Working Paper: CEPR ID: DP17366
Authors: W. Walker Hanlon; Brian Beach
Abstract: Digitized historical newspaper databases offer a valuable research tool. A rapidly expanding set of studies use these databases to address a wide range of topics. We review this literature and provide a toolkit for researchers interested in working with historical newspaper data. We provide a brief description of the evolution of historical newspapers, focusing on aspects that are likely to have implications for the design of empirical studies. We then review the main databases in use. We also discuss some key challenges in using these data, most importantly the fact that even the most extensive datasets contain only a selected sample of the universe of historical newspaper articles. We offer tools for evaluating the comprehensiveness of available newspaper datasets, show how to assess potential identification concerns, and suggest some solutions.
Keywords: historical newspapers
JEL Codes: N01
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| exposure to news reported in newspapers (G14) | influence on social behaviors and economic outcomes (F61) |
| newspaper coverage (Y30) | public awareness and subsequent actions (H12) |
| historical newspaper data (N00) | measure treatment effects during the 1918 influenza pandemic (I12) |
| newspaper reports (Y50) | public compliance with health measures (I18) |
| directory data (Y10) | identification of selection biases in newspaper coverage (C83) |