With critical midterm elections approaching this fall, do voters approve of the job Congress and the president are doing? Feel better or worse about the economy? Rate the pandemic, race relations or climate change as key to their vote?

It’s been two decades since the federally funded American National Election Studies (ANES) – the preeminent national election survey since 1948 – posed such questions before and after a midterm to assess voter preferences and policy priorities.

This year, the Cornell Center for Social Sciences (CCSS) will introduce the 2022 Collaborative Midterm Survey, implementing an innovative proposal awarded $2 million by the National Science Foundation's American National Election Study Competition. Engaging multiple partners using a combination of new and longstanding methodologies, the proposal seeks to generate election insights while also advancing the science of survey research, potentially establishing a model for future surveys.

“This is going to be one of the go-to sources for understanding the midterm elections,” said Peter K. Enns, professor of government and public policy in the College of Arts and Sciences and the Cornell Jeb E. Brooks School of Public Policy, and the Robert S. Harrison Director of CCSS. “The data and findings will be broadly available and accessible to the public. The unique study design will inform the future of survey research for election studies and beyond.”

Enns is principal investigator of the Collaborative Midterm Survey, which will study U.S. House, U.S. Senate and gubernatorial races, with a special focus on voters nationwide as well as three politically important and demographically diverse states: California, Florida and Wisconsin.

Co-principal investigators are Jonathon P. Schuldt, associate professor in the Department of Communication in the College of Agriculture and Life Sciences, and executive director of the Roper Center for Public Opinion Research; and Colleen L. Barry, inaugural dean of the Cornell Brooks School.

Election polling and other surveys have been hampered by declining participation rates and rapid changes in technology that have made obtaining nationally representative samples more challenging. Asking questions face-to-face at randomly selected addresses has been long considered the “gold standard” but is the costliest method – a reason for the pause in ANES midterm surveys. Other strategies include group panels, landline and mobile phone calls, text messages and online surveys. Some are probability-based, others not, with widely ranging sample sizes.

With so much variation and often limited transparency about methods, researchers struggle to determine which strategy is most accurate and cost-effective, Enns said. Are differences in results attributable to the questions, how or when they were asked, or to a slew of other potential factors?

“Our method is trying to solve that question,” Enns said. “Without a systematic, unified framework, we can’t reach strong conclusions about what approaches work best.”

To establish such a framework, the Cornell investigators plan to select up to three organizations, teams or researchers to collaborate on the midterm survey. Each will use two sampling methodologies, transparently disclosed, to ask a core set of identical questions during the same time period.

“This is really flipping the standard approach to surveys on its head,” Schuldt said. “We’re soliciting proposals that encourage methodological diversity and innovation.”

The total sample size of roughly 20,000 voters will be nearly 20 times larger than a typical nationally representative survey. The Cornell team said that would enable more effective analysis of specific segments of the electorate, such as Latinos or different age groups.

The proposal also anticipates partnering with other prominent election polls to ask some common questions across surveys, creating the opportunity for an even larger “super-poll,” Enns said.

Data collection is expected to begin about two weeks before the Nov. 8 election and continue for a month. As quickly as possible, the Roper Center will archive and make publicly available topline and individual-level data, making it easily accessible through visualizations and search functions. In January, the research team plans to host a data launch and hackathon event at Cornell Tech in New York City to share key findings and conduct additional analysis.

The revived midterm survey is not a horse-race poll, Enns said, but seeks to give researchers and the public insights into voter preferences and behavior, the effects of campaigns and media coverage, and democratic accountability. More broadly, the Cornell team said its approach could inform the design of government surveys across a range of topics, from economic and business conditions to health outcomes to crime.

“If we can improve the science of surveys,” Enns said, “this will have huge positive spillover benefits to the public.”