Sie sind auf Seite 1von 23

The Social Demography of Internet

Dating in the United States n

Jessica M. Sautter, Duke University


Rebecca M. Tippett, University of Virginia
S. Philip Morgan, Duke University

Objective. The objective of this article is to identify the sociodemographic cor-


relates of Internet dating net of selective processes that determine who is ‘‘at risk.’’
We also examine the role of computer literacy, social networks, and attitudes
toward Internet dating among single Internet users. Methods. We use multivariate
logistic regression to analyze 3,215 respondents from the first nationally represen-
tative U.S. survey of Internet dating. Results. Sociodemographic factors have
strong effects on Internet access and single status but weak effects on use of Internet
dating services once the sample is conditioned on these factors. For this ‘‘at-risk’’
subpopulation, computer literacy and social networks strongly influence the like-
lihood of Internet dating. Conclusions. Internet dating is a common mate selection
strategy among the highly selective subpopulation of single Internet users and may
continue to grow through social networks. Material and virtual elements of the
digital divide have direct and indirect effects on Internet dating.

The Emergence of Internet Dating

In the last decade, searching for romantic partners via Internet dating
sites has become a visible and common strategy for mate selection;
this search strategy emerged amid important social, demographic, and
technological changes (Espinoza, 2009; Tulsiani, Best, and Card, 2008).
n
Direct correspondence to Rebecca M. Tippett, Weldon Cooper Center for Public Ser-
vice, University of Virginia, PO Box 400206, Charlottesville, VA 22904-4206 hrtip-
pett@virginia.edui. All authors contributed equally and share authorship of this article. Data
and coding used in this article are available upon request for those wishing to replicate this
study. This research was partially supported by a contract, ‘‘Designing New Models for
Explaining Family Change and Variation’’ (N01 HD-3-3354; PI. S. Philip Morgan) with the
National Institute of Child Health and Human Development. Data collection was carried
out and funded by the Pew Foundation with partial support from Duke University. The
authors thank Emilio A. Parrado, Seth Sanders, Lee Rainie, John Horrigan, Lynn Smith-
Lovin, Amanda Lenhart, Mary Madden, Douglas Downey, Linda K. George, and Nathan D.
Martin for helpful comments and suggestions. Early versions of this article were presented at
the 2005 Southern Demographic Association Annual Meeting and the 2006 Southern So-
ciological Society Annual Meeting.

SOCIAL SCIENCE QUARTERLY, Volume 91, Number 2, June 2010


r 2010 by the Southwestern Social Science Association
Social Demography of Internet Dating 555
Specifically, Internet dating is growing rapidly in terms of clientele, business
returns, and presence in media and pop culture. While it may be too early
to assess the impact of this innovation on union formation, we can
identify who is adopting this search strategy, with specific focus on the
sociodemographic and network correlates of Internet dating use. We will
show that Internet dating is highly selective—only a subset of the total
population are potential participants, that is, ‘‘at risk’’ of using this
technology. In our analyses, individuals must have Internet access and must
be seeking a partner to be considered ‘‘at risk’’ of Internet dating. Elements
of the ‘‘digital divide’’—inequality in both physical resources and virtual
schemas—and partner/union status condition the likelihood of Internet
dating.
We define Internet dating as the use of websites that provide a database
of potential partners—typically in close geographical proximity—that
one can browse and contact, generally for a fee. These Internet dating
services facilitate connections that may eventually lead to face-to-face
contact and in-person relationships. This is quite different from the
earliest forms of online interaction on which much previous research
focused. In those studies, singles met in chat rooms and carried out
long-distance relationships in cyberspace, often never meeting in person
(Ben-Ze’ev, 2004; Cooper and Sportolari, 1997). Extant work explores
a variety of topics, providing important insights about presentation of
self and identity (Ellison, Heino, and Gibbs, 2006; Whitty, 2008),
the psychological correlates of online relationships (Anderson, 2005;
Donn and Sherman, 2002), and the utility of mate selection theories
of homophily (Fiore and Donath, 2005) and complementary needs
(Hitsch, Hortaçsu, and Ariely, 2006). This research, however, focuses
solely on individuals who use the Internet or are already active Internet
daters.
Sociological research has focused more on macro-level social factors (e.g.,
increasing singles population, reduced dating networks, and growing mo-
bility) and their impact on Internet dating (Brym and Lenton, 2001;
Whitty, 2006); Internet daters identify these social factors and the desire to
find love as stronger motivations to turn to Internet dating than personality
factors (Barraket and Henry-Waring, 2008). Drawing on recent data from
the Pew Internet & American Life Project’s first nationally representative
survey about use of Internet dating and attitudes toward this technology
(Madden and Lenhart, 2006), this article expands on sociodemographic
research by modeling the selective processes leading to use of Internet dat-
ing. Specifically, we identify sociodemographic correlates of being ‘‘at risk’’
of using Internet dating, that is, having access to the Internet and being
single in the past five years during the technology’s growth. We then model
the likelihood of Internet dating net of these selective processes, examining
the impact of sociodemographic factors, computer literacy, and social net-
work effects.
556 Social Science Quarterly
The Structured Growth of the Internet and Internet Dating

Internet dating has grown rapidly. As of 1999, only 2 percent of American


singles had used some form of online personals services. By 2002, one in
four singles had used Internet dating services in their search for a potential
partner and Internet dating had grown into a $304 million business (Orr,
2004). A $1 billion industry in 2008, Internet dating is projected to grow an
average of 10 percent annually through 2013 (Tulsiani, Best, and Card,
2008); in spite of global economic recession, the industry continues to grow
(Carpenter, 2008).
This growth in Internet dating services and patterns of utilization reflect
three factors: (1) technological change making Internet dating available and
efficient, coupled with growing computer literacy that makes it increasingly
accessible; (2) demographic change producing a greater number and variety
of persons who might be searching for romantic partners (people ‘‘at risk’’ of
using this technology); and (3) social change making Internet dating more
acceptable.

Technological Change and the Digital Divide

A ‘‘perfect storm’’ of events produced the rapid growth of Internet dating:


widespread Internet access, growing computer literacy, improved real-time
chat programs, and digital cameras. Internet access and use have increased
dramatically since 1997 (Fox, Anderson, and Rainie, 2005; Morales, 2009).
According to Current Population Survey data, 37 percent of U.S. house-
holds had at least one computer in 1997; by 2003, this number was 62
percent. Over the same time period, households with Internet access in-
creased from 18 percent to 55 percent (Day, Janus, and Davis, 2005). As the
Internet becomes an increasingly important part of many Americans’ lives,
used as a tool for everyday activities such as shopping, finding information
about new products and places, and a myriad of other tasks (Horrigan and
Rainie, 2006; Victory and Cooper, 2002), it is logical that individuals may
turn to the Internet to search for a partner.
Although proliferation of computers and related technology has been
dramatic, it has also been unequal. This stratified system of computer and
Internet access, marked by racial and class-based divisions, is known as the
‘‘digital divide’’ (Day, Janus, and Davis, 2005; Morgan and Welsh, 2006).
The result of the digital divide is unequal access to information and tech-
nology that have the potential to improve individuals’ lives. The unequal
distribution and growth of technology use across sociodemographic groups
is well documented in social science research (e.g., Fountain, 2005) and
evidence shows that gaps in Internet usage remain structured along socio-
demographic lines (Morales, 2009). Those who are most likely to be online
have higher incomes and education, are more likely to live in a suburb or a
Social Demography of Internet Dating 557
city, and are more likely to be white (Morgan and Welsh, 2006; Orr, 2004;
Rainie and Bell, 2004).
Disparities in material resources are fundamental drivers of the ‘‘digital
divide,’’ but recent critiques of this concept point out that differences in
Internet use are not merely a function of differential access to technological
hardware such as computers and high-speed modems (DiMaggio et al.,
2001; Peter and Valkenburg, 2006; van Dijk, 2006; Warschauer, 2003).
Internet use is also stratified by computer skills, which shape whether in-
dividuals are able to effectively use Internet technology. New material re-
sources, such as computers and the Internet, must be linked to schemas,
ways of knowing and understanding how to use material resources that
individuals often implement without noticing (Johnson-Hanks et al., 2006;
Katz and Rice, 2002; Sewell, 1992). Computer literacy, or virtual schemas
for computer/Internet use, must be acquired: a set of concepts must be
learned (e.g., logging on, uploading files) and strategies formed (e.g., ap-
propriate search queries, saving and archiving located information) (Moss-
berger, Tolbert, and Stansbury, 2003; Warschauer, 2003). Similar to
material resources, virtual schemas that facilitate Internet use are also struc-
tured by income, education, and race (Kvasny, 2006; Peter and Valkenburg,
2006).
Both material and virtual components of the digital divide—access to
computers, propensity toward use, and strength of computing skills—in-
dicate that breadth of Internet use will be strongly conditioned by social and
economic variables (Horrigan, 2008; van Dijk, 2006; Willis and Tranter,
2006; Zillien and Hargittai, 2009). As a technology becomes more common
and diffuses throughout the population, disparities become less prevalent
(Fountain, 2005). We argue, however, that use of Internet dating websites is
still in the early stages of diffusion, and thus will be stratified by socio-
economic factors.

Demographic Change and the Single Population

Since 1970, the general U.S. singles population, comprised of individuals


who have never been married and those who are currently divorced or
widowed, has grown for both men and women. As of 2003, 48 percent of
women and 45 percent of men were single compared to 40 percent and 35
percent, respectively, in 1970. This growth is due in part to steady increases
in the median age at first marriage, rising from 20.8 to 25.3 years for women
and 23.2 to 27.1 years for men between 1970 and 2003 (U.S. Census
Bureau, 2004).
Concomitant with the delay of first marriage has been increasing geo-
graphic mobility as individuals travel to college and abroad and frequently
move to meet the job demands of a global economy (Brym and Lenton,
2001; Furstenberg, 2003; Rosenfeld, 2007). Relocation moves individuals
558 Social Science Quarterly
away from established social networks and traditional avenues for mate
selection. A majority of individuals currently in a committed relationship
report meeting their partner or spouse through in-person social networks
such as family, friends, work, or school (Madden and Lenhart, 2006). As
individuals face new life demands that may remove them from these more
traditional ways of meeting potential partners, they may become more re-
liant on intermediaries to facilitate union formation.

Social Networks, Stigma, and Diffusion

Although the use of intermediaries in the marriage market is not new


(Ahuvia and Adelman, 1992; Sindberg, Roberts, and McClain, 1972), many
people hold negative attitudes toward formal marriage market intermediaries
that may be anchored in U.S. cultural understandings of love and partner-
ing. Dominant ideologies purport that love ‘‘happens,’’ it cannot be
planned, bought, or found through a search engine (Hardey, 2002) and it is
supposedly distinct from the commercial activities that permeate capitalist
societies (Zelizer, 2005). Internet dating, like other approaches that create a
rationalized shortcut to finding a compatible mate, is at odds with this
traditional notion of romantic, spontaneous love (Hollander, 2004).
Negative attitudes specific to Internet dating are related to safety and
deception on the Internet and the type of people who use Internet dating.
Many Americans express these concerns. Madden and Lenhart’s (2006)
analysis of data from the Pew Internet & American Life Project finds that
55.7 percent of respondents believe that a lot of people who use online
dating lie about whether they are married and 67.1 percent of respondents
think that Internet dating is dangerous because it puts your personal in-
formation on the Internet. Though fewer respondents indicated concerns
about the quality of Internet dating users, 29 percent agreed that people who
use online dating are desperate. Results from a recent cross-national survey
show similar levels of stigma. Of the respondents who claim they will never
use an online personal advertisement or online dating service, 30 percent
report they refuse to do so because ‘‘it could be dangerous,’’ while 22
percent claim that the Internet should not ‘‘be used to find someone a date’’
(Barnhoorn, 2007).
Stigma against Internet dating may be a deterrent for some, particularly if
it originates from close friends and family members (e.g., Wildermuth,
2004). Just as social networks can spread stigma and discourage individuals
from Internet dating, networks with enthusiastic members may do the op-
posite, facilitating adoption of Internet dating. Orr (2004) and Fallows
(2004) both note that the early users of new services and technologies serve
as evangelists, spreading the good word to the rest of society. This is con-
sistent with Rindfuss et al.’s (2004) analysis of family change in Japan.
Drawing from literature on network diffusion and social movements, the
Social Demography of Internet Dating 559
authors explain how innovative family behaviors characteristic of the second
demographic transition spread throughout the population. Following Rind-
fuss and colleagues (2004), we conceptualize Internet dating as an initially
rare and marginalized family behavior that is in transition to becoming more
prevalent and socially acceptable. New behavioral norms spread as more
people know someone who has engaged in new behaviors, changing atti-
tudes toward those behaviors. In this way, micro-level interpersonal inter-
actions link demographic processes and social change. This theory is
applicable to the spread of Internet dating: as a changing population of
singles turns to the Internet as a mate selection strategy and tells others about
their success, Internet dating may become an increasingly normative path to
union formation.

Data and Measures


Data

We use data from a December 2005 national telephone survey sponsored


by the Pew Internet & American Life Project. The Pew Internet & Amer-
ican Life Project collects and analyzes data on the impact of the Internet on
family, community, and public life. Princeton Survey Research Associates
International, an independent research company, conducted 3,215 tele-
phone interviews using standard list-assisted random-digit-dialing method-
ology. Detailed information about sampling and methodology, including
sample design, contact procedures, weighting, design effects, and response
rates, can be found in Madden and Lenhart (2006).
Unlike prior work on Internet dating, our description and analysis are
based on a nationally representative sample, not a sample drawn solely from
an Internet dating site or Internet users (e.g., Fiore and Donath, 2005). We
argue that sociodemographic factors select who will be at risk for Internet
dating. By analyzing both respondents who have used Internet dating and
those who have not, our sample allows us to examine the sociodemographic
factors associated with Internet dating net of the selection processes leading
one to be at risk of its use.

Measures

Outcome Variables. ‘‘SINGLE’’ STATUS respondents are: (1) currently di-


vorced, separated, widowed, or never married, or (2) married or in a com-
mitted relationship for fewer than five years. This captures a range of people
who could have used Internet dating since its rapid expansion in 2000,
including people who are currently married.
INTERNET USERS are respondents who have ever used either Internet or
560 Social Science Quarterly
email at least occasionally. SINGLE INTERNET USERS are Internet users who
have been single in the past five years.
An INTERNET DATER is a respondent who has ever gone to/used an online
dating website where you can meet someone online. A ‘‘yes’’ response in-
dicates that an individual has used an Internet dating website at some point;
our data do not allow us to distinguish between current and previous users.
Respondents who have never used an online dating website, and those who
do not use the Internet, are coded 0. Internet dating site is not defined in the
survey questionnaire; it was up to the respondent to decide whether sites she
or he used qualified as dating sites. Respondents reported using a wide range
of sites, including traditional pay sites (match.com), specialized sites (black-
planet.com, gay.com), social networking sites (myspace.com), and general
chat rooms.

Sociodemographic Variables. GENDER is measured dichotomously. RACE


categories include white, black, and other (including Asian or Pacific Is-
landers, mixed race, Native American, and other). AGE is measured con-
tinuously; we also include a squared term to capture nonlinear effects of age
based on likelihood ratio tests of model fit. EDUCATION is measured with
four categories: less than high school, high school, some college (including
technical, trade, or vocational school), and college degree or higher. INCOME
refers to total family income from all sources before taxes in 2004. As is
typical in survey research, we had a great many missing responses on the
income measure (21.5 percent). To prevent loss of these cases, we included
missing as an analytical category instead of imputing. Categories include
missing, less than $30,000, $30,000–$75,000, and $75,000 and higher.
COMMUNITY TYPE is measured categorically and includes urban, suburban,
and rural. RELIGION is measured categorically and includes Protestant,
Catholic, and other religion. We also include indicators of being currently
DIVORCED and being a part-time or full-time STUDENT.

Computer Literacy. First, we create a measure of how often respondents


use the Internet from home or work. FREQUENT USERS are respondents who
use the Internet once a day or more from home or work. MEDIUM USERS are
respondents who use the Internet between two and five days a week from
home or work. Respondents who use the Internet frequently from one
location and less frequently at the other are classified as frequent users.
Respondents who are neither frequent nor medium users are categorized as
INFREQUENT USERS. Next, we create an ACTIVITY INDEX that represents the
number of distinct activities for which the respondents used the Internet.
Respondents were asked if they used the Internet for the following seven
activities: sending or reading email; participating in chat rooms; creating or
reading a blog; rating a product, person, or service; using online classified
Social Demography of Internet Dating 561
sites; or using online social or professional networking sites. The index
ranges from 1–7, with higher scores indicating that the respondent uses the
Internet for a wider range of activities. Lastly, DURATION is the number of
years the respondent has been using the Internet. Less than one year of use is
coded as 0.

Social Networks. To address social network effects on the likelihood of


Internet dating, we create dichotomous measures of whether respondents
know someone who is an Internet dater. All respondents were asked whether
they know anyone (aside from themselves) who has used an online dating
website (KNOW). Respondents who know someone who has used Internet
dating were further asked if they know someone who has (1) gone on a date
with someone they met through a dating website (KNOW DATER), or (2) been
in a long-term relationship with or married someone they met through a
dating website (KNOW LONG-TERM).

Internet Dating Attitudes. All Internet users were asked whether they
agree or disagree with six statements about Internet dating: (1) ‘‘Online
dating is a good way to meet people,’’ (2) ‘‘Online dating allows people to
find a better match for themselves because they can get to know a lot more
people,’’ (3) ‘‘Online dating is easier and more efficient than other ways of
meeting people,’’ (4) ‘‘People who use online dating are desperate,’’ (5) ‘‘A
lot of people who use online dating lie about whether they are married,’’ and
(6) ‘‘Online dating is dangerous because it puts your personal information
on the internet.’’ We create a scale of overall INTERNET DATING ATTITUDES to
summarize these responses; this scale ranges from 0 (all negative attitudes) to
6 (all positive attitudes).

Methods

First, we report descriptive statistics for the full sample and subsamples of
Internet users and Internet using singles, providing chi-square tests for
differences between groups. We then report sequential multivariate logistic
regressions to examine three dependent variables that characterize the se-
lective processes leading to Internet dating. We test the associations of so-
ciodemographic factors with the likelihood of Internet dating in the general
population and then model the structural factors that condition being at risk
of Internet dating: factors associated with likelihood of being an Internet
user and likelihood of being single among Internet users. Finally, we model
sociodemographic, computer literacy, social network, and attitudinal factors
associated with likelihood of Internet dating among the at-risk population of
single Internet users.
562 Social Science Quarterly
Results

The Pew Internet & American Life Project provides the first ever na-
tionally representative data on the prevalence of Internet dating in the U.S.
population. Figure 1 illustrates our approach and estimates the prevalence of
Internet dating. Only 5.6 percent of the full sample has ever used Internet
dating. We divide the full sample into those who use a computer (72
percent) and those who do not (28 percent). Among computer users, almost
all use the Internet or email (97 percent). Only 70 percent of the original
sample uses the Internet and is therefore eligible to be an Internet dater.
Among Internet users, just under half were single or married less than five
years. This highly selective process leaves 32 percent of the original sample—
single Internet users—at risk for Internet dating. Once we restrict analysis to
the population at risk for Internet dating, over 17 percent of single Internet
users have used Internet dating websites.

Descriptive Statistics

Table 1 presents sociodemographic characteristics of the full sample and


subsamples. We illustrate how the sociodemographic composition significantly
changes with increasingly selective samples. Compared to the full sample,
Internet users have a higher proportion of males, whites, suburban and urban
residents, respondents with a college degree or higher, and respondents who
make $75,000 or more per year. This indicates the presence of a material
digital divide structured by socioeconomic status. Internet users are also, on
average, five years younger than the full sample (47.51 vs. 52.34).

FIGURE 1
Conceptual Approach for Estimating Use of Internet Dating Sites

Full Computer Internet or Single or Internet


Sample Use Email Use Committed <5 Daters
n = 3,215 Years

Yes Yes Yes


Yes
n =2,313 n =2,252 n =1,026
n =180
71.94% 97.36% 45.56%
17.54%
(5.6% of full)

No No No
n =902 n =59 n =1,228 No
28.06% 2.55% 54.53% n =846
TABLE 1
Sociodemographic Characteristics of Full Sample and Subsamples

Sample Population Between Group Differencesa

All Internet Users Single Internet Users Internet Daters All vs. I All vs. SI I vs. SI SI vs. ID
Sample size (N) 3,215 2,252 1,026 180
nnn
Male 45.7% 48.3% 48.2% 54.4%
Race
n nnn nnn
White 83.7% 84.8% 79.0% 79.7%
nnn nnn nnn
Black 10.3% 8.9% 13.2% 13.5%
nn nn
Other 6.0% 6.3% 7.8% 6.8%
nnn nnn nnn n
Ageb 52.34 47.51 42.44 39.2
Social Demography of Internet Dating

(19.5) (17.6) (19.3) (15.7)


Education
nnn nnn nnn
o HS 9.9% 4.4% 6.5% 7.9%
nnn
HS/GED 31.8% 25.6% 25.8% 24.1%
nnn nn nn
Some college 25.9% 28.6% 31.4% 30.9%
nnn nnn nnn
BA1 32.4% 41.4% 36.3% 37.1%
Income
nnn nnn nnn
o $30K 26.2% 18.2% 29.0% 27.2%
nnn
$30–75K 31.0% 35.3% 35.9% 40.0%
nnn nnn nnn n
$75K1 21.7% 29.2% 18.7% 23.9%
nnn
Missing 21.2% 17.4% 16.5% 8.9%
Residence
nnn
Rural 22.2% 19.8% 18.3% 13.9%
n nn nn
Suburban 50.3% 51.7% 48.2% 50.5%
n nnn nnn
Urban 27.5% 28.5% 33.5% 35.5%
Religion
nn
Protestant 63.0% 61.0% 59.6% 62.8%
563
TABLE 1—continued
564

Sample Population Between Group Differencesa

All Internet Users Single Internet Users Internet Daters All vs. I All vs. SI I vs. SI SI vs. ID
nn nn n
Catholic 21.4% 21.6% 19.0% 13.7%
nnn nnn nnn
Other 15.7% 17.3% 21.4% 23.4%
nn nnn nnn nn
Divorced 11.3% 10.3% 21.1% 29.9%
nnn nnn nnn
Student 10.3% 13.1% 20.4% 20.7%
a
Significance tests indicate chi-square tests for deviance between the two indicated groups. All represents full sample, I the Internet users, SI the single
Internet users, and ID the Internet daters. p values: npo0.05; n npo0.01; n n npo0.001.
b
Reported as mean and standard deviation.
Social Science Quarterly
Social Demography of Internet Dating 565
Examining the subsample at risk for Internet dating—single Internet us-
ers—we see that they are significantly different from both the full sample
and Internet users. For example, even though Internet users have a higher
proportion of whites (84.8 percent) compared to the full sample (83.7
percent), there is a lower percentage of white respondents among single
Internet users (79 percent), reflecting racial differences in likelihood of being
single. Single Internet users are also younger (42.44 vs. 47.51), reflecting age
patterns of marriage, and have nearly twice the proportion of respondents
currently enrolled in school.
Few sociodemographic characteristics distinguish Internet daters from
single Internet users. Internet daters are, on average, 2.5 years younger than
single Internet users and have a higher proportion of both respondents who
make $75,000 or more per year and respondents who are currently divorced.
Thus, among a relatively younger at-risk population pursuing higher ed-
ucation but with moderate income, those who are younger with higher
earnings are more prevalent in the population choosing Internet dating to
find a mate.
Table 2 presents percentages and means of computer literacy, social net-
works, and attitudinal measures among Internet users, single Internet users,
and Internet daters. Compared to all Internet users, single Internet users
report a higher average activity index but shorter duration of use. These
patterns most likely reflect cohort differences in Internet usage, with younger
cohorts having less time (in years) exposure to the Internet but greater
integration of it into their daily lives. A greater proportion of single Internet
users know someone who has used Internet dating (46.6 percent vs. 30
percent of all Internet users) and they report, on average, more positive
attitudes about Internet dating. These differences may reflect the influence
of both age and social networks.
The distributions of computer literacy, social networks, and attitudes are
significantly different between single Internet users and Internet daters.
Compared to single Internet users, more Internet daters use the Internet
once or more a day (80.9 percent vs. 63 percent), use the Internet for a
greater average number of activities (3.15 vs. 2.2), and have a longer du-
ration of Internet use (8.29 years vs. 7.18 years). More Internet daters know
someone who has used Internet dating and they hold more positive attitudes
about Internet dating. Internet daters are distinct from Internet users and
single Internet users in that they have high computer literacy, have social
networks that include Internet daters, and hold more positive attitudes to-
ward Internet dating.

Multivariate Analyses

We use weighted multivariate logistic regression to examine the selective


processes that make Internet daters a demographically distinct subsample of
TABLE 2
566

Computer Literacy and Social Network Distributions

Between Group
Sample Population Differencesa

Internet Users Single Internet Users Internet Daters SI vs. I ID vs. SI


Sample size (N) 2,252 1,026 180
Computer literacy
nnn
Frequent Internet user 63.9% 63.0% 78.9%
nnn
Medium Internet user 24.0% 22.9% 11.7%
nnn
Infrequent Internet user 12.2% 14.1% 9.4%
nnn nnn
Activity indexb 1.99 2.2 3.15
(1.3) (1.4) (1.5)
nnn nnn
Durationb 7.55 7.18 8.28
(4.4) (4.1) (4.3)
Social networks
nnn nnn
Know 37.9% 46.6% 76.7%
nnn nnn
Know – Date 33.1% 39.7% 67.8%
nnn nnn
Know – Long Term 19.0% 22.4% 45.6%
Attitudes
nnn nnn
ID attitudesb 2.51 2.82 3.81
(1.8) (1.8) (1.6)
a
Significance tests indicate chi-square tests for deviance between the two indicated groups. I represents the Internet users, SI the single Internet users, and
ID the Internet daters. p values: npo0.05; n npo0.01; n n npo0.001.
b
Reported as mean and standard deviation.
Social Science Quarterly
Social Demography of Internet Dating 567
the general population. Table 3 presents odds ratios for the effects of so-
ciodemographic variables on the likelihood of: (1) ever being an Internet
dater (in the full sample); (2) being an Internet user; (3) being single in the
past five years, conditional on being an Internet user; and (4) ever being an
Internet dater, conditional on being an Internet user and being single. Note
that the results for each dependent variable are based on separate logistic
equations. The first column shows that a number of variables are signifi-
cantly associated with the likelihood of being an Internet dater. As previ-
ously discussed, however, use of Internet dating is contingent on Internet
access and being single. We turn now to an examination of the sociode-
mographic correlates of these selective processes.
Analyzing Internet use among the full sample (Column 2 of Table 3)
provides strong evidence for the digital divide. Both education and income
are positively associated with likelihood of Internet use. Even after control-
ling for socioeconomic status, black respondents are less likely than white
respondents to use the Internet or email at least occasionally (OR 5 0.59),
reflecting racial disparities in access to technology. Younger individuals are
more likely to use the Internet than are older respondents (OR 5 0.94 per
year of age), a finding consistent with ideas that younger people may be
more willing to adopt newer technologies. This finding is also in sync with
theories of the importance of computer literacy or virtual schemas for
effective technology use. Not only have younger individuals been exposed to
computers and related technologies for a larger proportion of their lives,
many have also received formal computer education, unlike their older
counterparts, thus enhancing their abilities to utilize these technologies.
The third column of Table 3 highlights the second selective process
leading individuals to be at risk of Internet dating—odds of being single
(i.e., not currently in a committed relationship or entered one within the last
five years) among Internet users. The odds of being single vary dramatically
with age—they are very high at young ages, decline until later middle-age,
and then increase again as individuals become more likely to experience
divorce or the death of their partner. Although education and income are
highly correlated, they exert distinct effects on the probability of being
single. Respondents with higher education have a higher likelihood of being
single than the reference group of high school graduates, reflecting delays in
marriage due to education (students are much more likely to be single than
nonstudents, OR 5 1.48). In contrast, higher income is associated with
decreased likelihood of being single.
The fourth column of Table 3 estimates likelihood of Internet dating net
of these selective processes. When we restrict analyses to the population at
risk—single Internet users—sociodemographic factors are no longer asso-
ciated with likelihood of Internet dating. The only significant predictor of
Internet dating is being divorced (OR 5 2.10), perhaps because divorced
respondents are still embedded in social networks created during their mar-
riage and are seeking new avenues to find a partner. Ultimately, patterns of
568 Social Science Quarterly
TABLE 3
Odds Ratios for Internet Dating and Selective Processes: Sociodemographic
Variables, Weighted Multivariate Logistic Regression

Outcome Internet Internet Single in Internet


Variable Dater Use Past 5 Years Dater
Sample Population All All Internet Single Internet
Users Users
Sample size (N) 3,031 3,030 2,125 957
Gender
Male 1.24 0.85 1.24 n 1.31
Female — — — —
Race
White — — — —
Black 0.85 0.59 n n 1.43 0.92
Other 0.81 0.62 0.87 0.92
Age 0.93 n 0.94 n n n 0.82 n n n 1.01
Age2 1.00 1.00 1.001 n n n 0.999
Education
Less than HS 0.87 0.44 n n n 1.72 1.35
HS/GED — — — —
Some college 1.37 2.74 n n n 1.56 n n 1.02
BA/BS1 1.49 5.25 n n n 1.39 n 1.02
Income
o $30,000 0.86 0.39 n n n 2.60 n n n 0.73
$30–75,000 — — — —
$75,0001 1.03 3.28 n n n 0.51 n n n 0.97
Missing 0.57 0.48 n n n 0.88 0.64
Community type
Urban 1.65 n 1.25 1.62 n n 1.50
Suburban 1.35 1.2 1.22 1.45
Rural — — — —
Religion
Protestant 0.98 1.22 0.76 1.20
Catholic 0.52 n 1.26 0.67 n 0.63
Other — — — —
Divorceda 4.48 n n n 0.89 — 2.10 n n
Student 1.07 2.02 n 1.62 n 1.00
 2LL  717.108  1229.421  1157.368  442.115
df 17 17 16 17
a
Divorced is excluded from the equation estimating single in the past 5 years as all individuals
who are currently divorced are, by definition, single.
n
po0.05; n npo0.01; n n npo0.001.

Internet dating are influenced by extant patterns of stratification: age, ed-


ucation, income, and race are associated with both Internet use and like-
lihood of being single. Once we control for Internet access and single status,
however, these stratifying forces no longer predict who utilizes Internet
dating to search for a partner.
Social Demography of Internet Dating 569
Next, we turn to an examination of computer literacy, social networks,
and attitudes toward Internet dating to better understand how other social
factors affect differential propensity toward Internet dating among our sub-
sample at risk. As previously discussed, computer literacy, a virtual com-
ponent of the digital divide, should be positively related to likelihood of
Internet dating, as should social networks that include Internet daters and
positive attitudes toward Internet dating. Table 4 presents results of
weighted multivariate logistic regressions that incorporate these proximate
determinants in the estimation of the likelihood of Internet dating among
single Internet users. The first three columns show these determinants
modeled individually; in Model 4 their effects are estimated simultaneously.
Model 1 of Table 4 indicates a significant effect of computer literacy on
likelihood of Internet dating. For each additional activity the respondent
does on the Internet, the odds of Internet dating increase by 1.81, indicating
that the more individuals use the Internet for a wide variety of day-to-day
activities, the more likely they are to also turn to the Internet as a tool for
finding potential partners. Duration of Internet use shows no significant
relationship with likelihood of Internet dating, suggesting that factors
other than mere exposure to technology may influence who utilizes Internet
dating.
Model 2 of Table 4 demonstrates the influence of social networks on
Internet dating. Respondents who know someone who has used Internet
dating websites and those who know individuals who have been in a long-
term relationship with someone they have met through an Internet dating
website have significantly greater odds of ever Internet dating themselves
(OR of 3.48 and 2.04, respectively). Individuals who are embedded in social
networks in which Internet dating is an acceptable and successful behavior
are much more likely to turn to Internet dating.
Last, Model 3 of Table 4 shows that, not surprisingly, stigma toward
Internet dating is associated with decreased likelihood of Internet dating; the
more positive attitudes toward Internet dating one holds, the more likely
one is to Internet date (OR 5 1.47). The effect of divorce status remains
significant across these three models, but no other sociodemographic cov-
ariates show consistently significant effects.
In the final model, we include the full set of these proximate determi-
nants. In general, the effects described above persist. Specifically, no socio-
demographic variables significantly influence the likelihood of Internet
dating. These results confirm that while sociodemographic variables strongly
condition those who are at risk of Internet dating, ultimately, they do not
predict whether a respondent adopts this technology or not. In contrast, the
more proximate variables—using the Internet for a variety of activities
(OR 5 1.55), knowing someone who has used Internet dating websites
(OR 5 2.19), and holding positive attitudes toward Internet dating
(OR 5 1.32)—show independent positive effects on the likelihood of In-
ternet dating.
570 Social Science Quarterly
TABLE 4
Odds Ratios for Use of Internet Dating Within Subsample at Risk:
Sociodemogrpahic and Proximate Determinants,
Weighted Multivariate Logistic Regressions

Single Internet Users

Model 1 Model 2 Model 3 Model 4


Sample size (N) 957
Gender
Male 1.21 1.54 n 1.12 1.31
Female — — — —
Race
White — — — —
Black 1.07 1.25 1.41 1.91
Other 1.03 1.10 0.93 1.12
Age 1.07 1.01 0.99 1.03
Age2 0.99 0.99 1.00 1.00
Education
Less than HS 1.28 1.36 1.59 1.43
HS/GED — — — —
Some college 0.72 0.87 1.00 0.66
BA/BS1 0.59 0.68 0.83 0.46
Income
o $30,000 0.72 0.83 0.67 0.72
$30–75,000 — — — —
$75,0001 0.74 0.8 1.02 0.72
Missing 0.73 0.67 0.65 0.68
Community type
Urban 1.44 1.34 1.73 1.64
Suburban 1.78 n 1.42 1.65 1.87
Rural — — — —
Religion
Protestant 1.47 1.28 1.29 1.59
Catholic 0.96 0.80 0.82 1.38
Other — — — —
Divorced 1.94 n n 1.85 n 2.06 n 1.75
Student 0.98 1.07 0.82 0.88
Computer literacy
Activity index 1.81 n n n 1.55 n n n
Duration 1.04 1.03
Social networks
Know 3.48 n n n 2.19 n
Know – Long term 2.04 n n 1.66
Attitudes 1.47 n n n 1.32 n n n
 2LL  374.01  401.52  281.42  239.64
df 19 19 18 22

po0.05;
n
po0.01;
nn
po0.001.
nnn
Social Demography of Internet Dating 571
Discussion

Our analyses show that the prevalence and correlates of Internet dating
change dramatically when considering the various factors leading one to be
at risk of Internet dating. Initial analysis of all respondents, regardless of
Internet use and marital status, indicates that 5.6 percent of all respondents
have used Internet dating, and that some sociodemographic variables—
being male, having education beyond the high school level, and living in an
urban or suburban community—are associated with higher likelihood of
ever using Internet dating websites. When we focus analysis on Internet use,
strong evidence of the digital divide emerges: race, education, and income
are all significantly associated with the likelihood of Internet use. Further,
among the sample of Internet users, race, education, and income are as-
sociated with likelihood of being single. Among the subsample at risk of
Internet dating, 17.5 percent of Internet using singles have used Internet
dating. What is most compelling, however, is that once the digital divide
and relationship status are taken into account, variables previously associated
with likelihood of Internet dating are no longer significant predictors of its
use. Sociodemographic characteristics ultimately have indirect effects on
likelihood of Internet dating through their direct effects on Internet use and
single status. They define the populations at risk for Internet dating, but fail
to determine who dates online and who does not among the at-risk pop-
ulation. By analyzing a sample of all Americans, not just Internet users or
Internet daters, this study highlights the distinct and sometimes counter-
balancing effects of sociodemographic factors on likelihood of Internet use,
likelihood of being single, and likelihood of Internet dating.
We find support for both material and virtual aspects of the digital divide.
First, we find that Internet use among the full sample is stratified by so-
cioeconomic status and race. However, we find no significant socioeconomic
predictors of Internet dating within the subpopulation of Internet using
singles. Beyond the digital divide, socioeconomic status does not determine
use of Internet dating among the at-risk population. Second, our analyses
show that computer literacy, a nonmaterial component of the digital divide,
is positively associated with use of Internet dating websites among single
Internet users. More experienced Internet users, that is, those who use the
Internet for a variety of tasks, may be more comfortable using Internet
dating websites, which require skills such as creating an account and logging
on, uploading pictures, and searching a database. Additionally, experienced
users may be more comfortable with using the Internet in the task of finding
a partner because it is a tool they utilize in their daily activities.
We also find indirect support for the network diffusion of novel family
practices through social networks and changing attitudes (Rindfuss et al.,
2004). Knowing someone who has used Internet dating and holding positive
attitudes about Internet dating were both strongly and positively associated
with use of Internet dating websites. The existence of social network effects
572 Social Science Quarterly
on likelihood of Internet dating extends prior research on predictors of
Internet dating that focused more exclusively on personality characteristics
(e.g., Whitty, 2006). Although we cannot address the causal order of these
relationships, our results are consistent with the idea that positive attitudes
and stories of success are spread through social networks, increasing par-
ticipation in Internet dating among people who already know an Internet
dater.

Conclusion

We show that social, economic, and demographic variables are associated


with whether a person has ever used Internet dating websites. These effects
operate primarily on the conditioning variables, whether a person is an
Internet user and whether he or she is single. Net of these variables that
define the ‘‘at-risk’’ population, the distal social, economic, and demo-
graphic variables have modest effects. Instead, proximate variables, such as
computer literacy, social networks, and attitudes toward Internet dating
technology, strongly influence an individual’s propensity to use Internet
dating. Our results concerning social networks suggest that Internet dating
will continue to grow as Internet daters tell friends and family about their
experiences, thus encouraging other Internet using singles to use Internet
dating. Stratification and inequality in Internet dating due to the material
and virtual aspects of the digital divide will continue but may attenuate as
Internet use becomes more common.
Although we use data from the first study of Internet dating from a
nationally representative sample of Americans, there are still some limita-
tions to our analyses, including cross-sectional data and lack of information
about relationships that are initiated through Internet dating. Future lon-
gitudinal data that sample both Internet daters and non-Internet daters will
be valuable because they will allow one to assess the temporal order of key
variables like knowing an online dater and the respondent’s reported use.
Also, longitudinal data would allow description of how Internet-facilitated
relationships differ from other relationships in outcomes such as satisfaction,
homogamous matching, and longevity. Given that participation in Internet
dating shows no signs of slowing in the upcoming years, it is important that
researchers devote more attention to understanding whether the digital di-
vide will have increasing influence on rates of union formation and whether
Internet-facilitated mate selection processes differ substantively from more
traditional ones.

REFERENCES

Ahuvia, Aaron C., and Mara B. Adelman. 1992. ‘‘Formal Intermediaries in the Marriage
Market: A Typology and Review.’’ Journal of Marriage and the Family 54(2):452–63.
Social Demography of Internet Dating 573
Anderson, Traci L. 2005. ‘‘Relationships Among Internet Attitudes, Internet Use, Romantic
Beliefs, and Perceptions of Online Romantic Relationships.’’ CyberPsychology & Behavior
8(6):521–31.
Barnhoorn, Catherine. 2007. ‘‘Desperately Seeking Love Online.’’ Synovate. Available at
hhttp://www.synovate.comi.
Barraket, Jo, and Millsom S. Henry-Waring. 2008. ‘‘Getting it On(line): Sociological Per-
spectives on E-Dating.’’ Journal of Sociology 44(2):149–65.
Ben-Ze’ev, Aaron. 2004. Love Online: Emotions and the Internet. New York: Cambridge
University Press.
Brym, Robert J., and Rhonda L. Lenton. 2001. Love Online: A Report on Digital Dating in
Canada. Available at hhttp://www.nelson.com/nelson/harcourt/sociology/newsociety3e/love-
online.pdfi.
Carpenter, Susan. 2008. No Recession for Online Dating Sites. Available at hhttp://www.la
times.com/features/lifestyle/la-ig-dating28-2008dec28,0,1563805.storyi.
Cooper, Alvin, and Leda Sportolari. 1997. ‘‘Romance in Cyberspace: Understanding Online
Attraction.’’ Journal of Sex Education and Therapy 22(1):7–14.
Day, Jennifer C., Alex Janus, and Jessica Davis. 2005. Computer and Internet Use in the
United States: 2003. Washington, DC: U.S. Census Bureau.
DiMaggio, Paul J., Eszter Hargittai, W. Russell Neuman, and John P. Robinson. 2001.
‘‘Social Implications of the Internet.’’ Annual Review of Sociology 27:307–36.
Donn, Jessica E., and Richard C. Sherman. 2002. ‘‘Attitudes and Practices Regarding the
Formation of Romantic Relationships on the Internet.’’ CyberPsychology & Behavior
5(2):107–23.
Ellison, Nicole, Rebecca Heino, and Jennifer Gibbs. 2006. ‘‘Managing Impressions Online:
Self-Presentation Processes in the Online Dating Environment.’’ Journal of Computer-Me-
diated Communication 11(2):415–41.
Espinoza, Javier. 2009. Online Dating Sites Flirt with Record Growth. Available at hhttp://
www.forbes.com/facesinthenews/2009/01/06/online-dating-industry-face-markets-
cx_je_0105autofacescan01.htmli.
Fallows, Deborah. 2004. The Internet and Daily Life. Pew Internet & American Life Project.
Available at hwww.pewinternet.orgi.
Fiore, Andrew T., and Donath, Judith S. 2005. ‘‘Homophily in Online Dating: When Do
You Like Someone Like Yourself?’’ Presented at CHI ‘05 Extended Abstracts on Human
Factors in Computing Systems. Portland, OR.
Fountain, Christine. 2005. ‘‘Finding a Job in the Internet Age.’’ Social Forces 83(3):1235–62.
Fox, Susannah, Janna Q. Anderson, and Lee Rainie. 2005. The Future of the Internet. Pew
Internet & American Life Project. Available at hwww.pewinternet.orgi.
Furstenberg, Frank F. 2003. ‘‘Growing Up in American Society: Income, Opportunities, and
Outcomes.’’ Pp. 211–34 in Victor Marshall and Walter R. Heinz, eds., Social Dynamics of the
Life Course: Transitions, Institutions, and Interrelations. Edison, NJ: Aldine Transaction.
Hardey, Michael. 2002. ‘‘Life Beyond the Screen: Embodiment and Identity Through the
Internet.’’ Sociological Review 50(4):570–85.
Hitsch, Günter J., Ali Hortaçsu, and Dan Ariely. 2006. What Makes You Click?: Mate
Preferences and Matching Outcomes in Online Dating. MIT Sloan Research Paper No. 4603-
06. Available at hhttp://ssrn.com/abstract=895442i.
574 Social Science Quarterly
Hollander, Paul. 2004. ‘‘The Counterculture of the Heart.’’ Society 41(2):69–77.
Horrigan, John. 2008. Home Broadband Adoption 2008. Pew Internet & American Life
Project. Available at hwww.pewinternet.orgi.
Horrigan, John, and Lee Rainie. 2006. The Internet’s Growing Role in Life’s Major Moments.
Pew Internet & American Life Project. Available at hwww.pewinternet.orgi.
Johnson-Hanks, Jennifer, Christine Bachrach, Hans-Peter Kohler, and S. Philip Morgan.
2006. Social Structure, Social History, and the American Family. EFC Parenthood Group.
Available at hhttp://www.ssc.upenn.edu/soc/courses/2005/spring/soc796_hpkohler/readings/
john06s.pdfi.
Katz, James E., and Ronald E. Rice. 2002. Social Consequences of Internet Use: Access, In-
volvement, and Interaction. Cambridge, MA: MIT Press.
Kvasny, Lynette. 2006. ‘‘Cultural (Re)Production of Digital Inequality in a US Community
Technology Initiative.’’ Information, Communication & Society 9(2):160–81.
Madden, Mary, and Amanda Lenhart. 2006. Online Dating. Pew Internet & American Life
Project. Available at hhttp://www.pewinternet.orgi.
Morales, Lymari. 2009. Nearly Half of Americans Are Frequent Internet Users. Available at
hhttp://www.gallup.com/poll/113638/Nearly-Half-Americans-Frequent-Internet-Users.aspxi.
Morgan, S. Philip, and Whitney Welsh. 2006. ‘‘Stability and Change in the Digital Terrain.’’
Unpublished manuscript. Duke University.
Mossberger, Karen, Caroline J. Tolbert, and Mary Stansbury. 2003. Virtual Inequality:
Beyond the Digital Divide. Washington, DC: Georgetown University Press.
Orr, Andrea. 2004. Meeting, Mating, and Cheating: Sex, Love, and the New World of Online
Dating. Upper Saddle River, NJ: Reuters.
Peter, Jochen, and Patti M. Valkenburg. 2006. ‘‘Adolescents’ Internet Use: Testing the
‘Disappearing Digital Divide’ Versus the ‘Emerging Digital Differentiation’ Approach.’’
Poetics 34:293–305.
Rainie, Lee, and Peter Bell. 2004. ‘‘The Numbers That Count.’’ New Media and Society
6(1):44–54.
Rindfuss, Ronald R., Minja K. Choe, Larry L. Bumpass, and Noriko O. Tsuya. 2004. ‘‘Social
Networks and Family Change in Japan.’’ American Sociological Review 69(6):838–61.
Rosenfeld, Michael J. 2007. The Age of Independence: Interracial Unions, Same-Sex Unions,
and the Changing American Family. Cambridge, MA: Harvard University Press.
Sewell, William H. 1992. ‘‘A Theory of Structure: Duality, Agency, and Transformation.’’
American Journal of Sociology 98:1–29.
Sindberg, Ronald M., Allyn F. Roberts, and Duane McClain. 1972. ‘‘Mate Selection
Factors in Computer Mediated Marriages.’’ Journal of Marriage and the Family 34(4):
611–14.
Tulsiani, Bobby, Mark Best, and David Card. 2008. US Paid Content Forecast, 2008 to 2013.
Available at hhttp://www.jupiterresearch.comi.
U.S. Census Bureau. 2004. America’s Families and Living Arrangements: 2003. Washington,
DC: U.S. Census Bureau.
van Dijk, Jan A. G. M. 2006. ‘‘Digital Divide Research, Achievements and Shortcomings.’’
Poetics 34:221–35.
Social Demography of Internet Dating 575
Victory, Nancy J., and Kathleen B. Cooper. 2002. A Nation Online: How Americans Are
Expanding Their Use of the Internet. Washington, DC: National Telecommunications and
Information Administration, Economics and Statistics Administration, U.S. Department of
Commerce.
Warschauer, Mark. 2003. Technology and Social Inclusion: Rethinking the Digital Divide.
Cambridge, MA: MIT Press.
Whitty, Monica. 2006. Cyberspace Romance: The Psychology of Online Relationships. New
York: Pallgrave MacMillan.
———. 2008. ‘‘Revealing the ‘Real’ Me, Searching for the ‘Actual’ You: Presentations of Self
on an Internet Dating Site.’’ Computers in Human Behavior 24:1707–23.
Wildermuth, Susan M. 2004. ‘‘The Effects of Stigmatizing Discourse on the Quality of
On-Line Relationships.’’ Cyberpsychology & Behavior 7(1):73–84.
Willis, Suzanne, and Bruce Tranter. 2006. ‘‘Beyond the ‘Digital Divide’: Internet Diffusion
and Inequality in Australia.’’ Journal of Sociology 42(1):43–59.
Zelizer, Viviana A. 2005. The Purchase of Intimacy. Princeton, NJ: Princeton University
Press.
Zillien, Nicole, and Eszter Hargittai. 2009. ‘‘Digital Distinction: Status-Specific Types of
Internet Usage.’’ Social Science Quarterly 90(2):274–91.
Copyright of Social Science Quarterly (Blackwell Publishing Limited) is the property of Wiley-Blackwell and
its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.

Das könnte Ihnen auch gefallen