Experiments Are Not Just for Scientists Anymore

Marvin JS Ferrer

The 4th Toronto Political Behaviour Workshop answers big questions about why citizens do what they do

US Republicans who are aware that the fast-food chain Wendy’s donates to members of the Republican Party, will choose to eat at Wendy’s more often than those who aren’t. Democrats who are aware of the association, however, will choose to eat at Wendy’s less frequent.

This interesting tidbit is one of many research findings presented at the 4th Toronto Political Behaviour Workshop hosted at the School of Public Policy and Governance at the University of Toronto.

“Political behaviour explores how people think about, and engage with, politics generally,” said Rebecca Wallace, Ph.D. candidate at Queen’s University.  “It asks and answers challenging questions about how and why citizens think, act and vote in the ways they do.”

“Specialists delve into complex queries about attitudes, knowledge, and opinions to better understand the implications of policies and political communications on the voting public,” Wallace said.

The conference explored concepts that have the potential to help explain recent events, and inform policy-makers about the problems they will face and the decisions they will make in the future.

Did people lie to pollsters in America?

Social desirability bias is a phenomenon in polling where people misrepresent their opinions on sensitive topics in order to be viewed more positively by others.  For example, Michelle Dion, Associate Professor at McMaster University, presented research showing that when asked if they supported same-sex marriage, Argentinians show a higher level of support than they in fact hold; Argentinians will say they support same-sex marriage even if they don’t, because they don’t want to be judged negatively by the surveyor.

questionnaireSince few polls predicted a Donald Trump victory in the recent US Presidential election, Dion weighed in on whether social desirability bias might have played a factor.  “If you look at national polls, the outcome was within the margin of error,” Dion said.  “I suspect it had more to do with problems with sampling than social desirability bias.  Some polls are automatically done by robot dialers and not live people.  But there wasn’t a gap between the two kinds of polls.”  If people had been lying to pollsters in order to be more liked, there would have been a difference between the results of automatic surveys and live surveys.

Why did women and Democrats vote for Donald Trump?

Mark Pickup, Associate Professor at Simon Fraser University, presented his paper about individuals who vote “against their interests.”  Regarding voter motivations, Pickup said that although data from the recent US election was still fresh and hard to find, ideological identity might have been something with which people struggled.

“During the campaign I was thinking about people whose identities are conflicting, particularly Republican women,” Pickup said.

“One might feel a Republican can’t vote for a Democrat, but as a woman they might find it difficult to vote for Trump,” Pickup continued.  “How they resolved that, I don’t know.  Conflict might have been difficult given gender was such an issue in this election.”

Data, data, and more data

Presenters repeatedly demonstrated that randomized controlled experiments and natural experiments can play a big role in helping policy makers learn how people react to policies in intended and unintended ways:

  • Christopher Dawes, Assistant Professor at New York University, presented evidence that “stop and frisk,” a policy used by New York City police to stop and search people suspected of committing a crime, had a negative effect on the voting behaviour of black and Hispanic males; these are groups that the policy disproportionately affected.
  • Mark Buntaine, Assistant Pofessor at the University of California in Santa Barbara, presented results of a randomized controlled experiment showing that good news about budgetary management in Uganda can lead to increased vote share for incumbent candidates, and how bad news can lead to decreased vote share.
  • Henning Finseraas, Professor at the Institute for Social Research in Oslo, presented evidence from a natural experiment showing that the sooner immigrants can vote after arrival, the more likely they will be to vote in future elections.

Experiments can answer the big questions facing Canada today

Peter Loewen, Director of the School of Public Policy and Governance and Associate Professor of Political Science at the University of Toronto, maintained that experiments can shed light on the major policy questions with which Canada is currently grappling.

“We are constantly implementing policies and making claims about what the effects of those policies are, or what they should be, without making sufficient measurements about what those effects are, or even specifying what the mechanisms are by which those effects come about,” Loewen said.  “These problems are solvable by good experimentation.”

“We can do a lot more experimentation in government, both metaphorically but also literally, and learn more, iterate faster, and create better policy,” he continued.

Elizabeth Goodyear-Grant, Associate Professor at Queen’s University and Director of the Canadian Opinion Research Archive, agreed.  “The big take-away is really methodological and research-design oriented,” she said.

“I am thinking about effects especially as we engage in one of those rare periods of institutional change, like Senate reform, reform to the appointment of Supreme Court justices, and electoral reform,” Goodyear-Grant said.

Still, experiments have some drawbacks, she warned.  Experimental conditions might not always be “externally valid;” in other words, they may not translate well in practice.  “It forces us to think about ways in which we can design experiments to strike the optimal balance between tightly controlled experiments and those that have strong external validity as well,” she said.

Nevertheless, Goodyear-Grant praised the experimental method.  “This workshop has demonstrated the power of experimental research designs to get at causal mechanisms and relationships,” she concluded.

The research presented at the workshop decisively confirmed that experiments and quantitative methods are the future of sound policymaking.

Marvin is a member of the Master of Public Policy Class of 2018 at the University of Toronto. He previously completed his master’s and doctoral degree in the cell biology of reproduction and fertility at Queen’s University, where he helped many Canadians start new families. His policy interests include science and research policy, industry-government relations, and health policy.  As a politically-minded scientist, he would like to advance the use of the scientific method to improve evidence-based decision making.

One Comment Add yours

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s