Working Paper: NBER ID: w22744
Authors: Shane Greenstein; Yuan Gu; Feng Zhu
Abstract: Do online communities segregate into separate conversations about “contestable knowledge”? We analyze the contributors of biased and slanted content in Wikipedia articles about U.S. politics, and focus on two research questions: (1) Do contributors display tendencies to contribute to topics with similar or opposing bias and slant? (2) Do contributors learn from experience with extreme or neutral content, and does that experience change the slant and bias of their contributions over time? Despite heterogeneity in contributors and their contributions, we find an overall trend towards less segregated conversations. Contributors tend to edit articles with slants that are the opposite of their own views, and the slant from experienced contributors becomes less extreme over time. The experienced contributors with the most extreme biases decline the most. We also find some significant differences between Republicans and Democrats.
Keywords: Wikipedia; Ideological Segregation; Online Collaboration; Political Bias
JEL Codes: L17; L3; L86
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Contributors' article slant (D30) | Contributors' editing behavior (D16) |
| Contributors' political viewpoint (D72) | Editing behavior towards opposite slant (Y60) |
| Contributors' interaction with extreme content (D16) | Contributors' bias (J15) |
| Contributors' years of activity (B31) | Contributors' slant (F01) |