- Email: [email protected]
Effects of relationship duration, cohabitation, and marriage on the frequency of intercourse in couples: Findings from German panel data
Social Science Research 52 (2015) 72–82
Contents lists available at ScienceDirect
Social Science Research journal homepage: www.elsevier.com/locate/ssresearch
Effects of relationship duration, cohabitation, and marriage on the frequency of intercourse in couples: Findings from German panel data Jette Schröder a,⇑, Claudia Schmiedeberg b a b
GESIS – Leibniz Institute for the Social Sciences, P.O. Box 122155, 68072 Mannheim, Germany Ludwig Maximilian University, Institute of Sociology, Konradstraße 6, 80801 Munich, Germany
a r t i c l e
i n f o
Article history: Received 7 February 2014 Revised 8 December 2014 Accepted 11 January 2015 Available online 30 January 2015 Keywords: Sexual intercourse Relationship duration Marriage Cohabitation Panel analysis Germany
a b s t r a c t Research into the changes in the frequency of sexual intercourse is (with few exceptions) limited to cross-sectional analyses of marital duration. We investigate the frequency of intercourse while taking into account relationship duration as well as the duration of cohabitation and marriage, effects of parenthood, and relationship quality. For the analysis we apply fixed effects regression models using data from the German Family Panel (pairfam), a nationwide randomly sampled German panel survey. Our findings imply that the drop in sex frequency occurs early in the relationship, whereas neither cohabitation nor marriage affects the frequency of intercourse to a significant extent. Sex frequency is reduced during pregnancy and as long as the couple has small children, but becomes revived later on. Relationship quality is found to play a role as well. These results are contrary to the honeymoon effect found in earlier research, but indicate that in times of postponed marriage an analogous effect may be at work in the initial period of the relationship. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction Sexual interaction is an important component of intimate relationships (Edwards and Booth, 1994). Sexuality is related to wellbeing (Blanchflower and Oswald, 2004; Rosen and Bachmann, 2008), relationship quality (Donnelly, 1993; Young et al., 1998), and union stability (Yabiku and Gager, 2009). Given the high prevalence of unhappy marriages and divorces (González and Viitanen, 2009), it seems imperative to study how sexuality develops in committed relationships. It is, therefore, hardly surprising that research on sexual interaction for couples, and, as one special aspect, the frequency of sexual intercourse, has a long tradition. Following the Kinsey reports (Kinsey et al., 1948, 1953), the frequency of intercourse in intimate relationships has been in the focus of research (see Christopher and Sprecher (2000) for a review). Confirmed by many studies is the fact that the frequency of intercourse decreases over time (Call et al., 1995; James, 1981; Westoff, 1974; Udry, 1980), but the reasons for this effect are not yet fully understood. Research has been mainly limited to marital sexuality, without taking into account the premarital phases of a relationship. This limited focus might have been negligible in the past, but premarital sex and cohabitation are increasingly common phenomena (Cohen and Manning, 2010; Lichter et al., 2010; Yucel and Gassanov, 2010), and, what is more, differentiating
⇑ Corresponding author. E-mail addresses: [email protected] (J. Schröder), [email protected] (C. Schmiedeberg). http://dx.doi.org/10.1016/j.ssresearch.2015.01.009 0049-089X/Ó 2015 Elsevier Inc. All rights reserved.
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
73
between relationship duration and marital duration may shed some further light on the mechanisms causing the decline in the frequency of intercourse over time. In this study we address the following question: at which point of the relationship does the decline in the frequency of intercourse takes place? Does it start right from the beginning of the relationship, even if the couple is not (yet) cohabiting, or does it start when the couple moves in together? Moreover, does marriage make a difference, or is McCarthy’s (2003) statement true that ‘‘it is not marriage which kills sexual desire’’ (p. 2)? This question cannot be answered by simply comparing married and unmarried couples of the same relationship duration in a cross-sectional setting, as selection effects may be at work. Therefore, we use longitudinal data and employ fixed effects models, which have the advantage that the estimation is based on individual changes in the frequency of intercourse over time, and is therefore not biased by time-constant unobserved heterogeneity. Our analysis is based on data from the German Family Panel (pairfam), a large, randomly sampled German panel survey with interviews conducted annually beginning in 2008, which focuses on the processes of relationship formation and development. We use a sample of 2855 heterosexual individuals from birth cohorts 1971–1973 and 1981–1983 who had the same partner for at least two waves. This dataset is well suited for our purposes, as all participants were questioned about the frequency of intercourse in each panel wave starting from the second wave, irrespective of their relationship status. The U.S. and Germany are rather similar with respect to the prevalence of premarital cohabitation (Heuveline and Timberlake, 2004; Wilcox and Marquardt, 2011) and intercourse (Finer, 2007), although they differ in their societal norms toward premarital sex, as the U.S. tends toward more conservative values (Widmer et al., 1998). Hence, we expect to find patterns in our data similar to those reported in U.S. studies. 2. Contributions and limitations of past research Over the past decades, a considerable number of empirical studies have investigated how the frequency of intercourse develops over the course of a relationship (for a review, see Christopher and Sprecher, 2000). Findings in the 1970s and 1980s were largely consistent insofar as, in cross-sectional analyses, older respondents and couples married for longer periods of time reported lower frequencies (Blumstein and Schwartz, 1983; Edwards and Booth, 1994; Trussell and Westoff, 1980; Westoff, 1974). These effects were confirmed by early longitudinal research (Greenblat, 1983; James, 1981; Udry, 1980), but as these studies were based on small convenience samples, their generalizability is limited.1 To give some examples, James (1981) reported diary data from 21 married couples, Greenblat (1983) findings from interviews with a sample of 80 persons, and Udry (1980) a non-representative panel of 500 couples. More recent research has attempted to disentangle the effects of age and marital duration, mostly based on large-scale surveys. Relying on the National Survey of Families and Households (NSFH) with a national probability sample of 12,008 respondents, Call et al. (1995) found that couples engaged in sexual intercourse significantly more frequently in the first two years of marriage but after that, marital duration played no role when controlling for time-variant factors. Cheung et al. (2008) found an effect for ageing, but not for marital duration in a randomly sampled survey in Hong Kong. Drawing on cross-sectional data from the Demographic and Health Survey (DHS) of 91,744 women in nineteen countries (in Asia, Africa, and America), Brewis and Meyer (2005) reported a decline in sex frequency with marital duration, but found varying effects of husband’s and wife’s age across countries. We see three limitations of past research. First, as mentioned above, most of the existing research has been targeted toward married couples, without taking into account premarital relationships and their duration. In fact, all of the above cited studies relied on samples of married couples or individuals.2 This may be at least partly due to changing norms, as only married couples were considered to be ‘real’ couples in the past, and cohabitation or even living apart together (LAT) were not as widespread and accepted living arrangements as they are today (Levin, 2004). As a consequence, data on sexual intercourse was often only available for married individuals or couples. Due to this focus, it remains unclear at which point in a relationship the decline in the frequency of intercourse begins: is it right from the beginning of the relationship, is it when the couple moves in together, or does marriage matter? Stafford et al. (2004) do not find changes in the frequency of sex associated with the transition from cohabitation to marriage, which implies that relationship duration or duration of cohabitation rather than marriage matters for the issue of sex frequency. It is remarkable, though, that even in the National Survey of Families and Households (NSFH) with more than 10,000 respondents surveyed in 1987 and 1988 (wave 1) and 1992–1994 (wave 2), only less than 115 individuals were cohabiting with the same partner in both waves, and 163 cohabited in wave 1 and were married in wave 2, whereas thousands of respondents were married without prior cohabitation (Stafford et al., 2004). Today, 25 years later, the situation is different (Finer, 2007; Wilcox and Marquardt, 2011), but investigations using current data have not yet been published. Second, to our knowledge, no analysis of this nature exists using a large, randomly sampled longitudinal dataset. In their decade review, Christopher and Sprecher (2000) addressed the need for long-term longitudinal analyses to investigate the development of sexual activity over the life course. Nevertheless, panel analyses on couples’ sexuality are still sparse. 1 It has been shown that volunteer samples using college student participants are biased in areas such as sexual experience and sexual attitudes (Wiederman, 1999). 2 Greenblat (1983) considered premarital cohabitation and intercourse, but only as control variables, without accounting for premarital relationship duration. Call et al. (1995) reported age effects for both married and cohabiting couples, but then limited their analysis to the effect of marital duration.
74
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Third, existing research has been restricted to the U.S. Exceptions are Brewis and Meyer (2005), who investigated data from Asia, Africa, and the Americas, Cheung et al. (2008), who focused on married couples in Hong Kong, and O’Fallon et al. (2003), who reported findings for Nigeria (see Schneidewind-Skibbe et al. (2008) for a literature review). European countries have not yet been the focus of such research.
3. Theoretical framework Our hypotheses are based on the concept of passion and intimacy by Baumeister and Bratslavsky (1999), who theorize that sexual activity depends on the level of passion, which is in turn influenced by the development of intimacy. According to their theory, intimacy encompasses ‘‘mutual disclosure of personal information,’’ leading to mutual understanding, a ‘‘strong favorable attitude’’ for each other, and verbal and non-verbal ‘‘communication of affection’’ (p. 51). Passion is defined as ‘‘involving strong feelings of attraction’’ (p. 52) including sexual desire for the partner. While intimacy is a characteristic of the relationship, passionate love can be regarded as an emotion (Berscheid, 1983), which fluctuates over time. Passion arises when intimacy increases. Whereas intimacy accumulates over time, passion does not: If the rise in intimacy is rapid at a given time, passion will be high, whereas if there is a slower rise in intimacy, passion will be lower. The expected development of intimacy over the course of a relationship is a fast increase at first, which then slows down over time. As a consequence, passion, and therefore the frequency of intercourse, will be highest at the beginning and decrease over the course of the relationship, even if absolute intimacy is still increasing (Ainsworth and Baumeister, 2012; Baumeister and Bratslavsky, 1999). Nevertheless, passion can rise again in stable relationships as a consequence of temporary increases of intimacy, e.g. by sharing new experiences (Baumeister and Bratslavsky, 1999). Accordingly, our first hypothesis is that coital frequency decreases from the beginning of the relationship, with the steepest decline at the beginning. Our second hypothesis regards the effect of marriage on intercourse frequency. In previous literature, the decline in intercourse frequency in the beginning of a marriage, which has been found to exist in several past studies, is often referred to as the ‘‘honeymoon effect’’ (James, 1981, p. 114), and explained by habituation to sexual activities and by the rising predictability of sexual behavior during this period (Call et al., 1995). The implicit assumption of this argumentation is that the sexual relationship starts with marriage, so that marital duration is equal with the duration of the sexual relationship. In countries such as Germany, however, where today the majority of couples engage in premarital sex, this assumption does not hold. Thus, instead of marital duration, only the duration of the (sexual) relationship should play a role in sex frequency. As a consequence, this theoretical approach has led to an alternative explanation for the expected negative effect of relationship duration on sex frequency. But we propose a different type of honeymoon effect, derived from Baumeister and Bratslavsky (1999), which may also exist in the case of premarital sexual activities: If the decision to get married increases intimacy, passion will rekindle around the date of the wedding and as a consequence, the frequency of intercourse will rise. This uptick, however, will not be lasting, as after the wedding intimacy does not continue to rise at the same rate, so that after a while passion will abate again and sex frequency will decline. Our third hypothesis concerns the effect of cohabitation. Prior research has not focused on this topic, neither with regard to entering cohabitation nor to cohabitation duration. We would, therefore, apply the argument we have already made for marriage: If the decision to move in together causes intimacy to rise at a faster rate, an increase in passion will be triggered and sex frequency will rise when the couple begins cohabiting. Later on, the rise in intimacy will slow down, and passion and sex frequency will decrease again. A further reason for the steadily declining frequency of intercourse over time is age. With increasing age, sex drive and sexual capacities decrease (Ainsworth and Baumeister, 2012). Deteriorating health may also play a role (Eisenberg et al., 2010; Karraker et al., 2011). An explanation for the age effect is a decrease in hormone production (Addis et al., 2006), but also here the findings are not definite (Mazur et al., 2002). Udry et al. (1982) report findings implying that female age rather than male age influences sex frequency due to age-related changes of female androgen production. Another aspect in this regard is that the importance of frequency for perceived quality of sexual activities might differ across age groups (Christopher and Sprecher, 2000). In addition, sexual frequency may decrease with age due to societal attitudes against the sexuality of elderly people, causing couples in older age groups to restrict their sexual activities (Riportella-Muller, 1989). However, age plays a role only in the long term and cannot explain the rapid development in the first years of a relationship. Low frequency of intercourse, in particular of young or middle-aged couples, may also be caused by situational constraints. Jobs, housework, and children leave little time and energy for sexual activities (Call et al., 1995). The DINS (double income, no sex) dilemma (Olson and DeFrain, 1994) could be one of the consequences. The DINS phenomenon, however, was not confirmed by the results of Hyde et al. (1996), and Hyde et al., 2001 did not find effects of female employment. Also, the low sexual frequency of parents (Greenblat, 1983) might be due to situational constraints. The younger the children, the more pronounced this effect should be, as babies and toddlers demand more parental care and attention than do older children (Call et al., 1995). Charny and Asineli-Tal (2004) described sex-avoidance as a typical phenomenon among young couples in combination with an emphasis on family management and parenthood, in particular after the birth of a child. This effect is also explained by evolutionary biology, which postulates a trade-off between investments in mating and parenting (Lancaster, 1994) such that after a stage of putting effort into reproduction, the frequency of intercourse will decline when the couple has children and concentrates on parenting (Liu, 2000). However, the effects of pregnancy and childbirth on
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
75
sexual activity include more than this aspect. During pregnancy, decreases in sexual desire and sex frequency have been reported (Call et al., 1995; De Judicibus and McCabe, 2002), be it due to fear about the pregnancy, physical discomfort, or lower sexual interest (Calhoun et al., 1981; Erol et al., 2007; Hyde et al., 1996). Shortly after childbirth, intercourse is not immediately resumed, mainly because of physiological reasons such as low estrogen production postpartum and dyspareunia (Barrett et al., 1999; Botros et al., 2006; Hyde et al., 1996). Moreover, breastfeeding mothers are found to resume sexual activities later (Glazener, 1997). It should be considered that a declining frequency of intercourse can be a consequence of deteriorating relationship quality. A number of studies have found relationship satisfaction to be associated with sex frequency (Blumstein and Schwartz, 1983; Call et al., 1995; Donnelly, 1993). However, the causal direction between satisfaction with the relationship and sex frequency is not clear (Christopher and Sprecher, 2000; Dzara, 2010). We focus on communication styles and intimacy of communication as aspects of relationship quality, as we regard these two characteristics as not affected by a couple’s sexuality. Negative communication patterns between partners have been shown to be associated with relationship quality (Julien et al., 2003) and with sexual dysfunction (Metz and Epstein, 2002).
4. Data and model For our analysis we use waves 2–4 of the German Family Panel (pairfam), Release 4.0 (Nauck et al., 2013). The German Family Panel is a nationwide, randomly sampled longitudinal study of more than 12,000 individuals of the three birth cohorts 1971–1973, 1981–1983, and 1991–1993. Survey data focusing on relationships and family relations were collected annually in personal interviews, starting in 2008.3 For a detailed description of the panel, see Huinink et al. (2011). The dataset has a number of strengths, in particular with regard to the investigation of the sexual activities of couples over time. First, the duration of the relationship as well as of cohabitation and marriage, when applicable, are included, so that we can differentiate between these three stages in a relationship. Second, a large number of variables are available in the dataset, spanning from socio-economic information to psychological scales. Third, we do not expect large interviewer effects and reporting bias, as the questions on sexuality have been asked in a CASI section (computer-assisted self-interview). Castelo-Branco et al. (2010) have shown considerable over-reporting of sexual activity and the importance of sex in personal interviews as compared to anonymous questionnaires. Finally, sexuality is only a small part of the study, so that the sample should not be biased by selective participation. The data includes 3604 heterosexual respondents of the two older birth cohorts who had the same partner during at least two of the analyzed waves. Besides married and cohabiting couples, we also include non-coresiding couples, for which we use the term ‘‘living apart together’’ (LAT). The youngest cohort is excluded, as we expect adolescent sexuality and relationships to be different from sexuality in older age groups. Respondents of the oldest cohort were 35–39 years old in the second wave (in 2009), 36–40 in the third wave (in 2010) and 37–41 in the fourth wave (in 2011). Respondents of the middle cohort were 25–29 years old in the second wave, 26–30 in the third wave, and 27–31 in the fourth wave. After excluding observations with missing values on dependent and independent variables as well as cases with inconsistent data, we retain a total of 2855 respondents for our analyses, resulting in 7507 observations, as for each respondent we use observations of two or three waves. We apply fixed effects regression models to analyze changes in respondents’ relationships over time (for details of fixed effects estimation, see e.g. Brüderl and Ludwig, 2014; Wooldridge, 2010). The regressions are estimated with cluster-robust standard errors (modified Huber–White sandwich estimators are applied). The advantage of fixed effects regression is that only intra-individual changes over time are used for estimation. As a consequence, unobserved heterogeneity caused by time constant variables does not bias the estimations. Cross-sectional estimates of the effect of marriage on sex frequency might, for example, be biased due to unobserved differences between married and unmarried persons which affect sex frequency, e.g. if sexually more liberated couples are less likely to marry and have sex more often. Thus, unmarried couples would have a higher sex frequency than married couples, but the difference would not be caused by marriage. Consequently, OLS-regression would result in biased estimates of the effect of marriage on the frequency of sexual intercourse as long as such factors affecting both marriage status and sex frequency were not controlled for. In contrast, estimations with fixed effects regressions would not be biased, as they are not based on the comparison of married and unmarried respondents, but rather on the comparison of respondents’ sex frequency before and after their marriage. As time constant variables are by definition the same before and after marriage for every respondent, it is neither possible nor necessary to control these variables in fixed effects estimation. A summary of the variables is given in Table 1. The first column shows mean values with standard deviation in brackets for metric variables and percentage of observations for dichotomous variables based on all three waves. The second column gives the share of respondents who were in the respective group in at least one of the three waves. The last column indicates the percentage of respondents who changed their status between waves. For instance, in 24.2% of all the observations across the three waves, respondents were cohabiting (see first column). The number of respondents who cohabited in at least one of the three waves was higher, at 30.8% (second column). The reason is that part of the respondents were cohabiting in one of 3 The reported response rates of the German Family Panel are as follows: 37% in wave 1, 75% in wave 2, 81% in wave 3, and 85% in wave 4. For details of calculation see Arránz Becker et al. (2013).
76
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Table 1 Descriptive statistics. Percent/Mean (SD)
Percent of respondents in the group in at least 1 wave
Percent of respondents with change between waves
Frequency of sexual intercourse per month (range: 0–30)
5.4 (5.3)
Relationship duration 0–5 months 6–11 months 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P 16 years
66.8
1.1 2.2 5.4 5.0 4.8 5.0 5.5 12.5 12.1 10.6 8.3 6.4 21.0
2.8 5.7 13.7 12.9 12.5 12.8 14.1 22.6 22.1 19.7 15.1 11.8 23.0
2.8 5.6 13.2 12.7 12.4 12.6 14.0 20.5 20.1 18.2 14.0 11.2 5.5
Relationship status Living apart together (LAT) Cohabitation Marriage
9.8 24.2 66.0
14.3 30.8 68.3
7.0 12.3 7.3
Duration of cohabitation 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P16 years
2.1 2.7 6.0 5.4 4.8 5.7 5.9 12.0 10.6 8.4 7.2 6.7 12.5
5.5 7.0 15.0 13.9 12.4 14.7 15.0 22.1 19.8 15.5 13.0 12.6 15.1
5.5 7.0 14.8 13.8 12.4 14.5 14.9 20.3 18.4 14.6 12.4 11.9 6.1
Duration of marriage 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P16 years
1.8 2.4 4.4 4.6 4.8 5.1 4.6 8.6 7.3 6.3 5.8 4.6 5.6
4.7 6.2 11.4 12.0 12.6 13.1 12.1 15.8 13.4 11.8 10.3 8.2 7.3
4.7 6.2 11.3 11.9 12.6 13.0 12.1 14.6 12.5 11.2 9.7 7.5 3.3
Age of youngest child living in household No child 0–3 months 4 months–1 year 1–3 years 3–6 years 6–10 years 10–14 years P14 years Breastfeeding Expecting a child Tried to conceive since last interview Intimacy of communication (range: 1–5) Negative conflict style (range: 1–5) Living distance: more than 45 min
32.2 2.3 5.6 17.1 17.9 15.2 6.8 2.9 4.6 5.9 11.7 3.8 (0.8) 2.1 (0.6) 2.5
36.8 5.9 14.6 30.5 29.1 21.2 9.9 4.1 10.7 14.6 20.5
4.0
6.8 5.9 14.6 27.7 22.2 12.6 6.5 2.3 10.4 14.5 16.3 75.4 96.0 2.4
Woman’s employment status Full-time employment
30.7
40.0
17.7
77
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82 Table 1 (continued) Percent/Mean (SD)
Percent of respondents in the group in at least 1 wave
Percent of respondents with change between waves
Self-employment Other employment Unemployment No employment Education/vocational training
4.8 34.4 3.5 21.1 5.5
6.8 47.0 6.9 33.6 9.5
4.0 26.1 5.9 23.2 6.9
Man’s employment status Full-time employment Self-employment Other employment Unemployment No employment Education/vocational training
75.8 9.5 4.9 3.8 1.8 4.2
82.7 12.4 8.3 6.9 3.4 7.5
15.0 6.1 6.4 5.4 2.6 5.7
Woman’s age
33.0 (5.6)
98.7
Man’s age
35.8 (6.4)
98.6
Wave Wave 2 Wave 3 Wave 4
32.0 35.4 32.6
Number of respondents Number of observations
2855 7507
84.1 93.0 85.8
84.1 93.0 85.8
the waves but had another relationship status in other waves. Indeed, 12.3% of the respondents started or stopped cohabiting between waves (last column). Our main variable of interest is the self-reported frequency of sexual intercourse during the past 3 months with their partner. The answer format was 7 categories (never/not in the past 3 months – once per month or less – 2–3 times per month – once per week – 2–3 times per week – more than 3 times per week – daily). We recoded the answers to indicate frequency per month, so that we could apply linear regression models. The resulting categories are: 0, 0.67, 2.5, 4, 10, 20 and 30 times per month. The following descriptive statistics illustrate the prevalence of premarital sex. Only 7 (out of 2855) respondents reported in one or more waves that they never had sex in their life. Of the 400 cases of respondents reporting not to have had sex in the past three months, 84 (21%) are observations of cohabiting respondents and 300 (75%) of married respondents. To put it differently, of the 1114 respondents who are not married (in at least one of the observed waves), 98% report having intercourse regularly at least in one wave. Mean frequency is 7.9 times per month for non-cohabiting couples (living apart together), 5.6 for cohabiters and 4.9 for married couples. In addition, we calculated the duration of the relationship when couples enter cohabitation or marriage and the duration of previous cohabitation at the time of a marriage. As our dataset does not include these variables, we use the difference between relationship duration and marriage duration (or cohabitation duration, respectively), although we have to account for potential errors caused by couples who are separated (but not divorced) for a time and then later reunite. We report median values instead of means to take this into account. The resulting figures confirm our reasoning that it is worth investigating the development of the frequency of intercourse also before marriage. Among the married couples in our sample, the median relationship duration before marriage is 3.8 years and median cohabitation duration is 2.0 years (which is in line with the figures reported for Germany by Heuveline and Timberlake (2004)). Among the cohabiters, the median relationship duration before moving in together is 1.25 years. Duration of relationship, cohabitation, and marriage are categorized and included as a set of dummy variables in order to fully capture nonlinear effects, as the decline is expected to be stronger in the first years. We do not include duration of cohabitation and duration of marriage in the same model in order to avoid problems of collinearity. Instead, we estimate one model that includes relationship duration and cohabitation duration, and another model that includes relationship duration and marital duration. To test our hypothesis on the honeymoon peak, we compare sex frequency in the first 6 months after the wedding with the time before marriage and later stages of marriage. The honeymoon effect would be confirmed if newlywed couples had sex significantly more often in these 6 months compared with periods before and after. At the beginning of cohabitation there might be a peak in intercourse frequency as well, indicated by a significantly higher sex frequency in the first six months after starting cohabitation. In order to capture the effect of childbearing on the frequency of intercourse, we include a set of variables for the age of the youngest child in the household, a variable indicating if the woman is pregnant at the time of the interview, and a variable measuring whether the couple has tried to conceive since the last interview. Note that these variables are not mutually exclusive, so that e.g. for couples who have a toddler and expect their second child, the effects may accumulate. Regarding the age of the children, we concentrate on the youngest child in the household instead of including all existing children, as the youngest child poses the strongest situational constraints. In addition, we control for effects of relationship quality on sex frequency focusing on conflict styles and intimacy. We refrain from including relationship satisfaction in the model because this variable would most likely be endogenous due
78
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
to reciprocal causality between relationship satisfaction and frequency of intercourse. Using a 12-item scale we measure the constructiveness of a couple’s conflict style as follows. Respondents rated both their own and their partner’s conflict style in the relationship with the same six items (e.g. frequency of yelling at the partner, remaining silent or insulting the partner when having a disagreement). To measure intimacy of communication we use two items with ratings on how often they share their thoughts and secrets/private feelings with the partner (adapted from the intimacy scale of the Network of Relationships Inventory (NRI) (Furman and Burmester, 1985)). As additional control variables we include age and employment status for both partners, a variable indicating if the respondent lives more than 45 min away from her partner, and a dummy variable for each wave. Note that it is neither necessary nor possible to include time-constant variables such as birth cohort, migration status, religiosity, or sexual experiences prior to the current relationship in the fixed-effects model, as the estimation is based on the individual changes of independent and dependent variables over time. 5. Results Table 2 shows the results of our regression analyses.4 The interpretation of coefficients is straightforward: times per month more or less intercourse than the reference group. For instance, the negative coefficient of pregnancy ( 0.84 in the baseline model) indicates that when couples are expecting, they have intercourse about one time less per month than in times when the woman is not pregnant. In the baseline model (1), relationship duration and dummies for cohabitation and marriage are included, whereas in the second and the third models we add the duration of cohabitation and marriage, respectively. We find a significant decrease in the frequency of intercourse with relationship duration, as the significant negative coefficients in all three models imply. The drop over the first 4 years is remarkable. In the fourth year of a relationship, there are about 7 intercourse occurrences less per month than in the first six months of the relationship. The largest decline is found in the second year of the relationship such that compared to the first six months couples have just under 5 times less intercourse per month. We do not find a significant difference between the first six months of the relationship (the reference category) and the second half of the first year. After the fourth year, further decreases in sex frequency are small, and after the eighth year coefficients even begin to rise slightly. This rise might be a selection effect due to (non-random) separation of couples, however. For the dummies differentiating between LAT, cohabitation, and marriage, we do not find any effects when controlling for relationship duration. In addition, we do not find any effects of cohabitation duration (model 2) or marriage duration (model 3). We also tested the significance of the whole set of dummy variables for cohabitation duration (model 2) and marriage duration (model 3), respectively, using Wald tests. In both cases we do not find a significant effect of the duration variable. Note that when cohabitation duration is controlled for in model (2), the main effect of cohabitation shows the contrast between cohabitation duration of up to 6 months and the time when couples were living apart together. Similarly, in model (3) the main effect of marriage shows the contrast between the first six months after the wedding and the time when couples were living apart together. Our analyses confirm that parenthood depresses sexuality, as we find a significant and negative effect of the existence of small children in the household. In particular, in the first months postpartum, sex frequency is low, confirming the existing literature on this topic. The difference between having a newborn (not older than 3 months) and having an infant (4– 12 months) is significant. Sex frequency rises with the youngest child’s age; if the youngest child is 6 years or older, the negative effect is not significant anymore, and with each age category effect sizes diminish further. We interpret this as a sign that the depressing effect of parenthood on sex frequency is not lasting, which is in contrast to the hypothesis given by Call et al. (1995). A separate model (not reported) which included interactions with birth cohort for all variables showed, however, that the reported pattern holds only for the older of the two birth cohorts in our sample. In the younger birth cohort, only a newborn child of up to three month has a significant negative effect. The finding that couples report lower frequencies as long as the mother is breastfeeding can be regarded as support for the existing literature about the role hormonal levels play in (a mother’s) sexuality. As expected, pregnancy has a significant and negative effect, whereas for couples trying to conceive we find a significantly higher sex frequency. Relationship quality also has a significant effect on the frequency of intercourse. Both the respondent’s rating on intimacy of communication and the negative communication scale influence sex frequency in the expected way. Since the fixed effects model measures only intra-individual changes of the variables over time, the positive effect of the intimacy-variable can be interpreted as rising sexual frequency in the case of increasing intimacy of communication, and declining sexual frequency in the case of decreasing intimacy. This is in line with the argument put forward by Baumeister and Bratslavsky (1999) that changes in intimacy stimulate passion, leading to increased sexual desire.5 The effect of a negative communication style is significant and negative, i.e. if communication in a relationship deteriorates over time, the frequency of intercourse will decrease; and if communication patterns ameliorate, frequency will increase. 4 Due to the age difference, the two birth cohorts included in our analysis might differ regarding relationship stage. Therefore, we run the model also fully interacted with birth cohort, which is equivalent to estimating it separately for the two cohorts (results not reported). As we found substantial differences only for the age of the youngest child and female unemployment variables, we decided to report only the estimations on the entire sample. 5 As this variable captures one (small) aspect of intimacy as defined by Baumeister and Bratslavsky (1999), we run the models also without this variable. This did not yield substantially different results (not shown).
79
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82 Table 2 Fixed effects regression: Frequency of sexual intercourse. Model 1
Model 2
Model 3
Relationship duration 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P16
Ref. 1.88 4.77*** 6.19*** 7.05*** 7.06*** 7.65*** 7.89*** 7.33*** 6.94*** 6.98*** 7.07*** 6.59***
(1.04) (0.91) (0.98) (1.05) (1.13) (1.16) (1.22) (1.31) (1.39) (1.47) (1.54) (1.63)
Ref. 1.98 4.63*** 5.80*** 6.49*** 6.51*** 7.00*** 7.13*** 6.54*** 6.08*** 6.09*** 6.30*** 5.91***
(1.04) (0.91) (1.00) (1.11) (1.19) (1.25) (1.32) (1.40) (1.48) (1.55) (1.61) (1.67)
Ref. 1.89 4.69*** 6.06*** 6.85*** 6.81*** 7.37*** 7.59*** 7.04*** 6.67*** 6.70*** 6.79*** 6.37***
(1.04) (0.91) (0.98) (1.06) (1.14) (1.18) (1.25) (1.34) (1.43) (1.50) (1.56) (1.64)
Relationship status Living apart together (LAT) Cohabitation Marriage
Ref. 0.13 0.68
(0.49) (0.57)
Ref. 0.38 1.01
(0.64) (0.71)
Ref. 0.13 1.01
(0.49) (0.71)
Ref. 0.22 0.75 1.08 0.83 0.77 1.02 1.13 0.99 1.22 1.00 0.76 0.61
(0.69) (0.58) (0.65) (0.69) (0.73) (0.78) (0.81) (0.85) (0.91) (0.99) (1.08) (1.18) Ref. 0.52 0.22 0.12 0.27 0.24 0.45 0.21 0.84 0.69 0.87 1.42 1.72
(0.55) (0.52) (0.60) (0.68) (0.74) (0.80) (0.86) (0.94) (1.04) (1.14) (1.24) (1.36)
Duration of cohabitation 0–1/2 year 1/2 1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P 16 years Duration of marriage 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P 16 years Age of youngest child living in household No child 0–3 month 4 month 1 year 1–3 years 3–6 years 6–10 years 10–14 years P14 years Breastfeeding Expecting a child Tried to conceive since last interview Intimacy of communication (range: 1–5) Negative conflict style (range: 1–5) Living distance: more than 45 min
Ref. 3.22*** 1.56** 1.36** 1.20* 0.96 0.64 0.36 1.09** 0.84** 0.79** 0.29* 0.56** 1.92*
(0.59) (0.50) (0.45) (0.49) (0.54) (0.66) (0.96) (0.34) (0.26) (0.24) (0.13) (0.18) (0.79)
Ref. 3.17*** 1.53** 1.31** 1.16* 0.94 0.65 0.35 1.12** 0.85** 0.79** 0.29* 0.57** 1.85*
(0.59) (0.50) (0.46) (0.49) (0.54) (0.66) (0.97) (0.34) (0.26) (0.25) (0.13) (0.18) (0.80)
Ref. 3.21*** 1.51** 1.32** 1.17* 0.92 0.60 0.30 1.10** 0.85*** 0.81** 0.29* 0.57** 1.92*
(0.59) (0.50) (0.46) (0.49) (0.54) (0.67) (0.97) (0.34) (0.26) (0.25) (0.13) (0.18) (0.80)
Woman’s employment status Full-time employment Self-employment Other employment Unemployment
Ref. 0.15 0.07 0.45
(0.44) (0.26) (0.49)
Ref. 0.15 0.07 0.45
(0.44) (0.26) (0.49)
Ref. 0.15 0.08 0.42
(0.44) (0.26) (0.49)
(continued on next page)
80
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Table 2 (continued) Model 1
Model 2
Model 3
No employment Education/vocational training
0.55 0.23
(0.30) (0.35)
0.54 0.21
(0.30) (0.35)
0.54 0.20
(0.30) (0.35)
Man’s employment status Full-time employment Self-employment Other employment Unemployment No employment Education/vocational training
Ref. 0.28 0.71 0.25 0.97 0.16
(0.36) (0.38) (0.45) (0.59) (0.44)
Ref. 0.31 0.68 0.25 1.00 0.15
(0.36) (0.38) (0.45) (0.60) (0.44)
Ref. 0.28 0.69 0.26 0.99 0.13
(0.37) (0.38) (0.46) (0.60) (0.44)
Woman’s age
0.47
(0.28)
0.47
(0.28)
0.47
(0.28)
Man’s age
0.15
(0.24)
0.15
(0.24)
0.16
(0.25)
(0.38) (0.71)
Ref. 0.03 0.14
(0.38) (0.72)
Wave Wave 2 Wave 3 Wave 4
Ref. 0.04 0.29
Number of respondents Number of observations r2 (within)
2855 7507 0.132
(0.37) (0.71)
Ref. 0.03 0.28 2855 7507 0.134
2855 7507 0.134
Note: Standard errors in parentheses. * p < 0.05. ** p < 0.01. *** p < 0.001.
Our control variable for living more than 45 min driving distance apart proves significant as well. The coefficient implies that long-distance relationships are associated with lower sex frequency. The reason for this effect could be that in long-distance relationships, fewer opportunities for sexual activities arise because the partners spend less time together. The other control variables do not exhibit significant effects. We do not find a significant age effect, which we attribute to the fact that we have only middle-aged individuals in our sample covering 15 years across 3 panel waves and 2 birth cohorts. Pronounced age effects would be expected, in particular, for older couples, which are not included in the German Family Panel.
6. Discussion The changing frequency of sexual intercourse has been studied for decades, albeit mainly with a focus on marital intercourse. However, sexuality is not limited to the marital context, in particular in contemporary Western societies where marriage is a facultative and often postponed step in romantic relationships. Hence, our investigation contributes to the existing research by considering the duration of relationship as well as durations of cohabitation and marriage. Our study draws on the concept of intimacy and passion developed by Baumeister and Bratslavsky (1999). Their theory that an individual’s passion and sex drive will be low when increases in intimacy in the relationship are slow can be used to explain the pattern of declining intercourse frequency over the course of a relationship. In addition, their theory can be used to hypothesize an uptick in intercourse frequency around the date of the wedding, as well as when the couple begins cohabiting. With respect to past research, our analysis provides a more differentiated view on the honeymoon effect observed by prior studies. These studies focused on marital duration as opposed to relationship duration, and they relied on data collected in the 1980s, when marriage took place earlier in a relationship than it does now. They referred to the decline in sex frequency found in the beginning of marriage as the honeymoon effect. With the implicit assumption that marriage was the starting point of the (sexual) relationship, this type of honeymoon effect was explained by habituation or, according to Baumeister and Bratslavsky (1999), by a slow-down in intimacy growth accompanied by abating passion. If the sexual relationship starts before marriage, however, as is the case for most couples today, another type of honeymoon effect is possible: there could be an accelerated increase in intimacy during the special and emotional time around the wedding, which triggers passion and therefore leads to increased sex frequency. The implication would be a temporary uptick in sexual activity around the date of the wedding. As in Western countries today marriage is a facultative and postponed step in intimate relationships, we would argue that the first type of honeymoon effect should be independent of marriage and start at the beginning of the relationship, whereas the second type may still be found around the wedding. Our findings indicate that in contemporary Germany, the frequency of intercourse depends on relationship duration, but not on cohabitation or marriage. In support of our hypothesis, we find a steep drop in sexual frequency in the second year of the relationship and a slower decline afterwards, which we interpret as the modern analogy of the first type of honeymoon effect. On the other hand, contrary to our hypotheses, our estimations show that neither the event of the wedding nor moving in together play a role for sex frequency, once controlling for relationship duration, as intercourse frequency is not significantly higher in the first six months of marriage or cohabitation than before or after. Having said that, we cannot rule out
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
81
that there might be very short term effects of marriage or cohabitation which are not detectable with the data utilized. A panel study with shorter intervals between the waves could provide evidence in this regard. In addition, further research is necessary to investigate if and under what conditions cohabitation and marriage trigger intimacy and passion sufficiently to affect intercourse frequency. In accordance with past research, we find pronounced effects of parenthood on the frequency of intercourse, i.e. a higher frequency when couples try to conceive and, on the other hand, lower frequencies when couples are expecting a child or have small children, and also if the woman is breastfeeding. The effect of having children diminishes with the age of the youngest child and vanishes at the child’s age of 6, implying that the depressing effect is not lasting. However, we do not analyze the reasons for these effects. For instance, future research could test if the existence of children imposes situational constraints by measuring the free time the couple spends together (without children and housework) or the children’s sleeping behavior. Although the German Family Panel (pairfam) has several advantages, in particular regarding the large number of cases and observation of respondents over time, information with regard to sexuality is limited, which poses some limitations to the analysis. For instance, the effect of specific events cannot be captured exactly, as the frequency of intercourse is reported for a rather long period (3 months). Accordingly, for a specific analysis of the effects of incisive events such as a wedding, but also the birth of children or extra-marital affairs on sexual frequency, more specifically targeted surveys would be more suitable. Moreover, further studies could take an international perspective, comparing effects of relationship duration, cohabitation, and marriage across countries. Whereas biological mechanisms should be identical throughout the world, varying societal norms toward sexuality in relationships might cause differences in coital frequency. Acknowledgements This paper uses data from the German Family Panel pairfam, coordinated by Josef Brüderl, Karsten Hank, Johannes Huinink, Bernhard Nauck, Franz Neyer, and Sabine Walper. pairfam is funded as long-term project by the German Research Foundation (DFG). References Addis, I.B., Van Den Eeden, S.K., Wassel-Fyr, C.L., Vittinghoff, E., Brown, J.S., Thom, D.H., 2006. Sexual activity and function in middle-aged and older women. Obstet. Gynecol. 107 (4), 755–764. Ainsworth, S.E., Baumeister, R.F., 2012. Changes in sexuality: how sexuality changes across time, across relationships, and across sociocultural contexts. Clin. Neuropsy. 9 (1), 32–38. Arránz Becker, O., Brüderl, J., Buhr, P., Castiglioni, L., Fuß, D., Ludwig, V., Schmiedeberg, C., Schröder, J., & Schumann, N. (2013). The German Family Panel: Study Design and Cumulated Field Report (Waves 1 to 4). Pairfam Technical Paper Nr. 01, Release 4.0. Barrett, G., Pendry, E., Peacock, J., Victor, C., Thakar, R., Manyonda, I., 1999. Women’s sexuality after childbirth: a pilot study. Arch. Sex. Behav. 28 (2), 179– 191. Baumeister, R.F., Bratslavsky, E., 1999. Passion, intimacy, and time: passionate love as a function of change in intimacy. Pers. Soc. Psychol. Rev. 3 (1), 49–67. Berscheid, E., 1983. Emotion. In: Kelley, E.B.H.H., Christensen, A., Harvey, J.H., Huston, T.L., Levinger, G., McClintock, E., Peplau, L.A., Peterson, D.R. (Eds.), Close Relationships. Freeman, New York, pp. 110–168. Blanchflower, D.G., Oswald, A.J., 2004. Money, sex and happiness: an empirical study. Scand. J. Econ. 106 (3), 393–415. Blumstein, P., Schwartz, P., 1983. American Couples: Money, Work, Sex. Morrow, New York. Botros, S.M., Abramov, Y., Miller, J.J.R., Sand, P.K., Gandhi, S., Nickolov, A., Goldberg, R.P., 2006. Effect of parity on sexual function: an identical twin study. Obstet. Gynecol. 107 (4), 765–770. Brewis, A., Meyer, M., 2005. Marital coitus across the life course. J. Biosoc. Sci. 37 (04), 499–518. Brüderl, J., Ludwig, V., 2014. Fixed-effects panel regression. In: Best, H., Wolf, C. (Eds.), Sage Handbook of Regression Analysis and Causal Inference. Sage Publishers, London, pp. 327–356. Calhoun, L.G., Selby, J.W., King, H.E., 1981. The influence of pregnancy on sexuality: a review of current evidence. J. Sex Res. 17 (2), 139–151. Call, V., Sprecher, S., Schwartz, P., 1995. The incidence and frequency of marital sex in a national sample. J. Marriage Fam. 57 (3), 639–652. Castelo-Branco, C., Palacios, S., Ferrer-Barriendos, J., Alberich, X., 2010. Do patients lie? An open interview vs. a blind questionnaire on sexuality. J. Sex. Med. 7 (2pt2), 873–880. Charny, I.W., Asineli-Tal, S., 2004. Study of ‘‘Sex-Less’’ (Sex-Avoidant) young couples. J. Fam. Psychother. 15 (1–2), 197–217. Cheung, M.W.L., Wong, P.W.C., Liu, K.Y., Yip, P.S.F., Fan, S.Y.-S., Lam, T.-H., 2008. A study of sexual satisfaction and frequency of sex among Hong Kong Chinese couples. J. Sex Res. 45 (2), 129–139. Christopher, F.S., Sprecher, S., 2000. Sexuality in marriage, dating, and other relationships: a decade review. J. Marriage Fam. 62 (4), 999–1017. Cohen, J., Manning, W., 2010. The relationship context of premarital serial cohabitation. Soc. Sci. Res. 39 (5), 766–776. De Judicibus, M.A., McCabe, M.P., 2002. Psychological factors and the sexuality of pregnant and postpartum women. J. Sex Res. 39 (2), 94–103. Donnelly, D.A., 1993. Sexually inactive marriages. J. Sex Res. 30 (2), 171–179. Dzara, K., 2010. Assessing the effect of marital sexuality on marital disruption. Soc. Sci. Res. 39 (5), 715–724. Edwards, J.N., Booth, A., 1994. Sexuality, marriage, and well-being: the middle years. Sex. Across Life Course, 233–259. Eisenberg, M.L., Shindel, A.W., Smith, J.F., Breyer, B.N., Lipshultz, L.I., 2010. Socioeconomic, anthropomorphic, and demographic predictors of adult sexual activity in the United States: data from the national survey of family growth. J. Sex. Med. 7 (1), 50–58. Erol, B., Sanli, O., Korkmaz, D., Seyhan, A., Akman, T., Kadioglu, A., 2007. A cross-sectional study of female sexual function and dysfunction during pregnancy. J. Sex. Med. 4 (5), 1381–1387. Finer, L.B., 2007. Trends in premarital sex in the United States, 1954–2003. Public Health Rep. 122 (1), 73–78. Furman, W., Buhrmester, D., 1985. Children’s perceptions of the personal relationships in their social networks. Dev. Psychol. 21 (6), 1016–1024. Glazener, C.M.A., 1997. Sexual function after childbirth: women’s experiences, persistent morbidity and lack of professional recognition. Br. J. Obstet. Gynaecol. 104 (3), 330–335. González, L., Viitanen, T.K., 2009. The effect of divorce laws on divorce rates in Europe. Eur. Econ. Rev. 53 (2), 127–138. Greenblat, C.S., 1983. The salience of sexuality in the early years of marriage. J. Marriage Fam. 45 (2), 289–299. Heuveline, P., Timberlake, J.M., 2004. The role of cohabitation in family formation: the United States in comparative perspective. J. Marriage Fam. 66 (5), 1214–1230.
82
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Huinink, J., Brüderl, J., Nauck, B., Walper, S., Castiglioni, L., Feldhaus, M., 2011. Panel analysis of intimate relationships and family dynamics (pairfam): conceptual framework and design. Z. Familienforschung – J. Fam. Res. 23, 77–101. Hyde, J.S., DeLamater, J.D., Plant, E.A., Byrd, J.M., 1996. Sexuality during pregnancy and the year postpartum. J. Sex Res. 33 (2), 143–151. Hyde, J.S., DeLamater, J.D., Durik, A.M., 2001. Sexuality and the dual-earner couple, Part II: beyond the baby years. J. Sex Res. 38 (1), 10–23. James, W.H., 1981. The honeymoon effect on marital coitus. J. Sex Res. 17 (2), 114–123. Julien, D., Chartrand, E., Simard, M.-C., Bouthillier, D., Bégin, J., 2003. Conflict, social support and relationship quality: an observational study of heterosexual, gay male and lesbian couples’ communication. J. Fam. Psychol. 17 (3), 419–428. Karraker, A., DeLamater, J., Schwartz, C.R., 2011. Sexual frequency decline from midlife to later life. J. Gerontol. Ser. B: Psychol. Sci. Soc. Sci. 66B (4), 502–512. Kinsey, A., Pomeroy, W., Martin, C., 1948. Sexual Behavior in the Human Male. Saunders, Philadelphia. Kinsey, A., Pomeroy, W., Martin, C., Gebhard, P., 1953. Sexual Behavior in the Human Female. Saunders, Philadelphia. Lancaster, J., 1994. Human sexuality, life histories, and evolutionary ecology. In: Rossi, A. (Ed.), Sexuality across the Life Course. The University of Chicago Press, Chicago, pp. 39–62. Levin, I., 2004. Living apart together: a new family form. Curr. Sociol. 52 (2), 223–240. Lichter, D.T., Turner, R.N., Sassler, S., 2010. National estimates of the rise in serial cohabitation. Soc. Sci. Res. 39 (5), 754–765. Liu, C., 2000. A theory of marital sexual life. J. Marriage Fam. 62 (2), 363–374. Mazur, A., Mueller, U., Krause, W., Booth, A., 2002. Causes of sexual decline in aging married men. Int. J. Impot. Res. 14 (2), 101. McCarthy, B., 2003. Marital sex as it ought to be. J. Fam. Psychother. 14 (2), 1–12. Metz, M.E., Epstein, N., 2002. Assessing the role of relationship conflict in sexual dysfunction. J. Sex Marital Ther. 28 (2), 139–164. Nauck, B., Brüderl, J., Huinink, J., & Walper, S. (2013). Beziehungs- und Familienpanel (pairfam). GESIS Datenarchiv, Köln. ZA5678 Datenfile Version 4.0.0. http://dx.doi.org/10.4232/pairfam.5678.4.0.0. O’Fallon, J.B., Tsui, A., Adewuyi, A., 2003. Social and proximate determinants of sexual activity in rural Nigeria. J. Biosoc. Sci. 35, 585–599. Olson, D., DeFrain, J., 1994. Marriage and the Family: Diversity and Strengths. Mountain View, CA, Mayfield. Riportella-Muller, R., 1989. Sexuality in the elderly: a review. In: McKinney, K., Sprecher, S. (Eds.), Human Sexuality: The Societal and Interpersonal Context. Ablex, Norwood, NJ, pp. 210–236. Rosen, R.C., Bachmann, G.A., 2008. Sexual well-being, happiness, and satisfaction in women: the case for a new conceptual paradigm. J. Sex Marital Ther. 34, 291–297. Schneidewind-Skibbe, A., Hayes, R.D., Koochaki, P.E., Meyer, J., Dennerstein, L., 2008. The frequency of sexual intercourse reported by women: a review of community-based studies and factors limiting their conclusions. J. Sex. Med. 5 (2), 301–335. Stafford, L., Kline, S.L., Rankin, C.T., 2004. Married individuals, cohabiters, and cohabiters who marry: a longitudinal study of relational and individual wellbeing. J. Soc. Pers. Relat. 21 (2), 231–248. Trussell, J., Westoff, C.F., 1980. Contraceptive practice and trends in coital frequency. Fam. Plann. Perspect. 12 (5), 246–249. Udry, J.R., 1980. Changes in the frequency of marital intercourse from panel data. Arch. Sex. Behav. 9 (4), 319–325. Udry, J.R., Deven, F.R., Coleman, S.J., 1982. A cross-national comparison of the relative influence of male and female age on the frequency of marital intercourse. J. Biosoc. Sci. 14 (01), 1–6. Westoff, C.F., 1974. Coital frequency and contraception. Fam. Plann. Perspect. 6 (3), 136–141. Widmer, E.D., Treas, J., Newcomb, R., 1998. Attitudes toward nonmarital sex in 24 countries. J. Sex Res. 35 (4), 349–358. Wiederman, M.W., 1999. Volunteer bias in sexuality research using college student participants. J. Sex Res. 36 (1), 59–66. Wilcox, W.B., Marquardt, E. (2011). The state of our unions: Marriage in America 2010. The National Marriage Project and the Center for Marriage and Families at the Institute for American Values, University of Virginia.
Contents lists available at ScienceDirect
Social Science Research journal homepage: www.elsevier.com/locate/ssresearch
Effects of relationship duration, cohabitation, and marriage on the frequency of intercourse in couples: Findings from German panel data Jette Schröder a,⇑, Claudia Schmiedeberg b a b
GESIS – Leibniz Institute for the Social Sciences, P.O. Box 122155, 68072 Mannheim, Germany Ludwig Maximilian University, Institute of Sociology, Konradstraße 6, 80801 Munich, Germany
a r t i c l e
i n f o
Article history: Received 7 February 2014 Revised 8 December 2014 Accepted 11 January 2015 Available online 30 January 2015 Keywords: Sexual intercourse Relationship duration Marriage Cohabitation Panel analysis Germany
a b s t r a c t Research into the changes in the frequency of sexual intercourse is (with few exceptions) limited to cross-sectional analyses of marital duration. We investigate the frequency of intercourse while taking into account relationship duration as well as the duration of cohabitation and marriage, effects of parenthood, and relationship quality. For the analysis we apply fixed effects regression models using data from the German Family Panel (pairfam), a nationwide randomly sampled German panel survey. Our findings imply that the drop in sex frequency occurs early in the relationship, whereas neither cohabitation nor marriage affects the frequency of intercourse to a significant extent. Sex frequency is reduced during pregnancy and as long as the couple has small children, but becomes revived later on. Relationship quality is found to play a role as well. These results are contrary to the honeymoon effect found in earlier research, but indicate that in times of postponed marriage an analogous effect may be at work in the initial period of the relationship. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction Sexual interaction is an important component of intimate relationships (Edwards and Booth, 1994). Sexuality is related to wellbeing (Blanchflower and Oswald, 2004; Rosen and Bachmann, 2008), relationship quality (Donnelly, 1993; Young et al., 1998), and union stability (Yabiku and Gager, 2009). Given the high prevalence of unhappy marriages and divorces (González and Viitanen, 2009), it seems imperative to study how sexuality develops in committed relationships. It is, therefore, hardly surprising that research on sexual interaction for couples, and, as one special aspect, the frequency of sexual intercourse, has a long tradition. Following the Kinsey reports (Kinsey et al., 1948, 1953), the frequency of intercourse in intimate relationships has been in the focus of research (see Christopher and Sprecher (2000) for a review). Confirmed by many studies is the fact that the frequency of intercourse decreases over time (Call et al., 1995; James, 1981; Westoff, 1974; Udry, 1980), but the reasons for this effect are not yet fully understood. Research has been mainly limited to marital sexuality, without taking into account the premarital phases of a relationship. This limited focus might have been negligible in the past, but premarital sex and cohabitation are increasingly common phenomena (Cohen and Manning, 2010; Lichter et al., 2010; Yucel and Gassanov, 2010), and, what is more, differentiating
⇑ Corresponding author. E-mail addresses: [email protected] (J. Schröder), [email protected] (C. Schmiedeberg). http://dx.doi.org/10.1016/j.ssresearch.2015.01.009 0049-089X/Ó 2015 Elsevier Inc. All rights reserved.
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
73
between relationship duration and marital duration may shed some further light on the mechanisms causing the decline in the frequency of intercourse over time. In this study we address the following question: at which point of the relationship does the decline in the frequency of intercourse takes place? Does it start right from the beginning of the relationship, even if the couple is not (yet) cohabiting, or does it start when the couple moves in together? Moreover, does marriage make a difference, or is McCarthy’s (2003) statement true that ‘‘it is not marriage which kills sexual desire’’ (p. 2)? This question cannot be answered by simply comparing married and unmarried couples of the same relationship duration in a cross-sectional setting, as selection effects may be at work. Therefore, we use longitudinal data and employ fixed effects models, which have the advantage that the estimation is based on individual changes in the frequency of intercourse over time, and is therefore not biased by time-constant unobserved heterogeneity. Our analysis is based on data from the German Family Panel (pairfam), a large, randomly sampled German panel survey with interviews conducted annually beginning in 2008, which focuses on the processes of relationship formation and development. We use a sample of 2855 heterosexual individuals from birth cohorts 1971–1973 and 1981–1983 who had the same partner for at least two waves. This dataset is well suited for our purposes, as all participants were questioned about the frequency of intercourse in each panel wave starting from the second wave, irrespective of their relationship status. The U.S. and Germany are rather similar with respect to the prevalence of premarital cohabitation (Heuveline and Timberlake, 2004; Wilcox and Marquardt, 2011) and intercourse (Finer, 2007), although they differ in their societal norms toward premarital sex, as the U.S. tends toward more conservative values (Widmer et al., 1998). Hence, we expect to find patterns in our data similar to those reported in U.S. studies. 2. Contributions and limitations of past research Over the past decades, a considerable number of empirical studies have investigated how the frequency of intercourse develops over the course of a relationship (for a review, see Christopher and Sprecher, 2000). Findings in the 1970s and 1980s were largely consistent insofar as, in cross-sectional analyses, older respondents and couples married for longer periods of time reported lower frequencies (Blumstein and Schwartz, 1983; Edwards and Booth, 1994; Trussell and Westoff, 1980; Westoff, 1974). These effects were confirmed by early longitudinal research (Greenblat, 1983; James, 1981; Udry, 1980), but as these studies were based on small convenience samples, their generalizability is limited.1 To give some examples, James (1981) reported diary data from 21 married couples, Greenblat (1983) findings from interviews with a sample of 80 persons, and Udry (1980) a non-representative panel of 500 couples. More recent research has attempted to disentangle the effects of age and marital duration, mostly based on large-scale surveys. Relying on the National Survey of Families and Households (NSFH) with a national probability sample of 12,008 respondents, Call et al. (1995) found that couples engaged in sexual intercourse significantly more frequently in the first two years of marriage but after that, marital duration played no role when controlling for time-variant factors. Cheung et al. (2008) found an effect for ageing, but not for marital duration in a randomly sampled survey in Hong Kong. Drawing on cross-sectional data from the Demographic and Health Survey (DHS) of 91,744 women in nineteen countries (in Asia, Africa, and America), Brewis and Meyer (2005) reported a decline in sex frequency with marital duration, but found varying effects of husband’s and wife’s age across countries. We see three limitations of past research. First, as mentioned above, most of the existing research has been targeted toward married couples, without taking into account premarital relationships and their duration. In fact, all of the above cited studies relied on samples of married couples or individuals.2 This may be at least partly due to changing norms, as only married couples were considered to be ‘real’ couples in the past, and cohabitation or even living apart together (LAT) were not as widespread and accepted living arrangements as they are today (Levin, 2004). As a consequence, data on sexual intercourse was often only available for married individuals or couples. Due to this focus, it remains unclear at which point in a relationship the decline in the frequency of intercourse begins: is it right from the beginning of the relationship, is it when the couple moves in together, or does marriage matter? Stafford et al. (2004) do not find changes in the frequency of sex associated with the transition from cohabitation to marriage, which implies that relationship duration or duration of cohabitation rather than marriage matters for the issue of sex frequency. It is remarkable, though, that even in the National Survey of Families and Households (NSFH) with more than 10,000 respondents surveyed in 1987 and 1988 (wave 1) and 1992–1994 (wave 2), only less than 115 individuals were cohabiting with the same partner in both waves, and 163 cohabited in wave 1 and were married in wave 2, whereas thousands of respondents were married without prior cohabitation (Stafford et al., 2004). Today, 25 years later, the situation is different (Finer, 2007; Wilcox and Marquardt, 2011), but investigations using current data have not yet been published. Second, to our knowledge, no analysis of this nature exists using a large, randomly sampled longitudinal dataset. In their decade review, Christopher and Sprecher (2000) addressed the need for long-term longitudinal analyses to investigate the development of sexual activity over the life course. Nevertheless, panel analyses on couples’ sexuality are still sparse. 1 It has been shown that volunteer samples using college student participants are biased in areas such as sexual experience and sexual attitudes (Wiederman, 1999). 2 Greenblat (1983) considered premarital cohabitation and intercourse, but only as control variables, without accounting for premarital relationship duration. Call et al. (1995) reported age effects for both married and cohabiting couples, but then limited their analysis to the effect of marital duration.
74
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Third, existing research has been restricted to the U.S. Exceptions are Brewis and Meyer (2005), who investigated data from Asia, Africa, and the Americas, Cheung et al. (2008), who focused on married couples in Hong Kong, and O’Fallon et al. (2003), who reported findings for Nigeria (see Schneidewind-Skibbe et al. (2008) for a literature review). European countries have not yet been the focus of such research.
3. Theoretical framework Our hypotheses are based on the concept of passion and intimacy by Baumeister and Bratslavsky (1999), who theorize that sexual activity depends on the level of passion, which is in turn influenced by the development of intimacy. According to their theory, intimacy encompasses ‘‘mutual disclosure of personal information,’’ leading to mutual understanding, a ‘‘strong favorable attitude’’ for each other, and verbal and non-verbal ‘‘communication of affection’’ (p. 51). Passion is defined as ‘‘involving strong feelings of attraction’’ (p. 52) including sexual desire for the partner. While intimacy is a characteristic of the relationship, passionate love can be regarded as an emotion (Berscheid, 1983), which fluctuates over time. Passion arises when intimacy increases. Whereas intimacy accumulates over time, passion does not: If the rise in intimacy is rapid at a given time, passion will be high, whereas if there is a slower rise in intimacy, passion will be lower. The expected development of intimacy over the course of a relationship is a fast increase at first, which then slows down over time. As a consequence, passion, and therefore the frequency of intercourse, will be highest at the beginning and decrease over the course of the relationship, even if absolute intimacy is still increasing (Ainsworth and Baumeister, 2012; Baumeister and Bratslavsky, 1999). Nevertheless, passion can rise again in stable relationships as a consequence of temporary increases of intimacy, e.g. by sharing new experiences (Baumeister and Bratslavsky, 1999). Accordingly, our first hypothesis is that coital frequency decreases from the beginning of the relationship, with the steepest decline at the beginning. Our second hypothesis regards the effect of marriage on intercourse frequency. In previous literature, the decline in intercourse frequency in the beginning of a marriage, which has been found to exist in several past studies, is often referred to as the ‘‘honeymoon effect’’ (James, 1981, p. 114), and explained by habituation to sexual activities and by the rising predictability of sexual behavior during this period (Call et al., 1995). The implicit assumption of this argumentation is that the sexual relationship starts with marriage, so that marital duration is equal with the duration of the sexual relationship. In countries such as Germany, however, where today the majority of couples engage in premarital sex, this assumption does not hold. Thus, instead of marital duration, only the duration of the (sexual) relationship should play a role in sex frequency. As a consequence, this theoretical approach has led to an alternative explanation for the expected negative effect of relationship duration on sex frequency. But we propose a different type of honeymoon effect, derived from Baumeister and Bratslavsky (1999), which may also exist in the case of premarital sexual activities: If the decision to get married increases intimacy, passion will rekindle around the date of the wedding and as a consequence, the frequency of intercourse will rise. This uptick, however, will not be lasting, as after the wedding intimacy does not continue to rise at the same rate, so that after a while passion will abate again and sex frequency will decline. Our third hypothesis concerns the effect of cohabitation. Prior research has not focused on this topic, neither with regard to entering cohabitation nor to cohabitation duration. We would, therefore, apply the argument we have already made for marriage: If the decision to move in together causes intimacy to rise at a faster rate, an increase in passion will be triggered and sex frequency will rise when the couple begins cohabiting. Later on, the rise in intimacy will slow down, and passion and sex frequency will decrease again. A further reason for the steadily declining frequency of intercourse over time is age. With increasing age, sex drive and sexual capacities decrease (Ainsworth and Baumeister, 2012). Deteriorating health may also play a role (Eisenberg et al., 2010; Karraker et al., 2011). An explanation for the age effect is a decrease in hormone production (Addis et al., 2006), but also here the findings are not definite (Mazur et al., 2002). Udry et al. (1982) report findings implying that female age rather than male age influences sex frequency due to age-related changes of female androgen production. Another aspect in this regard is that the importance of frequency for perceived quality of sexual activities might differ across age groups (Christopher and Sprecher, 2000). In addition, sexual frequency may decrease with age due to societal attitudes against the sexuality of elderly people, causing couples in older age groups to restrict their sexual activities (Riportella-Muller, 1989). However, age plays a role only in the long term and cannot explain the rapid development in the first years of a relationship. Low frequency of intercourse, in particular of young or middle-aged couples, may also be caused by situational constraints. Jobs, housework, and children leave little time and energy for sexual activities (Call et al., 1995). The DINS (double income, no sex) dilemma (Olson and DeFrain, 1994) could be one of the consequences. The DINS phenomenon, however, was not confirmed by the results of Hyde et al. (1996), and Hyde et al., 2001 did not find effects of female employment. Also, the low sexual frequency of parents (Greenblat, 1983) might be due to situational constraints. The younger the children, the more pronounced this effect should be, as babies and toddlers demand more parental care and attention than do older children (Call et al., 1995). Charny and Asineli-Tal (2004) described sex-avoidance as a typical phenomenon among young couples in combination with an emphasis on family management and parenthood, in particular after the birth of a child. This effect is also explained by evolutionary biology, which postulates a trade-off between investments in mating and parenting (Lancaster, 1994) such that after a stage of putting effort into reproduction, the frequency of intercourse will decline when the couple has children and concentrates on parenting (Liu, 2000). However, the effects of pregnancy and childbirth on
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
75
sexual activity include more than this aspect. During pregnancy, decreases in sexual desire and sex frequency have been reported (Call et al., 1995; De Judicibus and McCabe, 2002), be it due to fear about the pregnancy, physical discomfort, or lower sexual interest (Calhoun et al., 1981; Erol et al., 2007; Hyde et al., 1996). Shortly after childbirth, intercourse is not immediately resumed, mainly because of physiological reasons such as low estrogen production postpartum and dyspareunia (Barrett et al., 1999; Botros et al., 2006; Hyde et al., 1996). Moreover, breastfeeding mothers are found to resume sexual activities later (Glazener, 1997). It should be considered that a declining frequency of intercourse can be a consequence of deteriorating relationship quality. A number of studies have found relationship satisfaction to be associated with sex frequency (Blumstein and Schwartz, 1983; Call et al., 1995; Donnelly, 1993). However, the causal direction between satisfaction with the relationship and sex frequency is not clear (Christopher and Sprecher, 2000; Dzara, 2010). We focus on communication styles and intimacy of communication as aspects of relationship quality, as we regard these two characteristics as not affected by a couple’s sexuality. Negative communication patterns between partners have been shown to be associated with relationship quality (Julien et al., 2003) and with sexual dysfunction (Metz and Epstein, 2002).
4. Data and model For our analysis we use waves 2–4 of the German Family Panel (pairfam), Release 4.0 (Nauck et al., 2013). The German Family Panel is a nationwide, randomly sampled longitudinal study of more than 12,000 individuals of the three birth cohorts 1971–1973, 1981–1983, and 1991–1993. Survey data focusing on relationships and family relations were collected annually in personal interviews, starting in 2008.3 For a detailed description of the panel, see Huinink et al. (2011). The dataset has a number of strengths, in particular with regard to the investigation of the sexual activities of couples over time. First, the duration of the relationship as well as of cohabitation and marriage, when applicable, are included, so that we can differentiate between these three stages in a relationship. Second, a large number of variables are available in the dataset, spanning from socio-economic information to psychological scales. Third, we do not expect large interviewer effects and reporting bias, as the questions on sexuality have been asked in a CASI section (computer-assisted self-interview). Castelo-Branco et al. (2010) have shown considerable over-reporting of sexual activity and the importance of sex in personal interviews as compared to anonymous questionnaires. Finally, sexuality is only a small part of the study, so that the sample should not be biased by selective participation. The data includes 3604 heterosexual respondents of the two older birth cohorts who had the same partner during at least two of the analyzed waves. Besides married and cohabiting couples, we also include non-coresiding couples, for which we use the term ‘‘living apart together’’ (LAT). The youngest cohort is excluded, as we expect adolescent sexuality and relationships to be different from sexuality in older age groups. Respondents of the oldest cohort were 35–39 years old in the second wave (in 2009), 36–40 in the third wave (in 2010) and 37–41 in the fourth wave (in 2011). Respondents of the middle cohort were 25–29 years old in the second wave, 26–30 in the third wave, and 27–31 in the fourth wave. After excluding observations with missing values on dependent and independent variables as well as cases with inconsistent data, we retain a total of 2855 respondents for our analyses, resulting in 7507 observations, as for each respondent we use observations of two or three waves. We apply fixed effects regression models to analyze changes in respondents’ relationships over time (for details of fixed effects estimation, see e.g. Brüderl and Ludwig, 2014; Wooldridge, 2010). The regressions are estimated with cluster-robust standard errors (modified Huber–White sandwich estimators are applied). The advantage of fixed effects regression is that only intra-individual changes over time are used for estimation. As a consequence, unobserved heterogeneity caused by time constant variables does not bias the estimations. Cross-sectional estimates of the effect of marriage on sex frequency might, for example, be biased due to unobserved differences between married and unmarried persons which affect sex frequency, e.g. if sexually more liberated couples are less likely to marry and have sex more often. Thus, unmarried couples would have a higher sex frequency than married couples, but the difference would not be caused by marriage. Consequently, OLS-regression would result in biased estimates of the effect of marriage on the frequency of sexual intercourse as long as such factors affecting both marriage status and sex frequency were not controlled for. In contrast, estimations with fixed effects regressions would not be biased, as they are not based on the comparison of married and unmarried respondents, but rather on the comparison of respondents’ sex frequency before and after their marriage. As time constant variables are by definition the same before and after marriage for every respondent, it is neither possible nor necessary to control these variables in fixed effects estimation. A summary of the variables is given in Table 1. The first column shows mean values with standard deviation in brackets for metric variables and percentage of observations for dichotomous variables based on all three waves. The second column gives the share of respondents who were in the respective group in at least one of the three waves. The last column indicates the percentage of respondents who changed their status between waves. For instance, in 24.2% of all the observations across the three waves, respondents were cohabiting (see first column). The number of respondents who cohabited in at least one of the three waves was higher, at 30.8% (second column). The reason is that part of the respondents were cohabiting in one of 3 The reported response rates of the German Family Panel are as follows: 37% in wave 1, 75% in wave 2, 81% in wave 3, and 85% in wave 4. For details of calculation see Arránz Becker et al. (2013).
76
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Table 1 Descriptive statistics. Percent/Mean (SD)
Percent of respondents in the group in at least 1 wave
Percent of respondents with change between waves
Frequency of sexual intercourse per month (range: 0–30)
5.4 (5.3)
Relationship duration 0–5 months 6–11 months 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P 16 years
66.8
1.1 2.2 5.4 5.0 4.8 5.0 5.5 12.5 12.1 10.6 8.3 6.4 21.0
2.8 5.7 13.7 12.9 12.5 12.8 14.1 22.6 22.1 19.7 15.1 11.8 23.0
2.8 5.6 13.2 12.7 12.4 12.6 14.0 20.5 20.1 18.2 14.0 11.2 5.5
Relationship status Living apart together (LAT) Cohabitation Marriage
9.8 24.2 66.0
14.3 30.8 68.3
7.0 12.3 7.3
Duration of cohabitation 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P16 years
2.1 2.7 6.0 5.4 4.8 5.7 5.9 12.0 10.6 8.4 7.2 6.7 12.5
5.5 7.0 15.0 13.9 12.4 14.7 15.0 22.1 19.8 15.5 13.0 12.6 15.1
5.5 7.0 14.8 13.8 12.4 14.5 14.9 20.3 18.4 14.6 12.4 11.9 6.1
Duration of marriage 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P16 years
1.8 2.4 4.4 4.6 4.8 5.1 4.6 8.6 7.3 6.3 5.8 4.6 5.6
4.7 6.2 11.4 12.0 12.6 13.1 12.1 15.8 13.4 11.8 10.3 8.2 7.3
4.7 6.2 11.3 11.9 12.6 13.0 12.1 14.6 12.5 11.2 9.7 7.5 3.3
Age of youngest child living in household No child 0–3 months 4 months–1 year 1–3 years 3–6 years 6–10 years 10–14 years P14 years Breastfeeding Expecting a child Tried to conceive since last interview Intimacy of communication (range: 1–5) Negative conflict style (range: 1–5) Living distance: more than 45 min
32.2 2.3 5.6 17.1 17.9 15.2 6.8 2.9 4.6 5.9 11.7 3.8 (0.8) 2.1 (0.6) 2.5
36.8 5.9 14.6 30.5 29.1 21.2 9.9 4.1 10.7 14.6 20.5
4.0
6.8 5.9 14.6 27.7 22.2 12.6 6.5 2.3 10.4 14.5 16.3 75.4 96.0 2.4
Woman’s employment status Full-time employment
30.7
40.0
17.7
77
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82 Table 1 (continued) Percent/Mean (SD)
Percent of respondents in the group in at least 1 wave
Percent of respondents with change between waves
Self-employment Other employment Unemployment No employment Education/vocational training
4.8 34.4 3.5 21.1 5.5
6.8 47.0 6.9 33.6 9.5
4.0 26.1 5.9 23.2 6.9
Man’s employment status Full-time employment Self-employment Other employment Unemployment No employment Education/vocational training
75.8 9.5 4.9 3.8 1.8 4.2
82.7 12.4 8.3 6.9 3.4 7.5
15.0 6.1 6.4 5.4 2.6 5.7
Woman’s age
33.0 (5.6)
98.7
Man’s age
35.8 (6.4)
98.6
Wave Wave 2 Wave 3 Wave 4
32.0 35.4 32.6
Number of respondents Number of observations
2855 7507
84.1 93.0 85.8
84.1 93.0 85.8
the waves but had another relationship status in other waves. Indeed, 12.3% of the respondents started or stopped cohabiting between waves (last column). Our main variable of interest is the self-reported frequency of sexual intercourse during the past 3 months with their partner. The answer format was 7 categories (never/not in the past 3 months – once per month or less – 2–3 times per month – once per week – 2–3 times per week – more than 3 times per week – daily). We recoded the answers to indicate frequency per month, so that we could apply linear regression models. The resulting categories are: 0, 0.67, 2.5, 4, 10, 20 and 30 times per month. The following descriptive statistics illustrate the prevalence of premarital sex. Only 7 (out of 2855) respondents reported in one or more waves that they never had sex in their life. Of the 400 cases of respondents reporting not to have had sex in the past three months, 84 (21%) are observations of cohabiting respondents and 300 (75%) of married respondents. To put it differently, of the 1114 respondents who are not married (in at least one of the observed waves), 98% report having intercourse regularly at least in one wave. Mean frequency is 7.9 times per month for non-cohabiting couples (living apart together), 5.6 for cohabiters and 4.9 for married couples. In addition, we calculated the duration of the relationship when couples enter cohabitation or marriage and the duration of previous cohabitation at the time of a marriage. As our dataset does not include these variables, we use the difference between relationship duration and marriage duration (or cohabitation duration, respectively), although we have to account for potential errors caused by couples who are separated (but not divorced) for a time and then later reunite. We report median values instead of means to take this into account. The resulting figures confirm our reasoning that it is worth investigating the development of the frequency of intercourse also before marriage. Among the married couples in our sample, the median relationship duration before marriage is 3.8 years and median cohabitation duration is 2.0 years (which is in line with the figures reported for Germany by Heuveline and Timberlake (2004)). Among the cohabiters, the median relationship duration before moving in together is 1.25 years. Duration of relationship, cohabitation, and marriage are categorized and included as a set of dummy variables in order to fully capture nonlinear effects, as the decline is expected to be stronger in the first years. We do not include duration of cohabitation and duration of marriage in the same model in order to avoid problems of collinearity. Instead, we estimate one model that includes relationship duration and cohabitation duration, and another model that includes relationship duration and marital duration. To test our hypothesis on the honeymoon peak, we compare sex frequency in the first 6 months after the wedding with the time before marriage and later stages of marriage. The honeymoon effect would be confirmed if newlywed couples had sex significantly more often in these 6 months compared with periods before and after. At the beginning of cohabitation there might be a peak in intercourse frequency as well, indicated by a significantly higher sex frequency in the first six months after starting cohabitation. In order to capture the effect of childbearing on the frequency of intercourse, we include a set of variables for the age of the youngest child in the household, a variable indicating if the woman is pregnant at the time of the interview, and a variable measuring whether the couple has tried to conceive since the last interview. Note that these variables are not mutually exclusive, so that e.g. for couples who have a toddler and expect their second child, the effects may accumulate. Regarding the age of the children, we concentrate on the youngest child in the household instead of including all existing children, as the youngest child poses the strongest situational constraints. In addition, we control for effects of relationship quality on sex frequency focusing on conflict styles and intimacy. We refrain from including relationship satisfaction in the model because this variable would most likely be endogenous due
78
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
to reciprocal causality between relationship satisfaction and frequency of intercourse. Using a 12-item scale we measure the constructiveness of a couple’s conflict style as follows. Respondents rated both their own and their partner’s conflict style in the relationship with the same six items (e.g. frequency of yelling at the partner, remaining silent or insulting the partner when having a disagreement). To measure intimacy of communication we use two items with ratings on how often they share their thoughts and secrets/private feelings with the partner (adapted from the intimacy scale of the Network of Relationships Inventory (NRI) (Furman and Burmester, 1985)). As additional control variables we include age and employment status for both partners, a variable indicating if the respondent lives more than 45 min away from her partner, and a dummy variable for each wave. Note that it is neither necessary nor possible to include time-constant variables such as birth cohort, migration status, religiosity, or sexual experiences prior to the current relationship in the fixed-effects model, as the estimation is based on the individual changes of independent and dependent variables over time. 5. Results Table 2 shows the results of our regression analyses.4 The interpretation of coefficients is straightforward: times per month more or less intercourse than the reference group. For instance, the negative coefficient of pregnancy ( 0.84 in the baseline model) indicates that when couples are expecting, they have intercourse about one time less per month than in times when the woman is not pregnant. In the baseline model (1), relationship duration and dummies for cohabitation and marriage are included, whereas in the second and the third models we add the duration of cohabitation and marriage, respectively. We find a significant decrease in the frequency of intercourse with relationship duration, as the significant negative coefficients in all three models imply. The drop over the first 4 years is remarkable. In the fourth year of a relationship, there are about 7 intercourse occurrences less per month than in the first six months of the relationship. The largest decline is found in the second year of the relationship such that compared to the first six months couples have just under 5 times less intercourse per month. We do not find a significant difference between the first six months of the relationship (the reference category) and the second half of the first year. After the fourth year, further decreases in sex frequency are small, and after the eighth year coefficients even begin to rise slightly. This rise might be a selection effect due to (non-random) separation of couples, however. For the dummies differentiating between LAT, cohabitation, and marriage, we do not find any effects when controlling for relationship duration. In addition, we do not find any effects of cohabitation duration (model 2) or marriage duration (model 3). We also tested the significance of the whole set of dummy variables for cohabitation duration (model 2) and marriage duration (model 3), respectively, using Wald tests. In both cases we do not find a significant effect of the duration variable. Note that when cohabitation duration is controlled for in model (2), the main effect of cohabitation shows the contrast between cohabitation duration of up to 6 months and the time when couples were living apart together. Similarly, in model (3) the main effect of marriage shows the contrast between the first six months after the wedding and the time when couples were living apart together. Our analyses confirm that parenthood depresses sexuality, as we find a significant and negative effect of the existence of small children in the household. In particular, in the first months postpartum, sex frequency is low, confirming the existing literature on this topic. The difference between having a newborn (not older than 3 months) and having an infant (4– 12 months) is significant. Sex frequency rises with the youngest child’s age; if the youngest child is 6 years or older, the negative effect is not significant anymore, and with each age category effect sizes diminish further. We interpret this as a sign that the depressing effect of parenthood on sex frequency is not lasting, which is in contrast to the hypothesis given by Call et al. (1995). A separate model (not reported) which included interactions with birth cohort for all variables showed, however, that the reported pattern holds only for the older of the two birth cohorts in our sample. In the younger birth cohort, only a newborn child of up to three month has a significant negative effect. The finding that couples report lower frequencies as long as the mother is breastfeeding can be regarded as support for the existing literature about the role hormonal levels play in (a mother’s) sexuality. As expected, pregnancy has a significant and negative effect, whereas for couples trying to conceive we find a significantly higher sex frequency. Relationship quality also has a significant effect on the frequency of intercourse. Both the respondent’s rating on intimacy of communication and the negative communication scale influence sex frequency in the expected way. Since the fixed effects model measures only intra-individual changes of the variables over time, the positive effect of the intimacy-variable can be interpreted as rising sexual frequency in the case of increasing intimacy of communication, and declining sexual frequency in the case of decreasing intimacy. This is in line with the argument put forward by Baumeister and Bratslavsky (1999) that changes in intimacy stimulate passion, leading to increased sexual desire.5 The effect of a negative communication style is significant and negative, i.e. if communication in a relationship deteriorates over time, the frequency of intercourse will decrease; and if communication patterns ameliorate, frequency will increase. 4 Due to the age difference, the two birth cohorts included in our analysis might differ regarding relationship stage. Therefore, we run the model also fully interacted with birth cohort, which is equivalent to estimating it separately for the two cohorts (results not reported). As we found substantial differences only for the age of the youngest child and female unemployment variables, we decided to report only the estimations on the entire sample. 5 As this variable captures one (small) aspect of intimacy as defined by Baumeister and Bratslavsky (1999), we run the models also without this variable. This did not yield substantially different results (not shown).
79
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82 Table 2 Fixed effects regression: Frequency of sexual intercourse. Model 1
Model 2
Model 3
Relationship duration 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P16
Ref. 1.88 4.77*** 6.19*** 7.05*** 7.06*** 7.65*** 7.89*** 7.33*** 6.94*** 6.98*** 7.07*** 6.59***
(1.04) (0.91) (0.98) (1.05) (1.13) (1.16) (1.22) (1.31) (1.39) (1.47) (1.54) (1.63)
Ref. 1.98 4.63*** 5.80*** 6.49*** 6.51*** 7.00*** 7.13*** 6.54*** 6.08*** 6.09*** 6.30*** 5.91***
(1.04) (0.91) (1.00) (1.11) (1.19) (1.25) (1.32) (1.40) (1.48) (1.55) (1.61) (1.67)
Ref. 1.89 4.69*** 6.06*** 6.85*** 6.81*** 7.37*** 7.59*** 7.04*** 6.67*** 6.70*** 6.79*** 6.37***
(1.04) (0.91) (0.98) (1.06) (1.14) (1.18) (1.25) (1.34) (1.43) (1.50) (1.56) (1.64)
Relationship status Living apart together (LAT) Cohabitation Marriage
Ref. 0.13 0.68
(0.49) (0.57)
Ref. 0.38 1.01
(0.64) (0.71)
Ref. 0.13 1.01
(0.49) (0.71)
Ref. 0.22 0.75 1.08 0.83 0.77 1.02 1.13 0.99 1.22 1.00 0.76 0.61
(0.69) (0.58) (0.65) (0.69) (0.73) (0.78) (0.81) (0.85) (0.91) (0.99) (1.08) (1.18) Ref. 0.52 0.22 0.12 0.27 0.24 0.45 0.21 0.84 0.69 0.87 1.42 1.72
(0.55) (0.52) (0.60) (0.68) (0.74) (0.80) (0.86) (0.94) (1.04) (1.14) (1.24) (1.36)
Duration of cohabitation 0–1/2 year 1/2 1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P 16 years Duration of marriage 0–1/2 year 1/2–1 year 1–2 years 2–3 years 3–4 years 4–5 years 5–6 years 6–8 years 8–10 years 10–12 years 12–14 years 14–16 years P 16 years Age of youngest child living in household No child 0–3 month 4 month 1 year 1–3 years 3–6 years 6–10 years 10–14 years P14 years Breastfeeding Expecting a child Tried to conceive since last interview Intimacy of communication (range: 1–5) Negative conflict style (range: 1–5) Living distance: more than 45 min
Ref. 3.22*** 1.56** 1.36** 1.20* 0.96 0.64 0.36 1.09** 0.84** 0.79** 0.29* 0.56** 1.92*
(0.59) (0.50) (0.45) (0.49) (0.54) (0.66) (0.96) (0.34) (0.26) (0.24) (0.13) (0.18) (0.79)
Ref. 3.17*** 1.53** 1.31** 1.16* 0.94 0.65 0.35 1.12** 0.85** 0.79** 0.29* 0.57** 1.85*
(0.59) (0.50) (0.46) (0.49) (0.54) (0.66) (0.97) (0.34) (0.26) (0.25) (0.13) (0.18) (0.80)
Ref. 3.21*** 1.51** 1.32** 1.17* 0.92 0.60 0.30 1.10** 0.85*** 0.81** 0.29* 0.57** 1.92*
(0.59) (0.50) (0.46) (0.49) (0.54) (0.67) (0.97) (0.34) (0.26) (0.25) (0.13) (0.18) (0.80)
Woman’s employment status Full-time employment Self-employment Other employment Unemployment
Ref. 0.15 0.07 0.45
(0.44) (0.26) (0.49)
Ref. 0.15 0.07 0.45
(0.44) (0.26) (0.49)
Ref. 0.15 0.08 0.42
(0.44) (0.26) (0.49)
(continued on next page)
80
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Table 2 (continued) Model 1
Model 2
Model 3
No employment Education/vocational training
0.55 0.23
(0.30) (0.35)
0.54 0.21
(0.30) (0.35)
0.54 0.20
(0.30) (0.35)
Man’s employment status Full-time employment Self-employment Other employment Unemployment No employment Education/vocational training
Ref. 0.28 0.71 0.25 0.97 0.16
(0.36) (0.38) (0.45) (0.59) (0.44)
Ref. 0.31 0.68 0.25 1.00 0.15
(0.36) (0.38) (0.45) (0.60) (0.44)
Ref. 0.28 0.69 0.26 0.99 0.13
(0.37) (0.38) (0.46) (0.60) (0.44)
Woman’s age
0.47
(0.28)
0.47
(0.28)
0.47
(0.28)
Man’s age
0.15
(0.24)
0.15
(0.24)
0.16
(0.25)
(0.38) (0.71)
Ref. 0.03 0.14
(0.38) (0.72)
Wave Wave 2 Wave 3 Wave 4
Ref. 0.04 0.29
Number of respondents Number of observations r2 (within)
2855 7507 0.132
(0.37) (0.71)
Ref. 0.03 0.28 2855 7507 0.134
2855 7507 0.134
Note: Standard errors in parentheses. * p < 0.05. ** p < 0.01. *** p < 0.001.
Our control variable for living more than 45 min driving distance apart proves significant as well. The coefficient implies that long-distance relationships are associated with lower sex frequency. The reason for this effect could be that in long-distance relationships, fewer opportunities for sexual activities arise because the partners spend less time together. The other control variables do not exhibit significant effects. We do not find a significant age effect, which we attribute to the fact that we have only middle-aged individuals in our sample covering 15 years across 3 panel waves and 2 birth cohorts. Pronounced age effects would be expected, in particular, for older couples, which are not included in the German Family Panel.
6. Discussion The changing frequency of sexual intercourse has been studied for decades, albeit mainly with a focus on marital intercourse. However, sexuality is not limited to the marital context, in particular in contemporary Western societies where marriage is a facultative and often postponed step in romantic relationships. Hence, our investigation contributes to the existing research by considering the duration of relationship as well as durations of cohabitation and marriage. Our study draws on the concept of intimacy and passion developed by Baumeister and Bratslavsky (1999). Their theory that an individual’s passion and sex drive will be low when increases in intimacy in the relationship are slow can be used to explain the pattern of declining intercourse frequency over the course of a relationship. In addition, their theory can be used to hypothesize an uptick in intercourse frequency around the date of the wedding, as well as when the couple begins cohabiting. With respect to past research, our analysis provides a more differentiated view on the honeymoon effect observed by prior studies. These studies focused on marital duration as opposed to relationship duration, and they relied on data collected in the 1980s, when marriage took place earlier in a relationship than it does now. They referred to the decline in sex frequency found in the beginning of marriage as the honeymoon effect. With the implicit assumption that marriage was the starting point of the (sexual) relationship, this type of honeymoon effect was explained by habituation or, according to Baumeister and Bratslavsky (1999), by a slow-down in intimacy growth accompanied by abating passion. If the sexual relationship starts before marriage, however, as is the case for most couples today, another type of honeymoon effect is possible: there could be an accelerated increase in intimacy during the special and emotional time around the wedding, which triggers passion and therefore leads to increased sex frequency. The implication would be a temporary uptick in sexual activity around the date of the wedding. As in Western countries today marriage is a facultative and postponed step in intimate relationships, we would argue that the first type of honeymoon effect should be independent of marriage and start at the beginning of the relationship, whereas the second type may still be found around the wedding. Our findings indicate that in contemporary Germany, the frequency of intercourse depends on relationship duration, but not on cohabitation or marriage. In support of our hypothesis, we find a steep drop in sexual frequency in the second year of the relationship and a slower decline afterwards, which we interpret as the modern analogy of the first type of honeymoon effect. On the other hand, contrary to our hypotheses, our estimations show that neither the event of the wedding nor moving in together play a role for sex frequency, once controlling for relationship duration, as intercourse frequency is not significantly higher in the first six months of marriage or cohabitation than before or after. Having said that, we cannot rule out
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
81
that there might be very short term effects of marriage or cohabitation which are not detectable with the data utilized. A panel study with shorter intervals between the waves could provide evidence in this regard. In addition, further research is necessary to investigate if and under what conditions cohabitation and marriage trigger intimacy and passion sufficiently to affect intercourse frequency. In accordance with past research, we find pronounced effects of parenthood on the frequency of intercourse, i.e. a higher frequency when couples try to conceive and, on the other hand, lower frequencies when couples are expecting a child or have small children, and also if the woman is breastfeeding. The effect of having children diminishes with the age of the youngest child and vanishes at the child’s age of 6, implying that the depressing effect is not lasting. However, we do not analyze the reasons for these effects. For instance, future research could test if the existence of children imposes situational constraints by measuring the free time the couple spends together (without children and housework) or the children’s sleeping behavior. Although the German Family Panel (pairfam) has several advantages, in particular regarding the large number of cases and observation of respondents over time, information with regard to sexuality is limited, which poses some limitations to the analysis. For instance, the effect of specific events cannot be captured exactly, as the frequency of intercourse is reported for a rather long period (3 months). Accordingly, for a specific analysis of the effects of incisive events such as a wedding, but also the birth of children or extra-marital affairs on sexual frequency, more specifically targeted surveys would be more suitable. Moreover, further studies could take an international perspective, comparing effects of relationship duration, cohabitation, and marriage across countries. Whereas biological mechanisms should be identical throughout the world, varying societal norms toward sexuality in relationships might cause differences in coital frequency. Acknowledgements This paper uses data from the German Family Panel pairfam, coordinated by Josef Brüderl, Karsten Hank, Johannes Huinink, Bernhard Nauck, Franz Neyer, and Sabine Walper. pairfam is funded as long-term project by the German Research Foundation (DFG). References Addis, I.B., Van Den Eeden, S.K., Wassel-Fyr, C.L., Vittinghoff, E., Brown, J.S., Thom, D.H., 2006. Sexual activity and function in middle-aged and older women. Obstet. Gynecol. 107 (4), 755–764. Ainsworth, S.E., Baumeister, R.F., 2012. Changes in sexuality: how sexuality changes across time, across relationships, and across sociocultural contexts. Clin. Neuropsy. 9 (1), 32–38. Arránz Becker, O., Brüderl, J., Buhr, P., Castiglioni, L., Fuß, D., Ludwig, V., Schmiedeberg, C., Schröder, J., & Schumann, N. (2013). The German Family Panel: Study Design and Cumulated Field Report (Waves 1 to 4). Pairfam Technical Paper Nr. 01, Release 4.0. Barrett, G., Pendry, E., Peacock, J., Victor, C., Thakar, R., Manyonda, I., 1999. Women’s sexuality after childbirth: a pilot study. Arch. Sex. Behav. 28 (2), 179– 191. Baumeister, R.F., Bratslavsky, E., 1999. Passion, intimacy, and time: passionate love as a function of change in intimacy. Pers. Soc. Psychol. Rev. 3 (1), 49–67. Berscheid, E., 1983. Emotion. In: Kelley, E.B.H.H., Christensen, A., Harvey, J.H., Huston, T.L., Levinger, G., McClintock, E., Peplau, L.A., Peterson, D.R. (Eds.), Close Relationships. Freeman, New York, pp. 110–168. Blanchflower, D.G., Oswald, A.J., 2004. Money, sex and happiness: an empirical study. Scand. J. Econ. 106 (3), 393–415. Blumstein, P., Schwartz, P., 1983. American Couples: Money, Work, Sex. Morrow, New York. Botros, S.M., Abramov, Y., Miller, J.J.R., Sand, P.K., Gandhi, S., Nickolov, A., Goldberg, R.P., 2006. Effect of parity on sexual function: an identical twin study. Obstet. Gynecol. 107 (4), 765–770. Brewis, A., Meyer, M., 2005. Marital coitus across the life course. J. Biosoc. Sci. 37 (04), 499–518. Brüderl, J., Ludwig, V., 2014. Fixed-effects panel regression. In: Best, H., Wolf, C. (Eds.), Sage Handbook of Regression Analysis and Causal Inference. Sage Publishers, London, pp. 327–356. Calhoun, L.G., Selby, J.W., King, H.E., 1981. The influence of pregnancy on sexuality: a review of current evidence. J. Sex Res. 17 (2), 139–151. Call, V., Sprecher, S., Schwartz, P., 1995. The incidence and frequency of marital sex in a national sample. J. Marriage Fam. 57 (3), 639–652. Castelo-Branco, C., Palacios, S., Ferrer-Barriendos, J., Alberich, X., 2010. Do patients lie? An open interview vs. a blind questionnaire on sexuality. J. Sex. Med. 7 (2pt2), 873–880. Charny, I.W., Asineli-Tal, S., 2004. Study of ‘‘Sex-Less’’ (Sex-Avoidant) young couples. J. Fam. Psychother. 15 (1–2), 197–217. Cheung, M.W.L., Wong, P.W.C., Liu, K.Y., Yip, P.S.F., Fan, S.Y.-S., Lam, T.-H., 2008. A study of sexual satisfaction and frequency of sex among Hong Kong Chinese couples. J. Sex Res. 45 (2), 129–139. Christopher, F.S., Sprecher, S., 2000. Sexuality in marriage, dating, and other relationships: a decade review. J. Marriage Fam. 62 (4), 999–1017. Cohen, J., Manning, W., 2010. The relationship context of premarital serial cohabitation. Soc. Sci. Res. 39 (5), 766–776. De Judicibus, M.A., McCabe, M.P., 2002. Psychological factors and the sexuality of pregnant and postpartum women. J. Sex Res. 39 (2), 94–103. Donnelly, D.A., 1993. Sexually inactive marriages. J. Sex Res. 30 (2), 171–179. Dzara, K., 2010. Assessing the effect of marital sexuality on marital disruption. Soc. Sci. Res. 39 (5), 715–724. Edwards, J.N., Booth, A., 1994. Sexuality, marriage, and well-being: the middle years. Sex. Across Life Course, 233–259. Eisenberg, M.L., Shindel, A.W., Smith, J.F., Breyer, B.N., Lipshultz, L.I., 2010. Socioeconomic, anthropomorphic, and demographic predictors of adult sexual activity in the United States: data from the national survey of family growth. J. Sex. Med. 7 (1), 50–58. Erol, B., Sanli, O., Korkmaz, D., Seyhan, A., Akman, T., Kadioglu, A., 2007. A cross-sectional study of female sexual function and dysfunction during pregnancy. J. Sex. Med. 4 (5), 1381–1387. Finer, L.B., 2007. Trends in premarital sex in the United States, 1954–2003. Public Health Rep. 122 (1), 73–78. Furman, W., Buhrmester, D., 1985. Children’s perceptions of the personal relationships in their social networks. Dev. Psychol. 21 (6), 1016–1024. Glazener, C.M.A., 1997. Sexual function after childbirth: women’s experiences, persistent morbidity and lack of professional recognition. Br. J. Obstet. Gynaecol. 104 (3), 330–335. González, L., Viitanen, T.K., 2009. The effect of divorce laws on divorce rates in Europe. Eur. Econ. Rev. 53 (2), 127–138. Greenblat, C.S., 1983. The salience of sexuality in the early years of marriage. J. Marriage Fam. 45 (2), 289–299. Heuveline, P., Timberlake, J.M., 2004. The role of cohabitation in family formation: the United States in comparative perspective. J. Marriage Fam. 66 (5), 1214–1230.
82
J. Schröder, C. Schmiedeberg / Social Science Research 52 (2015) 72–82
Huinink, J., Brüderl, J., Nauck, B., Walper, S., Castiglioni, L., Feldhaus, M., 2011. Panel analysis of intimate relationships and family dynamics (pairfam): conceptual framework and design. Z. Familienforschung – J. Fam. Res. 23, 77–101. Hyde, J.S., DeLamater, J.D., Plant, E.A., Byrd, J.M., 1996. Sexuality during pregnancy and the year postpartum. J. Sex Res. 33 (2), 143–151. Hyde, J.S., DeLamater, J.D., Durik, A.M., 2001. Sexuality and the dual-earner couple, Part II: beyond the baby years. J. Sex Res. 38 (1), 10–23. James, W.H., 1981. The honeymoon effect on marital coitus. J. Sex Res. 17 (2), 114–123. Julien, D., Chartrand, E., Simard, M.-C., Bouthillier, D., Bégin, J., 2003. Conflict, social support and relationship quality: an observational study of heterosexual, gay male and lesbian couples’ communication. J. Fam. Psychol. 17 (3), 419–428. Karraker, A., DeLamater, J., Schwartz, C.R., 2011. Sexual frequency decline from midlife to later life. J. Gerontol. Ser. B: Psychol. Sci. Soc. Sci. 66B (4), 502–512. Kinsey, A., Pomeroy, W., Martin, C., 1948. Sexual Behavior in the Human Male. Saunders, Philadelphia. Kinsey, A., Pomeroy, W., Martin, C., Gebhard, P., 1953. Sexual Behavior in the Human Female. Saunders, Philadelphia. Lancaster, J., 1994. Human sexuality, life histories, and evolutionary ecology. In: Rossi, A. (Ed.), Sexuality across the Life Course. The University of Chicago Press, Chicago, pp. 39–62. Levin, I., 2004. Living apart together: a new family form. Curr. Sociol. 52 (2), 223–240. Lichter, D.T., Turner, R.N., Sassler, S., 2010. National estimates of the rise in serial cohabitation. Soc. Sci. Res. 39 (5), 754–765. Liu, C., 2000. A theory of marital sexual life. J. Marriage Fam. 62 (2), 363–374. Mazur, A., Mueller, U., Krause, W., Booth, A., 2002. Causes of sexual decline in aging married men. Int. J. Impot. Res. 14 (2), 101. McCarthy, B., 2003. Marital sex as it ought to be. J. Fam. Psychother. 14 (2), 1–12. Metz, M.E., Epstein, N., 2002. Assessing the role of relationship conflict in sexual dysfunction. J. Sex Marital Ther. 28 (2), 139–164. Nauck, B., Brüderl, J., Huinink, J., & Walper, S. (2013). Beziehungs- und Familienpanel (pairfam). GESIS Datenarchiv, Köln. ZA5678 Datenfile Version 4.0.0. http://dx.doi.org/10.4232/pairfam.5678.4.0.0. O’Fallon, J.B., Tsui, A., Adewuyi, A., 2003. Social and proximate determinants of sexual activity in rural Nigeria. J. Biosoc. Sci. 35, 585–599. Olson, D., DeFrain, J., 1994. Marriage and the Family: Diversity and Strengths. Mountain View, CA, Mayfield. Riportella-Muller, R., 1989. Sexuality in the elderly: a review. In: McKinney, K., Sprecher, S. (Eds.), Human Sexuality: The Societal and Interpersonal Context. Ablex, Norwood, NJ, pp. 210–236. Rosen, R.C., Bachmann, G.A., 2008. Sexual well-being, happiness, and satisfaction in women: the case for a new conceptual paradigm. J. Sex Marital Ther. 34, 291–297. Schneidewind-Skibbe, A., Hayes, R.D., Koochaki, P.E., Meyer, J., Dennerstein, L., 2008. The frequency of sexual intercourse reported by women: a review of community-based studies and factors limiting their conclusions. J. Sex. Med. 5 (2), 301–335. Stafford, L., Kline, S.L., Rankin, C.T., 2004. Married individuals, cohabiters, and cohabiters who marry: a longitudinal study of relational and individual wellbeing. J. Soc. Pers. Relat. 21 (2), 231–248. Trussell, J., Westoff, C.F., 1980. Contraceptive practice and trends in coital frequency. Fam. Plann. Perspect. 12 (5), 246–249. Udry, J.R., 1980. Changes in the frequency of marital intercourse from panel data. Arch. Sex. Behav. 9 (4), 319–325. Udry, J.R., Deven, F.R., Coleman, S.J., 1982. A cross-national comparison of the relative influence of male and female age on the frequency of marital intercourse. J. Biosoc. Sci. 14 (01), 1–6. Westoff, C.F., 1974. Coital frequency and contraception. Fam. Plann. Perspect. 6 (3), 136–141. Widmer, E.D., Treas, J., Newcomb, R., 1998. Attitudes toward nonmarital sex in 24 countries. J. Sex Res. 35 (4), 349–358. Wiederman, M.W., 1999. Volunteer bias in sexuality research using college student participants. J. Sex Res. 36 (1), 59–66. Wilcox, W.B., Marquardt, E. (2011). The state of our unions: Marriage in America 2010. The National Marriage Project and the Center for Marriage and Families at the Institute for American Values, University of Virginia.