Home Internet Use In
Low-Income Families:
Frequency, Nature, and
Correlates of early Internet Use in the HomeNetToo Project
Linda A. Jackson
Gretchen Barbatsis
Frank Biocca
Yong Zhao
Alexander von Eye
Hiram E. Fitzgerald
Michigan State University
HomeNetToo is an NSF-funded project designed to
understand the antecedents and consequences of home Internet use among
low-income families. In this report the
results of the first six-months of server-logged measures of Internet use, and
survey and demographic measures are presented, along with ethnographic accounts
of participants' experiences with the Internet. Findings indicate that Web activities are more popular than
e-mail, and that race, age, and education influence the frequency of Internet
use. Participants' descriptions of
their experiences with the Internet speak to the importance of universal access
and the need to design technology better adapted to the user.
KEYWORDS: Digital divide, Internet use
APPROXIMATE WORD COUNT: 7900
SECTION: Alternate Track: Global Community
Linda A. Jackson
Department of Psychology
Michigan State University
East Lansing, MI 48824
517-353-8690
HomeNetToo is a research project funded by the
National Science Foundation (NSF-ITR, #0085348) to investigate the antecedents
and consequences of Internet use by families on the "other side" of
the digital divide.1 In December 2000, computers and Internet
service were provided to 90 low-income families in the midwestern U.S. who
agreed to have their Internet use continuously server-logged, and to complete
surveys at several points during an 18-month trial. This report focuses on the first 6 months of home Internet use by
adult family members, many of whom were first-time users.2 Specifically, it focuses on the
frequency and nature of Internet use, demographic correlates of use, and
quantitative and ethnographic accounts of experiences with the Internet.
The Carnegie Mellon University HomeNet project served as
a model for HomeNetToo (1). Until now,
it was the only research project that automatically logged frequency and nature
of home Internet use. Findings from the
first HomeNet study (1995-1996, 48 families) indicated the people used the
Internet about 1 hour/week in 2 sessions, visited 2 domains, and sent less than
one e-mail/week (0.47). Listserv and
newsgroup activity was infrequent. Comparable data for the HomeNet Study 2
(1998-1999; 216 families) have yet to be reported (2). Other findings from the project indicated
the variety of ways that people use the Internet at home, summarized by the
researchers as follows: "People
use the Internet for pleasure: to communicate with family, friends, and
strangers, to track sports and popular culture, to listen to music, to play
games, and to pursue specialized interests.
These pleasurable uses supplement and, for many people, are more
important than the practical uses of the Internet for jobs, school, and
shopping" (homenet.andrew.cmu.edu/progress/).
Two
other large-scale research projects have investigated how the
"average" American uses the Internet at home. Both are based on self-reported Internet
use. The Stanford University study,
released in early 2000, used a national probability sample of the general U.S.
population and Internet technology (InterSurvey) to administer surveys to both
users and nonusers (3). Based on the
responses of 4,113 adults (2,689 households), findings indicated that
approximately 66% of users reported using the Internet less than 5 hours/week. The most common use was to obtain
information (products, travel, hobbies, and general information; 100% of
users), followed by e-mail (90%).
Entertainment was a distant third (33%), followed by shopping online
(25%) and chat (20%), the latter activity dominated by "under 30"
users. Demographic characteristics
explained differences in access to the Internet, but not in its use, leading
the researchers to conclude: "once people are connected to the Net they
hardly differ in how much they use it and what they use it for, except for a
drop off after age 65." However,
statistics on demographic characteristics by amount and type of Internet use
were not provided, nor was information about refusal rates, which are likely to
be higher among nonusers than users.
In late 2000, UCLA published
its first Internet report (4), which was based on random digit dial (RDD) phone
interviews of 2,096 households of users and nonusers. UCLA researchers found that, on average, people spent 9.42 hours
per week on the Internet, with time online increasing with Internet
experience. For example, users with
more than 4 years of experience used the Internet 16.2 hours/week whereas those
with less than one year of experience used it 6.1 hours/week. The most frequent Internet activities were
Web surfing or browsing (82% of users) and e-mail (82%), followed by finding
hobby information (57%), reading news (57%), finding entertainment information
(54%), and buying online (52%). Less than
half of the interviewees reported using the Internet for travel information
(46%), instant messaging (40%), medical information (37%), or playing games
(33%). As in the Stanford report and
numerous other surveys (5, 6), education and income were related to access to
the Internet, although even the least educated and lowest income groups
reported using the Internet; 31% of adult respondents with less than a
high-school education, and 41% with incomes under $15K were Internet
users. However, as in the Stanford
report, statistics on demographic characteristics by amount and type of
Internet use were not reported, nor were racial/ethnic group differences. Gender and age differences were
reported. Men spent more time online
than did women and engaged in different types of activities (e.g., commerce,
games). Time online increased between
the ages of 12 and 35, and remained at or above average until the mid-50s, when
it began to decline.
Relationships between
demographic characteristics and Internet use have been of considerable interest
to both researchers and dot.com marketers.
Numerous surveys have focused on income, education, gender and race in
an effort to understand differences in the frequency and nature of Internet
activities among demographic groups (5-10).
Findings have been remarkably mixed.
For example, a recent report by comScore (5, July 17, 2001), an Internet
marketing research group, concluded that "It appears that the
less-educated use the Web to amuse themselves and their friends, while the
well-educated use the Web as part of their careers." In contrast, the Stanford study (discussed
earlier) concluded that once access is obtained, demographic characteristics
are of little importance to the amount and type of Internet use.
A recent report by the Pew
Internet & American Life Project described differences in the online
behavior of African-Americans and European Americans (6). Findings indicated that African Americans
were less likely to go online in a typical day than were European Americans
(36% versus 56%), and less likely to use e-mail (27% versus 49%). The nature of Internet activities also
differed for the two groups. African
Americans were more likely to listen to music, seek religious information, play
games, download music, seek information about jobs, seek information about a
place to live, and conduct school research and job training. They were more likely to believe that the
Internet helped them to get health care information and information about
hobbies, and less likely to believe that it helped them to connect to family
and friends. Again these results appear
to conflict with the conclusion of the Stanford researchers that demographic
characteristics do not matter once access is obtained.
Another report by the Pew
Internet & American Life Project (7) focused on gender differences in
Internet use. Findings indicated no
gender difference in e-mail use, chat, instant messages, browsing for fun,
school-related or job-related research, accessing popular culture (e.g., downloading
music), and arranging travel and banking.
However, women were more likely to seek health care information,
research new jobs, and play games online.
Men were more likely to seek news and financial information, shop,
participate in online stock trading and auctions, access government sites, and
search for sports news.
Marketing surveys frequently
find income-based differences in Internet use.
A study by Nielsen/NetRatings (8) found that low-income groups spend
more time surfing the Web at home than do more affluent groups. Similar results
were obtained in a study by Media Metrix, Inc. (9). Once again these results appear to conflict with conclusions from
Stanford study (3).
Results from the HomeNetToo
project, presented next, describe the Internet use of low-income adults using
server-logged as well as self-report measures of Internet use. Relationships between Internet use and
demographic characteristics are examined, namely race, education and
income. Ethnographic accounts of
participants' experiences with the Internet augment quantitative findings
regarding the frequency and nature of home Internet use.
Methods
Participants and Procedures
Participants were 117 adult
residents of a low-income, medium-size urban community in the mid-western
United States. In exchange for participating
in home visits, completing surveys, and allowing their Internet use to be
continuously server-logged, participants received a home computer, Internet
connection, and full technical support for the 18-month trial. Surveys were administered during home
visits at pre-trial, 1 month and 3 months, and completed online or on paper.
2
Demographic characteristics
of adult participants are presented in Table 1. Participants were primarily African American, female, never
married, working full-time, and earning less than $15,000 annually. The majority reported having some college
education or earning a college degree.
Average age of participants was 38.6 years old.
A sub-sample of adults from 30 of the 90 families
participated in 2-hour home interviews and observations at the interface. Interviews were conducted as a conversation
between interviewer and participant during which they explored the Internet
together. Interviews were recorded
unobtrusively using a small digital recorder and transcribed for content
analyses.
Table 1: Demographic characteristics of adult
participants in the HomeNetToo project (n=117)
|
Age: Mean:
38.57 years-old; Range: 19 years-old
to 75 years-old |
|
Sex: Male: 20% Female: 80% |
|
Race: African
American: 67% European American:
33% |
|
Income (annual household) |
|
Less than $10,000: 28% |
|
$10,000 - $14, 999: 21% |
|
$15,000 - $24.999: 26% |
|
$25,000 - $34.999: 17% |
|
$35,000 - $49, 999: 7% |
|
$50,000 - $75,000: 1% |
|
Greater than $75,000: 0% |
|
Education: |
|
8th grade or less: 4% |
|
Some high school but did not graduate: 10% |
|
High school graduate or Graduate Equivalency
Degree: 24% |
|
Some college:
49% |
|
College graduate of above: 13% |
|
Marital Status |
|
Never married:
42% |
|
Married, living with spouse: 25% |
|
Other (divorced, separate, widowed): 33% |
Measures
Internet use. A list of
server-logged measures of Internet use is presented in Table 2.3 For the analyses, use measures were divided
into two time periods: Time 1: 1-3
months; Time 2: 4-6 months.
Table 2: Server-logged measures of Internet use (per
day)
|
Time
on-line (minutes) |
|
# of
session (log-ins) |
|
#
of unique domain Web sites visited |
|
#
of e-mail messages sent |
|
#
of e-mail messages received |
|
#
listserv messages posted |
|
#
listserv messages received |
|
#
of newsgroup postings |
|
#
newsgroups read |
|
Total time in chats |
|
#
of chats visited |
Survey Measures. Participants
completed surveys at pre-trial and 3 months that included the following measures
considered in this report: 1) Prior Internet experience (e.g., How would
you rate the extent of your experience with the Internet? 1=no experience, 5=a
great deal of experience.); 2) Self-reported Internet use (e.g., hours online,
previous week); 3) Uses of the Internet (e.g., communicating with family); 4)
Demographic characteristics. In addition, a survey administered at 1 month
contained the self-report Internet use measures (e.g., hours online during the
previous week).
Ethnographic Measures. Content analyses of
the 2-hour home interviews were guided by an interest in identifying how these
adults made sense of the Internet as it intersected their already established
and ongoing lives.
General categories of "sense-making" were identified.
Results
The frequency and nature of Internet use based on
server-logged measures are summarized in Table 3. Omitted from the table are measures of
listserv, newsgroup and chat activities, which were essentially zero. Thus, the Internet
activities of these first-time home users focused on Web activities and, to a
lesser extent, e-mail.4
As indicated in Table 3, participants initially spent an average of 41.51
minutes/day online in a single session, visiting about 9 domains. Time online did not
change significantly from Time 1 to Time 2. Sessions became somewhat shorter, and the
number of domains visited became somewhat greater, but these differences were
not statistically significant.
E-mail activity during the first 6-months of home Internet user was
infrequent.
During both time periods, participants sent about 3 e-mails per week.
Also evident from Table 3 is the high variability in Internet use among
project participants.
For example, 25% of participants spent essentially no time online,
whereas another 25% spent over 36 minutes/day online. About half the
participants never used e-mail.
Table 3: Internet use
|
|
00 |
Time 1 | ||||
|
|
00 |
Time online (minutes) |
#
of sessions |
#
of domains visited |
#
of e-mails sent |
#
of e-mails received |
|
Mean |
00 |
41.51 |
1.00 |
9.05 |
.39 |
1.78 |
|
Std.Dev. |
00 |
87.79 |
1.47 |
13.40 |
1.07 |
3.97 |
|
%
tile |
25 |
2.45 |
.09 |
.66 |
.00 |
.12 |
|
|
50 |
12.25 |
.47 |
3.48 |
.02 |
.41 |
|
|
75 |
36.47 |
1.31 |
11.14 |
.22 |
1.22 |
|
|
00 |
Time 2 | ||||
|
Mean |
00 |
43.53 |
.74 |
10.94 |
.36 |
3.31 |
|
Std.Dev. |
00 |
96.15 |
1.03 |
17.06 |
1.36 |
8.11 |
|
%
tile |
25 |
1.21 |
.04 |
.50 |
.00 |
.14 |
|
|
50 |
12.70 |
.34 |
4.24 |
.00 |
.47 |
|
|
75 |
46.25 |
0.98 |
13.24 |
.22 |
2.14 |
Note: Values are averages/day. All measures were
automatically server-logged. Std.Dev.=standard deviation.
The correlation between self-reported time online and
server-logged measures was .49. The correlation between self-reported and
server-logged number of sessions was .71. These findings are comparable to those of the
HomeNet project (.55 and .42) and, taken together, suggest that self-report
measures of Internet use are moderately reliable.
Participants' self-reported Internet activities are presented in Table
4. Consistent
with server-logged measures, none of the activities was reported as
"frequent."
Five activities were engaged in "sometimes:" getting information about
interests/hobbies, getting information about a product, e-mailing friends,
playing games, and listening to music. Note, however, that the latter two activities
may be computer rather than Internet activities. The least frequently engaged in
activities were viewing pornography, creating a Web page and getting job
training.
Table 4:Mean level of Internet activities
|
Type of Activity |