Communication Chains: A Methodology for Assessing the Effects of the Internet on Communication and Travel

Although numerous researchers have investigated the impact on travel of specific telecommunications applications like telecommuting, only rarely has a broader look at the impact of electronic communications on multiple communications media, including travel, been attempted. This is no doubt due in part to the measurement challenges associated with such an attempt. Time-use diaries, activity diaries, or communication logs can provide opportunities for cross-sectional analysis, typically at a single point (or small interval) of time. However, these tools generally do not identify the chain of communication events cascading from a specific message, and thus are unable to capture behavioral linkages between events. Thus, for example, if a cross-sectional study using such a tool finds that greater Internet use is associated with more travel, it has not been established that the Internet use caused the travel; a third-party variable such as income, or a gregarious or variety-seeking personality, may be responsible for both effects separately. On the other hand, following communication chains in a precise and quantifiable way is a daunting task for both the researcher and the respondent. The methodology proposed in this study offers a practical middle ground between no data and perfect data on causal linkages. Specifically, for a given Internet activity, we simply ask the respondent to identify its causal antecedent and its likely communication consequences, by checking off the appropriate responses from a list. Obviously, by sacrificing precision and detail we lose the ability to quantify those consequences (e.g., in terms of miles of travel generated or substituted). However, we argue that the qualitative view of causal chains provided by this approach can offer valuable insights not possible from single cross-sectional studies alone. We analyze the relative strengths of effects of a given Internet activity in three ways. Using a 1994 sample of 148 early Internet adopters to illustrate the application of the methodology, we first tabulate the total presence, within the sample, of specific types of effects, to assess the net effect of Internet activities on other communication media. Second, we tabulate the frequency with which specific combinations of effects, or “pattern vectors,” occur in the sample. Finally, we use cluster analysis to identify latent types of communication chains having relatively similar causal relationships between Internet activities and other media. To establish a theoretical background and context for this study, we briefly summarize the relevant literature and develop a conceptual framework for understanding interactions among communications media. After introducing the survey instrument and methods of analysis, we apply the proposed methodology and discuss key empirical findings.