Home Research Reports Researchers Affiliates Participants Page

TABLE OF CONTENTS

1. HOMENET
2. Research model
3. Homenet is distinctive
4. Project implementation
   1. A diverse population
   2. Equipment
   3. Reducing barriers
   4. Data collection
5. EARLY FINDINGS
   1. Interest in the net is strong, but cloudy
   2. The net is still too hard to use
   3. Use is vigorous
   4. Use is varied
6. MORE TO COME ...
7. PROJECT TEAM
8. CORPORATE PARTNERS

HOMENET

Figure 1: Percentage of US households with personal computers by time and income. (Data from the Current Population Survey)

HomeNet is a research program at Carnegie Mellon University that tries to understand the public's use of Internet-based electronic services. We are carefully studying how members of households actually use such services. Our research goals are to model the motivations for and barriers to using electronic services and to identify how their use changes people's lives. HomeNet placed its first families on the Internet in mid-February, 1995.

Current initiatives to offer on-line services are based on surprisingly little information about people's preferences. Almost all of the systematically collected, non-proprietary information about the use of electronic services has focused on use in businesses and universities. Most studies over-represent the higher income, better educated young males who are now the dominant users of the Internet and commercial on-line services. As Figure 1 shows, richer Americans are much more likely to have a home computer than poorer Americans, and this gap has been increasing over time.

As a result, there is a desperate need for studies in residential settings with average citizens of both sexes, to understand how to build, deploy, and make on-line services valuable to households. Similarly, the rapid growth of the Internet and on-line services has been accompanied by very little research on social impact.

The HomeNet project attempts to fill these gaps by combining detailed data and analyses to identify both preferences for and impact of on-line services. Our analyses will help companies and public institutions across the country in their decisions to provide on-line services. It should inform the policy debate about the value of these services, as well.

In the HomeNet project families in four diverse Pittsburgh neighborhoods are using Internet-based communication, entertainment, and information services. The project has developed a service architecture and operational procedures that allow detailed but confidential monitoring of usage, usability, and social impact.

RESEARCH MODEL

We believe that what influences use will change over time. Initially, drivers of Internet use include 1) individual differences among participants, such as their innovativeness or their demographic characteristics, 2) the behavior and attitudes of others in their households, schools, and workplaces, 3) characteristics of the services themselves including their degree of interactivity, their ease of use, and whether they provide access to information, entertainment, or communication, and 4) the competition from other media and activities, including stand alone computer applications, TV viewing, newspapers, and visits with friends. We expect that these factors will influence both the perceived utility and costs of using the Internet, and subsequently its use. We measured these drivers, before the field trial started and continue to measure them throughout its operation.

We expect that people's early experience with the Internet and the experiences of others around them will influence their subsequent use. Thus, we will focus on time-based models, since patterns of use are likely to change over time and outcomes are likely to feed back.

In addition to predicting use, we are attempting to understand the outcomes that result from using the Internet. These outcomes are both immediate (e.g., the satisfaction, enjoyment, or other states that users experience while they are using the service) and longer term (e.g., the potential broadening of interests and social networks users might achieve by roaming the networks).

Figure 2: Model of Internet Use

HOMENET
IS DISTINCTIVE

Second, our research strategy is to simulate a time when personalized, on-line services will be as ubiquitous and easy to use as the telephone or cable TV. In the HomeNet project, families receive a personal computer, a high speed modem, a dedicated telephone line for connecting to the Internet, and a software configuration that allows them to use Internet services without learning the details of any computer operating system. Their Internet services are personalized, so that they receive information tailored to their personal interests, as well as more generic electronic communication services.

Third, with our detailed data collection, we can describe how people's motivations for and use of the Internet and its impact on their lives changes over time. By combining automated measurements, longitudinal surveys, and in-home interviews, we capture a rich history of how each participant in the project adapts to the on-line world.

PROJECT
IMPLEMENTATION

Race                       76% white                  24% minority               
Gender                     43% male                   57% female                 
Age                        42% < 19                   58% >= 19                  
Family income              25% < $35,000              75% >= $35,000             

A diverse population

By selecting school newspaper staff, we hoped to ensure that at least one person in each family had some reason to use the Internet: to communicate with other newspaper staff and to research articles. We also hoped that the common bond would give students from different schools something to talk to each other about.

We have discovered since an additional benefit of including journalism students in the sample. Some of them are attempting to put their newspapers and other information on-line. They are information providers on the Internet, not just consumers. (See the Langley home page, that one student created after using the Internet for less than three weeks.)

Equipment

The families were given a choice of using the computer at no charge for the duration of the project, or buying the computer at approximately half price. A payment plan was offered to make the purchase easier and to allow the project to select participants independent of their income.

Reducing barriers

Once they set up their computers, they were able to connect to the Internet through three mouse clicks, without needing to learn details of the Macintosh operating system.

To allow customization of the content of their Internet services, each family member filled out a form describing their interests. This information formed the basis of a customized "home page" for each family that pointed to resources on the World Wide Web about their interests. For instance, one family of sports fans found football statistics and a family of religious Jews had Old Testament commentary when they launched Netscape (the World Wide Web browser installed on their machines). Most family members were excited to find information tailored to their interests available from their computer.

We also conducted training sessions on the days that the families picked up their computers. These were 3 hour classes, divided into sections on 1) introductory Macintosh skills (using the mouse, putting things in folders, etc.), 2) using electronic mail and reading news groups, and 3) browsing the World Wide Web.

The object of the training was to get people comfortable enough with computers and with the Internet so that they would not be afraid to log on themselves at home. Although the training was not mandatory, a great majority of the participants attended the class.

After the training, we offered families three kinds of help when they had computer troubles. First, they could call a help desk staffed by students working on the project. Use of the help desk has dropped off with time, but still averages several calls per evening. Second, they could send email to a designated address and receive an electronic response from one of the project team. They could also post questions to a news group accessible to all project participants, where they could get help from peers as well as paid staff.

Although the participants have used the help desk over the other options in the first month of operation, we believe their patterns of help seeking will change with time.

Data collection

Computer generated usage records. Custom designed logging programs track the project participants' actual use of the Internet. Confidential records are kept of how much email subjects send and receive and to whom, what news groups they read and how often, what World Wide Web sites they visit, and how much time they spend logged on to the Internet. We also capture the programs they run locally on their personal computers. From this data, we'll be able to answer questions such as "Is sophisticated use of stand alone applications a prerequisite for frequent Internet use?" or "Do people who use the Internet to get information also use it for communication?".

Questionnaires. Each family member participating in the project filled out a detailed pre-study questionnaire, which asked about things like their attitudes towards and previous experience with computers, the other media they use, their use of time, their innovativeness, their social networks, and their demographic characteristics. Each month users will be filling out shorter questionnaires about what they use the computer and the Internet for, how this use makes them feel, how valuable they perceive this use to be, what services they predict they will use in the future, and related questions.

Public messages. We have established a number of news groups for the HomeNet participants, and are keeping archives of all messages posted to them. From this record, we will be able to reconstruct the evolution of some on-line communities.

Interviews. Each family has agreed to be interviewed in their home during the project. Our goal is to obtain a rich qualitative understanding of the impact of the system on participants' lives. How do they integrate electronic services into their lives, if at all?

Help logs. Each time the help desk personnel respond to a caller, they write a detailed log about the problem prompting the call and how it was solved.

EARLY FINDINGS

Interest in the net is strong, but cloudy

When given on option to use the Internet with a borrowed computer or to purchase their own on an installment plan at approximately 50% of current retail prices, half of the families decided to buy. Surprisingly, richer families were less likely to buy, presumably because many of them already had computers at home.

The net is still too hard to use

Potential Internet users have a wide range of skills and experiences. Among the HomeNet participants, a few families already had computers with modems, while other people had never touched a keyboard before volunteering for HomeNet and did not know how to roll a mouse. Inexperience, coupled with common physical disabilities, such as poor eyesight or hand tremors, made even a relatively easy-to-use interface to the Internet difficult for some potential users.

As one participant plaintively wrote, asking for advice, "Are any of the other adults in the thing having as much trouble getting to use, or learning to use, this #$@*&# computer as I am?" Another responded, "I am frustrated too!! This computer just doesn't like me!! I can't get it to do what I want it to."

Unreliable software and hardware makes the difficulty of using the Internet even greater. Software companies are releasing new versions of their software with significant bugs in them. Hardware companies, such as modem manufacturers, are upgrading their equipment to give it greater capabilities, but they do so at the expense of precise compatibility with older versions already in the field. The incompatibilities with the installed base of equipment raises problems both for service providers and consumers.

When the software or hardware breaks, some inexperienced participants attributed the failure to themselves rather than to the technology. "How can I be so stupid?" was their response to buggy software.

The participants required a moderate amount of help. The 48 families had called the help desk 122 times over a 6 week period, or approximately 1 call per family every 3 weeks.

"Getting started" problems were a major cause for contacting the help desk. People had troubles connecting to the Internet for a wide variety of reasons -- bad telephone lines and busy signals, broken modems, illegal passwords because a `CapsLock' key was mistakenly hit, or because it was forgotten, erased login scripts, and many more. When confronted with these problems, they had few skills in diagnosing them.

Connecting to the Internet requires that people correctly navigate a complex sequence of actions, from buying and setting up a computer, to estabilishing telephone service, to learning application software, to defining and remembering a password. While these procedures were automated for HomeNet participants as much as possible, the resulting methods were somewhat brittle. In addition, HomeNet participants didn't have robust models of how components of this action sequence operate, including the telephone network, their modems, their computers, their software or the Internet itself. As a result, when something went wrong, they were often unable to isolate the source of the problem. Their problems were more likely to be solved quickly if they had other family members or friends who were more sophisticated than they or if they felt comfortable revealing their ignorance to the strangers on the help desk.

The unreliable software and hardware discussed previously was the other major reason that participants called the HomeNet help desk.

Use is vigorous

Figure 3: Internet users per day

While Carnegie Mellon University operates 110 phone lines for about 2000 students and staff who access the Internet and the campus network by modem, HomeNet's 10 phone lines soon became inadequate and additional modems had to be installed to handle the demand from its 90 active users per week. It is still too early to judge, though, whether this high level of use will be sustained.

As Figure 4 shows, use varies by time of day and day of the week. On weekdays, use starts to become substantial in late afternoon, when students return from school, and peaks in the early evening, during the prime-time TV hours, when both students and parents have free time. On weekends, people spread their use of the Internet more evenly during the day, although their heaviest hours of use are still the prime-time hours. Surprisingly, adults do not use the Internet more heavily on weekends than on weekdays.

Figure 4: HomeNet Use by Time of Day

Students and parents use the Internet differently. The students log on to the Internet about twice as frequently as their parents, with a typical student going on-line every other day.

Unlike the early days of television, when viewing was a family activity, the household computer tends to be a one-person-at-a-time appliance. This means that students and parents compete for machine time. As one parent said, "The biggest problem I have is trying to beat the kids to the system. It seems when I have time to sit down one of them has it in use."

Despite this competition for machine time, Internet use is communal in the sense that use by one member of the family makes it easier for others to get on-line as well. Our first observations in the household were of young teenagers configuring the home computer for the rest of the family.

All is not cooperation in the family, however. Traditional conflicts across generation and gender crop up in the the world of computers as well. One adult, for example, was seeking help from other adults, because "if I do get my hands on [the computer, my daughter] hovers over me and punches buttons in front of me and I get lost...... I'd like to hear from someone over 30... I need to leave because my daughter is sipping her hot chocolate LOUDLY in my ear and correcting my spelling." Later analyses will describe this dynamic in more detail.

Use is varied

Sometimes the insults can get out of hand, as students attempt to defend their conversational turf from was what they termed were "invaders" from other schools. The students have asked for controlled-access communication areas exclusively for the use of their local high school. This desire for sheltered communication areas that reinforce local bonds is likely to be common.

Despite what appears to be some xenophobia, the students were also more likely than their parents to meet "strangers" -- students from other area high schools -- on the net, and to use these strangers for social support. For example, teenagers from different high schools exchanged information about the easiest places in the Pittsburgh area to pass the exam for obtaining a driving license.

While most Internet use so far has been to exchange electronic mail and to read electronic bulletin boards, students have attempted to put their newspapers on-line and have publicized their schools by creating home pages.

MORE TO COME ...

In its first year, we expect to study two major issues. First, we will learn the motivations and barriers to using on-line electronic services. Second, we will explore how satisfied the participants are with their use of the Internet and how their experience may change their social life.

The HomeNet platform will help us investigate a range of important issues in the future. We list only a few of these questions here: How does easy access to price and product information through the net affect people's buying habits? Will the easy availability of new people and information on the net change people's social networks and their interests? Do the groups that form on the net have the same social consequences of face-to-face groups? We will also conduct service and pricing experiments using our platform in the future. In addition, we will study alternative software architectures for community services that both support a broad range of capabilities and address the barriers to scale-up.

The HomeNet project creates an urban laboratory to measure actual usage of on-line services by a diverse group of people over an extended period of time. With an increase of the pool of participants and an extension of the length of the trial, we believe this laboratory will produce a wealth of findings not possible to generate by any other means.

PROJECT TEAM

Sara Kiesler is Professor of Social and Decision Sciences. Over the last 15 years, she has studied how computing and computer networking changes group dynamics and communication in organizations. She is co-author of Connections: New Ways of Working in the Networked Organization (MIT Press), has published more than 45 research papers on electronic communication, and has served on five National Academy of Sciences panels concerned with technology.

Robert Kraut is Professor of Social Psychology and Human-Computer Interaction. He was previously Director of Interpersonal Communications Research at Bellcore, where he supervised a multi-disciplinary study of residential telecommunications needs and designed and managed a field trial for studying the use of video telephony. He has published widely on the topic of the social impact of new technology, served on relevant national committees, and is associate editor for two journals in the area.

Tridas Mukhopadhyay is Associate Professor of Information Systems at the Graduate School of Industrial Administration at CMU. His research is in the economics of information technology. He has worked with leading corporations and agencies, including General Motors, Texas Instruments, LTV Steel, and the United States Postal Service, in assessing the business value of information technology.

William Scherlis is a Senior Research Scientist in the Computer Science Department at CMU, doing research in information management and software technology. He was previously at the Advanced Research Projects Agency, where he managed research programs in software technology and helped lead the initiation of the federal high performance computing program. Scherlis is on the steering committee for a National Research Council study related to high performance computing and information infrastructure.

Jane Aronson is HomeNet Project Director. She recently completed a Ph.D. in philosophy at Stanford University, where she studied analytic philosophy, and wrote a dissertation about levels of meaning in interpersonal communication. She has developed World Wide Web pages for both universities and industry.

CORPORATE PARTNERS

HomeNet Contact Information:

Vicki Lundmark,
Social and Decision Sciences
5000 Forbes Avenue
Pittsburgh, PA 15213
412-268-7505 (voice) lundmark@andrew.cmu.edu