See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261599275

Social network sites, marriage well-being and divorce: Survey and state-level
evidence from the United States

Article in Computers in Human Behavior · July 2014

DOI: 10.1016/j.chb.2014.03.034

CITATIONS READS

103 4,128

3 authors:

Sebastián Valenzuela Daniel Halpern

Pontificia Universidad Católica de Chile Pontificia Universidad Católica de Chile
82 PUBLICATIONS 8,207 CITATIONS 38 PUBLICATIONS 1,314 CITATIONS

SEE PROFILE SEE PROFILE

James Katz
Boston University
153 PUBLICATIONS 7,940 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Turnout in Chile View project

All content following this page was uploaded by James Katz on 12 June 2019.

The user has requested enhancement of the downloaded file.

Electronic copy available at: https://ssrn.com/abstract=4146895
 Computers in Human Behavior 36 (2014) 94–101

Contents lists available at ScienceDirect

Computers in Human Behavior

journal homepage: www.elsevier.com/locate/comphumbeh

Social network sites, marriage well-being and divorce:

Survey and state-level evidence from the United States
Sebastián Valenzuela a,⇑, Daniel Halpern a, James E. Katz b
a
Pontificia Universidad Católica de Chile, School of Communications, Alameda 340, Santiago 8331150, Chile
b
Boston University, College of Communication, Division of Emerging Media Studies, Boston, MA 02215, United States

a r t i c l e i n f o a b s t r a c t

Article history: This study explores the relationship between using social networks sites (SNS), marriage satisfaction and
Available online 14 April 2014 divorce rates using survey data of married individuals and state-level data from the United States. Results
show that using SNS is negatively correlated with marriage quality and happiness, and positively corre-
Keywords: lated with experiencing a troubled relationship and thinking about divorce. These correlations hold after
Social network sites a variety of economic, demographic, and psychological variables related to marriage well-being are taken
Facebook into account. Further, the findings of this individual-level analysis are consistent with a state-level anal-
Marriage well-being
ysis of the most popular SNS to date: across the U.S., the diffusion of Facebook between 2008 and 2010 is
Divorce
positively correlated with increasing divorce rates during the same time period after controlling for all
time-invariant factors of each state (fixed effects), and continues to hold when time-varying economic
and socio-demographic factors that might affect divorce rates are also controlled. Possible explanations
for these associations are discussed, particularly in the context of pro- and anti-social perspectives
towards SNS and Facebook in particular.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction ‘‘evidence’’ derived from Facebook (AAML, 2010). Further, websites

have been developed to aid in detection of cheating on the social
On May 17, 2012 Facebook became the first social network site network site. FacebookCheating.com, for instance, provides tips
to hold a public offering, valuing the company at $104 billion, the on how to catch a spouse having an extramarital affair using the
largest valuation to date for a newly listed public company (Tangel social network site.
& Hamilton, 2012). Few months later, Facebook—created with the Anecdotal evidence notwithstanding, the issue of whether Face-
vision to make the world more connected and help users to dis- book affects negatively marriage satisfaction and increases the
cover what’s going on in their world—announced to have more likelihood of divorce is an empirical claim that to our best knowl-
than one billion active users (Facebook, 2013). Despite its pro- edge has not been put to test yet. This study is a first step towards
fessed mission to help people to connect each other, the company meeting that goal. It does so in two ways. First, it analyzes the
has been accused of damaging the relationship of thousands of aggregate-level relationship between Facebook penetration and di-
couples. Circumstantial evidence, including information described vorce rates across 43 U.S. states between 2008 and 2010 control-
in the popular media and law firms, suggests that Facebook may be ling for a host of variables, including stable differences across
responsible for causing divorce in one out of five divorces in the states. Second, using individual-level survey data of a representa-
U.S. (Gardner, 2013). tive sample of married individuals in the U.S., it examines the rela-
The first report was in 2009, when an executive of Divorce- tionships between using social network sites (SNS) such as
Online in the U.K., Mark Keenan, found that the word ‘‘Facebook’’ Facebook and indicators of relationship satisfaction, also control-
appeared in 989 of the company’s 5000 most recent divorce peti- ling for potential confounds.
tions (Keenan, 2009). Similarly, a 2010 survey by the American To be clear, both data sets can provide solid evidence of the
Academy of Matrimonial Lawyers (AAML) found that four out of existence of an association between Facebook use and marriage
five lawyers reported an increasing number of divorce cases citing quality and, in the case of the state-level data, some evidence of
the temporal ordering of the variables. However, the study does
⇑ Corresponding author. Tel.: +56 2 2354 1959. not establish a cause-and-effect relationship because that would
E-mail addresses: [email protected] (S. Valenzuela), [email protected] (D. Halpern), require longitudinal and/or experimental data. In fact, as we
[email protected] (J.E. Katz). explain below, a negative correlation between Facebook and

http://dx.doi.org/10.1016/j.chb.2014.03.034
0747-5632/Ó 2014 Elsevier Ltd. All rights reserved.

Electronic copy available at: https://ssrn.com/abstract=4146895

 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101 95

marriage well-being can be explained by either a causal link or the partner, may increase the potential for jealousy and suspicion.
through self-selection. Therefore, the current study is a necessary, Further, Muise, Christofides, and Desmarais (2009) corroborated
though not sufficient, step towards understanding the role of SNS, empirically this jealousy-provoking information between partners
especially Facebook, in marriage well-being. using Facebook, which instead creates a feedback loop whereby
Following, we review extant claims on the relationship between heightened jealousy leads to increased surveillance of a partner’s
SNS and marriage quality. Afterward, we present the data and Facebook page, causing even more suspicion between the partners,
methods employed to test this association. Further, we report the which ultimately affects negatively the relationship (for online sur-
findings of the individual- and aggregate-level statistical analyses. veillance behaviors within married couples, see also Helsper &
Finally, we discuss the implications and limitations of the study, Whitty, 2010).
and present directions for future research. Third, previous research has noted that substantial decline in
partner search costs could lead to higher levels of divorce (Kendall,
1.1. Theoretical overview 2011). Kendall explains that when people manage more informa-
tion about others and it is easy to search for partners after mar-
A negative relationship between SNS use and marriage well- riage, the expected benefit from a new match may outweigh the
being could be explained by two, very different, perspectives: (1) cost of dissolving the old one, fuelling divorce rates. Facebook in
Using SNS weakens marriage and causes divorce (the negative ef- particular has a series of unique affordances that has helped to re-
fect hypothesis); or (2) divorcees and people in troubled relation- duce these searching costs and consequently may contribute to
ships use SNS such as Facebook more often (the self-selection cheating. First, Facebook’s search options and capabilities make
hypothesis). Whereas both views predict an association between cheating easier. If someone is trying to find another person, it is
using the social networks site and marriage dissatisfaction, their possible to search by name, email address, company/workplace,
implications for the social effects of Facebook and other SNS are or common friends, and even get narrow results by indicating only
opposite. In the first case, SNS are conceived as an anti-social med- hometown or school, making it still easier to find, for example, an
ia. In the second, SNS are perceived as a pro-social force that helps ex. Similarly, with the ‘‘event’’ invitations feature, it is easy for
people with a bad marriage experience to find social support. In the users to monitor and determine if a certain person of potential
next two sections, we review the rationale of each perspective. interest will be attending a particular event.
The mutual and suggested friends features may also facilitate
1.2. The case for negative effects potential cheating since users can search through their friends’
friends to find someone in whom they may be interested. Facebook
Before trying to explain why using sites like Facebook may be also suggests friend based on mutual friends and interests, conse-
negatively associated to marriage satisfaction, it is relevant to sum- quently, if the user is already predisposed to being interested in
marize prior research on the basis of strong, satisfying marriages someone else, it is more likely they will become close friends with
(Bergner & Bridges, 2002; Davis, 1985; Roberts, 1982). Manning them and open a venue for an extramarital or separation.
(2006) has noted that a long-term romantic relationship entails se- In addition, Facebook allows users to have multiple profiles: a
ven characteristics: (a) investment in the well-being of the be- person could have a profile for family and friends which lists them
loved, (b) respect, (c) admiration, (d) sexual desire, (e) intimacy, as married but also a secondary profile which lists them as single
(f) commitment, (g) exclusivity, and (h) understanding. Bergner and being interested in forming relationships. Consequently, Face-
and Bridges (2002) hold that when one or more of these aspects book and other SNS make easier finding another romantic partner
are violated by a romantic partner, the other partner is likely to feel for those so inclined to do so.
unloved, causing that spouse to reevaluate the relationship.
There are many reasons why Facebook and SNS in general 1.3. The case for self-selection
might be negatively affecting one or more of these characteristics
and, consequently, marriage quality. First, excessive use of social The discussion summarized earlier is consistent with a causal
media has been associated with ‘‘dependency’’ or compulsive use relationship between SNS and marriage quality. But there is a
(Raacke & Bonds-Raacke, 2008), creating psychological, social, counterargument: individuals in unhappy marriages may use SNS
school and/or work difficulties in a person’s life (Kuss & Griffiths, such as Facebook more often because it proves beneficial to them
2011). Lee, Cheung, and Thadani (2012) describe social features and, thus, it is self-selection what would explain the negative cor-
that could be highly problematic. Citing research by Sickface- relation between SNS use and marriage well-being. In this in-
book.com, an anti-Facebook blog, the authors argue that over 350 stance, individuals may turn to services like Facebook more
million users suffer of Facebook Addiction Disorder (FAD). The frequently after they get divorced for social support and/or to en-
term, introduced by American psychologists, has been considered hance their (newly single) social lives. Thoits (1982) defines social
by some an addiction since individuals who use social network support as the degree to which a person’s basic social needs are
sites (SNS) excessively present several addictive symptoms such gratified through interaction with others. Consistent with this def-
as neglect of personal life, mental preoccupation, escapism, mood inition, several studies suggest that groups formed in Facebook are
modifying experiences, tolerance, and concealing the addictive acting as these ‘‘support’’ places where users go to find emotional
behavior (Kuss & Griffiths, 2011). support (Ellison, Steinfield, & Lampe, 2011), sense of belonging
Second, SNS create an environment with potential situations (Bender, Jimenez-Marroquin, & Jadad, 2011), and encouragement
that may evoke feelings of jealousy between partners, harming (Greene, Choudhry, Kilabuk, & Shrank, 2011), in addition to instru-
the quality of their relationship (Elphinston & Noller, 2011). More- mental aid (Newman, Lauterbach, Munson, Resnick, & Morris,
over, SNS facilitate users reconnecting with a variety of people 2011). Other research suggests that online services, in general,
with whom they have had a past relationship (Ellison, Steinfeld, can provide social support when a personal or family transition oc-
& Lampe, 2007), creating the potential for jealousy in current rela- curs (Mikal, Rice, Abeyta, & DeVilbiss, 2013).
tionships. Similarly, SNS also support users’ maintenance of rela- Previous research has also associated SNS use with bonding so-
tionships that may otherwise be only transitory, but could cial capital (i.e., emotional support from close friends). Ellison and
become problematic when juxtaposed to the marital relationship. her colleagues (2007) argue that given the searching capabilities
Elphinston and Noller (2011) explain that exposing one’s partner afforded by SNS to form groups based on individuals with related
to all of these individuals, many of whom may be unknown to interests and needs, Facebook makes it easier to find others in

Electronic copy available at: https://ssrn.com/abstract=4146895

96 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101

similar situations and get emotional support, social support, a as registered by the World Internet Stats website (www.internet-
sense of belonging and companionship. In fact, compared with worldstats.com) in 2008, 2009 and 2010 was divided by the states’
other Internet users, Facebook users report significantly higher lev- U.S. Census population estimates for the same years. This measure
els of social support (Hampton, Sessions Goulet, Rainie, & Purcell, does not, of course, provide information on which individuals
2011). Additionally, SNS may reduce the coordination costs associ- (married or not) use Facebook. Therefore, this measure captures
ated with interacting both directly and indirectly with individuals the availability of Facebook within the population, rather than
and groups of users, supporting relationship maintenance behav- Facebook access of married individuals only. While this may be
iors among close friends, which, in turn, could enable individuals thought as a limitation of the study, it has the advantage of being
to increase the social and emotional support they may receive in a more exogenous measure than married individuals’ Facebook
case of need (Vitak, Ellison, & Steinfield, 2011). For instance, Face- usage.
book’s numerous communication channels (e.g., status updates, In addition, a number of covariates related to both divorce rates
wall posts, inbox messages, chat) are helpful for individuals look- and Facebook penetration were used in the statistical models.
ing for some forms of support and for engaging in generalized rec- Internet access was measured as the proportion of people in a state
iprocity by responding to others’ requests. As Hampton et al. relative to the state’s population that use the Internet at any loca-
(2011) explained to put the finding that Facebook users get more tion, as measured by the Internet Use Supplement of the CPS. In-
support into perspective, ‘‘someone who uses Facebook multiple come was measured as the real per capita personal income of a
times per day gets about half the boost in total support that some- state in a given year measured in 2005 U.S. dollars, as reported
one receives from being married or living with a partner’’ [p. 35]. by the U.S. Bureau of Economic Analysis. The unemployment rate
In conclusion, both the negative effect perspective and the self- was the state’s annual average unemployment rate, obtained from
selection perspective predict that there is a negative relationship the Bureau of Labor Statistics. The rest of the control variables were
between Facebook use and marriage well-being. This study seeks all taken from the American Community Survey conducted by the
to empirically test the existence, size, and sign of this relationship Census Bureau. Level of education was measured as the share of
as a first step towards understanding the role played by SNS on the state population in a given year with a Bachelor’s degree or
marriage. Absent a statistically discernible association between higher. Race and ethnicity was measured as the share of the state
SNS use, marriage well-being and divorce rates, any claim about population at a given year that self-identified as African American
the socially negative effects of Facebook on marriages is un- and Hispanic. The mean household size was the average number of
founded. On the other hand, if an association is established, then people living in a single household located within the state, and
future research can address the causality quandary. age distribution was gauged as the proportion of the states’ adult
population (aged 18–64) relative to the states’ total population. Ta-
ble 1 shows descriptive statistics of all state-level variables
2. Methods
analyzed.

Borrowing from the work of Kendall (2011) on Internet access

2.2. Individual-level data
and divorce rates, we followed a two-pronged approach of con-
ducting aggregate- and individual-level data analyses of SNS and
A secondary analysis of survey data was employed to study the
Facebook use, marriage quality indicators and divorce rates in
individual-level relationship between SNS use and marriage well-
the United States. Whereas cross-sectional survey data allows us
being. The survey was conducted online on a probability sample
to control for a larger variety of confounds and to create a richer
of adults aged 18 to 39 by Knowledge Networks (KN) on behalf
battery of psychological variables related to marriage well-being,
of the University of Texas at Austin’s Population Research Center
state-level data can be used to specify statistical models that incor-
(http://www.prc.utexas.edu/nfss/index.html), which is not respon-
porate the time ordering of the variables of interest.
sible for the analyses and interpretations presented here. To our
knowledge, this is the only publicly available representative survey
2.1. State-level data in the U.S. that contains questions about both SNS use and indica-
tors of marriage well-being. Participants were recruited via ran-
Considering that Facebook became available to any person dom-digit-dialing and address-based sampling methods. Data
worldwide in 2006, all state-level data was compiled annually were collected between July 2011 and February 2012. Due to the
from 2008 to 2010 (the most current year with official state-level objectives of the current study, only data from the subsample of
divorce data available). The key dependent variable was the married individuals (N = 1160) were analyzed. To make the results
state-level divorce rate as measured by Bitler, Gelbach, Hoynes, generalizable to the population of married individuals in the U.S.,
and Zavodny (2004) and Kendall (2011), which is calculated as all calculations presented here employ KN’s postratification
the count of new divorces in a state accumulated between January weights.
and December of each year as measured by the National Center for Four dependent variables gauging different aspects of marriage
Health Statistics (NHCS) divided by the number of married women satisfaction were created. Current marriage quality was an additive
in a state at the same year as measured by the Annual Demo- scale (a = .96) of responses to six items asking the respondent’s
graphic Supplement of the Current Population Survey (CPS). Unfor- level of agreement with statements such as ‘‘We have a good
tunately, six states (California, Georgia, Hawaii, Indiana, Louisiana relationship’’ and ‘‘My relationship with my partner is very
and Minnesota) do not report divorce data to the NHCS and thus healthy.’’ Happiness in current marriage was measured asking
had to be excluded from the analysis. Nevada and Washington, the respondent to rate on a scale from 1 to 10 the degree of
D.C., were excluded, too, because they were extreme outliers (Ne- happiness in his/her marriage. A scale of trouble in the current
vada doubled the average state-level divorce rates, whereas in D.C. marriage (a = .75) was computed by averaging three items on
the population of Facebook users almost doubled the population of the frequency with which the respondent has thought the
residents). In addition, considering that divorce rates are positively relationship is on trouble. Lastly, the likelihood of divorce was
skewed, in the estimations the variable was transformed using the gauged with a dichotomic variable asking whether in the past year
natural logarithm. the respondent has thought about leaving his/her spouse.
Facebook penetration rate was a composite of two measures. The key independent variable, SNS use, was measured with a
For each state, the total number of Facebook accounts in each state question on how much time the respondent spends on a typical

Electronic copy available at: https://ssrn.com/abstract=4146895

 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101 97

Table 1
Descriptive statistics for state-level data.

M Mdn SD Min Max Valid N

Divorces per 100 married couples 1.777 1.762 0.370 0.963 2.716 129
Ln (divorces per 100 married couples) 0.553 0.567 0.215 -0.037 0.999 129
Facebook penetration rate 0.236 0.203 0.146 0.028 0.594 129
Share with Internet access 0.697 0.700 0.057 0.550 0.803 129
Real per capita personal income ($2005) 35,771.061 35,182.404 5,348.820 27,671.532 52,048.401 129
% With Bachelor’s degree or higher 0.200 0.193 0.040 0.112 0.310 129
% Of unemployment 0.074 0.074 0.023 0.030 0.134 129
% Who are African American 0.099 0.069 0.090 0.005 0.374 129
% Who are Hispanic 0.091 0.065 0.093 0.004 0.437 129
Average household size 3.075 3.040 0.179 2.700 3.880 129
% Aged 18–64 years 0.627 0.627 0.015 0.589 0.668 129

weekday ‘‘on social networking sites like Facebook, Twitter, or Thus, regression coefficients are obtained from variance in
MySpace.’’ Responses were measured on a 7-point scale ranging Facebook penetration within states over time.
from ‘‘not at all’’ to ‘‘four hours or more.’’ At the time of the Table 3 shows the results of four different models. The first
survey, Facebook had 145.5 million unique visitors in the U.S., model shows that Facebook penetration is a positive predictor
compared to Twitter’s 32.8 million and MySpace’s 33.1 million of divorce rates. In substantive terms, a 20% annual increase in
(Martin, 2013). the share of a state’s population with a Facebook account (i.e.,
As was the case with the state-level data, and based on prior re- the median growth rate of Facebook in the period under study)
search (Bitler et al., 2004; Kendall, 2011; Lehrer, 2008) a battery of is associated with a 2.18% increase in the divorce rate. This rela-
control variables were measured to account for potential con- tionship becomes more robust when a series of time-varying
founds: whether the respondent ever had extramarital sex, variables related to divorce rates are added, as shown in model
whether he or she lived together with both parents while growing 2. In this case, the model predicts that a 20% annual increase in
up, unemployment, total household income, educational attain- Facebook penetration rates is associated with an average 4.32%
ment, religiosity, and number of children. In order to control for growth in divorce rates. In model 3, we take advantage of the
respondents who use SNS more frequently because they happen longitudinal nature of the state-level data and include a lagged
to have more spare time, time spent on television and Internet ac- term of the dependent variable as an explanatory variable. In
cess at home were gauged with separate variables. Lastly, a set of this case, the model still controls for all unobserved time-con-
demographic variables were included as covariates: age (in years), stant differences among states but adds past factors not directly
race and ethnicity. Descriptive statistics of all variables are observed that shape current divorce rates (i.e., factors filtered
displayed in Table 2. through the lagged term). Again, the correlation between Face-
book penetration and divorce rates is positive and statistically
3. Results significant. In this case, a 20% increase in the share of Facebook
users in a given state is associated with a 4.00% increase in the
3.1. State-level data findings divorce rate in the following year. This relationship holds in
Model 4, which adds a set of covariates. Thus, under diverse
For the analysis we pooled the data to form a panel and then ran specifications, the state-level data shows that Facebook penetra-
a set of fixed-effects regression models. This type of estimator, tion rates and divorce rates across the states have a positive,
contrary to pooled OLS regression, allows us to control for all statistically significant relationship. As noted before, this should
time-invariant differences across the states that may be related not be interpreted as a causal effect necessarily. It does,
to divorce rates and Facebook usage, which are not controlled for however, suggest the need to further probe the relationship
in the models, through the inclusion of a state-specific constant. using individual-level data.

Table 2
Descriptive statistics for individual-level data.

M Mdn SD Min Max Cronbach’s a Valid N

Scale of current marriage quality 4.113 4.333 0.943 1.000 5.000 0.962 1109
Degree of happiness in current marriage 7.604 8.000 2.147 1.000 10.000 1153
Scale of trouble in current marriage 1.592 1.333 0.548 1.000 3.000 0.753 1134
In the past year, respondent has thought about leaving spouse 0.238 0.000 0.426 0.000 1.000 1156
Frequency of SNS use 2.730 2.000 1.656 1.000 7.000 1144
Internet access at home (yes) 0.880 1.000 0.324 0.000 1.000 1160
Household Income 12.610 13.000 3.823 1.000 19.000 1158
Educational attainment 3.025 3.000 0.923 1.000 4.000 1160
Currently unemployed (yes) 0.070 0.000 0.255 0.000 1.000 1160
Race/ethnicity (African American) 0.083 0.000 0.276 0.000 1.000 1160
Race/ethnicity (Hispanic) 0.174 0.000 0.379 0.000 1.000 1160
Number of children 1.383 1.000 1.430 0.000 5.000 1119
Age 31.726 32.000 4.917 18.000 39.000 1160
Religious attendance 2.290 2.000 2.027 0.000 5.000 1152
Lived with both parents entire time until age 18 (yes) 0.441 0.000 0.497 0.000 1.000 1160
Time spent watching TV 4.280 4.000 1.590 1.000 7.000 1147
Ever had extramarital sex (yes) 0.186 0.000 0.389 0.000 1.000 1147

Electronic copy available at: https://ssrn.com/abstract=4146895

98 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101

Table 3
Regression analysis of divorce rates in 43 U.S. states (2008–2010).

Dependent variable: ln (divorces per 100 married couples)

Model 1 fixed effects Model 2 fixed effects Model 3 fixed effects Model 4 fixed effects with
with covariates with lagged DV lagged DV and covariates
Facebook penetration rate 0.109*** (0.033) 0.216*** (0.064) 0.199*** (0.049) 0.238*** (0.087)
Internet access 0.272 (0.215) 0.704** (0.274)
Ln (real per capita personal income) [$2005] 0.685 (0.627) 0.678 (1.447)
Share with Bachelor’s degree or higher 1.108 (0.752) 1.210 (0.853)
Unemployment rate 1.053 (0.688) 3.898* (2.178)
Share who are African American 2.021 (3.860) 1.456 (4.043)
Share who are Hispanic 0.301 (0.856) 0.929 (1.029)
Mean household size 0.373*** (0.132) 0.445** (0.179)
Share aged 18–64 years 0.063 (0.857) 0.448 (1.318)
Ln (divorces per 100 married couples) [lagged] 0.547*** (0.160) 0.449*** (0.139)
(constant) 0.527*** (0.008) 5.625 (6.681) 0.785*** (0.093) 9.599 (14.853)
Adjusted R2 0.916 0.930 0.949 0.959
(Observations) (129) (129) (86) (86)

Notes: The data set is comprised of 129 state-level (excluding District of Columbia) observations for 2008, 2009, and 2010. Of the possible 153 observations, 24 are unusable
due to missing data on the number of divorces. Data from Nevada and Washington, D.C., extreme outliers in marriage patterns and Facebook penetration, are also excluded.
Entries report regression coefficients with robust (HAC) standard errors in parentheses.
*
p < 0.10.
**
p < 0.05.
***
p < 0.01 (two-tailed).

3.2. Individual-level data findings experiencing trouble in the marriage and to having thought about
separating.
Following, we present the results of the analysis of the survey The first model shows that frequency of SNS use is a negative,
sample of married individuals aged 18–39 years. Contrary to the statistically significant predictor of the index of marriage quality,
state-level data, the survey data allows us to measure the associa- although the relationship is rather weak. Holding all other vari-
tion between SNS use and a host of psychological variables related ables constant at their means, a respondent that does not use
to marriage well-being. Also, the statistical power is greater, which SNS scores 4.22 in the scale of current marriage quality—a 6.96%
allows for the inclusion of additional covariates. difference with a respondent who uses SNS four hours or more
Table 4 shows the estimates of four regression models, one for (score = 3.87). More robust is the negative relationship between
each dependent variable studied, along with a host of demo- the SNS use and degree of happiness in current marriage (second
graphic, economic, and social factors that prior research and theory model). In this case, the model predicts that a nonuser is 11.40%
suggest are related to marriage well-being, SNS and Facebook use. more happy with his/her marriage than a heavy user (predicted
The expectation was that the key predictor variable, frequency of score of 7.99 vs. predicted score of 6.85 on a 1 to 10 scale). The
SNS use, would be negatively related to marriage quality and nature of these associations—that using sites like Facebook more
marriage happiness and, conversely, positively associated to frequently is associated with diminished marriage wellbeing—is

Table 4
Regression analysis of indicators of Marriage Satisfaction among U.S. married individuals (2011).

Dependent variables
Scale of current marriage Degree of happiness in current Scale of trouble in current In past year respondent has thought
quality (range: 1–5) marriage (range: 1–10) marriage (range: 1–3) about leaving spouse (range: 0–1)
OLS regression OLS regression OLS regression Logistic regression
Frequency of SNS use 0.058** (0.018) 0.190*** (0.040) 0.035*** (0.010) 0.146** (0.048)
Internet access at home (yes) 0.069 (0.095) 0.306 (0.207) 0.086 (0.052) 0.731* (0.284)
Household Income 0.009 (0.009) 0.028 (0.020) 0.004 (0.005) 0.037 (0.025)
Educational attainment 0.007 (0.038) 0.066 (0.082) 0.049* (0.021) 0.237* (0.101)
Currently unemployed (yes) 0.185 (0.122) 0.342 (0.261) 0.013 (0.064) 0.009 (0.321)
Race/ethnicity (African 0.061 (0.109) 0.023 (0.238) 0.052 (0.060) 0.357 (0.291)
American)
Race/ethnicity (Hispanic) 0.119 (0.076) 0.468** (0.166) 0.093* (0.041) 0.751*** (0.195)
Number of children 0.055* (0.022) 0.047 (0.047) 0.026* (0.012) 0.017 (0.058)
Age 0.012 (0.006) 0.032* (0.013) 0.002 (0.003) 0.010 (0.017)
Religious attendance 0.009 (0.015) 0.076* (0.032) 0.014 (0.008) 0.007 (0.042)
Lived with both parents entire 0.093 (0.061) 0.155 (0.131) 0.099** (0.032) 0.356* (0.171)
time until age 18 (yes)
Time spent watching TV 0.001 (0.019) 0.033 (0.041) 0.007 (0.010) 0.016 (0.053)
Ever had extramarital sex (yes) 0.387*** (0.076) 1.030*** (0.164) 0.338*** (0.041) 1.522*** (0.183)
(Constant) 4.704*** (0.227) 9.286*** (0.492) 1.463*** (0.121) 1.767** (0.617)
Adjusted R2 0.059 0.084 0.148 0.136
(Observations) (1047) (1084) (1070) (1083)

Notes: Entries report regression coefficients with standard errors in parentheses. Data is about married individuals aged 18–39 years. Adjusted R2 for Model 4 is Cox and Snell.
*
p < 0.10.
**
p < 0.05.
***
p < 0.001 (two-tailed).

Electronic copy available at: https://ssrn.com/abstract=4146895

 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101 99

confirmed by the last two models. Regressing the index of trouble recommendations on this basis. Under these collaborative models,
in the current marriage on SNS use and a battery of control vari- the system automatically retrieves and filters data by considering
ables yields that a one-unit increase in SNS use is associated with the feedback given by other users to the documents. On the other
a .035 increase in marriage trouble. In substantive terms, moving hand, social media in general and Facebook in particular have facil-
from the lowest to the highest score on the SNS use measure— itated the creation of a ‘‘profile culture’’ where millions of users
and holding all other variables constant—is associated with a 7% have generated their own persona on the web (Utz, 2010). Home-
increase in the index of trouble in current marriage. town, phone number, email, language, football team, favorite book,
The last model is a logistic regression predicting the likelihood best movie, political affiliation, religion, marital status, type and
that the respondent has thought about leaving his or her spouse in number of friends, job, preferred networks, photos, and videos
the last 12 months. The model shows that the SNS measure is a are but a fraction of the huge amount of data that users may utilize
strong, positive predictor of thinking about such course of action. to define themselves. In this way—and powered by collaborative
The predicted probabilities of average respondents who differ only systems—, SNS can today tie millions of geographically dispersed
in SNS use show that the likelihood of thinking about separating is users who have common elements in their profiles. Consequently,
16.34% for a nonuser and 31.93% for a heavy user. The individual- individuals are grouped based on commonalities and it is possible
level analysis, thus, is consistent with the results of the state-level to expect that divorcees may use the same recommendations
analysis: use of SNS such as Facebook is associated with lower developed by these platforms in order to look for people in similar
marriage satisfaction and a higher likelihood of divorce or, con- situations and get support from them. Such a mechanism afforded
versely, respondents in troubled relationships use SNS, including by social network sites could also explain why individuals may
Facebook, more often. turn to Facebook more frequently after they get divorced for social
support and/or to enhance their (newly single) social lives.
Another explanation for the positive relationship between
4. Discussion experiencing a troubled relationship or divorce and SNS use, fol-
lowing the self-selection perspective, is that this social network
We explored the relationship in the U.S. between SNS use and site may be useful for reducing uncertainty following the termina-
relationship satisfaction among a nationally representative sample tion of a romantic relationship. In contrast to traditional communi-
of married individuals, as well as the association between Face- cation, SNS such as Facebook allow people to reduce uncertainty
book penetration and divorce rates at a state-level. Survey results covertly. Tong and Walther (2011) explain that Facebook obviates
revealed a positive correlation between more frequent use of SNS direct communication between ex-partners or mutual friends by
and the variables that reflected lower marriage quality, marriage allowing individuals to gather information through social search
unhappiness, experiencing a troubled relationship, and thinking of profiles and newsfeeds. This according to the authors, allows
about separating. This was consistent with state-level analyses, ex-partners to avoid the social disapproval from friends in their
in which we found that across the U.S., the diffusion of Facebook network that could arise from direct question-and-answer strate-
between 2008 and 2010 is positively correlated with divorce rates, gies, while still being able to reduce their own uncertainty in a less
a relationship that held in the presence of numerous control detectable way. In fact, Tong and Walther (2011) found that after
variables. breaking up, ex-partners rely on Facebook to monitor their ex’s so-
Two possible explanations may account for this negative corre- cial activities, to detect if their ex has a new romantic partner, and
lation. From a self-selection perspective, this phenomenon can be to communicate directly with their ex. Thus, the termination of a
understood by the fact that it is not so much that social network relationship may lead users to spend more time in Facebook in or-
services such as Facebook causes problematic relationships be- der to monitor their ex-partner’s behavior.
tween couples or cause divorce, but that divorcees and individuals From a cause-and-effect perspective, SNS may reduce marriage
in unhappy marriages use Facebook and SNS more often because it well-being through habituation or addiction, sparking feelings of
proves beneficial to them by providing emotional support. Accord- jealousy between partners, or facilitating having extramarital af-
ing to this first view, Facebook would be fulfilling its raison d’être fairs. We offered three different reasons to explain this relation-
(or at least its vision) as it connects people with friends, family, ship. First, excessive use of social media has been associated with
and other strong ties. That means that divorcees or people going compulsive use, which may create psychological, social, school
through difficult moments in their marriage would choose this so- and/or work difficulties in a person’s life. These phenomena, in
cial network site to communicate with their close contacts, trying turn, may trigger marriage unhappiness and, ultimately, divorce.
to found the psychological well-being that often flows from bond- Second, Facebook in particular creates an environment with poten-
ing social capital. tial situations that may evoke feelings of jealousy between part-
There are various reasons to explain why SNS, particularly Face- ners, harming the quality of their relationship. And third, we
book, would be useful to facilitate reciprocity, emotional support, noted that services like Facebook have unique affordances that
and companionship from close contacts. Online services such as may help partners to reduce searching costs for extra-matrimonial
Facebook have several features that support relationship mainte- affairs and consequently may contribute to cheating. Conse-
nance among close friends, which, in turn, could enable individuals quently, it is important to note that based on the data we analyzed
to accrue bonding social capital. The multiple communication it is also plausible to argue that the capabilities of Facebook also
channels reduce the coordination costs and simplify the process enable certain negative social consequences such as cheating or
through which individuals can request some form of support (Vitak deteriorating a marriage. In fact, previous research has shown that
et al., 2011). Similarly—and based in our results—it is possible to pervasive technology often leads to unintended consequences,
argue that also SNS present several affordances that might lead such as infidelity and threats to privacy (Iachello & Hong, 2007;
users to connect with people that are going through similar prob- Hand et al., 2013; Mileham, 2007; Pankoke-Babatz & Jeffrey,
lems in order to receive emotional support. 2002). As Debatin, Lovejoy, Horn, and Hughes (2009) explain, spe-
Importantly, Facebook has exploited the use of collaborative cific privacy concerns of online social networking include inadver-
filtering systems (user-to-user interactions) to increase the tent disclosure of personal information, damaged reputation due to
amount of social interaction between users. Unlike content-based rumors and gossip, unwanted contact and harassment or stalking,
recommendation methods, collaborative recommender systems surveillance-like structures due to backtracking functions, use of
try to match people with similar interests and then make personal data by third-parties, and hacking and identity theft.

Electronic copy available at: https://ssrn.com/abstract=4146895

100 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101

However, absent longitudinal and experimental data, the cur- whereas the state-level analysis is based on a coarse measure of
rent study cannot determine the causal direction of these associa- Facebook use, assesses marriage quality using divorce rates only,
tions. It may well be that for some individuals Facebook and SNS and covers three years only. As more publicly available data on
use in general creates opportunities that may end up in divorce divorce rates becomes available, the current study should be repli-
whereas for individuals who have recently divorced or are experi- cated in order to test the robustness and generalizability of its
encing a bad marriage, Facebook provides social and emotional findings.
support. The media effects perspective and the self-selection per-
spective are not mutually exclusive, thus, future research needs 6. Conclusion
to address the possibility of reinforcing spirals. Furthermore, the
key theoretical issue would not be the relationship between partic- The data presented in this study provide evidence that Face-
ular features and particular outcomes, but rather how people use book use is correlated with reduced marital satisfaction and di-
this social network site and what they try to accomplish with it. vorce rates. Although it may seem surprising that a Facebook
This approach would be consistent with what the literature in profile, a relatively small factor compared to other drivers of hu-
communication has called the ‘‘rational actor’’ perspective man behavior, could have a significant statistical relationship with
(Markus, 1994), which holds that impacts result not from the tech- divorce rates and marital satisfaction, it nonetheless seems to be
nology itself but from the choices individuals make about how to the case. This relationship holds up at both the individual and state
use it. According to this perspective, there would be ‘‘good uses’’ levels. If the preliminary findings in this study are sustained, it
that result in positive social outcomes, within the constraints im- would represent an important step forward in the study of SNS
posed by technological characteristics, and ‘‘bad uses’’ that result and human behavior. It would also raise profound questions about
in negative social outcomes. the role of social media in daily lives. Finally, it would spur new
Markus (1994) argues that this perspective should lead lines of research in understanding the role of Facebook in divorce
researchers to understand under what conditions it would be ra- and marital satisfaction, prompting a host of policy-oriented re-
tional for users to behave in ways that result in negative effects. search endeavors by social scientists.
Following this logic, the approach suggests two broad answers.
First, ‘‘bad uses’’ of electronic communication technology might
References
be rational when users want to achieve ‘‘negative social impacts’’
using it. For instance, if users in this particular case would not have American Academy of Matrimonial Lawyers (AAML). (2013). Big surge in social
a good marriage and they would want to ‘‘escape’’ from their networking evidence says survey of nation’s top divorce lawyers. <www.aaml.org/
about-the-academy/press/press-releases/e-discovery/big-surge-social-network-
spouses and achieve some social distance in their relationships
ing-evidence-says-survey>. Accessed 03.03.13.
with them, Facebook would offer them a good alternative for that. Bender, J. L., Jimenez-Marroquin, M. C., & Jadad, A. R. (2011). Seeking support on
Following this logic, partners may spend more time in this social Facebook: A content analysis of breast cancer groups. Journal of Medical Internet
network connecting with friends in order to avoid communicating Research, 13(1), e16.
Bergner, R. M., & Bridges, A. J. (2002). The significance of heavy pornography
with their respective spouses. involvement for romantic partners: Research and clinical implications. Journal
Markus (1994) explains that ‘‘bad uses’’ might be also rational of Sex & Marital Therapy, 28, 193–206.
when, despite the fact that they generate negative social effects, Bitler, M. P., Gelbach, J. B., Hoynes, H. W., & Zavodny, M. (2004). The impact of
welfare reform on marriage and divorce. Demography, 41, 213–236.
they also produce benefits that outweigh these negative effects. Davis, K. (1985). Near and dear: Friendship and love compared. Psychology Today,
One insight of this perspective is that even though partners may 19, 22–30.
be aware of the negative social effects of Facebook, such as experi- Debatin, B., Lovejoy, J. P., Horn, A. K., & Hughes, B. N. (2009). Facebook and online
privacy: Attitudes, behaviors, and unintended consequences. Journal of
encing a troubled relationship, this may be the outcomes deliber- Computer-Mediated Communication, 15(1), 83–108.
ately intended by the social media users since they are really Ellison, N. B., Steinfeld, C., & Lampe, C. (2007). The benefits of Facebook ‘‘friends’’:
thinking about cheating. That means that users may continue to Social capital and college students’ use of online social network sites. Journal of
Computer-Mediated Communication, 12, 1143–1168.
engage in a behavior that has the potential for negative impacts, Ellison, N. B., Steinfield, C., & Lampe, C. (2011). Connection strategies: Social capital
because it has other, positive effects that users desire to achieve. implications of Facebook-enabled communication practices. New Media &
As rational actors, spouses may know that particular uses of SNS Society, 13, 873–892.
Elphinston, R. A., & Noller, P. (2011). Time to face it! Facebook intrusion and the
and Facebook, such as lurking their ex partners behaviors and
implications for romantic jealousy and relationship satisfaction.
activities, entail risks and potential problems at home. However, Cyberpsychology, Behavior, and Social Networking, 14, 631–635.
they may also value the benefits of using social media in these Facebook (2013). Facebook’s mission. <www.facebook.com/facebook?v=info>
ways, such as the ability to escape from their marriages in order Accessed 04.04.13.
Gardner, D. (2013). The marriage killer: One in five American divorces now involve
to feed their fantasies. In such cases, the rational actor perspective Facebook. <www.dailymail.co.uk/news/article-1334482/The-marriage-killer-
suggests that users, who are aware of the potential negative conse- One-American-divorces-involve-Facebook.html> Accessed 03.03.13.
quences that their actions may have, will take active measures to Greene, J. A., Choudhry, N. K., Kilabuk, E., & Shrank, W. H. (2011). Online social
networking by patients with diabetes: A qualitative evaluation of
minimize them. Markus (1994) explains that humans have a un- communication with Facebook. Journal of General Internal Medicine, 26,
ique ability to anticipate that an action they plan to take has the 287–292.
possibility of negative consequences, and accordingly they employ Hampton, K., Sessions Goulet, L., Rainie, L., & Purcell, K. (2011). Social networking
sites and our lives. pewinternet.org/Reports/2011/Technology-and-social-
measures to counteract, or compensate, for these consequences. networks.aspx Accessed 15.03.13.
Hand, M. M., Thomas, D., Buboltz, W. C., Deemer, E. D., & Buyanjargal, M. (2013).
Facebook and romantic relationships: Intimacy and couple satisfaction
associated with online social network use. Cyberpsychology, Behavior, and
5. Limitations Social Networking, 16, 8–13.
Helsper, E. J., & Whitty, M. T. (2010). Netiquette within married couples: Agreement
There are a variety of limitations that affect the validity and about acceptable online behavior and surveillance between partners. Computers
in Human Behavior, 26, 916–926.
generalizability of the data. One of these is that the data on individ-
Iachello, G., & Hong, J. (2007). End-user privacy in human–computer interaction.
ual actions is taken from self-reports. Although effort was made to Foundations and Trends in Human–Computer Interaction, 1(1), 1–137.
be sure the sample was representative, doubtless there are sources Keenan, M. (2009). Facebook is bad for your marriage – research finds. <http://
of distortion due to social desirability biases and other forms of www.journalism.co.uk/press-releases/facebook-is-bad-for-your-marriage—
research-finds/s66/a536960/> Accessed 03.03.13.
inaccuracies endemic to this methodology. Also, the survey is Kendall, T. D. (2011). The relationship between Internet access and divorce rate.
cross-sectional, which cannot address issues of temporal order, Journal of Family and Economic Issues, 32, 449–460.

Electronic copy available at: https://ssrn.com/abstract=4146895

 S. Valenzuela et al. / Computers in Human Behavior 36 (2014) 94–101 101

Kuss, D. J., & Griffiths, M. D. (2011). Online social networking and addiction: A Proceedings of the ACM 2011 conference on Computer Supported Cooperative Work
review of the psychological literature. International Journal of Environmental (CSCW) (pp. 341–350).
Research and Public Health, 8, 3528–3552. Pankoke-Babatz, U., & Jeffrey, P. (2002). Documented norms and conventions on
Lee, Z. W., Cheung, C. M., & Thadani, D. R. (2012). An investigation into the the Internet. International Journal of Human–Computer Interaction, 14(2),
problematic use of Facebook. In Proceedings of the 45th Hawaii International 219–235.
Conference on System Sciences (HICSS) (pp. 1768–1776). Raacke, J., & Bonds-Raacke, J. (2008). MySpace and Facebook: Applying the uses and
Lehrer, E. (2008). Age at marriage and marital instability: The Becker–Landes– gratifications theory to exploring friend-networking sites. Cyberpsychology,
Michael hypothesis revisited. Journal of Population Economics, 21, 463–484. Behavior, and Social Networking, 11, 169–174.
Manning, J. C. (2006). The impact of internet pornography on marriage and the Roberts, M. (1982). Men and women: Partners, lovers and friends. Advances in
family: A review of the research. Sexual Addiction & Compulsivity, 13, 131–165. Descriptive Psychology, 2, 57–78.
Markus, M. L. (1994). Finding a happy medium: Explaining the negative effects of Tangel, A., & Hamilton, W. (May 17, 2012). Stakes are high on Facebook’s first day of
electronic communication on social life at work. ACM Transactions on trading. The Los Angeles Times. Retrieved 17.05.12.
Information Systems (TOIS), 12(2), 119–149. Thoits, P. A. (1982). Conceptual, methodological, and theoretical problems in
Martin, S. (2013). Facebook, Twitter up despite Google+. content.usatoday.com/ studying social support as a buffer against life stress. Journal of Health and Social
communities/technologylive/post/2011/08/facebook-twitter-users-up-despite- behavior, 23, 145–159.
google/1#.UUsA0Rw3uuI Accessed 21.03.13. Tong, S. T., & Walther, J. B. (2011). Facebook and relational termination: Connection,
Mikal, J. P., Rice, R. E., Abeyta, A., & DeVilbiss, J. (2013). Transition, stress and isolation, and stalking. Symposium on relationships and the Internet, Oxford
computer-mediated social support. Computers in Human Behavior, 29, A40–A53. Internet Institute, Oxford, UK, 2011. <www.msu.edu/~stong/files/oxford_2011.
Mileham, B. L. A. (2007). Online infidelity in Internet chat rooms: An ethnographic pdf> Accessed 07.07.13.
exploration. Computers in Human Behavior, 23, 11–31. Utz, S. (2010). Show me your friends and I will tell you what type of person you are:
Muise, A., Christofides, E., & Desmarais, S. (2009). More information than you ever How one’s profile, number of friends, and type of friends influence impression
wanted: Does Facebook bring out the green-eyed monster of jealousy? formation on social network sites. Journal of Computer-Mediated Communication,
Cyberpsychology, Behavior, and Social Networking, 12, 441–444. 15, 314–335.
Newman, M. W., Lauterbach, D., Munson, S. A., Resnick, P., & Morris, M. E. (2011). It’s Vitak, J., Ellison, N. B., & Steinfield, C. (2011). The ties that bond: Re-examining the
not that I don’t have problems, I’m just not putting them on Facebook: relationship between Facebook use and bonding social capital. In Proceedings of
Challenges and opportunities in using online social networks for health. In the 44th Hawaii International Conference on System Sciences (HICSS) (pp. 1–10).

Electronic copy available at: https://ssrn.com/abstract=4146895

View publication stats