. 1987;48(05):450. doi: 10.15288/jsa.1987.48.450. http://doi.org/http://dx.doi.org/10.15288/jsa.1987.48.450. [DOI] [PubMed] [Google Scholar]
  18. Holmila M, Raitasalo K. Gender differences in drinking: why do they still exist? Addiction (Abingdon, England) 2005;100(12):1763–1769. doi: 10.1111/j.1360-0443.2005.01249.x. http://doi.org/10.1111/j.1360-0443.2005.01249.x. [DOI] [PubMed] [Google Scholar]
  19. Horn EE, Xu Y, Beam CR, Turkheimer E, Emery RE. Accounting for the physical and mental health benefits of entry into marriage: a genetically informed study of selection and causation. Journal of Family Psychology. 2013;27(1):30–41. doi: 10.1037/a0029803. http://doi.org/10.1037/a0029803. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jackson J, Miller R, Oka M, Henry R. Gender Differences in Marital Satisfaction : A Meta-analysis. Journal of Marriage and Family. 2014 Feburary;76:105–129. http://doi.org/10.1111/jomf.12077. [Google Scholar]
  21. Johnson W, McGue M, Krueger RF, Bouchard TJ. Marriage and personality: a genetic analysis. Journal of Personality and Social Psychology. 2004;86(2):285–294. doi: 10.1037/0022-3514.86.2.285. http://doi.org/10.1037/0022-3514.86.2.285. [DOI] [PubMed] [Google Scholar]
  22. Kannel WB. Blood Pressure as a Cardiovascular Risk Factor. JAMA. 1996;275(20):1571. http://doi.org/10.1001/jama.1996.03530440051036. [PubMed] [Google Scholar]
  23. Kaprio J. Is the association between alcohol use and coronary artery disease causal? Evidence from a long-term twin study. The American Journal of Clinical Nutrition. 2015;102(1):1–2. doi: 10.3945/ajcn.115.114348. http://doi.org/10.3945/ajcn.115.114348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kaprio J, Koskenvuo M, Langinvainio H, Romanov K, Sarna S, Rose RJ. Genetic influences on use and abuse of alcohol: a study of 5638 adult Finnish twin brothers. Alcoholism, Clinical and Experimental Research. 1987;11(4):349–356. doi: 10.1111/j.1530-0277.1987.tb01324.x. http://doi.org/10.1111/j.1530-0277.1987.tb01324.x. [DOI] [PubMed] [Google Scholar]
  25. Kenny DA. Cross-lagged panel correlation: A test for spuriousness. Psychological Bulletin. 1975;82(6):887–903. http://doi.org/10.1037/0033-2909.82.6.887. [Google Scholar]
  26. Kiecolt-Glaser JK, Newton TL. Marriage and health: his and hers. Psychological Bulletin. 2001;127(4):472–503. doi: 10.1037/0033-2909.127.4.472. http://doi.org/10.1037/0033-2909.127.4.472. [DOI] [PubMed] [Google Scholar]
  27. Leonard KE, Eiden RD. Marital and family processes in the context of alcohol use and alcohol disorders. Annual Review of Clinical Psychology. 2007;3:285–310. doi: 10.1146/annurev.clinpsy.3.022806.091424. http://doi.org/10.1146/annurev.clinpsy.3.022806.091424. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Leonard KE, Mudar P. Peer and partner drinking and the transition to marriage: A longitudinal examination of selection and influence processes. Psychology of Addictive Behaviors. 2003;17(2):115–125. doi: 10.1037/0893-164x.17.2.115. http://doi.org/10.1037/0893-164X.17.2.115. [DOI] [PubMed] [Google Scholar]
  29. Liang W, Chikritzhs T. Brief report: marital status and alcohol consumption behaviours. Journal of Substance Use. 2012;17(1):84–90. http://doi.org/10.3109/14659891.2010.538463. [Google Scholar]
  30. Linsky AS, Straus MA, Colby JP., Jr. Stressful Events, Stressful Conditions and Alcohol Problems in the United States: A Partial Test of Bales's Theory. Journal of Studies on Alcohol and Drugs. 1985;46(01):72. doi: 10.15288/jsa.1985.46.72. http://doi.org/http://dx.doi.org/10.15288/jsa.1985.46.72. [DOI] [PubMed] [Google Scholar]
  31. McGue M, Osler M, Christensen K. Causal Inference and Observational Research: The Utility of Twins. Perspectives on Psychological Science : A Journal of the Association for Psychological Science. 2010;5(5):546–56. doi: 10.1177/1745691610383511. http://doi.org/10.1177/1745691610383511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Miller-Tutzauer C, Leonard KE, Windle M. Marriage and alcohol use: a longitudinal study of “maturing out”. Journal of Studies on Alcohol. 1991;52(5):434–440. doi: 10.15288/jsa.1991.52.434. http://doi.org/http://dx.doi.org/10.15288/jsa.1991.52.434. [DOI] [PubMed] [Google Scholar]
  33. Monin JK, Clark MS. Why Do Men Benefit More from Marriage Than Do Women? Thinking More Broadly About Interpersonal Processes That Occur Within and Outside of Marriage. Sex Roles. 2011;65(5-6):320–326. http://doi.org/10.1007/s11199-011-0008-3. [Google Scholar]
  34. Muthén L, Muthén B. Mplus user's guide. 6th ed. Muthén & Muthén; Los Angeles, CA: 2010. [Google Scholar]
  35. Osler M, McGue M, Lund R, Christensen K. Marital status and twins’ health and behavior: an analysis of middle-aged Danish twins. Psychosomatic Medicine. 2008;70(4):482–487. doi: 10.1097/PSY.0b013e31816f857b. http://doi.org/10.1097/PSY.0b013e31816f857b. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Power C, Rodgers B, Hope S. Heavy alcohol consumption and marital status: disentangling the relationship in a national study of young adults. Addiction (Abingdon, England) 1999;94(10):1477–87. doi: 10.1046/j.1360-0443.1999.941014774.x. http://doi.org/10.1046/j.1360-0443.1999.941014774.x. [DOI] [PubMed] [Google Scholar]
  37. Prescott CA, Hewitt JK, Heath AC, Truett KR, Neale MC, Eaves LJ. Environmental and Genetic Influences on Alcohol Use in a Volunteer Sample of Older Twins. Journal of Studies on Alcohol. 1994;156(1):34–40. doi: 10.15288/jsa.1994.55.18. http://doi.org/http://dx.doi.org/10.15288/jsa.1994.55.18. [DOI] [PubMed] [Google Scholar]
  38. Prescott CA, Kendler KS. Genetic and environmental contributions to alcohol abuse and dependence in a population-based sample of male twins. The American Journal of Psychiatry. 1999;156(1):34–40. doi: 10.1176/ajp.156.1.34. http://doi.org/http://dx.doi.org/10.1176/ajp.156.1.34. [DOI] [PubMed] [Google Scholar]
  39. Prescott CA, Kendler KS. Associations between marital status and alcohol consumption in a longitudinal study of female twins. Journal of Studies on Alcohol. 2001;62(5):589–604. doi: 10.15288/jsa.2001.62.589. http://doi.org/http://dx.doi.org/10.15288/jsa.2001.62.589. [DOI] [PubMed] [Google Scholar]
  40. Rehm J, Room R, Graham K, Monteiro M, Gmel G, Sempos CT. The relationship of average volume of alcohol consumption and patterns of drinking to burden of disease: an overview. Addiction. 2003;98(9):1209–1228. doi: 10.1046/j.1360-0443.2003.00467.x. http://doi.org/10.1046/j.1360-0443.2003.00467.x. [DOI] [PubMed] [Google Scholar]
  41. Rodriguez LM, Neighbors C, Knee CR. Problematic alcohol use and marital distress: An interdependence theory perspective. Addiction Research & Theory. 2014;22(4):294–312. http://doi.org/10.3109/16066359.2013.841890. [Google Scholar]
  42. Schone BS, Weinick RM. Health-Related Behaviors and the Benefits of Marriage for Elderly Persons. The Gerontologist. 1998;38(5):618–627. doi: 10.1093/geront/38.5.618. http://doi.org/10.1093/geront/38.5.618. [DOI] [PubMed] [Google Scholar]
  43. Simon RW. Revisiting the Relationships among Gender, Marital Status, and Mental Health. American Journal of Sociology. 2002;107(4):1065–1096. doi: 10.1086/339225. http://doi.org/10.1086/339225. [DOI] [PubMed] [Google Scholar]
  44. Stockwell T, Donath S, Cooper-Stanbury M, Chikritzhs T, Catalano P, Mateo C. Under-reporting of alcohol consumption in household surveys: a comparison of quantity-frequency, graduated-frequency and recent recall. Addiction (Abingdon, England) 2004;99(8):1024–1033. doi: 10.1111/j.1360-0443.2004.00815.x. http://doi.org/10.1111/j.1360-0443.2004.00815.x. [DOI] [PubMed] [Google Scholar]
  45. Strachan E, Hunt C, Afari N, Duncan G, Noonan C, Schur E, Buchwald D. University of Washington twin registry: Poised for the next generation of twin research. Twin Research and Human Genetics. 2013;16(1):455–462. doi: 10.1017/thg.2012.124. http://doi.org/10.1017/thg.2012.124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Thompson KD, Stockwell T, MacDonald S. Is there a “low-risk” drinking level for youth? The risk of acute harm as a function of quantity and frequency of drinking. Drug and Alcohol Review. 2012;31(2):184–193. doi: 10.1111/j.1465-3362.2011.00378.x. http://doi.org/10.1111/j.1465-3362.2011.00378.x. [DOI] [PubMed] [Google Scholar]
  47. Trumbetta SL, Markowitz EM, Gottesman II. Marriage and genetic variation across the lifespan: not a steady relationship? Behavior Genetics. 2007;37(2):362–375. doi: 10.1007/s10519-006-9132-1. http://doi.org/10.1007/s10519-006-9132-1. [DOI] [PubMed] [Google Scholar]
  48. Turkheimer E, Harden KP. Handbook of Research Methods in Social and Personality Psychology. Cambridge University Press; 2013. Behavior genetic research methods: Testing quasi-causal hypotheses using multivariate twin data. pp. 159–187. [Google Scholar]
  49. Uecker JE. Marriage and mental health among young adults. Journal of Health and Social Behavior. 2012;53(1):67–83. doi: 10.1177/0022146511419206. http://doi.org/10.1177/0022146511419206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Waite L, Gallagher M. The Case for Marriage: Why Married People are Happier, Healthier and Better Off Financially. Random House LLC; 2002. [Google Scholar]
  51. Wald A. Tests of statistical hypotheses concerning several parameters when the number of observations is large. Transactions of the American Mathematical Society. 1943;54(3):426–426. http://doi.org/10.1090/S0002-9947-1943-0012401-3. [Google Scholar]

RESOURCES