Epidemic Diseases and Newspaper Reporting Behavior

The recent pandemic highlighted the importance of understanding media reporting behavior with respect to mortality fluctuations. This paper investigates the relationship between mortality by cause and related newspaper reporting in a historical setting with many high-incidence epidemic and non-epidemic causes of death: the United States, 1900-1909. I use measures of disease mortality share which do not require population data but reflect the relative risk of death from each disease in a county, urban area, or rural portion of a county. For a sample of eight epidemic diseases, I find a clear positive association between related reporting and each of: (i) mortality share; (ii) bad news on mortality share; and (iii) outbreaks of disease. Bad news on mortality share is more strongly associated with newspaper reporting than good news. Urban and rural areas report on good news in different ways but otherwise respond similarly to fluctuations in epidemic mortality. By contrast, I do not find clear relationships for a sample of five non-epidemic conditions. Across epidemic diseases, the estimated relationship between mortality and reporting is heterogeneous but does not appear related to the overall percentage of deaths caused by the disease in the sample. I find the strongest evidence of a positive association for diphtheria, smallpox, typhoid, and measles. Finally, while death by cause and race data are unavailable, I use peaks in the relative overall mortality by race to find that outbreaks of epidemic disease coinciding with peaks in nonwhite relative mortality have increased related reporting in urban areas.