Pseudoscience Conspiracy Dies Hard: Evidence from the MMR-Autism Controversy in the United States 1998-2011
We combine data on individual-level immunization record with state-level information exposures to investigate differential impacts of information on health decisions by mother’s education level. From the National Immunization Survey 1998-2011, we obtain vaccination records and demographics for children aged 19-to-35 months. We also assemble state-level information exposures, including passive and active ones. For passive exposures, we obtain relevant disease prevalence rates, from the Office of Special Education Programs and various issues of Morbidity and Mortality Weekly Report, and news counts from LexisNexis Academic. For active exposures, we collect online search intensity from Google Trends. These exposures are further categorized into groups based on their attitudes, contents, and sources, which helps to identify heterogeneous responses to certain features of information.
By exploiting variations in differential responses to information by mother’s education, we find that strong biased beliefs among mothers with college education are responsible for the persistent non-uptake rate of MMR. Such beliefs cause asymmetric responses to new information based on its attitude and content, which in turn intensify the strength of the beliefs. One-standard-deviation increase in information exposures indicating vaccine may not be safe leads to 40% increase in the belief coefficient. In contrast, increase in information exposures encouraging for immunization generally results in small and insignificant impact on MMR non-uptake rate. Moreover, although authorities’ words in newspaper help to decrease the non-uptake rate, overall impact of main stream media is limited. In contrast, online search results are more influential to mothers when making immunization decisions for their children.
Our results provide empirical evidence on confirmatory bias. It is important to keep prudent when convey information to the public. The impact of an initial misinformation may take a long time to be fully addressed because people may suffer from such bias when processing new information. Moreover, the study also provides implications on how to efficiently and effectively communicate public policy, research results, or even science to the mass. By targeting at a large and more interested audience, web turns out to be a more effective medium than traditional newspaper for the authorities’ opinions to be heard.