Working Paper: NBER ID: w20681
Authors: Yosh Halberstam; Brian Knight
Abstract: In this paper, we investigate political communications in social networks characterized both by homophily–a tendency to associate with similar individuals–and group size. To generate testable hypotheses, we develop a simple theory of information diffusion in social networks with homophily and two groups: conservatives and liberals. The model predicts that, with homophily, members of the majority group have more network connections and are exposed to more information than the minority group. We also use the model to show that, with homophily and a tendency to produce like-minded information, groups are disproportionately exposed to like-minded information and the information reaches like-minded individuals more quickly than it reaches individuals of opposing ideologies. To test the hypotheses of our model, we analyze nearly 500,000 communications during the 2012 US elections in a social network of 2.2 million politically-engaged Twitter users. Consistent with the model, we find that members of the majority group in each state-level network have more connections and are exposed to more tweets than members of the minority group. Likewise, we find that groups are disproportionately exposed to like-minded information and that information reaches like-minded users more quickly than users of the opposing ideology.
Keywords: Homophily; Political Information; Social Networks; Twitter
JEL Codes: D7; D8
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
Cause | Effect |
---|---|
increased group size (C92) | greater exposure to information (D83) |
members of larger groups (C92) | disproportionately exposed to likeminded information (C92) |
majority group (conservatives) (J15) | more network connections (D85) |
majority group (conservatives) (J15) | exposed to more information (D83) |
retweets flow more quickly to likeminded users (Z13) | efficiency of information transmission within homophilous networks (D85) |
exposure to likeminded information (C92) | greater than baseline exposure (C22) |