Measuring political information: novel method for estimating the information content of political communication

Communication is one of the fundamental dimensions of both domestic and international politics. How much information is transmitted among political agents, and what information, is a crucial determinant of political outcomes in such realms as negotiation and strategic interaction, reputation and trust-building, collective identity formation and community building, or institutionalized cooperation. In this paper, we propose a novel method for estimating information content of communication in politics, based on the cornerstones of information theory. By doing so, we provide a highly intuitive and applicable, yet theoretically and methodologically rigorous tool for measuring information flows in politics. Our framework applies and further elaborates the formalized treatment of communication developed by C. Shannon. His concept of message entropy, a key building block of information theory, provides a convenient way of estimating how much information is transmitted through communication, taking into account the prior levels of uncertainty on the part of the communicating agents. In this framework, transmitted information equals to uncertainty which the message helps to dissolve. Information flows, therefore, do not depend primarily on the quantity of signals communicated, but instead on the unique, unexpected information the signals carry. As a result, less voluminous exchange of messages may carry, and often carries, higher volumes of information than floods of uniform signals. Our adaptation of the original framework enables us to include in the analysis various features of communication that scholars in politics typically need to consider, such insincere communication, misunderstandings, political use of language, and communication channels limitations. We demonstrate the applicability of the framework with the use of a large original dataset mapping the flows of political information in online media worldwide. The dataset is based on automated analysis of the content of more than 20 000 online media outlets from close to 200 countries, over the years 2019 and 2020. For this study, several country dyads are selected and a probabilistic sample of media-carried communications among them is drawn from the more than 20 million articles in the given period. Furthermore, we assess the prominence of media outlets with the use of web traffic data and apply automated content translation to a sample of non-English texts. Thanks to this robust data infrastructure, we obtain a uniquely rich and comprehensive picture of political information flows across these countries in online media. This enables us to provide an estimate of the information flows concerned, but also to demonstrate the advanced features of our methodological approach. The analysis presented in this paper is motivated particularly by our observations of international communication patterns concerning Covid-19 in the early period of the pandemic, especially in February 2020. Our data mapping the content of online media in that period, collected originally for a different purpose, show a pronounced gap in information flows concerning China and Covid-19 in the critical period before the massive spread of the disease worldwide. Our paper highlights how the framework presented here can shed light on this important dysfunction in global communication flows.