ArticlePDF Available

Cognition, Personality, and Individual Response to Technological Change: The Case of Internet Adoption

Authors:
Article

Cognition, Personality, and Individual Response to Technological Change: The Case of Internet Adoption

Center for Demography and Ecology
University of Wisconsin-Madison
Cognition, Personality, and Individual Response to
Technological Change: The Case of Internet Adoption
Jeremy Freese
Salvador Rivas
CDE Working Paper No. 2006-07
COGNITION, PERSONALITY, AND INDIVIDUAL RESPONSE TO
TECHNOLOGICAL CHANGE: THE CASE OF INTERNET ADOPTION
Jeremy Freese
Robert Wood Johnson Scholars in Health Policy Research Program, Harvard University
Department of Sociology, University of Wisconsin-Madison
Salvador Rivas
Department of Sociology, University of Wisconsin-Madison
Word Count: 13,895 (inclusive of abstract, main text, footnotes, and references), 8 tables
Running head: Cognition, Personality, and Internet Adoption
COGNITION, PERSONALITY, AND INDIVIDUAL RESPONSE TO
TECHNOLOGICAL CHANGE: THE CASE OF INTERNET ADOPTION
Abstract
Existing sociological and other literature on digital inequalities in the United States has
given little specific attention to the potential roles of cognitive ability and personality, especially
in terms of how effects of these variables may vary in predictable ways depending on
individuals’ social position. We consider the matter specifically with regard to whether late
midlife adults have become Internet users. Consistent with our predictions, we find that
cognitive ability and openness to experience are strongly and positively associated with Internet
adoption in this cohort, and that neuroticism is inversely associated with adoption. Including
these variables also attenuates the apparent causal effect of both education and income on
Internet adoption. We also find that the effect of cognitive ability is smaller for those with more
education, more wealth, and who have jobs that use the Internet. Openness and neuroticism are
importantly also mediated by social circumstances, although the latter more ambiguously. The
results underscore not only the importance of considering basic psychological characteristics in
understanding why people in similar social positions respond differently to change, but also how
social conditions create conditions under which psychological differences are more- or less-
strongly implicated in behavioral differences.
Cognition, Personality, and Internet Adoption - 1
A famous description of the sociological enterprise charges the discipline with
understanding the interaction of history and biography, including how large-scale social changes
are manifested in the everyday lives of individuals (Mills 1959). When discussing the individual
consequences of social change, sociologists commonly emphasize that such consequences
usually vary greatly depending on one’s social position. Sociologists also generally
acknowledge that psychological characteristics may have their own, autonomous influence on
how people respond to change, although such characteristics are less often and less elaborately
considered. An alternative position is that not only are the biographical consequences of basic
psychological characteristics interesting and fundamental in their own right, but the failure to
consider these characteristics can also limit and distort sociologists’ understanding of how social
position influences response to change. More importantly, whether and how much basic
psychological characteristics influence behavior and outcomes in the face of social change might
itself vary in systematic and sociologically explicable ways depending on individual
circumstances.
We expand upon this orienting premise here with regard to social change brought about
by technological change. In terms of technology transforming the character of everyday life in
the United States, nothing has received as much scholarly and popular attention in the past two
decades as has the spread of Internet into American homes (see DiMaggio, Hargittai, Neuman,
Robinson 2001; DiMaggio, Hargittai, Celeste, and Shafer 2004; Boulianne 2005 for overviews).
The Internet went quickly from a frontier inhabited by a relatively few professionals and
hobbyists to a widely used sphere for a broad and expanding array of activities. Even if some of
the headier claims about the changes the Internet would bring have not come to pass (as in, e.g.,
Negroponte 1995; Turkle 1995), it remains undeniable that, for many Americans, using the
Cognition, Personality, and Internet Adoption - 2
Internet at home has become an important part of their lives. Yet there are also many Americans
who never use the Internet and feel that they have no use for it. Why some people have become
Internet users while others have not has been a topic of much speculation and research (e.g., de
Haan 2004; DiMaggio et al. 2004; Boulianne 2005).
We are interested particularly in the case in which the historical phenomenon of the rapid
diffusion of Internet into American homes intersected with the biographies of those in the most
comfortable and settled segment of the life course. For Americans, one’s fifties are a period
marked by comparatively high residential, occupational, income, and marital stability, reduced
childrearing responsibilities, and relatively good health. This is especially so for those who do
not suffer the social disadvantages of being either non-white or without a high school education.
Such individuals were typically quite settled into lives without the Internet as its rapid diffusion
into homes began, but they were not yet elderly, either. How does this population respond to the
widescale introduction of a new technology with potentially manifold personal implications? As
it turns out, these cohorts evince some of the most dramatic variation in the home adoption of
Internet technology, and, excepting those near the bottom of the socioeconomic distribution, this
variation in Internet adoption is only modestly related to available financial resources (Fox 2004;
Freese and Rivas 2005). Moreover, despite various predictions that Internet diffusion was on an
S-curve toward relatively swift near-saturation (Leigh and Atkinson 2001), evidence indicates
that intracohort rates of Internet adoption have largely stalled, suggesting that the mix of heavy
users and complete non-users may well mark the remainder of the active lives of these cohorts
(DiMaggio et al. 2004).
For reasons we describe, we believe response to the rapid rise of the Internet among late
midlife adults represents exactly the response to social change that might be better understood by
Cognition, Personality, and Internet Adoption - 3
sociologists than it has to date through a more detailed consideration of the potential interaction
of basic psychological characteristics and social position. In pursuing this idea, we restrict
attention here to one indicator of personal Internet adoption, whether one uses the Internet from
home, and to two types of basic psychological characteristics: measures of cognitive ability and
personality. We present conjectures about the relationship between individual psychological
characteristics and household adoption, and we consider how we might expect the strength of
these relationships to vary by one’s familial, socioeconomic, and occupational circumstances.
Then, we interrogate these conjectures using a large, longitudinal sample of adults in this cohort.
Afterward, we discuss our findings in terms of what they might suggest more broadly about the
interaction of personality and social structure in the unfolding of biography, as well as what they
might signify for questions about the causes and meaning of digital inequality in the United
States.
INTERNET ADOPTION
Between 1997 and 2001 alone, the percentage of American households with Internet
service increased from 18.6% to 50.3% (NTIA 2004). Yet while Leigh and Atkinson (2001:6)
projected 90% Internet penetration by 2003, the actual percentage of households with Internet by
that year had increased only to 54.3% (NTIA 2004). The possible leveling off of household
penetration has been observed in other studies (e.g., Lenhart et al. 2003), and the divergence
from S-curve-based predictions of rapid saturation might reflect Internet access being a
continuing expense rather than one-time purchase (Mueller and Schement 1996; DiMaggio et al.
2004). In any event, barring further changes in prices or modes of use, the best forecast appears
to be a protracted future in which a majority of people have and use the Internet at home but a
very substantial minority do not (Norris 2001; Martin 2003).
Cognition, Personality, and Internet Adoption - 4
Sociologists and others have long worried about the possibility of exactly such a stagnant
divide, especially if the result is users gaining considerable and varied benefits from the Internet
that non-users do not (NTIA 1999; Harris 2000; Norris 2001). For this reason, there has been
much interest in documenting the relationship between cleavages in Internet use and prominent
lines of existing inequality in the United States. Recent studies find little gender difference in
whether someone uses the Internet (NTIA 2002; Wasserman and Richmond-Abbott 2005; among
older adults, Fox 2004), and there have been conflicting findings about the extent to which racial
differences exist net of socioeconomic and other controls (see Rivas 2004). In contrast, existing
studies are said to have “made it clear that [years of education and income] are the main ways in
which Internet users differ from nonusers” (Robinson, Neustadtl, and Kestnbaum 2002: 285). In
2003, for American adults age 25 and above, 39% of those with only a high school diploma used
the Internet from home, as compared to 12% of without a high school diploma and 77% of those
with a college degree (authors’ CPS analyses). About 15% of those living in households with
annual incomes below $20,000 were Internet adopters, as opposed to almost 85% of those with
household incomes over $75,000, with this gradient being much steeper below the median than
above (authors’ CPS analyses). For reasons elaborated below, this study focuses only on the
matter of home Internet adoption, but many recent studies have also found considerable
sociodemographic differences in what Internet users do on the Internet (e.g., Howard, Rainie,
and Jones 2001; Loges and Jung 2001; Howard 2004).
While no one disputes the importance of documenting socioeconomic cleavages in
adoption or use, de Haan (2004) complains that too much work has had a “descriptive tendency”
focused mainly on documenting differences, perhaps taking their explanation as self-evident,
rather than working to understanding better why cleavages exist. In this respect, for example,
Cognition, Personality, and Internet Adoption - 5
education and income have been commonly discussed together as if they were two sides of the
same coin of “socioeconomic status” or “privileged households” (see especially the NTIA
reports, e.g., NTIA 2002). Separate education and income gaps are sometimes discussed as
though they were an inevitable feature of the adoption of new technologies involving
information, but analyses of the Current Population Survey indicate that there is no education
gradient for adoption of either cable/satellite television or cellular telephones when income is
held constant (Rogers 1995; Freese and Rivas 2005). de Haan (2004) elaborates what he calls a
“resource theory” for understanding variation in technology adoption that posits the influence of
different kinds of material, social, and psychological resources. For our purposes, the details of
this theory are less important than that it calls attention to major themes suggested elsewhere in
the literature: namely, the independent importance of money, motivation, and skill (e.g., Reddick
and Boucher 2002; Nurmela and Viherä 2004). The importance of motivation is made plain
from surveys showing that a large number of non-users cite a lack of interest rather than a lack of
funds as the primary reason for non-adoption (NTIA 2002).1 Skill has been emphasized because
differences in what individuals are capable of doing with the Internet (or expect that they will be
readily capable of learning to do) has been emphasized as a possibly important determinant of
their adoption (Hargittai 2002; van Dijk and Hacker 2003; Kubicek 2004; regarding age
differences in evinced skill, see also Hargittai 2004b).
One might consider Internet adoption decisions to be like many other prospective
expenditures of money and time in being influenced importantly by both how much individuals
have to spend and how much utility they expect to derive relative to other possible expenditures.
1 This pattern becomes stronger with age. In 2001, more persons under 35 cited expense rather
than non-interest as the reason for not having the Internet at home, while, for persons over 55,
69% cited non-interest and only 14% cited expense (NTIA 2002).
Cognition, Personality, and Internet Adoption - 6
Financial resources would seem especially important for lowering the threshold of how much
utility individuals must expect to derive from the Internet in order to become and remain
adopters. “Motivation” might be recast as expectations about the utility available to one from
Internet use, while “skills” might be considered to be expectations about one’s capacity to
actually extract available utility through use. In this view, even though education and income
are so often considered together in discussions of digital inequality, we might imagine any causal
role of income to be mainly as financial resources while any causal role of education net of
income to operate mainly though motivation, skill (and expectations about the development of
skill), or both. Importantly, emphasis on motivation and skill also raise the possibility that basic
psychological characteristics might also be a separate, and still largely undocumented, “main
way” that Internet users vary from nonusers, for reasons we develop further below.
One can easily find research about Internet adoption or use that has focused on children
(e.g., Livingstone 2003), adolescents (e.g., Gross 2004), college students (e.g., Landers and
Lounsbury in press), early midlife adults (e.g., Jackson et al. 2003), and the elderly (e.g., Fox
2004). Late midlife adults have received the least specific attention, which we regard as
especially unfortunate because we think they provide an especially interesting age group for
thinking about how psychological characteristics and social circumstances might interact in
Internet adoption decisions. The group is in a better position than any other to have financial
resources play a nondominant role in their adoption decision (NTIA 2002), and yet their
adoption and use varies greatly. In terms of motivation, the group was already well into midlife
when even personal computers became prominent, much less the Internet, but they are also
sufficiently young as to have a prospectively quite long time horizon in which Internet use will
presumably be a highly visible part of American culture. At the same time, very few had
Cognition, Personality, and Internet Adoption - 7
children living with them during the sharp rise of the Internet, so that while children may have
played an important role in some adoption decisions (i.e., as a source of communication and
informal technical support), their obtaining Internet access would rarely be household adoption
specifically for a child’s use rather than their own. In terms of skill, they did not “grow up with
the Internet” or have computers as part of their formal schooling, but they also have relatively
low rates of the physical and cognitive infirmities that may inhibit Internet-specific skill
acquisition in more elderly populations (NTIA 2000: 71). Apart from any such general
characterizations of late-midlife adults, their social circumstances of course vary tremendously.
Below we will develop predictions about how we expect this variation in circumstances to be
related to the extent to which basic psychological characteristics differentiate Internet users from
non-users, but first we briefly elaborate what we mean by basic psychological characteristics and
the two types of characteristics we consider.
BASIC PSYCHOLOGICAL CHARACTERISTICS
As discussed above, studies of digital inequality have focused mostly on standard
demographic cleavages and have given less consideration to psychology. As increasing interest
in “skill” and “motivation” suggest, however, attention to psychological concepts seems
indispensible for understanding variation in Internet adoption. Toward this end, we do not at all
question the usefulness of concepts such as Internet skills (Hargittai 2002), computer or Internet
anxiety (Marcoulides 1989; White and Scheb 2000), “informacy” (de Haan and Rijken 2002),
technophobia (Brosnan 1998), or a “venturesome” disposition toward innovations (Rogers 1995),
but we believe there is potential value for sociologists to think more in terms of basic
psychological characteristics as well. Not only do we think basic psychological characteristics
may contribute importantly to the development of domain-specific psychological constructs
Cognition, Personality, and Internet Adoption - 8
(e.g., “Internet anxiety” might substantially reflect a more general disposition toward anxiety
[Landers and Lounsbury in press]), but they may also influence adoption by influencing the
circumstances of individuals’ lives (e.g., their occupations). Basic psychological characteristics
also provide a better vocabulary for considering continuities and cross-influences in the
importance of psychology across different life domains of sociological interest. Additionally, as
opposed to the often haphazard work on the properties of domain-specific measures, basic
psychological characteristics have highly studied measurement properties and have already been
studied with regard to a plethora of other outcomes.
What we mean by “basic psychological characteristics” are those theoretically
conceptualized as domain-general in their potential consequences and empirically indicated to
have high rank-order stability over midlife, at least in populations under study. The domain-
generality of these characteristics makes them potentially causally relevant for behaviors and
outcomes across a variety of domains, and suggests that they may combine importantly with
experience in the development of more domain-specific (“higher-order, “superstructural”)
psychological constructs like skills or attitudes (see McCrae and Costa [2003: 184-205] for an
appealing dynamic model of such psychological adaptation with regard to personality). In
pointing also to high rank-order stability over midlife, we mean to highlight the empirical matter
that for some psychological characteristics, there is a high correlation between one’s intracohort
value on the construct at the start of midlife and one’s value at the end, regardless of how
variation in the characteristics arises and how much it changes from infancy through
adolescence. When a psychological characteristic has plausibly many consequences and also
high rank-order stability, this creates a useful asymmetry in which the influence of basic
psychological characteristics over midlife can be expected to be typically (and perhaps much)
Cognition, Personality, and Internet Adoption - 9
greater than the influence of particular events on the intracohort rank-ordering of the
psychological characteristics themselves.
The basic psychological characteristics considered in this paper are cognitive ability and
personality. Our experience suggests that the study of the consequences of basic psychological
characteristics is regarded with sufficient suspicion by enough sociologists as to require defense
before proceeding. Cognitive ability especially is one of the most controversial constructs in the
whole of behavioral science, as seen especially in the vigorous debates following publication of
Jensen (1969) and Herrnstein and Murray (1994) (see Fischer et al. 1996; Devlin, Fienberg,
Resnick, and Roeder 1997; Jacoby and Glauberman 1995; Gould 1996; Jencks and Phillips
1998). Central points of dispute in these debates have concerned why individual and group
differences in cognitive abilities exist, the extent to which cognitive ability is changed by
interventions, and the implications that particular views about the origins and malleability of
cognitive ability would have for various efforts at redressing economic inequality.2 Far less in
dispute is that cognitive measures vary within populations and that this variation is potentially
causally related to many life outcomes. Empirically, general measures of cognitive ability have
been found to be perhaps the most stable of all psychological constructs over midlife (Schaie
1996).3
2 Importantly contested matters also concern the conceptualization of cognitive ability and the
validity of existing measures, especially for intergroup comparisons, both of which we discuss
later in this paper.
3 Kohn, Schooler, and colleagues have published several well-known studies on the dynamically
reciprocal relationship between job characteristics and “intellectual flexibility” over the life
course (e.g., Kohn & Schooler, 1978, 1983; Schooler et al., 1999). The relationship between
their intellectual flexibility measure and more broadly deployed measures of cognitive ability
remains unsatifactorily clarified, and making it hard to evaluate how their findings complement
or contradict the broader literature on the rank-order stability of cognitive ability over midlife.
Cognition, Personality, and Internet Adoption - 10
Importantly, belief that cognitive measures capture some psychological reality that is
aptly characterized as an “ability” and that has subsequent life consequences does not commit
one to any position either about how the variation in cognitive ability that exists by midlife
originates or about how much cognitive ability can ultimately be changed by potential
interventions (regardless of its stability in studied populations) (Wahlsten 1997; Freese, Li, and
Wade 2003). Moreover, whatever the origins of cognitive and other psychological variation,
central to the argument of this paper is that the consequences of that variation reflect the
organization of societies and the positions of individuals within it. Social conditions create the
contexts in which psychological characteristics are more or less influential for outcomes, and
social processes also influence the development of new technologies and other innovations in
ways that can influence the extent to which their adoption is associated with psychological
characteristics. In short—and as we will emphasize again when we consider the policy
implications of our findings later—greater attention to the consequences of basic psychological
characteristics does not compel sociology onto some intellectual or ideological dark road down
which the discipline should resist venturing even at the expense of more complete understanding
of societal phenomena.
“Cognitive” refers to information processing and “ability” refers to a capacity or skill at
doing something, and so we use cognitive ability to refer to one’s demonstrable and general
internal capacity or skill at processing information.4 The seemingly intrinsic vagueness of
“processing information” suggests a wide range of diverse tasks might reflect cognitive
4 Some use the word “ability” to refer to innate ability—itself not a straightforward concept—
and thus take “cognitive ability” as providing assumptions about how differences in cognitive
performance measures arise. We emphatically do not intend “ability” in this sense. Some
literatures (e.g., Krosnick 1999) use “cognitive skills” with the same domain-general and
developmentally-neutral meaning as our use of “cognitive ability” here.
Cognition, Personality, and Internet Adoption - 11
performance. Considerable and continuing debate has been generated by the empirical
demonstration that even ostensibly quite diverse cognitive tasks are often substantially correlated
with one another, especially since this allows for the mathematical possibility that a single score
can capture much shared variation across very diverse cognitive tasks (Carroll 1993). Although
the data used in this study limits us to examining only such a summary score intended as a
general characterization of ability, this should not be interpreted as denying that cognitive ability
can be fruitfully divided into multiple measures of multiple abilities (Sternberg and Kaufman
1998). We would expect that multiple measures would likely be more illuminating than any
single measure—no matter how intendedly general—and this implies that our use of a single
score may systematically underestimate the ultimate causal effects of cognitive ability on
Internet adoption. Concern has long been raised about comparing summary cognitive ability
scores across cultural or ethnic groups; this may suggest the value of exploring effects of
measured cognitive ability in relatively ethnically and culturally homogeneous samples, which is
the approach of this study.
Like “cognitive ability,” the history of “personality” has been characterized by
considerable diversity of ideas about its proper conceptualization and measurement (Digman
1996). Personality is typically intended to encompass psychological characteristics that are not
abilities but are reflected in abstract styles of thinking and feeling. Over at least the past two
decades, however, personality psychology has gained increased vitality as a result of growing
consensus about the usefulness of the Five-Factor Model (FFM) as a characterization of the
structure of personality (Norman 1963; Borgatta 1964; John & Srivastava 1999; McCrae and
Costa 2003). The premise of the FFM is that five independent, superordinate categories can be
Cognition, Personality, and Internet Adoption - 12
used to reasonably collect the bulk of the concepts and measures that have been developed
toward describing personality.
These “Big Five” factors of personality are commonly referred to as Extraversion,
Agreeableness, Conscientiousness, Neuroticism (also sometimes referred to by its opposite,
emotional stability), and Openness to Experience. How these dimensions are instantiated in the
data used in this paper is presented in Table 1. As with cognitive ability, each of the factors of
the FFM can be usefully divided into subscales, but subscale measures are not available in the
data we use here.
TABLE 1 ABOUT HERE
Empirically, the premise that the FFM captures the main independent dimensions of
personality seems to work reasonably well for adults in the United States, even if there is
uncertainty about its replicability across cultures (De Raad, Perugini, Hrebickova, and Szarota
1998) or to children (John, Caspi, Robins, Moffitt, and Stouthamer-Loeber 1994; van Lieshout
and Haselager 1994). Research also indicates that subscales from other common personality
inventories do not measure major independent dimensions apart from those measured by the Big
Five (McCrae and Costa 2003: 52-57). Studies have shown associations between personality and
a wide array of psychological and social outcomes (see McCrae and Costa 2003: 216-233;
Roberts et al. 2003: 580 for reviews and references). There has been considerable debate about
the midlife stability of personality in various senses (see Costa and McCrae 1994; Roberts et al.
2003; Srivastava et al. 2003); important for our purposes is the intracohort rank-order stability of
personality. For this, it should be noted that even a review that gives much attention to studies of
effects of life course transitions on five-factor traits acknowledges that the “only psychological
constructs more consistent [over midlife] than personality traits are measures of cognitive
Cognition, Personality, and Internet Adoption - 13
ability” (Roberts et al. 2003). In other words, while studies have found some modest rank-order
personality change in response to specific life circumstances, the main story of personality as
observed over midlife for adults in the United States appears to be one of considerable stability.
DEVELOPING PREDICTIONS
The task of this paper is to consider when and how basic psychological characteristics
might be relevant for which late midlife adults have adopted the Internet. We agree with the
complaints of other Internet researchers that studies have focused too much on this binary
outcome of whether or not someone uses the Internet from home, as opposed to broader
measures differentiating use (see also Hargittai 2003, 2004a). However, we focus on it here
precisely because we think attention to basic psychological characteristics can yield more to be
said even about this well-studied and simple binary indicator of adoption. We also regard it as
best single indicator of whether Internet technology has become part of the respondents’ personal
lives.
In developing predictions, we avail ourselves of a broad range of findings and thought
both about psychological characteristics generally and about Internet adoption specifically. We
do not enumerate these expectations as hypotheses to avoid suggesting that they are derived from
some single theory or dueling set of competing theories. In this respect, we are less committed
to whether favored predictions are borne out than whether the cumulative result is that
combining basic psychological characteristics and indicators of social position results in a more
satisfactory understanding of Internet adoption for this population than what is available from
earlier studies. We describe first predictions regarding the main effects of basic psychological
characteristics on Internet adoption, and then how we might expect the effect of basic
Cognition, Personality, and Internet Adoption - 14
psychological characteristics to vary depending on respondents’ familial, socioeconomic, and
occupational circumstances.
Main effects of psychological characteristics on adoption. As noted, cognitive ability
has been shown to have various educational, economic, and other social consequences, and the
general expected direction of these associations and adoption would lead one to expect cognitive
ability to be positively related to Internet adoption. Possibly, whatever effect cognitive ability
has on Internet use could be entirely explained by its consequences for differential sorting into
life circumstances that are then predictive of adoption. In that case, we would expect estimates
of the effect of cognitive ability on Internet to be substantially or even entirely diminished when
social circumstances are included as mediating variables in the model.
Discussions and research on the importance of “cognitive resources” for adoption,
however, would lead us also to expect substantial cognitive selectivity in motivation to use the
Internet (de Haan 2004). As an informational medium, Internet use typically involves much
reading; as a communications medium, the Internet involves “writing” as well (de Haan 2004).
There is ample reason to expect cognitive ability to be related to tasks that involve intensive
literacy, even within populations (as in the WLS) for which practically all members can be
presumed to have at least basic literacy (Gottfredson 2002: 359-367). Second, following
Hargittai (2002) and others’ emphasis on skills, it is important to keep in mind that for this
population, Internet skills would have needed to be acquired in adulthood. Considerable
research indicates that cognitive ability is related to more rapid and complete skill development
in various domains, but especially those involving (as the Internet does) the manipulation and
processing of information (Gettinger 1984; Schmidt and Hunter 1998). In other words, cognitive
ability may be related to both expectations and reality about how easily the technology can be
Cognition, Personality, and Internet Adoption - 15
effectively used. For these reasons, we might instead expect cognitive ability to remain
significantly associated with adoption even when socioeconomic attainments are controlled, a
possibility consistent with findings of DiMaggio and Hargittai (2002) using the rough indicator
of ability provided by a 10-item vocabulary task on the General Social Survey.
Regarding personality traits, we think existing thought offers the most unequivocal
predictions for the traits of Openness and Neuroticism. As Openness can be defined as “a
receptiveness to new ideas, approaches, and experiences” (McCrae and Costa 2003:46), it seems
resonant with notions both that a dispositional preference for novelty characterizes early adopters
of new technologies and that a dispreference for novelty characterizes “laggards” (Rogers 1995).
By contrast, given that a main facet of Neuroticism is anxiety, it would seem to be the basic
psychological characteristic most implicated in “Internet anxiety” or “technophobia”, which have
been often discussed in explaining low motivation to Internet adoption, especially among older
adults (Laguna and Babcock 1998). Accordingly, we would expect Openness to be positively
and Neuroticism to be negatively associated with Internet adoption.
With respect to Agreeableness and Conscientiousness, we think implications for Internet
adoption follow less obviously. However, Landers and Lounsbury (in press) found that both
Agreeableness and Conscientiousness were inversely associated with use in a small and select
sample of college students. Landers and Lounsbury speculate post-hoc that lower Agreeableness
might be related to higher levels of Internet use because less agreeable persons might find the
Internet an attractive refuge from face-to-face interaction. They also conjecture that, since
Conscientiousness persons tend to be more time-conscious, they may be more averse to the
Internet if perceived as a source of distraction.
Cognition, Personality, and Internet Adoption - 16
Of the Big Five traits, Extraversion most readily lends itself to plausible contrary
predictions. As a communications medium, we might imagine the Internet to be attractive to
extraverted individuals as a way of providing them with another outlet (Kraut et al. 2002;
Robinson et al. 2000; Robinson and Kestnbaum 1999). At the same time, however, discourse
about the Internet has been long infused with the idea that the technology (and especially the
Web) might offer a sanctuary for the introverted, while extroverted people might still be
expected to maximize non-“virtual” interaction and to be less inclined toward the solitary
possibilities of websurfing. Nie and Ebring (2000) find Internet use to be associated with less
social interaction in cross-sectional data (but see Gershuny 2003), and Landers and Lounsbury
(in press) found Extraversion to be negatively related with Internet use among college students.
Psychological characteristics and familial circumstances. The majority of late midlife
adults are currently married and have lived with their spouses for a long time. Although whether
one uses the Internet at home is an individual outcome, obtaining and maintaining access in the
first place is a decision that is presumably made in different ways by different couples. Adoption
may be prompted by the expected benefits of both spouses, or expected benefits to one may be
enough. In the latter case, individuals who would not have the Internet but for their spouses’
interest might still use it once it is obtained, although data make clear that there are many adults
who have the Internet in their homes but do not use it (NTIA 2002).
At least among adults of the cohort we examine, we might expect a tendency toward
asymmetric power in spousal relationships such that husbands have more influence over
financial decisions than do wives (e.g., Denton 2004: 1164; Schaninger and Buss 1986: 132). If
so, the motivation of husbands might be more consequential for adoption than the motivation of
wives. Thus, if cognitive and personality characteristics influence motivation, then we might
Cognition, Personality, and Internet Adoption - 17
expect the psychological characteristics of married men to be more consequential than the
personality characteristics of married women in determining whether a household obtains
Internet access.
Social support has been commonly emphasized as important to Internet adoption
(DiMaggio et al. 2004). For adults in late midlife and older, children have been thought to play
an important role as sources of motivation—via direct encouragement and the prospect of being
able to communication with them online—as well as a source of assistance in developing skills
(Fox et al. 2001). We might expect the support of children to lessen the influence of
psychological factors in Internet adoption, especially any effect of Neuroticism and perhaps
more for mothers than fathers.
Psychological characteristics and socioeconomic circumstances. Educational
attainment, meanwhile, is strongly related to earlier measures of cognitive ability, including in
older cohorts for whom standardized tests did not play as strong a gatekeeping role (Sewell and
Hauser 1975; Snow and Yalow 1982). Toward this end, it seems possible that cognitive ability
measured prior to sample differentiation in educational attainment (i.e., adolescence) might
explain some of the “education effect” estimated by models in which cognitive measures are
excluded. This might be especially so for older cohorts that did not have computers (much less
the Internet) as part of their formal schooling.5
Even if this is the case, we might expect to observe also that any effect of cognitive
ability will be largest among those with the least education. For if at least part of the effect of
cognitive ability on later outcomes is the result of sorting into later-life outcomes, we can
5 A curious feature of the education gradient in Internet adoption in the Current Population
Survey is that, adjusting for cohort differences in the marginal distribution of education, the
effects of education on adoption do not diminish for older cohorts despite the lack of computers
in formal schooling (CPS analyses by authors).
Cognition, Personality, and Internet Adoption - 18
imagine the possibility of education plausibly providing a “substitution” or “compensatory”
effect on unmeasured components of social position that, in turn, attenuates the effect of
cognitive ability. For that matter, the same prediction would be implied to whatever extent that
continued education makes a lasting contribution to the realized abilities that are relevant for
adoption.
Quite apart from education, we might expect psychological characteristics to matter most
for those with the least wealth. For one, financial resources could be expected to diminish at
least the relevance of cognitive ability by increasing the capacity to afford compensatory
training, support, or software in the acquisition of skill. Additionally, one might conjecture that
motivation and skill differences—and any of their psychological correlates—will matter most for
those whose financial situation puts them closer to the margin of affordability. This last
argument should be regarded cautiously, however, for if the role of some psychological
characteristics is more in creating an active aversion for adoption, then we might imagine effects
will show up as largest precisely among those for whom the expense of the Internet would be
least at issue.
Psychological characteristics and occupational circumstances. The percentage of
American adults who used e-mail/Internet at work increased from 9.3% in 1993 to 42.2% in
2003 (authors’ CPS analyses). At younger ages, we would expect considerable active sorting of
individuals into jobs with varying degrees of Internet use. However, late midlife, until
retirement, tends to be a period of comparatively high occupational stability. Consequently,
although some Internet-related occupational sorting presumably has occurred in late-midlife
cohorts, it is presumably much more the case that the rise of the Internet in the workplace
intersected an established career trajectory. While there has been much concern that introducing
Cognition, Personality, and Internet Adoption - 19
computers imposed new skill demands on older workers that they could not meet, these fears are
now thought to have been overstated, in part because so many older workers were indeed able to
adapt to the introduction of new technologies into their jobs (Friedberg 2003). Importantly for
our purposes, not only may workers have been able to adapt to the Internet at work, but this
might also have increased their desire and facility for having the Internet at home (de Haan
2004).
Should we expect that the same psychological characteristics that predict Internet
adoption at home to also affect Internet use at work? We would under at least two quite distinct
conditions. First, to whatever extent individuals have influence over whether their jobs involve
using the Internet or not (whether by control over job conditions or by job switching), we would
expect that many of the same arguments for why cognitive and personality characteristics might
motivate home adoption would also apply to discretionary adoption in one’s workplace. Second,
psychological characteristics might actually be strongly predictive of career trajectories that
subsequently have been highly penetrated by Internet technology and, in turn, using the Internet
at work might be such a strong predictor of subsequent home adoption that this is actually the
primary reason for any relationship between psychological characteristics and home Internet
adoption in this cohort. In this second scenario (but not the first), the effect of psychological
characteristics on home adoption would mostly diminish when the use of the Internet at work is
taken into account.
Should we expect the effect of basic psychological characteristics to differ between those
who do and do not use the Internet at work? As noted, one possibility is that effects of
psychological characteristics in the full sample largely disappear for both groups when those
who do and do not use the Internet at work are treated separately, and this would suggest that the
Cognition, Personality, and Internet Adoption - 20
story of psychological characteristics affecting Internet adoption is really a story about sorting
into Internet-using jobs and the dominant effect of these jobs on home adoption. A reason to
think use at work could have such a dominant effect is that it would seem an obvious route to
gaining the familiarity with the Internet that would make its home use more appealing (not to
mention useful, perhaps for work but other purposes as well). If that is the case, however, a
separate possibility is that basic psychological characteristics would predict Internet use much
more strongly for those who do not use it at work than those who do. Use at work, in other
words, may importantly facilitate home adoption (especially by surmounting the skills hurdle),
and it may be that psychological characteristics especially matter when one does not have this
experience.
DATA AND MEASURES
Data We use data from the Wisconsin Longitudinal Study (WLS), whose participants
comprise a 1/3 sample of all Spring 1957 graduates from high schools in Wisconsin (original N
= 10317). The WLS has gathered information through surveys in 1957 (in school), 1964 (mail
survey of parents), 1975 (telephone), 1993 (telephone and mail), and 2003-4 (telephone and
mail, as well as telephone survey of spouses). Sample retention has remained high throughout:
76% of sample members who were alive in 2004 were successfully interviewed by telephone in
1993 and 2004 (N=6857).6 The WLS is the only data resource that combines a large population-
based sample, detailed and longitudinal measurement of cognition, personality, and financial and
sociodemographic covariates, and extensive measurement of Internet adoption and use.
As such, the WLS would seem almost ideally suited for the purposes of this study, so
long as key limitations of the sample are kept in mind. First, as a cohort sample, WLS
6 About 67% of those interviewed in 2004 still lived in Wisconsin, which has had Internet
penetration rates close to the median for all states (NTIA 2000).
Cognition, Personality, and Internet Adoption - 21
respondents are all approximately the same age, and so findings from the study cannot be
generalized to other ages and cohorts. For the theoretical purpose of looking at response to
change among individuals who are well entrenched in midlife when the change occurs, WLS
respondents would seem almost exactly the right age, and a cohort sample allows for within-
cohort exploration of the interaction of social structural and psychological variables with
statistical power that would require many times more cases in an age-diverse sample.
Second, all WLS sample members are high school graduates, so findings cannot be
generalized to those who did not complete high school (roughly 25-30% of adolescents in
Wisconsin in 1957 [Sewell and Hauser 1975]). Because cognitive test scores and completing
high school are substantially related in the WLS cohort, the expectation would be that any
observed association between test scores and Internet adoption will be less than the association if
non-truncated data were available. Inclusion only of high school graduates presumably means
that there are fewer respondents for whom finances directly preclude adoption, making the
sample perhaps better for considering why so many for whom Internet adoption is plausibly
affordable still do not use it.
Finally, given the composition of Wisconsin and patterns of high school completion in
1957, the WLS sample is almost entirely white. As such, our study will not contribute to the
debate over how much racial and ethnic disparities in Internet use can be resolved by other
variables, especially as the relevant interaction terms and statistical power for addressing the
issue might not even be available if the WLS did reflect the nationwide ethnic distribution of
1957 high school graduates (see Rivas 2004). As mentioned earlier, the relative homogeneity of
the WLS may actually be desirable in first explorations of effects of cognitive ability given
concerns about comparisons of ability measures across cultural/ethnic groups.
Cognition, Personality, and Internet Adoption - 22
Internet Adoption The primary outcome considered in this paper is whether the
respondent has the Internet in their home and uses it. We construct this binary variable from a
nested series of telephone survey items asking whether respondents (1) have a computer in their
home [74% of respondents], (2) have Internet access from their home computer [67% of
respondents, and 91% of those with a computer at home], (3) themselves access the Internet from
their home computer [57% of respondents, and 84% of those with household Internet access].7
The WLS does not specifically ask about WebTV and related devices, as these were expected to
be a very small percentage (< 1%) of sample members, but respondents were coded as having
Internet access from home when the use of these devices was volunteered.8
Cognitive Ability All Wisconsin high school students in the WLS cohort were
administered the Henmon-Nelson test of Mental Ability (hereafter H-N) at least once during high
school. Scores for WLS respondents were obtained from the Wisconsin State Testing Archive.
The measure we use here is based on respondents’ junior year score if available and freshman
year score otherwise. Scores were converted to standardized (z) scores based on the
corresponding percentile rank for all Wisconsin high school students. As already noted, the H-N
was intended as a general measure of ability and includes no subtests, which implies that it may
underestimate the total effective of cognitive abilities as measurable in adolescence on later-life
Internet adoption.9 We emphasize again that our analyses make no assumptions about the
developmental origins of measured cognitive ability.10
7 Supplementary analyses considering adoption in terms of transitions through these constituent
items do not yield results that add substantively to the conclusions presented here.
8 The WLS contains additional information about respondent Internet adoption, but interrogation
of these other measures is postponed to a subsequent study.
9 For a small subset of respondents (N=108), scores were obtained from archives on subtests of
Iowa Tests of Basic Skills. Analyses indicate that the quantitative skills, vocabulary, and verbal
Cognition, Personality, and Internet Adoption - 23
Personality In the WLS, personality was measured using a 39-item scale administered
across the 1993 telephone (10 items) and mail (29) surveys. Items are a subset of the BFI-54
(John & Srivastava 1999). Separate scales were intended to measure the five primary factors
prevailing in personality psychology: Extraversion (8 items), Agreeableness (8),
Conscientiousness (8), Neuroticism (7), and Openness to Experience (8) (see Table 1). Principal
factor analyses yield factors consistent with the intended scale structure. Multiple imputations
from other personality items, H-N test scores, and gender were used to account for nonresponse
to the mail survey as well as to individual items on either survey.11 All scales were standardized.
Personality measures derived from self-reports show substantial agreement with measures
derived from ratings by spouses and friends (see McCrae and Costa 2003: 37-51 for review).
That personality was not measured until 1993, when respondents were ~54 years old,
poses an obvious problem for treating personality as exogenous to educational, occupational, or
other attainments, as measured effects of personality may actually reflect the indirect effects of
the attainments through effects of the attainments on personality. Available research, however,
provides good reason to believe that personality selection into these attainments is high relative
to any long-term effect of attainments on variation in these traits (see Srivastava et al. 2003,
Roberts et al. 2003). For example, given the results that follow, the most obvious concern might
be the effect of educational attainment on Openness. A longitudinal study covering the college
expression subtests predict H-N scores to a roughly equal degree and together predict
performance as well or better than does an earlier administration of the H-N (Freese 2005).
10 Supplementary analyses examined models in which adolescent socioeconomic status measures
(parental education and income) were included as predictors of adoption, but these were not
significant when the cognitive ability and subsequent attainment measures were included. Such
a result has no implications for the developmental origins of ability but does imply that
adolescent status does not influence Internet adoption beyond its influence on cognitive ability
and later attainments.
11 Multiple imputations for this paper were done using the Amelia software package (see King,
Honaker, Joseph, and Scheve 2001).
Cognition, Personality, and Internet Adoption - 24
years, however, finds a corrected rank-order stability coefficient over the four years that is very
high (.90) and actually higher than the other Big Five traits (Robins et al. 2001).
Potential mediators and controls Educational attainment is based on 1993 self-report.
In the presented analyses, education is specified as a pair of dummy variables indicating some
college (13-15 years of education) and completed college (16+ years). Alternative specifications
fit no better and yield substantively similar results. Spouse educational attainment is taken from
the 2004 spouse survey when available (78% of cases in estimation sample) and from the WLS
respondent report otherwise.12
Income and wealth are based on individual reports and measured in logged dollars for the
models presented. In cases of refused or don’t know responses to individual reports, multiple
imputations were made using available financial and sociodemographic information.
Rural household is determined using the 1993 mailing address. Those households
designated as rural were located outside a census metropolitan statistical area.
Occupational education and occupational income for both the respondent and spouse are
based on the 1993 reported current/last occupation and on 1990 Census occupation information.
The occupational education score is derived from the percentage of Census respondents in the
occupational category who completed one year of college or more, while the occupational
earnings score is derived from the percentage of those in the category who earned at least $14.30
per hour in 1989. Both variables are standardized, with a constant imputed and dummy indicator
variable used for missing cases.
12 The latter is problematic because the WLS asks for spouse educational attainment at the time
of the marriage. While it might seem advisable to therefore provide some adjusted imputation
based on at least gender and spouse’s age at marriage, inspection of data suggest that this is only
one source of discrepant reporting between respondents and spouses about the spouse’s
education and that the consequence of such an adjustment would be minor and not necessarily
superior.
Cognition, Personality, and Internet Adoption - 25
Probability of a child online. The WLS does not measure whether children are online or
whether children directly provide support for respondents being online. We use information
about the number of respondents children and their education, along with estimates from 2003
CPS data, to create a proxy measure of the probability of a respondent having at least one child
who is online. Specifically, based on the October 2003 CPS, we used probabilities of .3 for a
child with a high school education or less, .6 with some college, or .8 with a college degree or
higher, and for the purposes of creating the proxy measure we assumed that childrens’
probabilities of being Internet users were independent. Attempting to refine probabilities by
adjusting further for the age of children did not improve ultimate results.
Job with Internet. Respondents were asked whether they used the Internet on their
current/last job; for respondents who had retired from a job that could be regarded as their
“career” job since 1993 and taken other employment, the question asked about the earlier job.
We do not know whether Internet use at work preceded Internet use at home for individuals, and
thus we cannot make decisive statements about work-to-home diffusion. Problems with trying to
measure reciprocal influence in cross-sectional data are well-known and are even less tractable
when one is talking about binary outcomes. Side information about the occupational stability
(excepting retirements) of this cohort, as well as the general rate of diffusion into workplaces
versus the home, would lead us to strongly expect that the extent to which respondents have the
Internet at work because of their experience using the Internet first at home is quite low
compared to the reverse. (That the item asks about one’s earlier job in the case of recent post-
retirement employment changes also likely reduces the magnitude of any endogeneity bias in this
regard.) Even so, while consider having a job with the Internet as a potential mediator of home
adoption, we emphasize caution in interpreting results.
Cognition, Personality, and Internet Adoption - 26
RESULTS
Main effects. Table 2 presents results from our key main effects models of Internet use
at home. Model 1 includes our cognitive and personality measures and only a control for gender.
Model 2 includes only sociodemographic characteristics; as such, it presents the estimates that
one would obtain from these data if, as in most previous studies of Internet adoption, basic
psychological characteristics were not considered at all. Model 3 includes both basic
psychological and sociodemographic characteristics, and Model 4 adds contemporaneous
measures of work status and whether the respondent uses the Internet in his or her current/last
job.
TABLE 2 ABOUT HERE
Looking first to Model 1, we observe a large effect of cognitive ability on Internet
adoption. We also observe significant effects in the expected direction for the two personality
traits we most expected to matter: Openness and Neuroticism, with the effect of Openness being
especially strong. Less expectedly, we also observed significant effects for Conscientiousness
and Agreeableness consistent with those found by Landers and Lounsbury (in press) among
college students. The trait for which would could most readily derive competing expectations,
Extraversion, was the only one not significantly related to adoption. In terms of predicted
probabilities, if we hold all variables to their means, then a standard deviation change in
cognitive ability centered on its mean is associated with a .124 increase in the predicted
probability of adoption (from .520 to .644); for the significant personality traits, the
correspondent effect of Openness = .092, Neuroticism = -.024, Agreeableness = -.016, and
Conscientiousness = -.014.
Cognition, Personality, and Internet Adoption - 27
The results of Model 2 largely affirm the extant literature about what sociodemographic
variables predict Internet adoption, although a few results merit highlighting. First, net worth
has a significant effect on adoption net of income. If true beyond this sample, this would
indicate that caution is well warranted in studies that wish to consider the effect of circumstances
or characteristics on adoption “controlling for socioeconomic status” but have only income and
education measures at their disposal.13 Second, occupational education has a significant
independent association with adoption but occupational income does not. As with wealth, this
speaks to the importance of a broader set of socioeconomic status measures in studies of
adoption for which SES is either the focus or a control. Third, our measure of the probability of
having a child online, despite its seemingly obvious shortcomings, has a strong significant
association, and, in ancillary analyses, fits better than number of children or number of college-
educated children (not shown). Even so, this can be taken as only weak but suggestive evidence
of child-to-parent diffusion without better measures, and the measure could be a proxy for other
aspects of social advantage not resolved by other measures.
Model 3 results indicate that all of the cognitive and psychological characteristics that
were predictors of adoption in Model 1 remain significant with the addition of the
sociodemographic controls. The effect of cognitive ability, however, is attenuated by 45% and
the effect of Openness is attenuated by 30%. By contrast, the effect size of the other significant
personality characteristics—although not large to begin with—are not diminished by the
inclusion of these other variables. The resulting estimates of the unmediated effects of cognitive
ability and Openness are much closer to being equal. Models adding sociodemographic
13 Given findings on where the difference between income and the concentration of wealth is
most pronounced (Conley 1999; Oliver and Shapiro 1995), the difference between the effects of
income and wealth might be an especially important distinction to pursue in samples less affluent
and more racially diverse than the WLS.
Cognition, Personality, and Internet Adoption - 28
characteristics sequentially reveal that 61% of the attenuation in the cognitive ability coefficient
can be accounted for by adding education alone, and 93% by education, income, net worth, and
occupational education and income. For the attenuation in Openness, these percentages are 75%
and 100%, respectively.
Excepting marital status, all regressors in Models 1-3 are measures in 1993 or before and
so presumably predate Internet adoption for all but a handful of the cohort. Model 4, on the
other hand, includes current work status and whether respondents used the Internet in their
current/last job. Because potential bias due to endogeneity is more of a concern for these
variables, we exclude them from the earlier specifications. However, one can see by comparing
the Model 3 and Model 4 results that their inclusion does not make much difference for the
estimation of the effect of psychological characteristics, other than reducing Agreeableness to
marginal significance.14
Taken together, these results lend support to the conclusion that an important part of why
cognitive ability is strongly related to Internet adoption is that it determines later-life attainments
and concomitant sorting that influences adoption. Even so, however, a substantial effect of
cognitive ability remains even after controlling for a more extensive range of sociodemographic
characteristics than most studies of Internet adoption. This is consistent with the conjecture that
the skill and literacy demands of the Internet provide an independent means of cognitive
selectivity. A smaller portion of Openness to Experience is attenuated by attainments, and the
effects of other personality characteristics are not attenuated at all. The ultimate result is strong
14 The reduction in sample size in Model 4 is due to the job characteristics items being asked of
only a 75% random subsample of WLS respondents. We do not limit the estimation samples for
Models 1-3 to this random subsample only because the Model 3 specification serves as the
baseline for later analyses, but supplementary analyses that do restrict the sample confirm that it
makes no practical difference for the magnitude of coefficients.
Cognition, Personality, and Internet Adoption - 29
evidence that heterogeneity in basic personality characteristics is important to understanding
variation in Internet adoption for this population of late midlife adults. Comparing Models 2 and
3, we see that the magnitude of the estimated effects of most sociodemographic characteristics is
not much changed with the exception of education and income, with the change in the education
coefficient being relatively larger even though the income coefficient is reduced to
nonsignficance. While we will consider the attenuation of the estimated education effect again
shortly, we note here that considering psychological characteristics indeed does reduce the
apparent influence of what have been regarded elsewhere as the two main cleavages in Internet
adoption.
Familial circumstances. We proposed that psychological characteristics might be more
strongly related to Internet adoption for men than women, if men in this cohort have relatively
more power over household financial decisions. Table 3 shows estimates of the effect of
cognitive and personality measures on Internet adoption for married men and married women.
Model 1 includes only the basic psychological characteristics, while Model 2 adds all the
sociodemographic variables (except sex) as in the full model of Table 3. In Model 1, we can see
the effect of H-N score on the log odds of adoption is about 26% larger for men than for women
(p = .06) and the effect of Openness is about 21% larger (p = .30).
TABLE 3 ABOUT HERE
While this result is consistent with what we would expect if husbands had more power
than wives in familial purchasing decisions (even if not to conventional standards of statistical
significance), we can see when we look to Model 2 that the cognitive ability effect difference
between the sexes is entirely accounted for by other sociodemographic variables. Further
analysis shows specifically that the differences are accounted for by the inclusion of the
Cognition, Personality, and Internet Adoption - 30
occupational measures, suggesting the importance of our closer consideration of occupation
below. There are also no significant differences in the estimated effects of basic psychological
characteristics between married and unmarried members of either sex (not shown). Taken
together, this would suggest that while sex differences in sorting into careers may be important
for understanding the role of psychological characteristics in Internet adoption, as we will
explore further below, we do not have reason to think that systematic differences in relative
household decision-making power between husbands and wives are important.
We also speculated that the effect of psychological characteristics on adoption
(especially, perhaps, that of Neuroticism) may be smaller for those with a higher probability of a
child online. Despite a relatively strong main effect of the probability of a child online and
adoption, our analyses found no evidence of a significant interaction between this probability and
any of the basic psychological characteristics (not shown). This may well indicate that the
relevance of psychological characteristics is not mediated by processes of child-to-parent
diffusion, but, as already noted, the indirectness of the child online measure would seem to
preclude the possibility of any decisive conclusion.
Socioeconomic circumstances. In the main effects models presented in Table 3, the
estimated effect of education was substantially attenuated by the inclusion of cognitive and
personality characteristics in the full model. This seemed the clearest example in which the
failure to consider basic psychological characteristics would actually distort our understanding of
the importance of the role of some particular sociodemographic characteristic in Internet
adoption. Because cognitive ability was measured prior to differentiation of educational
attainment in this sample, the reduction of the education coefficient can be regarded
straightforwardly as a correction for spuriousness, presumably of cognitive selectivity into
Cognition, Personality, and Internet Adoption - 31
education. Personality was not measured until the respondents’ mid-fifties, but the midlife
stability in personality found in other studies would seem to warrant strong suspicion that
attenuation due to the inclusion of personality measures may substantially indicate spuriousness
as well.
Table 4 examines the extent to which measured sources of spuriousness, potential
spuriousness, and mediation resolve the estimated education effect on adoption in this sample.
While 47% of women with only high school diplomas in this sample are Internet adopters, 74%
of the female college graduates are, and the difference is even larger among men (41% vs. 81%).
Over half this effect for women, and a quarter of the effect for men, is resolved by inclusion of
basic psychological characteristics. Virtually all of the remainder of the estimated effect, but
only 62% of the effect for men, is resolved by the mediating variables of income, wealth,
occupational characteristics, and spousal characteristics. Indeed, for women, there is no
significant effect of education on Internet adoption once these different variables are controlled.
Studies that have emphasized the importance of differences in education in models of Internet
adoption, even in samples that include older adults whose education predates the personal
computer, typically have only controlled for income among these various covariates. Our results
suggest that, especially for women, education cleavages in this cohort might be mostly or
entirely attributable to a combination of the spurious influence of cognitive and personality
characteristics and the consequences that education has for one’s finances, occupation, and
spouse characteristics.
TABLE 4 ABOUT HERE
We proposed that the effect of cognitive ability may be largest for those with the least
education, for a variety of reasons. We show the difference here in Table 5 by presenting results
Cognition, Personality, and Internet Adoption - 32
separately for respondents with different levels of education. Model 1 includes only basic
psychological characteristics and gender, while Model 2 includes all the same sociodemographic
characteristics (except education) as Model 3 of Table 2. The results show that the strongest
effects of cognitive ability are observed among those with only a high school education, while
the weakest effects are observed for those with a college degree. Against the possibility that this
difference reflects just a mismatch between the log odds metric of logistic regression and the
actual responsiveness of Internet adoption at different baseline probabilities, one can see that the
effect of Openness is not attenuated as education increases, and Neuroticism, Agreeableness, and
Conscientiousness have, if anything, larger coefficients as education increases. In terms of the
change in predicted probabilities, if we hold all other variables to their means, a standard
deviation change in cognitive ability centered on the mean is associated with an increase from
.399 to .467 for high school graduates, but only from .821 to .836 for those with a college degree.
TABLE 5 ABOUT HERE
We also conjectured that the effect of psychological characteristics on adoption might be
larger for individuals with less financial resources. Table 6 presents the effect of basic
psychological characteristics on Internet adoption as analyzed separately for respondents with
household net worth below and above the sample median (substantively similar results are
obtained if income is used instead of net worth). Model 1 includes the basic psychological
characteristics, gender, education, job status (working or not), and whether one’s job uses the
Internet, while Model 2 adds all sociodemographic characteristics included as Model 3 in Table
2.15 We can see that the estimated effect of cognitive ability is less for those with household
15 We include education and job variables in the first model here because of the significant
interactions we observe for these variables in the analyses of interactions of education earlier and
occupation later.
Cognition, Personality, and Internet Adoption - 33
wealth above the median than for those with wealth below the median. We can also see,
however, that this expected pattern is not observed for any of the personality measures. Indeed,
the results indicate that the entire effect of Conscientiousness on adoption observed in the full
sample is actually confined only to those with wealth above the median. The results would thus
seem inconsistent with the conjecture that motivation, broadly speaking, figures more
importantly in Internet adoption decisions for less wealthy individuals than more wealthy
individuals (or that variation in motivation is not importantly associated with variation in
personality measures). We cannot determine whether the interaction of wealth and cognitive
ability reflects aspects of motivation related to cognition but not personality or, as proposed, the
possibility that financial resources recompense for skill.
TABLE 6 ABOUT HERE
Occupational circumstances. We posited that the same psychological characteristics
that predict using the Internet at home might also predict using the Internet at work. Table 7
presents estimates of logistic regression models for whether the respondent uses the Internet at
work. Model 1 includes only the basic psychological characteristics, while Model 2 adds
whether respondents are married, their education, and whether they live in a rural area.16 We
present results for Model 2 for both all respondents and for only those who are currently
working.
TABLE 7 ABOUT HERE
The results indicate that Internet use at work is positively associated with both cognitive
ability and Openness. The differences are substantially reduced but not eliminated by the
16 As this is an outcome about the characteristic of the respondents’ employment, we include
only respondent characteristics and not spouse or household characteristics, or other
characteristics of their jobs. In supplemental analyses, including other job characteristics or all
sociodemographic characteristics does not change the conclusions of our results.
Cognition, Personality, and Internet Adoption - 34
addition of controls. The results indicate the importance of earlier sorting on cognitive ability
and personality into different careers but also the presence of additional psychological selectivity
into using computers with Internet at work.
The effects for both cognitive ability and Openness are stronger for women than for men,
although these differences are much reduced in Model 2, mainly as the result of the inclusion of
the additional controls for education. Unexpectedly, we also find that Conscientiousness is
positively associated with Internet use at work among women, even though it is negatively
associated with home use and is not associated with Internet use at work among men. The
differences seem likely to reflect the disproportionate sorting of women into administrative or
clerical jobs that have a relatively high proportion of computer or Internet use relative to the
average education of workers (see Losh 2004).17
Table 8 shows how the effects of basic psychological characteristics vary for those who
have jobs using the Internet, those currently working at jobs that do not use the Internet, and
those retired or otherwise not currently working. Consistent with expectations, the effects of
both cognitive ability and Openness are less for those with jobs that use the Internet. Keeping
in mind our earlier caveat about home and workplace Internet being cross-sectional measures,
the finding is consistent with the interpretation that having the Internet at work introduces
individuals to skills and benefits of Internet use that then diminish the relevance of cognitive
ability and Openness for adoption at home.
TABLE 8 ABOUT HERE
17 Even so, supplementary analyses indicate that the positive relationship between
Conscientiousness and Internet use is not confined to any one of the broad 1990 census
occupational categories (e.g., those classified as clerical jobs). As is well known, however, the
heterogeneity of job characteristics within those broad classifications is very large.
Cognition, Personality, and Internet Adoption - 35
At the same time, we also note that the effects of Conscientiousness, Neuroticism, and
Agreeableness are stronger for those using the Internet at work than for others (although these
differences vary in magnitude and statistical significance). This supports the idea that the
findings of attenuation for cognitive ability and Openness are not somehow artifacts of the higher
baseline probability of adoption for those who use the Internet at work. For Agreeableness and
Conscientiousness, the difference was not predicted but would not seem to contradict the earlier
conjectures about why these variables might be expected to affect Internet adoption, especially
since the workplace might provide a venue where the benefits of the solitary character of the
Internet (for those low in Agreeableness) or its potential for distraction (for those low in
Conscientiousness) are revealed. The same cannot be said for Neuroticism: if the inverse
relationship between Neuroticism and Internet adoption reflects “technophobia” or anxiety about
new technology, then we would expect the effect to be less for those with a potentially
compulsory workplace introduction. The difference in coefficients is not significant, but if it
does reflect a real difference in the circumstances when Neuroticism affects adoption, it suggests
that further work to understand why Neuroticism is inversely related to adoption is needed.
DISCUSSION
Even as technologies like the Internet may be regularly hailed for revolutionizing aspects
of everyday life, people vary considerably in the extent to which they are active participants in
the revolution. Social scientists have approached variation in Internet adoption in studies that
have relied heavily on sociodemographic characteristics, and their recognition of the need for
more attention to the psychology of adoption has been evinced mostly in highly domain-specific
concepts like Internet anxiety and Internet skill. Our study is not intended to question the
importance of such work, but rather we proposed that a fuller understanding of adoption might
Cognition, Personality, and Internet Adoption - 36
be achieved by also giving attention to aspects of psychology that are both substantively general
and relatively robust to changes in social circumstances over midlife. Basic psychological
characteristics, we believe, may hold much unfulfilled promise toward providing a common,
parsimonious, and interdisciplinary vocabulary in considering how psychological heterogeneity
and social position influence responses to change. We thought an especially interesting case for
Internet adoption was provided by those whose biographies intersected the rise of the technology
at a comparatively settled point in the life course.
We focused here on cognitive ability and personality for a sample of relatively privileged
late-midlife adults, and the predictions we developed worked out most cleanly for the former.
Even though cognitive ability in this sample was measured before the Internet was invented and
before any sample members had likely ever seen a computer, the measure very strongly predicts
whether respondents were Internet users almost fifty years later. Some of the effect of cognitive
ability is attributable to the consequences of cognitive ability for sorting into later life
circumstances, but much seems to reflect more direct cognitive selectivity into adoption, perhaps
due to the literacy and information-processing skill demands of the technology as it has
developed. Including our measure of cognitive ability from high school also diminishes the
extent to which Internet adoption appears to be causally affected by whether respondents
attended or completed college. Importantly, however, we also found that the influence of
cognitive ability is itself moderated by social position. Three separate dimensions of social
position that put one in a more privileged position in terms of the likelihood of adoption—higher
education, more wealth, and a job in which one uses the Internet—are all associated with a
reduced association between cognitive ability and Internet adoption.
Cognition, Personality, and Internet Adoption - 37
Meanwhile, of the Big Five personality traits, Openness is most strongly associated with
adoption and is moderated by having a job that uses the Internet (but not wealth or education)
similar to what was observed for cognitive ability. Neuroticism was inversely related to
adoption as predicted, but the stronger relationship between Neuroticism and adoption among
those with exposure to the Internet at work seems inconsistent with the idea that the relationship
reflects anxiety toward new technology. Conscientiousness and Agreeableness show results
consistent with other work, but Conscientiousness also shows interactions with some measures
of social position that are not obviously explained. Personality was also measured later in this
sample, so it does not have the epistemic luxury of temporal priority over the attainment
variables, and it is measured by relatively few items and thus has lower reliability than cognitive
ability. Whether better measures would have resulted in a more unambiguous picture is
uncertain, but, even with the observed ambiguities, the study does provide strong evidence for
the relevance of basic personality traits for Internet adoption. In sum, the findings for cognitive
ability and Openness especially emphasize not just that these characteristics are pertinent for
understanding adoption but that an appropriately nuanced account of their influence requires
sociological thinking about the kinds of social circumstances that make cognition and personality
more or less influential.
As such, we hope our study suggests the potential value of greater effort to include
measures of basic psychological characteristics in longitudinal surveys, as well as more attention
to those data sources that contain such measures. As things stand, our inquiry raises many more
questions than it answers. For one, of course, we do not know how far our findings extend
beyond the narrow outcome of Internet adoption and for the narrow population represented by a
sample of almost-entirely-white, late midlife adults with high school degrees. Our findings do
Cognition, Personality, and Internet Adoption - 38
lend themselves to further conjectures: for example, just as having a job that uses the Internet is
associated with a sharp reduction in the cognitive gradient of adoption, we might expect this
gradient to be less in younger cohorts that have exposure through schools. We also know very
little about the relationship between basic psychological characteristics and the more domain-
specific psychological characteristics that have been more commonly invoked in the literature.
For that matter, more detailed information about social circumstances might further elaborate our
understanding of when psychological characteristics matter more or less. As a notable example,
specific information on the psychological characteristics of both spouses would allow more
specific examination of how the psychological characteristics of each partner are implicated in
the household decision to obtain Internet access.
With these limitations, what might our findings imply for current debates and research
about digital inequality? Numerous researchers have rightly complained about the impoverished
picture of technology cleavages one gets from a focus on a narrow outcome rather than the
variegated spectrum of use (e.g., Hargittai 2003, 2004a). By focusing precisely on a narrow
outcome, however, we hope to show also the impoverished view that results from an overly
narrow consideration of explanatory variables. Even apart from our consideration of
psychological characteristics, we hope our study also underscores the potential value of
considering wealth alongside income, and occupational measures alongside educational
attainment. We recognize that many surveys do not collect information to construct these
measures, either, but, at the least, the practical absence of additional measures should cause
greater tentativeness in conclusions. While our own study finds the estimated education effect
on adoption is mostly accounted for by selection and posteducational sorting, rather than formal
schooling per se, the narrowness of our sample means the generalizability of the result is an open
Cognition, Personality, and Internet Adoption - 39
question—but we do emphasize that it is, truly, an open question. If digital inequalities are to be
redressed, then more complete knowledge of their causes would seem desirable, and such
knowledge will require more elaborate measures of both psychological and social characteristics.
The importance of redressing digital inequalities has been itself a matter of considerable
and much-politicized debate (Compaine 2001; Warschauer 2003). Such debates turn on the
consequences of non-adoption for non-adopters, which is outside the scope of this inquiry. What
is plain from looking at the recent course of Internet diffusion is that we are not presently on
some S-curve toward full saturation, even in younger cohorts. While many Americans do lack
the resources to affordably obtain the Internet in their homes (for that matter, many Americans
do not even have homes), it is also plain that for many Americans, not using the Internet is
mainly a matter of choice, a choice driven as much by their perceptions of it not being useful as
by constraints imposed by its expense. Changes in skill and the opportunity to acquire skill may
well change some choices. To whatever extent motivation is at issue, this still does not reduce
the imperative to motivate non-users if the consequences of non-use are negative. This point
may be especially important for Internet adoption, as its possible usefulness might not be as
apparent to individuals and one’s being online may have positive social externalities in addition
to whatever individual benefits of use. Such implications are not at all affected by findings of an
important role of basic psychological characteristics in understanding differences in technology
adoption choices, even as it may prove helpful in understanding how choices may be changed if
socially desirable. For that matter, the benefits of Internet use may themselves vary by basic
psychological characteristics (see, e.g., Kraut et al. 2002 regarding extraversion).
One reason basic psychological characteristics are often resisted in studies of inequalities
is that their relevance, when demonstrated, is often taken as implying the inevitability of the
Cognition, Personality, and Internet Adoption - 40
inequality. Not only does the observed midlife stability of psychological characteristics in
populations not imply some ultimate “natural” immutability of the characteristic (Freese, Li, and
Wade 2003), but, as our results suggest, the consequences of psychological characteristics can
also vary and so should not be construed as inevitable anyway. As for the case of Internet
adoption, sociology should resist thinking of technological evolution as an asocial phenomenon,
but instead as a matter open to investigation. Technologies co-evolve with the needs and
demands of users (DiMaggio et al. 2004), with the variously disadvantaged often providing the
least incentive for the expansion of markets. Home computers were once a paradigmatic
example of how the complexity of innovations could inhibit adoption (Rogers 1995: 243), and
the rise of the Internet is not just a story about a change of technical capacity but about making
the technology more “user-friendly.” If a new technology is cognitively selective in problematic
ways, one job for the sociologist may be to interrogate how and why the technology has come to
be as cognitively inclusive as it is and yet not more so. What we have found here is not some
natural relationship between mind and technology, but a socially contingent one. We
demonstrate social contingency here just in the sense of showing that the relevance of
psychological characteristics varies by social position. In closing, however, we wish to
encourage more thinking about the social contingencies of technological development in ways
that create, sustain, and reduce the relevance of basic psychological characteristics for a
technology’s adoption and use.
REFERENCES
Borgatta, E. F. 1964. "The structure of personality characteristics." Behavioral Science 9:8-17.
Cognition, Personality, and Internet Adoption - 41
Boulianne, Shelley J. 2005. "The Role of the Internet in Civic and Political Engagements."
Unpublished manuscript.
Brosnan, Mark J. 1998. Technophobia: The Psychological Impact of Information Technology.
New York, NY: Routledge.
Carroll, John B. 1993. Human cognitive abilities: a survey of factor-analytic studies. Cambridge:
Cambridge University Press.
Compaine, Benjamin M. (ed) 2001. The Digital Divide: Facing a Crisis or Creating a Myth?
Cambridge, Mass.: MIT Press.
Conley, Dalton. 1999. Being Black, Living in the Red: Race, Wealth, and Social Policy in
America. Berkeley, CA: University of California Press.
Costa, P. T., Jr. and R. R. McCrae. 1994. "Stability and change in personality from adolescence
through adulthood." in The developing structure of temperament and personality from
infancy to adulthood, edited by C. F. Halverson, Jr., G. A. Kohnstamm, and R. P. Martin.
Hillsdale, NJ: Erlbaum.
de Haan, Jos. 2004. "A Multifaceted Dynamic Model of the Digital Divide." IT&Society 1:66-88.
de Haan, Jos and Suzanne Rijken. 2002. "The Digital Divide in the Netherlands: The influence
of material, cogntive, and social resources on the possession and use of ITs." Electronic
Journal of Communication 12.
De Raad, B., M. Perugini, M. Hrebickova, and P. Szarota. 1998. "Lingua franca of personality:
Taxonomies and structures based on the psycholexical approach." Journal of Cross-
Cultural Psychology 29:212-232.
Denton, Melinda Lundquist. 2004. “Gender and Marital Decision Making: Negotiating
Religious Ideology and Practice.” Social Forces 82: 1151-1180.
Cognition, Personality, and Internet Adoption - 42
Devlin, Bernie, Stephen E. Fienberg, Daniel P. Resnick, and Kathryn Roeder. 1997. Intelligence,
Genes, and Success: Scientists Respond to The Bell Curve. New York: Copernicus.
Digman, John M. 1996. "The curious history of the five-factor model." in The Five-Factor
Model of Personality: Theoretical Perspectives, edited by J. S. Wiggins.
DiMaggio, Paul, Eszter Hargittai, Coral Celeste, and Steven Shafer. 2004. "Digital Inequality:
From Unequal Access to Differentiated Use." Pp. 355-400 in Social Inequality, edited by
K. Neckerman. New York: Russell Sage Foundation.
DiMaggio, Paul, Eszter Hargittai, W. Russell Neuman, and John P. Robinson. 2001. "Social
Implications of Internet." Annual Review of Sociology 27 307-36.
DiMaggio, Paul and Eszter Hargittai. 2002. "The New Digital Inequality: Social Stratification
Among Internet Users." in Annual Meetings of the American Sociological Assocation.
Chicago, Illinois.
Fischer, Claude S., Michael Hout, Martín Sánchez Jankowski, Samuel R. Lucas, Ann Swidler,
and Kim Voss. 1996. Inequality By Design: Cracking the Bell Curve Myth. Princeton,
NJ: Princeton University Press.
Fox, Susannah. 2004. "Older Americans and the Internet." http:/www.pewInternet.org.
Fox, S., L. Rainie, E. Larsen, J. Horrigan, A. Lenhart, T. Spooner, and C. Carter. 2001. "Wired
seniors: A fervent few, inspired by family ties." Pew Internet & American Life Project,
Washington, DC.
Freese, Jeremy. 2005. “Associations Between Henmon-Nelson Test Scores and Iowa Subtest
Scores for the Wisconsin Longitudinal Study Graduate Sample” Unpublished memo
(#169) available from the Wisconsin Longitudinal Study, Department of Sociology,
University of Wisconsin-Madison.
Cognition, Personality, and Internet Adoption - 43
Freese, Jeremy and Salvador Rivas. 2005. “Exploring Cognitive and Occupational Explanations
for Educational Differences in Home Internet Adoption among Older Adults.” Annual
Meetings of the American Sociological Association, Philadelphia.
Freese, Jeremy, Jui-Chung Allen Li, and Lisa D. Wade. 2003. "The Potential Relevances of
Biology to Social Inquiry." Annual Review of Sociology 29:233-256.
Friedberg, Leora. 2003. "The impact of technological change on older workers: Evidence from
data on computer use." Industrial and Labor Relations Review 56:511-529.
Gershuny, Jonathan. 2003. "Web Use and Net Nerds: A Neofunctionalist Analysis of the Impact
of Information Technology in the Home." Social Forces 82:141-168.
Gettinger, M. 1984. "Individual differences in time needed for learning: A review of literature."
Educational Psychologist 19:15-29.
Gottfredson, Linda S. 2002. "g: Highly general and highly practical." in The General Factor of
Intelligence: How General Is It?, edited by R. J. Sternberg and E. L. Grigorenko.
Mahwah, NJ: Lawrence Erlbaum.
Gould, Stephen Jay. 1996. The Mismeasure of Man. New York: Norton.
Gross, Elisheva F. 2004. "Adolescent Internet use: What we expect, what teens report." Journal
of Applied Developmental Psychology 25:633-649.
Hargittai, Eszter. 2002. "Second-level digital divide: Differences in people's online skills." First
Monday 7:http://firstmonday.org/issues/issue7_4/hargittai/.
Hargittai, Eszter. 2003. "The Digital Divide and What To Do About It." Pp. 821-839 in New
Economy Handbook, edited by D. C. Jones. San Diego, CA: Academic Press.
Hargittai, Eszter. 2004a. "Internet Access and Use in Context." New Media and Society 6:137-
143.
Cognition, Personality, and Internet Adoption - 44
Hargittai, Eszter. 2004b. How Wide a Web? Inequalities in Accessing Information Online.
Dissertation, Department of Sociology, Princeton University.
Harris, David R. 2000. "What's the Internet Got To Do with It? Race and Inequality in the 21st
Century." Focus 21:63-64.
Herrnstein, Richard J. and Charles Murray. 1994. The Bell Curve: Intelligence and Class
Structure in American Life. New York: The Free Press.
Howard, Philip N. 2004. Embedded Media: Who We Know, What We Know, and Society
Online. In Philip N. Howard and Steven Jones (Eds.), Society Online: The Internet in
Context. Thousand Oaks, CA: Sage Press.
Howard, P.E.N., L. Rainie, and S. Jones. 2001. "Days and Nights on the Internet: The Impact of
a Diffusing Technology." American Behavioral Scientist 45(3):383-404.
Jackson, Linda A., Alexander von Eye, Gretchen Barbatsis, Frank Biocca, Yong Zhao, and
Hiram E. Fitzgerald. 2003. "Internet attitudes and Internet use: some surprising findings
from the HomeNetToo project." International Journal of Human-Computer Studies
59:355–382.
Jacoby, Russell and Naomi Glauberman. 1995. The Bell Curve Debate. New York: Random
House.
Jencks, Christopher and Meredith Phillips. 1998. The Black-White Test Score Gap. Washington,
DC: Brookings Institution Press.
Jensen, Arthur R. 1969. "How Much Can We Boost IQ and Scholastic Achievement?" Harvard
Educational Review 39:1-123.
Cognition, Personality, and Internet Adoption - 45
John, O. P., A. Caspi, R. W. Robins, T. E. Moffitt, and M. Stouthamer-Loeber. 1994. "The 'Little
Five': Exploring the nomological network of the Five-Factor model of personality in
adolescent boys." Child Development 65:160-178.
John, Oliver P. and Sanjay Srivastava. 1999. "The big five trait taxonomy: History,
measurement, and theoretical perspectives." Pp. 102-138 in Handbook of Personality
Theory and Research, edited by L. A. Pervin. New York: Guilford.
King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. "Analyzing Incomplete
Political Science Data: An Alternative Algorithm for Multiple Imputation." American
Political Science Review 95:49-69.
Kohn, Melvin L., & Schooler, Carmi. 1978. The reciprocal effects of the substantive complexity
of work and intellectual flexibility: A longitudinal assessment. American Journal of
Sociology, 84(1), 24-52.
Kohn, Melvin L., & Schooler, Carmi. 1983. Work and Personality: An Inquiry into the Impact of
Social Stratification. Norwood, NJ: Ablex Publishing Corporation.
Kraut, Robert, Sara Kiesler, Bonka Boneva, Jonathon Cummings, Vicki Helgeson, and Anne M.
Crawford. 2002. "Internet Paradox Revisited." Journal of Social Issues 58: 49-74.
Krosnick, Jon A. 1999. "Survey Research." Annual Review of Psychology 50:537-567.
Kubicek, Herbert. 2004. "Fighting a moving target: Hard lessons from Germany's digital divide."
IT&Society 1(6):1-19.
Laguna, Kerrie and Renée L. Babcock. 1998. "Computer anxiety in young and older adults:
Implications for human-computer interactions in older populations." Computers in
Human Behavior 13:317-326.
Cognition, Personality, and Internet Adoption - 46
Landers, Richard N. and John W. Lounsbury. In press. "An investigation of Big Five and narrow
personality traits in relation to Internet usage." Computers in Human Behavior.
Leigh, Andrew and Robert D. Atkinson. 2001. "Clear Thinking on the Digital Divide." Policy
Report. Progressive Policy Institute.
Lenhart, Amanda, John Horrigan, Lee Rainie, Katherine Allen, Angie Boyce, Mary Madden,
Erin O'Grady. 2003. "The ever-shifting Internet population: A new look at Internet
access and the digital divide." Pew Internet and American Life Project.
http://www.pewinternet.org/
Livingstone, Sonia. 2003. "Children's use of the Internet: Reflections on the emerging research
agenda." New Media & Society 5:147-166.
Loges, William E., and Joo-Young Jung. 2001. "Exploring the Digital Divide: Internet
Connectedness and Age." Communication Research 28(4):536-562.
Losh, S. C. 2004. "Gender, educational, and occupational digital gaps: 1983-2002." Social
Science Computer Review 22:152-166.
Marcoulides, George A. 1989. "Measuring computer anxiety: the Computer anxiety scale."
Educational and Psychological Measurement 49:733-739.
Martin, Steven P. 2003. “Is the Digital Divide Really Disappearing? A Critique of Inequality
Measurement in A Nation Online.IT&Society 1(4):1-13.
McCrae, Robert R. and Paul T. Costa. 2003. Personality in Adulthood: A Five-Factor
Perspective, Second Edition. New York: Guilford Press.
Mills, C. Wright. 1959. The Sociological Imagination. New York: Oxford University Press.
Cognition, Personality, and Internet Adoption - 47
Mueller, Milton L. and Jorge Reina Schement. 1996. "Universal Service from the Bottom Up: A
Study of Telephone Penetration in Camden, New Jersey." The Information Society
12:273-292.
Negroponte, Nicholas. 1995. Being Digital. New York: Knopf.
Nie, Norman H. and Lutz Ebring. 2000. "Internet and Society: A Preliminary Report." Stanford
Institute for the Quantitative Study of Society, Stanford, CA.
Norman, W. T. 1963. "Toward an adequate taxonomy of personality attributes." Journal of
Abnormal and Social Psychology 66.
Norris, Pippa. 2001. Digital Divide: Civic Engagement, Information Poverty, and the Internet
Worldwide. New York, NY: Cambridge University Press.
NTIA (National Telecommunications and Information Administration). 1999. "Falling Through
the Net: Defining the Digital Divide." U.S. Department of Commerce, Washington, D.C.
NTIA (National Telecommunications and Information Administration). 2000. "Falling Through
the Net: Toward Digital Inclusion." U.S. Department of Commerce, Washington, D.C.
NTIA (National Telecommunications and Information Administration). 2002. "A Nation Online:
How Americans Are Expanding Their Use of the Internet." U.S. Department of
Commerce, Washington, D.C. Online:
(http://www.ntia.doc.gov/ntiahome/dn/Nation_Online.pdf).
NTIA (National Telecommunications and Information Administration). 2004. A Nation Online:
Entering the Broadband Age. Washington, D.C.: U.S. Department of Commerce.
Nurmela, Juha and Marja-Liisa Viherä. 2004. "Patterns of IT Diffusion in Finland: 1996-2002."
IT&Society 1:20-35.
Cognition, Personality, and Internet Adoption - 48
Oliver, Melvin L. and Thomas M. Shapiro. 1995. Black Wealth/White Wealth: A New
Perspective on Racial Inequality. New York: Routledge.
Reddick, A. J. and C. Boucher. 2002. Tracking the dual digital divide: EKOS Research
Associates.
Rivas, Salvador. 2004. "The U.S. Digital Divide: Race, SES, and Social Context as Predictors of
Computer Ownership, 1997-2001." Dissertation Thesis, Department of Sociology,
University of Michigan, Ann Arbor.
Roberts, Brent W., Richard W. Robins, Kali H. Trzesniewski, and Avshalom Caspi. 2003.
"Personality Trait Development in Adulthood." Pp. 579-595 in Handbook of the Life
Course, edited by J. T. Mortimer and M. J. Shanahan. New York: Kluwer.
Robins, Richard W., R. Chris Fraley, Brent W. Roberts, and Kali H. Trzesniewski. 2001. "A
Longitudinal Study of Personality Change in Young Adulthood." Journal of Personality
694:617-640.
Robinson, John P. and Meyer Kestnbaum. 1999. "The Personal Computer, Culture, and Other
Uses of Free Time." Social Science Computer Review 17:209-216.
Robinson, John P., Meyer Kestnbaum, Alan Neustadtl, and Anthony Alvarez. 2000. "Mass
Media Use and Social Life among Internet Users." Social Science Computer Review
18:490-501.
Robinson, John P., Alan Neustadtl, Meyer Kestnbaum. 2002. "The Online “Diversity Divide”:
Public Opinion Differences among Internet Users and Nonusers." IT&Society 1(1):284-
302.
Rogers, Everett M. 1995. Diffusion of Innovations. New York: Free Press.
Cognition, Personality, and Internet Adoption - 49
Schaie, K. Warner. 1996. Intellectual Development in Adulthood: The Seattle Longitudinal
Study. New York: Cambridge University Press.
Schaninger Charles M. and W. Christian Buss. 1986. A Longitudinal Comparison of
Consumption and Finance Handling between Happily Married and Divorced Couples.
Journal of Marriage and the Family. 48: 129-136.
Schmidt, F. L. and J. E. Hunter. 1998. "The validity and utility of selection methods in personnel
psychology: Practical and theoretical implications of 85 years of research findings."
Psychological Bulletin 124:262-274.
Schooler, C, Mulatu, MS, & Oates, G. (1999). The continuing effects of substantively complex
work on the intellectual functioning of older workers. Psychol Aging, 14(3), 483-506.
Sewell, William H. and Robert M. Hauser. 1975. Education, Occupation, and Earnings:
Achievement in the Early Career. New York: Academic Press.
Snow, R. E. and E. Yalow. 1982. "Education and intelligence." in Handbook of human
intelligence, edited by R. J. Sternberg. Cambridge: Cambridge University Press.
Srivastava, Sanjay, Oliver P. John, Samuel D. Gosling, and Jeff Potter. 2003. "Development of
Personality in Early and Middle Adulthood: Set Like Plaster or Persistent Change."
Journal of Personality 84:1041-1053.
Sternberg, Robert J. and James C. Kaufman. 1998. "Human Abilities." Annual Review of
Psychology 49:479-502.
Turkle, Sherry. 1995. Life on the Screen: Identity in the Age of the Internet. New York: Simon
& Schuster.
van Dijk, J. and K. Hacker. 2003. "The digital divide as a complex and dynamic phenomenon."
The Information Society 19:315-326.
Cognition, Personality, and Internet Adoption - 50
van Lieshout, C. F. and G. J. Haselager. 1994. "The Big Five personality factors in Q-sort
descriptions of children and adolescents." Pp. 293-318 in The developing structure of
temperament and personality from infancy to childhood, edited by C. F. Halverson, Jr.,
G. A. Kohnstamm, and R. P. Martin. Hillsdale, NJ: Erlbaum.
Wahlsten, Douglas. 1997. "The Malleability of Intelligence Is Not Constrained by Heritability."
Pp. 71-88 in Intelligence, Genes, and Success, edited by B. Devlin, S. E. Fienberg, D. P.
Resnick, and K. Roeder. New York: Copernicus.
Warschauer, Mark. 2003. Technology and Social Inclusion: Rethinking the Digital Divide.
Cambridge, Mass.: MIT Press.
Wasserman, Ira M. and Marie Richmond-Abbott. 2005. "Gender and the Internet: causes of
variation in access, level, and scope of use." Social Science Quarterly 86:252-270.
White, Candace and John M. II Scheb. 2000. "Impact of media messages about the Internet:
Internet anxiety as a factor in the adoption process in the USA." New Media & Society
2:181-194.
Table 1. Dimensions of the Five-Factor Model of Personality and Their Measurement in the
Wisconsin Longitudinal Study (1993 wave)
Dimension Items
Openness prefers work that is routine and simple*; inventive; prefers the
conventional, traditional*; values artistic, aesthetic experiences; has an
active imagination; wants things to be simple and clear-cut*;
sophisticated in art, music, and literature
Neuroticism worries a lot; relaxed and handles stress well*; can be tense;
emotionally stable, not easily upset*; remains calm in tense situations*;
gets nervous easily
Agreeableness tends to find fault with others*; has a forgiving nature; sometimes rude
to others*; generally trusting; can be cold and aloof*; considerate to
almost everyone; likes to cooperate with others
Conscientiousness easily distracted*; can be somewhat careless*; does a thorough job; a
reliable worker; tends to be disorganized*; lazy at times*; does things
efficiently
Extraversion tends to be quiet*; outgoing and sociable; talkative; reserved*; full of
energy; shy or inhibited*; generates a lot of enthusiasm
Asterisks denote items to be reverse scored for scaling. All items use stem “I see myself as a person who (is)” and
have six response categories ranging from “agree strongly” to “disagree strongly.” First item in list was
administered on both phone and mail survey, second item administered on phone only, the rest were administered
on mail only. Summated scales standardized for all analyses. Lower-bound reliability estimates (Cronbach’s α):
Openness = .69, Neuroticism = .83, Agreeableness = .74, Conscientiousness = .70, Extraversion = .82. The greater
homogeneity of the WLS sample implies lower reliabilities than would be expected for nationally representative
samples.
Table 2. Estimated Effects of Psychological and Sociodemographic Characteristics on Internet
Adoption
Model 1 Model 2 Model 3 Model 4
HN score 0.529*** 0.294*** 0.277***
(0.030) (0.033) (0.038)
Openness 0.383*** 0.265*** 0.249***
(0.031) (0.033) (0.039)
Neuroticism -0.100*** -0.103*** -0.110**
(0.029) (0.031) (0.036)
Agreeableness -0.068* -0.067* -0.060
(0.029) (0.031) (0.035)
Conscientiousness -0.063* -0.076* -0.102**
(0.028) (0.030) (0.035)
Extraversion 0.009 -0.048 -0.067
(0.028) (0.030) (0.034)
Female -0.041 0.237** 0.236** 0.211*
(0.055) (0.083) (0.085) (0.099)
No spouse -0.406*** -0.442*** -0.437***
(0.089) (0.090) (0.108)
Some college 0.489*** 0.321*** 0.323***
(0.077) (0.079) (0.092)
College degree 0.685*** 0.375*** 0.385***
(0.084) (0.088) (0.103)
Household income 0.107* 0.083 0.080
(0.045) (0.045) (0.050)
Net worth 0.092*** 0.088** 0.079*
(0.027) (0.028) (0.031)
Occupational education 0.466*** 0.394*** 0.259***
(0.059) (0.060) (0.070)
Occupational income 0.036 -0.023 0.052
(0.072) (0.074) (0.087)
Rural residence -0.382*** -0.349*** -0.242***
(0.060) (0.061) (0.071)
Online child 0.545*** 0.497*** 0.507***
(0.104) (0.106) (0.123)
Spouse age -0.001* -0.001* -0.001
(0.001) (0.001) (0.001)
Spouse no hs diploma -0.055 -0.037 -0.053
(0.110) (0.113) (0.134)
Spouse some college 0.254** 0.241* 0.220*
(0.093) (0.094) (0.109)
Spouse college 0.128 0.095 0.085
(0.095) (0.097) (0.111)
Spouse occupational education 0.221*** 0.179** 0.144*
(0.060) (0.061) (0.070)
Spouse occupational income -0.036 -0.006 0.053
(0.074) (0.076) (0.089)
Currently working -0.288***
(0.066)
Uses net at work 1.071***
(0.076)
N 6849 6849 6849 5281
bic 8604.428 8315.810 8156.083 6107.782
* p < .05, ** p < .01, *** p < .001 (two-tailed). Standard errors in parentheses. Models also include terms for
missing values on occupation, spouse occupation, or spouse education.
Table 3. Logistic regression estimates of the effect of psychological characteristics on Internet
adoption, by sex
WLS Females WLS Males
Model 1 Model 2 Model 1 Model 2
HN score 0.471*** 0.295*** 0.595*** 0.262***
(0.049) (0.053) (0.048) (0.055)
Openness 0.367*** 0.241*** 0.443*** 0.281***
(0.049) (0.053) (0.054) (0.059)
Neuroticism -0.104* -0.091 -0.133** -0.146**
(0.046) (0.048) (0.050) (0.053)
Agreeableness -0.067 -0.060 -0.103* -0.082
(0.049) (0.050) (0.047) (0.050)
Conscientiousness -0.084 -0.118* -0.073 -0.088
(0.045) (0.048) (0.047) (0.050)
Extraversion 0.008 -0.010 -0.010 -0.060
(0.044) (0.045) (0.047) (0.050)
N 3687 3687 3161 3161
BIC 4754.020 4631.907 3886.901 3653.655
* p < .05, ** p < .01, *** p < .001 (two-tailed). Standard errors in parentheses. Model 2 includes all regressors (except sex)
from Model 3 in Table 2.
Table 4. Logistic regression estimates of effect of educational attainment on log odds of
home Internet adoption
WLS females (N = 3687)
some college college graduates
bivariate .652*** ( 0%) 1.163*** ( 0%)
adding HN score .496*** ( 24%) .770*** ( 34%)
adding personality variables .363*** ( 44%) .545*** ( 53%)
adding income and wealth .297*** ( 54%) .446*** ( 62%)
adding occupational measures .196* ( 70%) .158* ( 86%)
adding spousal characteristics .148* ( 77%) .053 ( 95%)
WLS males (N = 3161)
some college college graduates
bivariate 1.086*** ( 0%) 1.810*** ( 0%)
adding HN score .914*** ( 16%) 1.413*** ( 22%)
adding personality variables .819*** ( 25%) 1.302*** ( 28%)
adding income and wealth .738*** ( 32%) 1.133*** ( 37%)
adding occupational measures .611*** ( 44%) .770*** ( 57%)
adding spousal characteristics
.575*** ( 47%) .683*** ( 62%)
* p < .05, ** p < .01, *** p < .001 (two-tailed). Percentage attenuation from bivariate
regression in parentheses. Respondents with high school diploma only are reference
category for estimates.
Table 5. Logistic regression coefficients of psychological and sociodemographic
characteristics on whether respondent used Internet at current/last job
High school only Some college College degree
Model 1 Model 2 Model 1 Model 2 Model 1 Model 2
HN score 0.433***§ 0.306***§ 0.302*** 0.225* 0.248** 0.124
(0.048) (0.052) (0.090) (0.095) (0.079) (0.086)
Openness 0.330*** 0.258*** 0.350*** 0.302** 0.287*** 0.274**
(0.049) (0.052) (0.094) (0.098) (0.082) (0.087)
Neuroticism -0.086 -0.085 -0.086 -0.070 -0.250** -0.204*
(0.044) (0.047) (0.087) (0.092) (0.077) (0.083)
Agreeableness -0.000 -0.005 -0.106 -0.126 -0.121 -0.113
(0.045) (0.048) (0.085) (0.088) (0.072) (0.077)
Conscientiousness -0.022 -0.064 -0.179* -0.214* -0.032 -0.090
(0.044) (0.047) (0.082) (0.087) (0.069) (0.075)
Extraversion -0.020 -0.067 -0.031 -0.082 0.056 -0.058
(0.043) (0.046) (0.079) (0.084) (0.066) (0.074)
Female 0.233** 0.278 -0.058 0.242 -0.373** 0.045
(0.085) (0.142) (0.160) (0.250) (0.136) (0.188)
N 2777 2777 818 818 1439 1439
BIC 3680.433 3524.548 1066.457 1106.388 1506.493 1464.691
* p < .05, ** p < .01, *** p < .001 (two-tailed). §Indicates p < .05 for interactions across categories.
Standard errors in parentheses. Model 2 includes all controls from Model 4 in Table 2.
Table 6. Logistic Regression Estimates of the Effects of Basic Psychological
Characteristics and Gender on Home Internet Adoption, by Wealth
Below median wealth Above median wealth
Model 1 Model 2 Model 1 Model 2
HN score 0.406***§ 0.335*** 0.250***§ 0.193**
(0.058) (0.060) (0.061) (0.063)
Openness 0.260*** 0.265*** 0.279*** 0.258***
(0.053) (0.056) (0.057) (0.059)
Neuroticism -0.093 -0.098 -0.114* -0.117*
(0.051) (0.052) (0.054) (0.055)
Agreeableness -0.032 -0.042 -0.055 -0.074
(0.051) (0.052) (0.053) (0.054)
Conscientiousness 0.003§ -0.006§ -0.180***§ -0.193***§
(0.050) (0.051) (0.053) (0.054)
Extraversion -0.036 -0.094 0.004 -0.045
(0.049) (0.052) (0.053) (0.055)
Female 0.101 0.296* 0.046 0.146
(0.098) (0.144) (0.104) (0.151)
N 2497 2497 2537 2537
BIC 3108.347 3082.386 2887.639 2925.896
* p < .05, ** p < .01, *** p < .005 (one-tailed). §Indicates p < .05 for interactions across
categories. Standard errors in parentheses. Model 1 includes education, job status, and Internet at
work as additional controls. Model 2 includes all controls from Model 4 in Table 2.
Table 7. Logistic regression coefficients of psychological and sociodemographic
characteristics on whether respondent used Internet at current or last job
WLS Females WLS Males
Model 1 Model 2 Model 2 Model 1 Model 2 Model 2
(full
sample)
(full
sample)
(current
workers)
(full
sample)
(full
sample)
(current
workers)
HN score 0.287***§ 0.154** 0.211** 0.553*** 0.240*** 0.179*
(0.047) (0.052) (0.076) (0.049) (0.056) (0.079)
Openness 0.278***§ 0.187*** 0.218** 0.453*** 0.329*** 0.359***
(0.050) (0.055) (0.077) (0.057) (0.060) (0.080)
Neuroticism -0.132** -0.107* -0.046 -0.082 -0.079 -0.088
(0.048) (0.049) (0.073) (0.052) (0.055) (0.077)
Agreeableness -0.043 -0.012 0.066 0.048 0.085 0.010
(0.048) (0.050) (0.071) (0.047) (0.050) (0.070)
Conscientiousness 0.143** 0.114* 0.217** 0.035 0.019 0.105
(0.046) (0.049) (0.071) (0.048) (0.050) (0.070)
Extraversion 0.035 0.040 -0.023 0.055 0.036 -0.004
(0.045) (0.047) (0.068) (0.047) (0.049) (0.069)
No spouse 0.231* 0.047 -0.289* -0.270
(0.097) (0.136) (0.142) (0.208)
Some college 0.154§ 0.174§ 0.792*** 0.968***
(0.122) (0.175) (0.131) (0.184)
College degree 0.158§ 0.265§ 1.132*** 1.198***
(0.118) (0.170) (0.119) (0.168)
Rural residence -0.330**§ -0.460** -0.482*** -0.501**
(0.103) (0.145) (0.111) (0.155)
N 2820 2820 1156 2464 2464 1222
BIC 3363.605 3208.590 1529.779 3013.352 2841.271 1477.085
* p < .05, ** p < .01, *** p < .001 (two-tailed). ). §Indicates p < .05 for difference between females and
males. Standard errors in parentheses.
Table 8. Logistic Regression Estimates of the Effects of Basic Psychological Characteristics
and Gender on Home Internet Adoption, by Whether Internet is Used at Work
Currently working Not currently working
Uses Internet
at work
No Internet at
work
Used Internet at
last job
No Internet at last
job
HN score 0.080§ 0.395***§ 0.220 0.299***
(0.100) (0.073) (0.114) (0.059)
Openness 0.113 0.382*** 0.400** 0.174**
(0.106) (0.075) (0.131) (0.058)
Neuroticism -0.229* -0.082 -0.050 -0.126*
(0.095) (0.069) (0.115) (0.053)
Agreeableness -0.214*§ -0.045§ -0.076 -0.011
(0.093) (0.066) (0.115) (0.055)
Conscientiousness -0.226* 0.018 -0.142 -0.129*
(0.093) (0.067) (0.118) (0.052)
Extraversion 0.015 -0.144* -0.242* -0.022
(0.084) (0.068) (0.109) (0.052)
N 1026 1351 775 2129
BIC 1128.104 1771.990 816.935 2776.005
* p < .05, ** p < .01, *** p < .001 (two-tailed). ). §Indicates p < .05 for interactions across categories.
Standard errors in parentheses. Model 1 includes education, job status, and internet at work as additional
controls. Model 2 includes all controls from Model 4 in Table 2.
Center for Demography and Ecology
University of Wisconsin
1180 Observatory Drive Rm. 4412
Madison, WI 53706-1393
U.S.A.
608/262-2182
FAX 608/262-8400
comments to: jfreese@ssc.wisc.edu
requests to: cdepubs@ssc.wisc.edu
... In the HRS, Internet access varies directly with educational attainment and inversely with age. Moreover, to a substantial degree, the effect of educational attainment in this context refl ects differences in cognitive functioning (Freese and Rivas 2005; Freese, Rivas, and Hargittai 2006). Thus, participants with Internet access are a highly selected subset of all participants. ...
Article
We review recent developments in longitudinal studies of aging, focusing on the Wisconsin Longitudinal Study (WLS) and the Health and Retirement Study (HRS). Both studies are part of a trend toward biosocial surveys in which biological measurement is joined with traditional survey techniques, and a related trend toward greater harmonization across studies. Both studies have collected DNA samples and are working toward genotyping that would allow broadly based association studies. Increased attention to psychological measurement of personality and of cognitive ability using adaptive testing structures has also been shared across the studies. The HRS has expanded its economic measurement to longitudinal studies of consumption and to broader-based measurement of pension and Social Security wealth. It has added biomarkers of cardiovascular risk. The WLS has developed an integrated approach to the study of death and bereavement and an innovative use of high school yearbook photographs to capture information about health in early life of its participants.
... Yet, because social engagement among older adults has been linked to higher levels of cognitive function [1] and reduced levels of cognitive decline [51], the effects of social media use may have some cognitive consequences. Prior studies have found that cognitive ability is a predictor of internet use in older adults [11,18], and that cognitive ability is associated with the amount and types of activities in which older adults engage [17]. Specifically, the executive function domains of cognitive flexibility, or the ability to shift between tasks or concepts, and speed of information processing have been found to be predictors of efficient technological use and proficiency in older adults [36] Older adults generally have lower levels of educational attainment than younger persons [31] and are less likely to subscribe to broadband internet [37], factors which may contribute to reduced internet use generally. ...
Conference Paper
Research on the effects of social media use at older ages has largely been focused on social benefits. Yet, participation in these new media forms may result in other favorable outcomes, such as improved cognitive functioning. Using a wait list control design, this study examines the effects of social media engagement among adult social media non-users, aged 65 and older, in four cognitive domains: attention, processing speed, working memory, and inhibitory control. Baseline and multiple post-tests indicate improvement of intervention participants in processing speed and inhibitory control. These findings demonstrate that the benefits of social media use at older ages extend beyond mere social engagement, and into other domains of everyday well-being.
Article
Full-text available
Research on the effects of social media use at older ages has largely focused on social benefits. Yet, participation in these new media forms may result in other favorable outcomes, such as improved cognitive functioning. Using a wait list-control design, this study examines the effects of social media engagement among novice adult social media users, aged 65 and older, in four cognitive domains: attention, processing speed, working memory, and inhibitory control. Baseline and multiple post-tests indicate improvement of intervention participants in inhibitory control. These findings demonstrate that the benefits of social media use at older ages extend beyond mere social engagement, and into other domains of everyday well-being.
Article
Full-text available
Data sets from seven original trait taxonomies from different languages, American English, Dutch, German, Hungarian, Italian, Czech, and Polish, are used for a crosscultural study. The taxonomic procedures, involving culling trait terms from the various lexicons and the construction Of representative samples of trait terms, are briefly discussed. Factor structures, presumably Big Five structures, within these languages, based on ratings from an average of about 640 subjects on an avenge of approximately 430 trait variables per language, are used for comparison. Congruence coefficients are calculated for the corresponding factors in the different languages, based on their independent positions and on their positions after rotations, using the American English solution as target In a relative sense, the congruences show replicability of the first four American English, Big Five factors in the other languages.
Article
Full-text available
This article summarizes the practical and theoretical implications of 85 years of research in personnel selection. On the basis of meta-analytic findings, this article presents the validity of 19 selection procedures for predicting job performance and training performance and the validity of paired combinations of general mental ability (GMA) and the 18 other selection procedures. Overall, the 3 combinations with the highest multivariate validity and utility for job performance were GMA plus a work sample test (mean validity of .63), GMA plus an integrity test (mean validity of .65), and GMA plus a structured interview (mean validity of .63). A further advantage of the latter 2 combinations is that they can be used for both entry level selection and selection of experienced employees. The practical utility implications of these summary findings are substantial. The implications of these research findings for the development of theories of job performance are discussed.
Book
This is author-approved bcc. If it is too long, delete the last sentence in each of the biographies. THE BELL CURVE by Richard Herrnstein and Charles Murray, a best selling book published in 1994, set off a hailstorm of controversy about the relationships among IQ, genetics, and various social outcomes, including welfare dependency, crime, and earnings. Much of the public reaction to the book was polemical and did not focus on the details of the science and in particular on the validity of the statistical arguments that underlie the books's conclusions. A detailed understanding of the arguments in THE BELL CURVE requires knowledge about (i) statistical models for genetic heritability, (ii) factor analysis, especially as it has been applied to the analysis of IQ tests, (iii) logistic regression and multiple regression analyses,and (iv) causal modelling and alternative statistical frameworks for making inference from longitudinal data. In this volume a group of statisticians and social scientists have assembled a scientific response to THE BELL CURVE. The sixteen chapters begin by presenting an overview of the scientific and statistical issues and summarize the material in Herrnstein and Murray's book. Then separate chapters by various experts deal with more focused issues, including reanalyses of data relied upon by the authors of THE BELL CURVE. The final chapters consider some of the implications of the work described in the book for American public policy and scientific research. BERNIE DEVLIN is Program Director of the Computational Genetics Program at the University of Pittsburgh Medical Center. He serves on
Chapter
In The Bell Curve, Herrnstein and Murray claim that a high value for heritability of intelligence limits or constrains the extent to which intelligence can be increased by changing the environment.1 In this chapter it is argued that the calculated numerical value of “heritability” has no valid implications for government policies and that evidence of a nonspecific genetic influence on human mental ability places no constraint on the consequences of an improved environment. On the contrary, a very small change in environment, such as a dietary supplement, can lead to a major change in mental development, provided the change is appropriate to the specific kind of deficit that in the past has impaired development. The results of adoption studies, the intergenerational cohort effect, and effects of schooling also reveal that intelligence can be increased substantially without the need for heroic intervention.
Book
Chapter American polygeny and craniometry before Darwin : blacks and Indians as separate, inferior species Notes in computer files
Article
According to the U.S. Department of Commerce Report "A Nation Online: How Americans are Expanding their Use of the Internet," computer ownership and Internet use are rapidly becoming more equally distributed across households in the United States. The authors of "A Nation Online" use two statistical arguments to support this claim: 1) annual rates of increase for computer and Internet use are increasing most quickly for poor households, and 2) "Gini" coefficients for inequality of computer use are decreasing. These analyses critique these arguments and show that patterns that the authors attribute to decreasing inequality are instead explained by two factors: 1) computer and Internet use is increasing, and 2) households with higher incomes began using computers and the Internet earlier than households with lower incomes. Reanalyzing these same data using odds ratios indicates that computer ownership and Internet use may actually be spreading less quickly among poorer households than among richer households. If current trends continue, poor households will eventually have the nearly universal levels of computer and Internet use currently seen among richer households, but this "catching-up" could take two decades.