A Taxonomy of Household Internet Consumption

Abstract

Despite an immense literature on Internet adoption and, to some extent, usage, there is still limited knowledge of how households consume information on the Internet. In this paper, we intend to add to this knowledge at a very basic level, by measuring and classifying Internet information consumption along time and space. Specifically, we aim to characterize household web surfing “types” according to the amount of time they spend online and how they distribute this time across different web domains. We plan to define these types according to the following dimensions of Internet use (or subset thereof): Number of days online per week, total time online per week, time per domain, and concentration of domains visited according to time and views. The last of these dimensions indicates whether a household’s visits across N domains are concentrated among just a few or more evenly spread across all N, as measured by time or visit frequency. Using these dimensions to partition the usage space, examples of types that would emerge include the “tourist,” who infrequently visits a handful of sites for a relatively short period of time, and the “lingerer,” who spreads a great deal of time online across a few domains.

We have assembled several years’ worth of data on web browsing behavior from ComScore to conduct this study. These data span the years 2008, 2009, 2012 and 2013, and track households over an entire year, recording all of their web browsing behavior on a home machine. The information collected includes the domains they visit, how long they spend at each domain, and the number of pages visited within the domain, along with several demographic measures, including income, education, and household size. Using these data, we can classify households along the aforementioned dimensions, thus identifying a distribution across web surfing types. Further, we can determine demographic predictors for each type; for example, we can measure the predictive power of income, education, household composition, etc. in determining whether a household is a “lingerer.” The methods employed to conduct these measures will be a mix of descriptive statistics and querying to establish the distribution over types, followed by standard linear regression and/or logit analysis for type prediction.

We have also amended the data to include a characterization of web domains visited by households as online video distribution (OVD) providers. Using this added information, we can determine whether a household consumes video via OVD, and characterize the general usage patterns specific to OVD consumers. This added measure can be informative toward better understanding consumers that are engaged in over-the-top (OTT) video consumption.

Keywords: Online content consumption, Household classification, Household online type distribution