Ballot Initiative Experiment

Dear Readers,

Next Tuesday, the state's primary voters will decide whether to change the state's constitution so the legislature can end a disaster emergency declaration without the governor's approval. This special edition newsletter discusses the results of an experiment we conducted to understand how the wording of one of these ballot questions influences the way people respond to it. Our findings show that the wording on the ballot question tends to reduce support for changing the constitution compared to a simpler question we developed.  And, although we do not attempt to understand whether or not the measure will pass, there are sizable differences in support for the measure by party and ideology, meaning the fate of this initiative will be determined by the partisan balance of primary voters.

Sincerely,

Berwood Yost

There has been significant interest in a pair of May primary ballot questions that would change the state’s constitutional procedures for terminating a disaster emergency declaration. The Pennsylvania Department of State has written the text of two ballot questions, but Republican officials believe the proposed ballot wording is unfairly biased. We designed an experiment to test whether the wording of one of these ballot questions influences how voters respond.

Survey participants were randomly assigned to get one of two different question forms--the ballot initiative as it will appear on the May primary ballot (written by the PA Dept. of State) and a common language alternative (written by the Center for Opinion Research). In addition to the different question forms, respondents were also randomized into a condition where they received some background information about disaster emergency declarations or no background information.

The likelihood that a survey respondent answered “yes” that they were willing to change the state constitution differed depending on the format and information they received about the question. One in four (25%) of those who received the Dept. of State wording (DoS) answered “yes,” while about one in three (37%) of those who received the Dept. of State wording and background information (Info & DoS) answered “yes.” Two in five (42%) of those who received the common language wording (CL) responded “yes,” regardless of whether or not they received the background information (Info & CL).

Collapsing the results into question format alone reveals that the Dept. of State wording produces a significantly lower “yes” vote than does the common language version. Fewer than one in three (31%) respondents select “yes” when presented with the Dept. of State wording, while more than two in five (42%) select “yes” when presented with the common language wording. The chances of voting “yes” tends to be higher for conservatives and Republicans regardless of wording, but is also likely to be lower for all groups when presented with the Dept. of State’s ballot question than if presented with a common language alternative.

Although this experiment did not try to determine if these measures will pass, it does suggest that the fate of these initiatives will be determined by the partisan balance of primary voters.

References & Resources

Franklin & Marshall College Poll: Pennsylvania May 2021 Ballot Initiative Experiment

There has been significant interest in a pair of May primary ballot questions that would change the state’s constitutional procedures for terminating a disaster emergency declaration. The Pennsylvania Department of State has written the text of two ballot questions, but Republican officials believe the proposed ballot wording is unfairly biased. We designed an experiment to test whether the wording of one of these ballot questions influences how voters respond.

Franklin & Marshall Poll - Ballot Initiatives: How Wording Matters

The Center conducted research on the 2016 initiative that showed how different wording on ballot initiatives produced different levels of support for the initiative. Survey researchers commonly use a tool known as a “split-ballot experiment” to understand the effect that wording has on how people respond to a question.