Measuring Usage and Satisfaction
Cable Modems and the Internet
Institute for Information Policy
Penn State University
University Park, Pennsylvania 16802 USA
Some of the most appealing Internet applications driving household adoption are bandwidth-intensive -- multimedia Web sites, videoconferencing and the downloading of software. Even at 56k transfer rates, the wait can frustrate users. Internet access via cable modems can be up to 500 times faster than with a 56k modem. Though not yet widely available, cable modems seem to have the potential to reduce frustration among household Internet users, to invite greater consumption and to improve satisfaction with the on-line experience. This article excerpts the results of a 1997 doctoral dissertation which investigated factors influencing household cable modem adoption and subsequent effects on Internet usage and satisfaction. The excerpts presented here focus on the measurement scheme for usage and satisfaction, which may be applicable to other studies of Internet usage, including digital libraries and electronic publishing.
Cable modem adopters use the Internet more, use more of its applications and use it to satisfy more needs. Not surprisingly, cable modem adopters are significantly more satisfied with the Internet than are their slower dial-up access counterparts. The overall implication is that cable modems are likely to live up to their potential -- they can lead to a very satisfying Internet experience in a household setting.
Table of Contents
- Data Collection
- Results & Discussion
The purpose of the study  from which this article is drawn was to investigate the factors influencing cable modem adoption in households and subsequent effects on Internet consumption and satisfaction. The research questions were, at bottom:
- Who adopts cable modems?
- How does a cable modem affect how the Internet and other media are used? and
- What is the impact of a cable modem on a user's satisfaction with the Internet?
The objective of this article is to describe both the measurement scheme for consumption and satisfaction and the attendant results. The description shows that Internet consumption and satisfaction are complicated, multi-faceted constructs. Further, the benefit of measuring them carefully is the capture of rich and revealing data that, in turn, scholars, practitioners and policymakers can use to refine our understanding of the broad and significant impact of Internet household usage. Moreover, the measurement method, particularly as it applies to characterizing user satisfaction, may be applicable to other domains, such as digital libraries and electronic publishing, in which satisficing (as opposed to optimizing or maximizing) is an important criterion.
2. Data Collection
A random sample survey (n=256) of both households with slow, dial-up Internet access (2400 to 33,600 bps) and others with cable modem Internet access (4 to 10 mbps) was conducted in November, 1996. All survey respondents lived in the same small, Midwestern market where cable modems had been available for three years, longer than almost anywhere else.
The survey instrument was designed to measure Internet consumption and satisfaction in multiple ways. For consumption, there were three approaches to measurement. First, the amount of time spent on-line was measured as was money spent on access. Second, respondents were asked how many "parts" of the Internet they used: E-mail, Usenet, Web, FTP, IRCs and MUDs, videoconferencing and teleconferencing.
Third, the intensity with which respondents used the Internet to complete more than 20 different tasks was measured. For example, subjects were asked how intensively they used the Internet for specific kinds of entertainment, education, socializing, shopping, research, information and software retrieval, time- management, telecommuting and even marketing goods and services.
Finally, consumption was measured for time and money spent on other media as well: newspapers, magazines, TV, home video, and the telephone.
For satisfaction, there were 18 different measures, from satisfaction with email, FTP and the Web to speed, reliability, cost and customer service.
Responses from the two groups were compared and statistically analyzed as part of the test of a larger theoretical model, much of which is beyond the scope of this account. However, analysis of just the consumption and satisfaction variables tells a very interesting story.
3. Results & Discussion
The first step in analyzing the results was to reduce the number of variables to their common factors. In the case of the consumption variables, the more than 20 Internet "tasks" were reduced to six underlying factors via principal components factor analysis. These six factors were "Acquisition of Information Uses," "Acquisition of Things Uses," "Diversion Uses," "Sociability/Friends & Family," "Sociability with Strangers," "Marketing/Work Uses," and "Time Management/Work Uses."
When factor analysis was applied to the 18 satisfaction measures, however, no underlying themes emerged. This implies that satisfaction is a complex, multi-faceted concept.
Next, t-tests and chi square analyses were generated to compare the two groups. A number of statistically significant differences emerged that are related to cable modem adoption. Both in terms of time spent on-line and money spent on Internet access, cable modem subscribers spent more. On average, 2.22 hours per day, versus three quarters of an hour for the slower access group (t=-5.712, p<.001).
This is a very meaningful difference! It goes to show that faster access does NOT lead to less time spent on-line as might be expected if the principal problem was slow downloading. Apparently, the increased speed leads to more efficient use of more Internet. Other consumption measures bear this out.
Nielsen , Simmons  and other researchers have been reporting for some time now that Internet adoption is related to reduced time spent with television. This study takes that trend one step further. Cable modem adopters spent, on average, 90 minutes with television the previous day while their dial-up modem counterparts spent 140 minutes (t=2.36, p<.05). Oddly, however, there seems to be a slight tendency among cable modem users to spend more time playing video games, reading more newspapers and magazines, ordering more PPV and renting more videos, but spending less time on the phone than the slow modem comparison group. The cable modem group seems to use more and a greater variety of media, except for TV and telephone. Thus, it is tempting to speculate that digital and print are complementary, at least among information intensive users like readers and researchers.
Table 1 shows the frequency of use of different "parts" of the Internet by modem type. Households with cable modems were significantly more likely to use the Web, the Usenet, FTP and tele-/videoconferencing, but not more likely to use email, or IRCs and MUDs. If the greater access capacity of cable modems improves data-intensive Internet "parts," then one would expect more frequent Web, FTP and conferencing use. Email, IRCs and MUDs are text-based and therefore would not be improved with speed. The survey results support this. However, since the Usenet is mostly text-based and asynchronous, speed would not improve it; it is therefore difficult to justify a causal link between cable modem adoption and use of more data-intense applications. Perhaps the two are jointly related to some attribute of the adopter.
The comparison of means for the two groups with respect to the six factors (categories) of Internet Tasks tells an interesting story. Table 2 shows the means, t-tests and significance levels for acquisition, diversion, work and sociability uses of the Internet. Cable modem adopters on average used the Internet more frequently for all these tasks. However, the results were meaningful and statistically significant for only the acquisition and work (including telecommuting) categories. This is very interesting, however, since it suggests that activities such as shopping, conducting research, downloading files and software, and performing a wide variety of job-related tasks are improved with high speed Internet access. Perhaps cable modems allow easier telecommuting.
Satisfaction, as mentioned earlier, appears to have several distinct facets. Table 3 lists the 18 different satisfaction-related items as they were measured on a five-point likert-type scale, and in the case of "overall satisfaction," on a 10-point scale. Table 4 gives the t-tests and significance levels. Cable modem users were more satisfied on 15 of the 18 facets (12 of them at statistically reliable levels) though the mean differences between the two groups is not very great.
In only one area, cost of service, were the dial-up users more satisfied, hardly a surprise for two reasons: A large number of the survey respondents received FREE Internet access from a local university or employer, and the local cable company charged nearly three times as much as AOL or Compuserve for Internet access.
When asked how satisfied a user was with the Internet as a way to "communicate," "learn new things," "be more efficient," "get information," "shop," and "work from home," the cable modem users were more satisfied than the slow modem group. Not surprisingly, they were more satisfied with speed. A bit of a surprise, the cable modem users were more satisfied with customer service which, the reader is reminded, was supplied by the local cable company, long the poor service leader.
Overall, it seems clear that cable modem users use the Internet more and are more satisfied with it than household users with slow access. It would be too great a leap to suggest that cable modems cause these user perceptions, but it is fairly safe to say that speed is positively related to use and satisfaction.
Cable modems are hot at the moment; they offer the promise of more and better Internet. Particularly for households, where Ethernet connections are generally impossible, cable modems open up a wide range of cyber-possibilities. On its face, the interesting story here may seem to be the differences in use and satisfaction that are related to cable modem adoption. (The conclusions stop short of inferring any causal links since the research design does not permit this.) However, of greater potential value is the lesson on measurement and research design. If this study had been done asking only a narrow set of consumption and satisfaction questions, many insights would have remained hidden. This example is offered with the suggestion that future social science inquiries on Internet use include even finer, more discrete measurement of variables.
 Hoag, Anne (1997). Speed and the Internet: Factors Influencing Choice of Access Capacity and Subsequent Effects on Media Consumption. Ph.D. dissertation, Michigan State University, East Lansing.
 Nielsen (1997). "Startling Increase" in Internet Shopping Reported in New CommerceNet/Nielsen Media Research Survey. In Nielsen Market Research (Ed.) World Wide Web www.nielsenmedia.com.
 Simmons Market Research (1997). Consumer Online Usage Study. New York: Simmons Market Research.
Copyright � 1998 Anne Hoag
Search | Author Index | Title Index | Monthly Issues
Previous Story | Next Story
Comments | E-mail the Editor