Sei sulla pagina 1di 12

CHIEN LIU Wagner College

A Theory of Marital Sexual Life

Why does the frequency of marital sex decrease Previous research reports that the frequency of
with marital duration? Does the probability of in- marital sex declines with marital duration (Blum-
volvement in extramarital sex increase or de- stein & Schwartz, 1983; Call, Sprecher, &
crease with marital duration, and why? I develop Schwartz, 1995; Greenblat, 1983; James, 1981;
a theory to answer these questions. I apply the Jasso, 1985; Kinsey, Pomeroy, & Martin, 1948;
law of diminishing marginal utility and human Kinsey, Pomeroy, Martin, & Gebhard, 1953; Mas-
capital theory to explain the basic observed de- ters, Johnson, & Kolodny, 1992; Trussell & Wes-
cline in the frequency of marital sex with marital toff, 1980; Udry, Deven, & Coleman, 1982). Al-
duration. Based on the above explanation, I de- though aging is correlated with this decline,
velop a discussion on the relationship between Blumstein and Schwartz (1983) find empirical ev-
marital duration and extramarital sex, and I de- idence that the impact of age and duration are ap-
rive two hypotheses, which are supported by sta- proximately equal. Moreover, it has been discov-
tistical analyses of the National Health and Social ered that the declining frequency of marital sex
Life Survey data. cannot be accounted for by ‘‘background’’ vari-
ables such as premarital sex, education, income,
and religion (Greenblat, 1983; Masters et al.,
It is well known that marital duration negatively 1992).
affects the frequency of marital sex. However, we Previous studies on marital sex frequency and
know little about how martial duration affects ex- extramarital sex are strong on empirical analysis
tramarital sex. Extramarital sex has enormous dis- but weak on theoretical explanation. Much of the
ruptive potential for marriages, and it is frequently previous research on marital sexual behavior has
cited by divorced persons as a prime reason for been in the tradition of ‘‘sexual demography’’, or
the dissolution of their marriages (South & Lloyd, a mapping of the relationships between sexual be-
1995). Therefore, a better understanding of the havior and social background variables (Davis,
factors influencing the frequency of marital and 1974; Edwards & Booth, 1976; Thompson, 1983).
extramarital sex should be of great theoretical and For example, in their studies of the frequency of
practical value. However, factors contributing to marital sex, most researchers focus mainly on in-
these two types of sexual behavior are not well dividual or social characteristics and on social
understood. This article develops a theory that ex- control, attributing the declining frequency of
plains the basic observed patterns of sex between marital sex to biological aging, attitudes about
marital partners and makes two predictions con- sex, the work status of a couple, presence of
cerning extramarital sex. young children, marital satisfaction, race, gender,
and religion (Call et al., 1995; Greenblat, 1983;
Department of Sociology/Anthropology, Parker Hall, Wag- James, 1981; Olson & DeFrain, 1994; Udry et al.,
ner College, Staten Island, NY 10301 (chienliu@wagner.
edu)
1982). The most commonly cited explanation of
this phenomenon is an intuitive one: The declining
Key Words: extramarital sex, gender, marital sex, theory. frequency of marital sex is due to the loss of nov-

Journal of Marriage and the Family 62 (May 2000): 363–374 363


364 Journal of Marriage and the Family

elty, which is often referred to as the ‘‘honeymoon tween marital duration and extramarital sex and
effect’’ or the ‘‘Coolidge effect’’ (Blumstein & derive two hypotheses. I test the derived hypoth-
Schwartz, 1983; Call et al., 1995; James, 1981; eses by a statistical analysis of the data from the
Jasso, 1985). Although these studies have contrib- National Health and Social Life Survey (NHSLS)
uted to our knowledge of the patterns of marital (Laumann, Gagnon, Michael, & Michaels, 1994).
sex, they hardly go beyond a description of these The theory developed in this article can im-
patterns to develop a theory of marital sexual be- prove our understanding of some patterns of mar-
havior. Specifically, in spite of an accumulation of ital sexual life and individuals’ motivation to en-
evidence that the frequency of marital sex declines gage in extramarital sex. Specifically, this theory
with marital duration, this observed phenomenon answers questions such as: Why, and how, does
remains theoretically unexplained; that is, a causal the frequency of marital sex decrease with marital
mechanism is not identified. (A theory is a set of duration? Does the probability of involvement in
explanations that forms a deductive system; a phe- extramarital sex increase or decrease with marital
nomenon is said to be explained if it can be de- duration, and why? Although the following anal-
duced from some initial conditions and at least ysis does not cover all aspects of marital sexual
one postulate. [See Homans, 1967; Jasso, 1985; life, the systematic treatment of two fundamental
Liu 1996; Markovsky, 1994; Popper, 1972.]) aspects perhaps justifies this paper’s use of the ti-
In recent years, some interesting hypotheses tle ‘‘A theory of marital sexual life.’’
concerning human sexual tendencies have been
derived from evolutionary perspectives such as
THEORY
sociobiology and evolutionary biology (for a re-
view see Lancaster, 1994; Posner, 1992). Evolu-
Explaining the Declining Frequency of Marital
tionary approaches argue that individuals are
Sex: Diminishing Marginal Utility and Human
faced with trade-offs in the allocation of resources
Capital Investment
between mating effort and parental investment. It
follows that once a couple has children, the part- Rational choice theory postulates that an actor
ners will shift their resource allocation, devoting chooses a course of action that maximizes his/her
less resources to marital sex and more to the nur- utility, subject to the constraints of this actor’s re-
ture and support of their offspring. Consequently, sources and social institutions. It is also common-
the frequency of marital sex declines. ly assumed that, other things being equal, the mar-
Though evolutionary theory provides some ginal utility (MU) of consuming a good or service
general hypotheses regarding human sexual ten- diminishes as the consumption of that good or ser-
dencies, it tells us little about how individuals vice increases (Coleman, 1990). This assumption
make their sexual choices. Human beings make is often called the law of diminishing marginal
choices and are able to resist the natural tendency utility (DMU). Below, I apply the postulate of ra-
developed in evolution (Posher, 1992). For in- tionality and the DMU assumption to explain the
stance, more and more couples intentionally re- phenomenon of the declining frequency of marital
main childless. As a result, evolutionary ap- sex with marital duration.
proaches might not be able to explain some Viewed as rational action, sexual behavior
empirical patterns of sexual behavior. For exam- shows substantial points of similarity to other
ple, previous research reports that the frequency types of behavior. First, sexual behavior is typi-
of marital sex declines with marital duration, net cally voluntary. Second, people have preferences.
of the effect of the presence of children (Call et Third, a market for sexual interaction exits. In ad-
al., 1995; Jasso, 1985). This empirical finding dition, sexual action provides a certain level of
would be hard for evolutionary theory to explain. utility. The problem that an actor faces is how to
By contrast, social scientists are mainly inter- maximize the utility derived from his or her mar-
ested in explaining individual behavior and how ital sexual actions and other actions subject to the
macrosocial phenomena emerge from individual constraints of his or her resources. The problem
actions (Coleman, 1990). In social sciences, ratio- can be formulated as: Max U (S, X), subject to G
nal choice theory analyzes how individuals (S, X) 5 C; where U denotes a utility function;
choose their courses of action. In what follows, I G denotes a constraint; S denotes marital sex; X
embed the ‘‘honeymoon effect’’ explanation with- denotes the consumption of another good or ser-
in a much broader rational choice approach to sex- vice produced in the family (e.g., listening to mu-
ual behavior. Then I analyze the relationship be- sic with a spouse); and C is a constant denoting
A Theory of Marital Sexual Life 365

the total amount of resources at an actor’s dispos- framework of rational choice theory. An equally
al. A necessary condition for maximizing U sub- plausible counterargument can also be derived
ject to G is: from the rational choice theory of marriage. Ac-
cording to Becker, Landes, and Michael (1977),
MUs /a 5 MUx /b (1) married persons invest in their marriage specific
capital, including houses, children, market and
where MUs denotes the marginal utility of S; MUx nonmarket skills, sexual adjustment with one’s
denotes the marginal utility of X; a and b are the spouse, and information. England and Farkas
units of physical, emotional, or time input re- (1986) argue that a marriage is a long-term rela-
quired for obtaining each unit of S and X (i.e., tionship and that both partners have incentives to
price or marginal cost). Equation (1) means that invest in the marriage specific human capital—
at the utility-maximizing point, each unit of S and abilities to provide empathy, companionship, sex-
X should yield the same amount of marginal util- ual pleasure, intellectual pleasure, social status, or
ity given one unit of input allocated to them. Oth- earnings. The accumulation of relationship-specif-
erwise, the total utility can be increased by allo- ic human capital raises the expected gain and
cating more resources to the good or service that makes both parties better off (England & Farkas,
brings about a larger amount of marginal utility. 1986).
Because of DMU, we know that, other things be- Clearly, this perspective suggests that, with an
ing equal, MUs decreases as S increases. As MUs increase in human capital stock—knowledge
decreases, equality (1) will become an inequality: about the spouse’s sexual preferences, desire, hab-
MUs/a , MUx/b, ceteris paribus. As a result, giv- its, and skills—the pleasure of marital sex could
en the constraint, the total utility could be in- actually increase with the knowledge gained from
creased by shifting resources from S to X; that is, past sexual experience between spouses. In other
more of X will be consumed, and S will become words, as marital sex increases, it might become
less frequent. This reallocation of resources will increasingly rewarding; that is, MUs could in-
continue until the equality MUs/a 5 MUx/b is re- crease with marital duration. This theory actually
stored. predicts that the effect of human capital invest-
In other words, marital sexual actions between ment will cause the frequency of marital sex to
a husband and a wife initially bring about a rel- increase with marital duration. For simplicity, this
atively high level of satisfaction; therefore, one explanation will be referred to as the ‘‘human cap-
can expect sexual activity to be more frequent. As ital explanation’’ or the ‘‘human capital effect.’’
marital sex increases, the level of satisfaction low- Adding the human capital effect to the DMU
ers; thus fewer resources will be allocated to it. explanation, one can better explain the empirical
Consequently, the frequency of marital sex de- pattern of the frequency of marital sex: Both the
clines. (A formal derivation is provided in the Ap- DMU effect and the human capital effect could
pendix.) If correct, this theory claims that, ceteris affect changes in the frequency of marital sex. The
paribus, the sole effect of the diminishing marital empirical finding that the frequency of marital sex
utility of marital sex will cause the frequency of declines less and less steeply with duration sug-
marital sex to decline with marital duration. For gests that the DMU effect dominates the human
simplicity, this explanation will be referred to as capital effect. The role played by the human cap-
the ‘‘DMU explanation’’ or the ‘‘DMU effect’’ in ital effect is to offset some of the DMU effect so
the remainder of this paper. that the frequency of marital sex declines less and
Although the DMU explanation predicts a cor- less steeply with marital duration. Thus, the phe-
rect direction of the changes in the frequency of nomenon of the declining frequency of marital sex
marital sex, it is not sufficient to explain the em- with marital duration is deduced from or ex-
pirical pattern of the frequency of marital sex. plained by rational choice theory.
Specifically, the DMU explanation generally pre-
dicts that the frequency of marital sex decreases
Two Hypotheses Concerning Extramarital Sex
at a constant rate (see Appendix). But the previous
research shows that, ceteris paribus, the frequency A married actor can have two types of partnered
of marital sex decreases less and less steeply sex, marital sex and extramarital sex. What is the
(James, 1981; Jasso, 1985). Thus other factors relationship between marital duration and extra-
need to be considered to improve this explanation. marital sex? The DMU and human capital expla-
The DMU explanation is not unique within the nations about marital sex have opposite implica-
366 Journal of Marriage and the Family

tions. Following the human capital explanation, (England & Farkas, 1986); in addition, physical
one would expect the likelihood of an actor’s en- pleasure is more important to men than to women
gaging in extramarital sex to decrease with marital in a sexual relationship (Masters et al., 1992). We
duration for two reasons. First, marital sex is in- can expect that, initially, a man has more incentive
vested with affective elements, ‘‘thereby creating to invest in his marriage specific human capital,
‘socioemotional closeness and exchange’ rather because the marginal return on the investment is
than just ‘psychophysiological pleasure and re- larger due to the lack of such capital, suggesting
lief’’’ (Posner 1992, p. 112), implying that marital that he is less likely to engage in extramarital sex.
sex provides a higher level of utility than extra- As the investment increases, its marginal return
marital sex. Second, the investment in the rela- will eventually decrease because of the law of di-
tionship-specific human capital, which makes minishing marginal returns (Becker, 1991; Ehr-
marital sex increasingly rewarding over time, neberg & Smith, 1994; Lipsey & Courant, 1996).
should make extramarital sex an increasingly less For example, Rubin (1992) observed that couples
attractive option. Therefore, the probability of ex- tried different ways (i.e., investing in the human
tramarital sex should become increasingly less capital) to enhance their marital sexual satisfac-
likely with marital duration. tion; however, the results were less and less sat-
By contrast, the DMU explanation suggests an isfactory. Therefore, we can expect that, at a point
opposite prediction. If the MUs decreases with
in the marriage (call this point a critical value of
marital duration, it can be hypothesized that, other
marital duration), MUs and marginal returns of hu-
things being equal, over the course of a marriage
man capital investment could diminish to such an
a rational actor would become more likely to en-
gage in extramarital sex for the following reason: extent that the expected payoff of extramarital sex
The DMU assumption suggests that, ceteris pari- becomes larger than its expected cost, implying
bus, the MU of a sexual action with an old partner that a man has stronger and stronger incentives to
is lower than that of a sexual action with a new engage in extramarital sex to seek a higher level
partner. As marital duration increases, MUs con- of satisfaction from sexual variety. Thus we can
tinues to decline so that it becomes increasingly predict:
lower than that of a sexual action with a new part-
ner. Therefore, a rational actor would become in-
Hypothesis 1: For men, the relationship be-
creasingly more likely to engage in extramarital
tween the likelihood of extramarital sex and
sex because the total utility can be increased by
allocating more resources to the good or service marital duration is convex (U-shaped); that
that yields a larger amount of MU. is, over the course of marriage, the likeli-
It is well known that in families men and wom- hood of extramarital sex decreases to a min-
en play different roles, have different preferences, imum at a point in marriage and thereafter
and express themselves sexually in different ways. increases.
These gender differences imply that the effects of
marital duration on extramarital sex might vary In contrast to married men, women tend to
by gender. Though it is nearly impossible to ex- make more relationship-specific investments
amine gender difference in the frequency of mar- (England & Farkas, 1986). It is also known that,
ital sex, gender differences in extramarital sex can for women, emotional attachment and satisfaction
be examined and are well known. The NHSLS are more important than physical pleasure in a
reports that 24.5% of men and 15% of women sexual relationship (Buss, 1994; Masters et al.,
have experienced extramarital sex (Laumann et 1992). Therefore, one can expect that, ceteris par-
al., 1994). In addition, extramarital sex incurs ibus, the longer a woman stays in a marriage, the
costs of negative internal and external sanctions more relationship-specific human capital she ac-
such as feelings of guilt and the possible disso- cumulates, the higher the expected cost of extra-
lution of the marriage if the extramartial sex is- marital sex, and, as a result, the less likely she
discovered. Therefore, adding gender and cost will engage in it. Thus we can predict:
considerations to the current analysis can help
specify the relationship between extramarital sex
and marital duration. Hypothesis 2: For women, the likelihood of
Previous research shows that men tend to in- extramarital sex decreases with marital du-
vest more in their career than in a relationship ration.
A Theory of Marital Sexual Life 367

METHOD Control Variables

Data The Master status variables. Prior research on


sexual behavior has identified several social and
The empirical analyses presented below are based demographic variables that influence sexual be-
on the NHSLS, which is a probability sample of havior. According to Laumann et al. (1994), gen-
3,432 noninstitutionalized U.S. residents ages 18 der, race, age, education, marital status, and reli-
to 59. The analyses are based on two sets of mar- gion are the basic components of individuals’
ried respondents. The first set consisted of those master status that organize their patterning of so-
who had not yet experienced divorce or separation cial relations, shape their understanding of the so-
by the time of the interview. The second set con- cial world, and influence their sexual behavior.
sisted of those who had experienced either divorce Among these master status variables, this study
or separation by the time of the interview. For the controlled for age, gender, race, education, and
latter set, the analyses focused on extramarital church attendance. Age was included as a contin-
sexual activity during the most recent marriage. uous variable, and a quadratic term of age was
Homosexual respondents were excluded. also incorporated into the model. Some research-
ers find that age and marital duration are linearly
Dependent Variable correlated. In this study, the value of the deter-
Hypotheses 1 and 2 were tested by estimating lo- minant of correlation between age and marital du-
gistic regression equations. To examine the hy- ration is .95, indicating no sign of severe multi-
pothesized relationship between marital duration collinearity. (The value of determinant 5 1
and the likelihood of extramarital sex, one needs indicates no multicollinearity at all, while the val-
a dependent variable that specifies when extra- ue of the determinant 5 0 indicates perfect mul-
marital sex occurred. The NHSLS provides such ticollinearity.) However, the age at marriage is lin-
information by asking respondents about the num- early correlated with marital duration; therefore,
ber of their sexual partners during the past year it was not included in the models (the value of
and the past 5 years. Because the selected respon- the determinant of correlation is .08). Laumann
dents had been living in a marriage during the past and his colleagues (1994) found the period effect
5 years, they undoubtedly had engaged in extra- of sexual revolution. In a separate analysis, the
marital sex if they had more than one sexual part- birth cohorts used by Laumann et al. were also
ner. The restriction that marital duration is equal incorporated into the analyses. The cohorts have
to or greater than 5 years was imposed in calcu- no significant effect on extramarital sex for either
lating the descriptive statistics and in the logistic gender. Results are available upon request.
regression analysis. (The percentage of respon- Race (White, Black, and other racial group)
dents who engaged in extramarital sex during the was included as a dummy variable. ‘‘Other racial
previous year was 3.6%, too small to create suf- group’’ served as the reference group. Education,
ficient variation in the dependent variable; there- measured by highest grade completed by the re-
fore, this variable was not used in the analysis. In spondent, was included as a continuous variable,
a separate analysis where this variable was used ranging from 1 (grade 8 or less) to 8 (other ad-
as the dependent variable, even though most of vanced degree).
the coefficients of the marital duration were con- Church attendance was included in the model
sistent with the two hypotheses, they were not sig- as a continuous variable, ranging from 0 (never)
nificant. Results are available upon request.) to 8 (several times a week). Church attendance
might affect an actor’s extramarital sexual behav-
ior in two ways. First, a person who belongs to
The Major Independent Variable and attends church is regularly exposed to sexual
Marital duration, the major explanatory variable, norms that proscribe extramarital sex; therefore,
was measured by years and was obtained by sub- he or she is more likely to internalize such norms
tracting the time when the marriage began or and adhere to them. Thus extramarital sex might
when the couple started living together from the incur a heavy cost of internal sanctions. Second,
time of the interview. Marital duration squared it can be said that those who regularly attend
was incorporated to model the hypothesized con- church services form a closure of social networks.
vex relationship between the dependent variable This closure is featured by frequent communica-
and duration of marriage. tion among members and is indispensable for
368 Journal of Marriage and the Family

sanctioning a target actor who violates norms spondents’ sexual interests were controlled for.
(Coleman, 1990). Taking these two factors into The NHSLS asked respondents how often they
account, one can hypothesize that those who go think about sex (Thinksex) and how often think-
to church regularly are less likely to engage in ing about sex makes them feel guilty (Guiltsex).
extramarital sex than those who do not. For Thinksex, the response categories range from
1 (once a month) to 6 (several times a day). For
Marital sex, children, and health. According to Guiltsex, the response categories range from 1
the theory developed above, declining physical (never) to 5 (nearly always). One can expect, oth-
pleasure of marital sex has a positive effect on the er things being equal, that individuals with strong
likelihood of extramarital sex. To examine this ef- sexual interests will be more likely to engage in
fect in the empirical analyses, I incorporated into extramarital sex, and individuals who feel guilty
the model a variable that measures the physical thinking about sex will be less likely to engage in
pleasure of marital sex. The NHSLS asked re- extramarital sex.
spondents how physically pleasurable they found Masturbation is another indicator of sexual in-
their relationship with their primary sexual partner terest and preference. Masturbation can be either
(i.e., spouse for the sample selected for this study) a complement to, substitute for, or independent of
to be: extremely pleasurable, very pleasurable, partnered sex (Laumann et al., 1994), suggesting
moderately pleasurable, slightly pleasurable, or that how masturbation affects extramarital sex is
not at all pleasurable. The categories were coded mainly an empirical question. Therefore, mastur-
from 1 to 5, with higher scores indicating higher bation should be incorporated into the model.
physical pleasure from marital sex. (Marital hap- However, the measurement of masturbation poses
piness was not included because was not available a limitation for this study. In the NHSLS, respon-
in the data set.) dents were asked to report masturbating for the
The number of children of respondents was past year. However, the dependent variable is mea-
also incorporated into the analyses. Childrearing sured by the extramarital sexual activity in the
is a form of human capital investment that in- past 5 years. Due to this temporal mismatch be-
creases a person’s stake in his or her family. tween the two variables, no masturbation variable
Therefore, one can expect the number of children was included in the model. In a separate analysis,
to reduce the likelihood of extramarital sex. Chil- the frequency of masturbation was incorporated
dren do increase a person’s stake in relationship- into the model. The results are similar to those
specific investments; however, because children reported in this paper. The frequency of mastur-
normally require an intense and continuing con- bation had no significant effect on extramarital sex
tribution in time and effort from both parents, for either gender. Results are available upon re-
their presence might also negatively affect a cou- quest.
ple’s sex life. For example, if one or both partners
devote too much attention to their children, the Attitudes about extramarital sex. The NHSLS also
frequency or quality of the couple’s sexual en- asked respondents about their attitudes about ex-
counters might be reduced. This would increase tramarital sex (Exmarsex). This variable was cod-
the probability of one or both of them becoming ed from 1 (always wrong) to 4 (not wrong at all).
involved in extramarital sex. The presence of chil- One can expect, other things being equal, that in-
dren might have either of the two opposing effects dividuals who think that extramarital sex is wrong
on extramarital sex. Thus, whether the presence will be less likely to engage in extramarital sex.
of children increases or decreases the probability
of extramarital sex is an empirical question, de- Employment characteristics. Because opportunity
pending on which effect is dominant. is an important constraint on an actor’s search for
In addition, this study controlled for the re- a sexual partner, the empirical analyses included
spondents’ health status. The NHSLS asked re- four variables that measured employment charac-
spondents to describe their general health. The re- teristics. The NHSLS asked respondents: ‘‘At
sponse categories are (1) excellent, (2) good, (3) your job, is it frequently the case that you are
fair or (4) poor. It is plausible to expect that a alone with one of your clients, customers or co-
healthier actor is more likely to engage in extra- workers?’’ (Wkalone); ‘‘Does your job require
marital sex, other things being equal. that you touch your clients, customers or co-work-
ers?’’ (Wktouch); and ‘‘Does your job require that
Sexual interests. Two variables that measure re- you discuss the personal concerns of your clients,
A Theory of Marital Sexual Life 369

TABLE 1. MEANS AND STANDARD DEVIATIONS FOR THE VARIABLE IN THE ANALYSIS: NHSLS

Variable Both Genders Male Female

Dependent variable
Extramarital sex (0 5 no, 1 5 yes) .099 .143 .066
(.298) (.350) (.249)
Independent variables
Marital duration 17.454 17.437 17.466
(9.246) (8.984) (9.444)
(Marital duration) 2
390.067 384.628 394.159
(391.582) (375.521) (403.431)
Control variables
Gender (0 5 female, 1 5 male) .429 — —
(.495)
Respondent’s age 41.268 42.689 40.230
(9.217) (8.848) (9.354)
(Respondent’s age)2 1,789.407 1,800.523 1,705.821
(783.620) (769.250) (784.430)
White (0 5 no, 1 5 yes) .829 .830 .822
(.379) (.376) (.382)
Black (0 5 no, 1 5 yes) .102 .100 .103
(.303) (.300) (.305)
Other (0 5 no, 1 5 yes) .072 .070 .074
(.259) (.255) (.262)
Education 4.247 4.384 4.171
(1.694) (1.767) (1.634)
Church attendance 4.216 3.885 4.466
(2.700) (2.682) (2.689)
Number of children 2.502 2.576 2.447
(4.005) (5.846) (1.555)
Physical pleasure of maritial sex 4.249 4.336 4.183
(.779) (.751) (.749)
Health 1.702 1.681 1.718
(.702) (.707) (.689)
Thinksex 3.055 3.438 2.768
(.938) (.878) (.877)
Guiltsex 1.208 1.217 1.201
(.580) (.569) (.589)
Exmarsex 1.305 1.380 1.248
(.671) (.766) (.583)
Central City (0 5 no, 1 5 yes) .180 .165 .192
(.384) (.371) (.394)
Wkalone (0 5 no, 1 5 yes) .664 .567 .739
(1.114) (.607) (1.383)
Wkdiscss (0 5 no, 1 5 yes) .419 .253 .548
(1.123) (.558) (1.402)
Wktouch (0 5 no, 1 5 yes) .322 .131 .472
(1.136) (.486) (1.437)
Total 1,372 589 783
Note: Means are in cells, and standard deviations are in parentheses.

customers or co-workers?’’ (Wkdiscss) One can a rural area; therefore, the place of residence was
expect that, other things being equal, those with also controlled for.
more opportunities (e.g., to be alone with, to dis-
cuss personal concerns with, or to touch clients,
RESULTS
customers or co-workers) will be more likely to
engage in extramarital sex. In addition, extramar- The descriptive statistics for the variables used in
ital sex is less costly, in terms of search and con- the empirical analyses are reported in Table 1. To
cealment, for people living and working in a large test the two hypotheses, separate logistic equa-
central city than for those living and working in tions controlling for other effects were estimated
370 Journal of Marriage and the Family

TABLE 2. COEFFICIENTS FOR THE LOGISTICS REGRESSION OF EXTRAMARITAL SEX DURING THE PAST 5 YEARS ON
SELECTED INDEPENDENT VARIABLES

Variable Both Genders Male Female

Marital duration 2.195*** 2.176** 2.227*


(.062) (.081) (.139)
(Marital duration)2 .004*** .005** .003
(.002) (.002) (.004)
Master status variables
Gender .627*** — —
(.247)
Respondent’s age .106 .315 .142
(.131) (.202) (.213)
(Respondent’s age)2 2.002 2.004* 2.002*
(.002) (.003) (.003)
White 2.409 2.241 2.522
(.413) (.579) (.646)
Black .979** .903 1.319*
(.471) (.665) (.743)
Education 2.201*** 2.178** 2.370***
(.071) (.089) (.133)
Church attendance 2.103** 2.114** 2.084
(.045) (.059) (.074)
Other control variables
Number of children 2.002 .088 2.236
(.076) (.094) (.167)
Physical pleasure of marital sex 2.513*** 2.712*** 2.373*
(.128) (.179) (.204)
Health 2.277 2.325 2.347
(.172) (.226) (.293)
Thinksex .423*** .490*** .263***
(.127) (.175) (.202)
Guiltsex .182 .085 .321
(.161) (.233) (.246)
Exmarsex .613*** .426*** 1.087***
(.133) (.151) (.262)
Central City .168 .075 .139
(.281) (.388) (.438)
Wkalone .436*** .380* .589
(.166) (.203) (.372)
Wkdiscss 2.091 .075 2.084
(.202) (.320) (.354)
Wktouch 2.216 2.048 2.338
(.213) (.253) (.439)
Constant 24.426* 28.841** 23.441
(2.432) (3.936) (3.700)
22 log likelihood 610.656 350.217 233.577
x2 137.925*** 63.770*** 81.665***
Degree of freedom 19 18 18
Critical value of marital duration 23.036 17.600 37.833
Total 1,198 518 680
Note: Standard errors are in parentheses.
*p , .10. **p , .05. *** p , .01 (two-tailed tests).

for each gender. The results are reported in Table varies nonmonotonically over the course of the
2. In the estimated model for men, the coefficient marriage, decreasing to a minimum at about the
for marital duration was -.176, and the coefficient 18th wedding anniversary (the critical value 5
for the quadratic term was .005; both were statis- 17.6), and thereafter increasing continuously. In
tically significant. This indicates that, as predicted the estimated model for women, the coefficient for
in Hypothesis 1, the likelihood of extramarital sex marital duration was -.227, and the coefficient for
A Theory of Marital Sexual Life 371

FIGURE 1. ODDS OF EXTRAMARITAL SEX DURING THE PAST 5 YEARS BY MARITAL DURATION AND BY GENDER

the quadratic term was .003; the coefficient for The analysis above shows that even though
marital duration was statistically significant, some of the control variables proved statistically
whereas the coefficient for the quadratic term was significant, they in no way negated the direct im-
not. This indicates that, for women, a critical val- pact of marital duration on extramarital sex. I thus
ue of marital duration does not exist, and the like- arrived at following sociological findings: For
lihood of extramarital sex decreases with marital men, the relationship between the likelihood of
duration at a constant rate. This result supports extramarital sex and marital duration is convex;
Hypothesis 2. The hypothesized effects of marital that is, the likelihood decreases to a minimum at
duration on extramarital sex for both genders are a point in the marriage, then increases thereafter.
shown by the two graphs presented in Figure 1. For women, the likelihood of engaging in extra-
Among the control variables, education had a marital sex decreases with marital duration.
negative effect on extramarital sex for both gen-
ders. This effect was greater for women than for
DISCUSSION
men. Similarly, for both genders, individuals who
had a stronger sexual interest and who thought I develop two explanations of marital sex fre-
that extramarital sex was not wrong were more quency to account for the declining frequency of
likely to engage in extramarital sex. Church atten- marital sex. I discuss the relationship between
dance had a negative effect on extramarital sex marital sex and extramarital sex. The theory de-
for men but not for women. The opportunity of veloped here not only formalizes what is known
working alone with customers, clients, or cowork- about marital sex frequency, but also generates
ers had a positive effect on extramarital sex for new predictions about the relationship between
men only. It is worth noting that for both genders, marital duration and extramarital sex. Although
those who found their marital sex less physically people engage in extramarital sex for various rea-
pleasurable were more likely to engage in extra- sons (e.g., an unhappy marriage, career advance-
marital sex; this effect was greater for men than ment, search for a new partner, etc.), this study
for women. This finding provides additional em- shows that marital duration is certainly an impor-
pirical evidence supporting the argument that de- tant factor that affects the likelihood of extramar-
clining physical pleasure of marital sex increases ital sex.
the likelihood of extramarital sex. The other con- The analysis above improves our understand-
trol variables had no significant effect on extra- ing of marital sexual behavior in the following
marital sex. three aspects:
372 Journal of Marriage and the Family

1. It improves our understanding of marital he is not likely to engage in extramarital sex, giv-
sex. Some researchers attribute the declining fre- en the fact that his wife’s threat of sanctioning him
quency of marital sex with marital duration to fa- is less credible. For example, suppose marital sex
tigue from work or to the fact that both husband gives a middle-aged man 5 units of utility, where-
and wife work full-time outside the home and as extramarital sex gives him only 2 units. In this
therefore do not have the interest or time for mar- case, he has no reason to engage in extramarital
ital sex (Olson & DeFrain, 1994). This effect is sex because he can maximize his utility by getting
referred to as the ‘‘DINS dilemma’’ (double in- 5 units from marital sex instead of 2 units from
come, no sex). Now, in light of the findings of extramarital sex. This example suggests that this
this paper, this explanation becomes questionable. middle-aged man will have incentive to engage in
Compared to marital sex, extramarital sex is much extramarital sex only if its marginal utility (say, 5
more time consuming and costly in terms of units) is larger than that of marital sex (say, 2
search and concealment. If couples who work full units), implying a necessary condition. It follows
time have no time for marital sex, they will have that only under such circumstance does the cred-
even less time to engage in extramarital sex. How- ibility of sanction have a role to play. Thus the
ever, the analysis above shows that the likelihood theory developed here suggests that the diminish-
of extramarital sex increases later in a couple’s ing marginal utility of marital sex is a deeper
marriage, implying that the DINS explanation is causal mechanism of extramarital sex than the
not supported by the analysis developed here. credibility of the threatened sanctions that will be
Moreover, Call et al. (1995) empirically ex- imposed on those who engage in extramarital sex.
amine the effects of work status on the frequency Like any other theoretical implications, this one
of marital sex. They report that the frequency of needs to be examined empirically in the future.
marital sex declines with marital duration when This study on individual sexual behavior has
the couples’ work status is held constant, and that an important implication for the institution of mar-
working full-time actually has significant and pos- riage. The declining marginal utility of marital sex
itive effects on the frequency of marital sex. As a over the course of a marriage suggests that mar-
result, they conclude that the DINS dilemma riage as a social institution for channeling sex for
might be a myth (Call et al., 1995). This finding married individuals, has a built-in defect that pos-
provides empirical evidence that falsifies the es a potential threat to marital stability. This po-
DINS explanation. Thus this study and the one by tential threat makes a marriage increasingly vul-
Call et al. rule out the DINS explanation. nerable to marital instability as a society moves
2. In contrast to some of the previous research from a primordial society to a modern industrial
that finds no significant gender differences in ex- society. In a primordial society, the family rela-
tramarital sex (Prins, Buunk, & VanYperen, 1993; tionship—the social capital provided by family
Treas & Giesen, 1996), this study shows that the (Coleman, 1990)—was vitally important for in-
effect of marital duration on the likelihood of ex- dividuals. The major functions of family included
tramarital sex varies by gender. For women, the economic production for subsistence, reproduc-
likelihood of marital sex decreases with marital tion, socialization, job training, health care, enter-
duration, whereas for men the relationship be- tainment, and protection. Therefore, a loss in sat-
tween the two variables is convex. isfaction with marital sex could be well offset by
3. Because ‘‘the reasons that motivate people other benefits of a marriage; as a result, the mar-
to engage in extramarital sex are only poorly un- riage could remain intact in spite of such a loss.
derstood at present’’ (Masters et al., 1992, p. 375), In a modern society, the social capital provided
this study contributes to the research on extra- by family has eroded considerably because mod-
marital sex by identifying one condition under ern industry has removed activities involving sub-
which extramarital sex is likely to occur. Previous sistence and exchange from the home (Coleman,
research indicates that middle-aged men are more 1990; England & Farkas, 1986). The family is be-
likely to engage in extramarital sex because their coming less and less important as a source of
wives’ threat of sanctioning their husbands be- goods and services and is therefore losing its pow-
comes less credible given both the wives’ heavy er and versatility. As a result, the loss in the qual-
investments in their marriage and their declining ity of marital sex might not be offset as much as
prospects for remarriage (Brown, 1991). However, it was in a primordial society. This implies that
for a middle-aged man, if the marginal utility of more and more married men and women might
marital sex is greater than that of extramarital sex, engage in extramarital sex for sexual pleasure. Be-
A Theory of Marital Sexual Life 373

cause extramarital sex is frequently cited by di- Davis, P. (1974). Contextual sex-saliency and sexual ac-
vorced persons as a reason for marital dissolution tivity: The relative effects of family and peer group
in the sexual socialization process. Journal of Mar-
(South & Lloyd, 1995), an increase in the extra- riage and the Family, 36, 196–202.
marital sex rate might cause the divorce rate to Edwards, J., & Booth, A. (1976). Sexual behavior in
rise. and out of marriage: An assessment of correlates.
A rational choice approach to sexual behavior Journal of Marriage and the Family, 38, 73–81.
Ehrneberg, R., & Smith, R. (1994). Modern labor eco-
is new and relatively undeveloped. This paper has nomics: Theory and public policy. New York: Harper
explored one theory within this framework and Collins.
has shown that this theory has fairly good explan- England, P., & Farkas, G. (1986). Households, employ-
atory and predictive power. The paper does not ment, and gender: A social, economic and demo-
explain all the variations in marital sex and extra- graphic View. New York: Aldine.
Greenblat, C. (1983). The salience of sexuality in the
marital sex. Indeed, other factors, such as the type early years of marriage. Journal of Marriage and the
of marriage, love, marital happiness, AIDS, dis- Family, 45, 289–298.
tribution of marital power, sex ratio, and social Homans, G. (1967). The nature of social science. New
norms, also affect these two types of sexual prac- York: Harcourt, Brace & World.
James, W. H. (1981). The honeymoon effect on marital
tices. Future studies are needed to examine, the- coitus. The Journal of Sex Research, 17, 114–123.
oretically as well as empirically, the effects of Jasso, G. (1985). Marital coital frequency and the pas-
these other factors on marital sexuality. For this sage of time: Estimating the separate effects of spous-
reason, this study must be viewed as a first step es’ ages and marital duration, birth and marriage co-
in developing a more comprehensive sociological horts, and period influences. American Sociological
Review, 50, 224–241.
theory of marital sexual life. Jasso, G. (1988). Principles of theoretical analysis. So-
ciological Theory, 6, 1–20.
NOTE Kinsey, A. C., Pomeroy, W., & Martin, C. E. (1948).
Sexual behavior in the human male. Philadelphia: W.
This research was supported by the American Sociolog- B. Saunders.
ical Association’s Fund for the Advancement of the Dis- Kinsey, A. C., Pomeroy, W., Martin, C. E., & Gebhard,
cipline Award sponsored by the American Sociological P. H. (1953). Sexual behavior in the human female.
Association and the National Science Foundation. Ad- Philadelphia: W. B. Saunders.
ditional support was provided by a faculty research fund Lancaster, J. (1994). Human sexuality, life histories, and
from Wagner College. I gratefully acknowledge the help evolutionary ecology. In A. Rossi, (Ed.), Sexuality
of my generous teachers and friends who commented across the life course (pp. 39–62). Chicago: The Uni-
on drafts of this paper, including Paul Beaumont, Barry versity of Chicago Press.
Hirsch, Larry Hazelrigg, Karl-Dieter Opp, and Huayin Laumann, E., Gagnon, J. H., Michael, R., & Michaels,
Wang. I appreciate help from James Cobbe, Yaohua Ji, S. (1994). The social organization of sexuality: Sex-
Frank Lopresti, Yangang Li, Larry Petersen, Edmund ual practices in the United States. Chicago: The Uni-
Worthy, Robert Yaffee, and Xiaoyan Zhang. Special versity of Chicago Press.
thanks go to Barry Markovsky for his detailed com- Lipsey, R., & Courant, P. (1996). Microeconomics.
ments on an earlier draft, his support and advice, and Harper Collins.
to the late James S. Coleman who kindly helped and Liu, C. (1996). On the desirability and possibility of a
encouraged me to pursue this line of research. positive sociology. International Journal of Sociology
and Social Policy, 16, 103–136.
REFERENCES Markovsky, B. (1994). The structure of theories. In M.
Foschi & E. J. Lawler (Eds.), Group process (pp. 3–
Becker, G. (1991). A treatise on the family. Cambridge, 24). Chicago: Nelson-Hall.
MA: Harvard University Press. Masters, W., Johnson, V., & Kolodny, R. (1992). Human
Becker, G., Landes, E. M., & Michael, R. T. (1977). An sexuality. New York: Harper Collins.
economic analysis of marital instability. Journal of Olson, D., & DeFrain, J. (1994). Marriage and the fam-
Political Economy, 85, 1141–1187. ily: Diversity and strengths. Mountain View, CA:
Blumstein, P., & Schwartz, P. (1983). American couples. Mayfield.
New York: William Morrow. Popper, K. (1972). Objective knowledge: An evaluation
Brown, E. (1991). Patterns of infidelity and their treat- approach. Oxford, England: Clarendon Press.
ment. New York: Brunner-Mazel. Posner, R. (1992). Sex and reason. Cambridge, MA:
Buss, D. (1994). The evolution of desire: Strategies of Harvard University Press.
human mating. New York: Basic Books. Prins, K., Buunk, B., & VanYperen, N. (1993). Equity,
Call, V., Sprecher, S., & Schwartz, P. (1995). The inci- normative disapproval and extramarital relationships.
dence and frequency of marital sex in a national sam- Journal of Social and Personal Relationships, 10,
ple. Journal of Marriage and the Family, 57, 639– 39–53.
652. Rubin, L. (1992). Sex and the coupled life. In A. S.
Coleman, J. (1990). Foundations of social theory. Cam- Skolnick & J. H. Skolnick (Eds.), Family in Transi-
bridge, MA: Belknap Press of Harvard University tion (pp. 137–150). New York: Harper Collins.
Press. South, S., & Lloyd, K. M. (1995). Spousal alternatives
374 Journal of Marriage and the Family

and marital dissolution. American Sociological Re- respectively. Suppose, for an actor, functions (A1) and (A2)
view, 60, 21–35. take the following forms:
Thompson, A. (1983). Extramarital sex: A review of the
research literature. The Journal of Sex Research, 19, max U(S0, S1, X0, X1) 5 S0a X0b 1 S1a X1b (A3)
1–22.
Treas, J., & Giesen, D. (1996, August). Sex, gender and st G1 5 aS0 1 bX0 5 C1 (A4)
rational choice: Determinants of infidelity among
married and cohabiting couples. Paper presented at G2 5 aS1 1 bX1 5 C2 (A5)
the annual meeting of the American Sociological As- G3 5 C1 1 C2 5 C (A6)
sociation, New York.
Trussell, J., & Westoff, C. (1980). Contraceptive prac- where, S0 . 0, S1 . 0, X0 . 0, X1 . 0, a . 0, b . 0, a
tice and trends in coital frequency. Family Planning 1 b 5 1. We interpret a and b as each unit of physical,
Perspectives, 12, 246–249. emotional, financial or time inputs required for obtaining
Udry, R., Deven, F., & Coleman, S. (1982). A cross- each unit of S0, S1, X0, and X1 (i.e., price or marginal cost).
national comparison of the relative influence of male Setting up the Lagrangian expression gives:
and female age on the frequency of marital inter-
course. Journal of Biosocial Science, 14, 1–6. L 5 Sa0 Xb0 1 Sa1 Xb1 1 l1(C1 – aS0 1 bX0)

1 l2(C1 – aS1 1 bX1)


THE APPENDIX
The problem can be formulated as: 1 l3(C – C1 – C2). (A7)

max U (S, X) (A1) Taking partial derivatives of (A7) with respect to S0, X0,
S1, X1, l1, l2, and l3, setting each derived equation to 0,
st G (S, X) 5 C (A2) and solving for S0 and S1, one obtains:

In the model, S denotes marital sex; X denotes the con- S*0 5 aC1 / a (A8)
sumption of another good or service produced in the fam-
ily; and C is a constant, denoting the total amount of re- S*1 5 aC2 / a. (A9)
sources at this actor’s disposal. S is decomposed to two
elements: S0 denoting past sexual experience with this ac- Since C 5 C1 1 C2 (A9) can be written as:
tor’s spouse at time t0 (the ‘‘stock’’ or cumulated marital
sex); and S1 denoting marital sex at time t1 (or current mar- S*1 5 a(C – C1) / a. (A10)
ital sex). Similarly, X can be decomposed to X0 and X1 that
denote the consumption of X at t0 and t1, respectively. Fi- Substituting (A8) into (A10) yields:
nally, C is decomposed to C1 and C2, representing the
amount of resources possessed by the actor at t0 and t1, S*1 5 aC / a – S*0. (A11)

Potrebbero piacerti anche