Notes on the Diffusion of the Internet
Bruce Bimber
University of California, Santa Barbara
July 22, 2003
Virtually
everyone with a claim to make about the social, political, or economic consequences
of the Internet must confront questions about the growth of the technology.
How many people now use the Internet and how many will in the future? This
question is basic to the marketing of products online, to concerns with
digital divides, to analysis of the social consequences of online behaviors,
to discussions of the prospects for the Internet to alter how democracy
works. Until Internet growth reaches an asymptotic plateau, as broadcast
television and radio did long ago, most analyses of the Internet as a socio-economic
or political phenomenon will have to attend to its changing population of
users. Following are some of the basic characteristics of Internet growth.
Most observers agree that the diffusion of the Internet to individuals is
following a path of the form 1-1/e^t, where t is some function of time.
Many earlier media, with the notable exception of telephones, diffused in
this way. Radio and television exhibit the most classic form of this diffusion.
Figure 1 shows the growth of newspaper circulation per household (in blue,
using the right axis), households with telephones (in red), radio (in light
blue) and television (in orange).
Figure 1. Diffusion of Mass Media in the U.S.
See Bimber 2003 below for sources.
Reliable assessments of the expansion of the Internet at the user-level require surveys based on random samples of the human population. Inferences drawn from counts of host computers or accounts of Internet service providers or other technology-side measures do not provide accurate information about human individuals, though they can provide an overall indicator of the business or complexity of the Internet. (Unfortunately no surveys are available prior to the mid-1990s, so figures prior to this time can only be estimated. See Clemente 1998 for an example.)
Survey-based efforts at measuring the expansion of the Internet at the user level vary somewhat in their results, though all scientific efforts produce a similar overall picture. Variation in question wording and definitions - as well as survey quality - mean that individual estimates of the extent of the Internet measured by number of users in the U.S. vary by as much as 12 to 15 percentage points from one another.
Figure 2 displays findings from six studies, each of which includes estimates of Internet use for at least three years. (Because of differences in measures, one-shot studies reporting a level of Internet use for a single year are not particularly helpful.) The vertical axis measures fraction of adults in the U.S. who are Internet users, though definitions and standards vary by study. No attempt is made here to combine these findings into a single curve, and readers are cautioned against over-reliance on any one analyst's attempts to provide a single definitive measure.
The most important aspect of these studies is the period since 2000, when the data show a roll-off in the Internet growth rate. Data from UCLA, Pew, Nielsen, and Bimber all show that the rapid growth of the Internet in number of users between 1995 and 2000 is not matched in the period since 2000, though these studies vary in their estimates of the size of the roll-off. This change in the slope of the curve may represent a temporary stall or may indicate that Internet diffusion is now approaching an asymptotic value at around 60% +/- 10% of the adult public in the U.S.
See individual studies for analysis and speculation about these trends.
Figure 2. Fraction of U.S. Adults who Use the Internet
