How Crowdsourcing Makes Local Government Data Accurate and Cheap

🔎 What This Paper Tackles

Local governments dominate U.S. public administration: of the country’s 90,106 governments, 99.9% are local. These jurisdictions vary widely in institutional features, descriptive representation, and policy-making power, but political scientists have been slow to exploit that variation because comprehensive local data are often hard to obtain—unavailable or costly, difficult to replicate, and rarely updated.

🧭 How Data Were Collected and Compared

• Two independent crowdsourced data-collection projects were carried out to assemble local political information.
• Multiple measures of coder agreement (consensus measures) were evaluated to assess reliability across nonexpert coders.
• The crowdsourced outputs were validated against elite and professional datasets to benchmark accuracy.

📊 Key Findings

• Crowdsourced local data show high levels of accuracy when compared with elite and professional sources.
• Different consensus measures across coders can be used to gauge and improve data quality.
• Crowdsourcing produces data that are easier to update and replicate than many existing local datasets.
• Nonexperts can effectively collect, validate, and update detailed local political data.

💡 Why It Matters

Reliable, inexpensive, and scalable crowdsourced data remove a major barrier to studying the large and diverse universe of U.S. local governments. That opens new possibilities for research on local institutions, representation, and policy variation, and offers a practical route for maintaining up-to-date, replicable data on subnational politics.