As with the telephone, radio, and television, the dramatic growth
in household computing has the potential to change how people
live. Use of computers is growing rapidly in homes. By 1993,
a third of all US households owned a computer and over 60% of
the richest quartile did so. By 1996, a significant proportion
of these home computers were being used for on-line connections
to the Internet. One estimate puts usage at about 12% of households
in 1996. The Internet is a new household technology, but its
growth is rapid. If the public press is to be believed, the widespread
use of the Internet portends large changes in the way people use
home computers and in the social effects of using computers at
home.
HomeNet is a field trial at Carnegie Mellon University whose purpose is to understand people's use of the Internet at home. The trial philosophy is to reduce economic and technological barriers to use, so that we can learn how a diverse sample of families use the Internet for the first time. Starting in 1995, we have provided families with hardware and connections and are carefully documenting how members of family use on-line services such as electronic mail, computerized bulletin boards, on-line chat groups, and the World Wide Web. Currently 100 families are part of the trial. Through detailed audit of trial data on Internet use, quantitative surveys and interviews with family members, we are able to measure the demand for and impact of electronic communication and telecommunication services over time.
Our research program attempts to answer questions crucial to companies
and government agencies developing advanced services:
What makes electronic services valuable? The success
of electronic services will be based in part on their capabilities,
their price, and the style with which they are operated. Our
previous research has shown the crucial importance of offering
services that allow people to communicate with each other and
that connect them with a critical mass of people and information
sources. Yet many crucial questions are still unanswered: What
is the relative preference for communication, information, and
entertainment among households of different types and among different
members of a single household? How should electronic services
integrate with each other and with non-electronic alternatives?
What kinds of user interfaces will make the complex array of services
easy and fun to use?
How can one predict consumer demand? New information
services present problems in forecasting, because consumers are
unlikely to understand the value they will receive without substantial
experience. Moreover, because the value of the services will
vary with the number of users and information suppliers, demand
is likely to change over time. Detailed data about the use of
communication, entertainment, and information services over time
will allow us to develop, different models to predict drop out
rates and use, and relate these to willingness to pay.
What impact will the services have in the household and the larger community? Like the telephone, the automobile, and the television, new electronic services may change patterns of communication and time use in households and in communities, which in turn will influence how the services are used and valued. Examining how these services fit into household routines and how these routines are changed by the services requires the longitudinal data that HomeNet supports.
Before participants in our sample had actually used the Internet,
they reported positive attitudes towards the Internet but only
vague ideas of what it would be good for. A large minority did
not know what downloading or Email was. Given their vague beliefs
and knowledge, it is probably not surprising that many had difficulty
getting started. We expected that people with more computer skill
and motivation would be likelier to overcome these difficulties,
and would, in turn, be more likely to use the Internet frequently.
However, we thought that as everyone learned how to use the computer
and what the Internet could do for them, the influence of their
initial computer skill would decline with time. We were wrong.
Even after a year of experience with the Internet, participant's
initial computer skill still constrained their Internet usage.
This result held across different gender and age groups.
Over 70% of the households called the help desk. Calls to the
help desk represented the behavior of some of the more sophisticated
users. Less sophisticated users dropped out once they hit usability
barriers. The kinds of problems logged by help desk staff included
problems in installing phone service, configuring the telecommunication
software, busy signals (users often blamed themselves!), buggy
software, inexperience with mice, keyboards, scroll bars, terminology,
radio buttons, and menus. Yet, in our home interviews, we noted
there were many more problems participants had not called about.
These included confusion using the Macintosh icons and features,
trouble finding live sites on the Web with useful or fun information,
and difficulties using Email.
Table 2 shows averages for various measures of Internet use, broken
down by generation and gender. Teenagers are much heavier users
of the Internet than their parents. Among teenagers, males are
heavier users than females, while among adults, women were heavier
users than men, especially for electronic mail. Although not
shown in Table 1, whites were heavier users of the Internet than
blacks, especially for electronic mail.
| Internet Use Statistic | ||||||||
| Teenage boys | Teenage girls | Adult men | Adult women | |||||
| Percent of weeks logged in at least once | ||||||||
| Internet sessions | ||||||||
| Hours connected to the Internet | ||||||||
| Session length (minutes) | ||||||||
| Electronic mail messages sent | ||||||||
| Electronic mail messages received | ||||||||
| Usenet newsgroups messages sent | ||||||||
| Usenet newsgroups messages read | ||||||||
| Unique World Wide Web sites visited | ||||||||
| Total N | ||||||||
| Winsorized N | ||||||||
Table 1: Weekly Internet use by different demographic groups.
Note: To compensate for skewed data, entries are Winsorized means,
from the middle 80% of the sample. The proportion logging in at
least once was not Winsorized.
Participants in the trial visited large numbers of Web sites, but the modal Web site appealed to only one percent of the sample. This finding implies that beyond a few highly popular services, people are looking for, and finding specific ("niche") services that match their particular interests. Figure 1 shows the distribution of web sites visited by different proportions of the sample during the first year of the trial. Of the 9,912 unique IP addresses visited, 55% were accessed by only a single individual and fewer than two percent of the sites were visited by even twenty percent of the sample.
Figure 1: Popularity of different WWW sites.
Currently, interpersonal communication via electronic mail and
information acquisition via the World Wide Web are the dominant
uses of the Internet. Email is a relatively mature technology
(circa 1969). Compared to the Web, electronic mail is personalized,
spontaneous, and interactive; the content of a particular message
is usually tailored to the recipient and often takes into account
their prior interactions. In contrast to Email, the World Wide
Web is more analogous to broadcast media like billboards, magazines,
or television. Information services on the Internet are generally
posted in a public place and are available to anyone who happens
by. They are minimally interactive. If their content is tailored
at all, it is to broad audience characteristics, not to particular
people.
For our sample, interpersonal communications via Email was both
more popular and more sustaining than information acquisition
via the Web. Participants used Email in at least 49% of their
Internet sessions, but the Web in only 38% of them. Furthermore,
during sessions in which participants accessed both their electronic
mail and the Web, they accessed electronic mail first 75% of the
time.
Moreover, time series analyses show that use of electronic mail
is more stable over time than use of the Web. Participants who
used electronic mail heavily in one week were very likely to use
it heavily in a subsequent week. Web use was moderately stable,
but substantially less so than Email use. Use of the Web during
a week is partially driven by external events (like whether other
family members are using during that week), while use of electronic
mail is not. In contrast, use of electronic mail is driven by
the availability of conversational partners. (For example, Email
use drops during vacation periods, while Web use does not.)
Furthermore, Email seems to drive continued use of the Internet
more than does the Web. That is, the people who were heavy electronic
mail users one week were more likely to use the Internet for more
hours in the subsequent week, even after controlling for the hours
they logged onto the Internet during the previous week. In contrast,
heavy Web use in one week reduces the number of hours
that they spend the Internet in a subsequent week.
Figure 2: Influence of electronic mail use and World Wide Web use on subsequent Internet use
Finally, use of electronic mail rather than the Web leads to longer survival on the Internet. (See Figure 3.) That is, about 85% of people who used electronic mail more heavily than they used the Web were still using the Internet after a year'. In contrast, less than 70% of people who used the Web more heavily than they used electronic mail were still using the Internet after a year.
Figure 3 Influence of relative electronic mail use and World Wide Web use on Internet survival.
In summary, among the HomeNet sample, electronic mail use was
more popular than use of the Web, more stable, and drove continued
use of the Internet overall. One reason is that Email sustains
ongoing dialogues and relationships. In contrast, the Web has
more bounded properties, in whichinformation gathering, for example,
for school assignments, purchase decisions, or paid employment
is satisfied with one or a few visits. In the abstract sense,
this is an argument that the Internet is a social and emotional
technology, and that it sustains social networks.
Research papers and other information on the HomeNet project are available at:
http://homenet.andrew.cmu.edu/progress/research.html.
Kiesler, S., Kraut, R., Lundmark, V., Scherlis, W., & Mukhopadhyay,
T. (1997, March 24-27) Usability, help desk calls, and residential
Internet usage. Proceedings, Human Factors in Computing Systems,
CHI '97 (Atlanta, GA,), New York: ACM.
Kraut, R. E. (1996, Dec.). Internet@home: Introduction to the
special section on the new home computing. Communications of
the ACM, 39, 32-35.
Kraut, R., Scherlis, W., Mukhopadhyay, T., Manning, J., &
Kiesler, S. (1996, Dec.). The HomeNet field trial of residential
Internet services. Communications of the ACM, 39, 55-63.
Kraut, R. E., Scherlis, W, Mukhopadhyay, T., Manning, J., Kiesler,
S. (1996). HomeNet: A field trial of residential Internet services.
Proceedings, Human Factors in Computing Systems, CHI' 96. P284-291