Case Study 38.1: The Push-Poll and the Pollster's Reputation

Case Study 38.1: The Push-Poll and the Pollster's Reputation

Background

Clearwater Opinion Research is a mid-sized polling firm with seventeen years of operation in the South and Midwest. The firm's founder and president, Margaret Holt, built Clearwater's reputation on competitive legislative and gubernatorial polling, and the firm has been cited favorably in state-level media coverage throughout its history.

In the spring of a gubernatorial election year, Clearwater is approached by Apex Strategy Group, a political consulting firm working on behalf of a gubernatorial candidate — the incumbent governor seeking re-election. Apex wants Clearwater to conduct what it describes as "a benchmark survey with strategic messaging testing."

The Assignment

The fielding specifications Apex provides describe a survey that includes standard benchmark items: candidate favorability, likely-vote questions, top-issue priorities, and economic satisfaction. These items are standard and Clearwater conducts them regularly.

The specifications also include a second section that Apex describes as "message testing." This section presents respondents with a series of statements about the challenger candidate — all framed as questions — and then measures the respondent's likelihood of voting for the challenger after exposure to each statement.

Among the "message testing" statements:

"The challenger has been accused by former staff members of creating a hostile workplace environment. Does this make you more or less likely to vote for him?"
"Records show the challenger accepted campaign contributions from out-of-state real estate developers. Does this make you..."
"If you knew that the challenger's economic plan had been characterized by independent analysts as 'likely to cost 40,000 jobs,' would that make you..."

Margaret's senior methodologist, reviewing the specifications, flags two concerns immediately.

Concern A: The "hostile workplace" statement uses the passive voice ("has been accused") without identifying the source of the accusation. A review of public records reveals that a single former staff member — one person — made an informal complaint that was resolved without finding. The word "accusations" in the plural implies a broader record than exists.

Concern B: The "40,000 jobs" characterization cannot be verified. Apex provides Margaret with a link to an op-ed column written by an economist associated with a think tank that receives funding from the incumbent governor's major donors. The column does include language about job loss risk, but the "40,000" figure does not appear in the column and the characterization "likely to cost" is not the language used by the economist.

The Push-Poll Question

In survey research ethics, a "push poll" is a technique that is not actually a poll — it is a mass message delivery operation disguised as a poll. Push polls contact thousands of voters not to measure opinion but to deliver negative information (often false or misleading) about an opponent under the cover of seeming like a neutral research exercise. The AAPOR Code explicitly prohibits members from conducting push polls while representing them as legitimate survey research.

The proposed survey has some characteristics of a push poll (strategic use of leading statements to damage an opponent's image) and some characteristics of legitimate message testing (it does measure opinions and does include genuine benchmark items). This ambiguity is common in real cases.

Margaret's Decision

Margaret meets with Apex's contact, a senior consultant named Darren Walsh. She raises both concerns.

Darren's response to Concern A: "The wording comes from our legal team. It's technically accurate — there was an accusation. We're not saying he was found guilty of anything."

Darren's response to Concern B: "We have other research that supports the figure. We can't share it with you — it's proprietary. But you can take our word for it."

Margaret asks for a 48-hour delay to review the specifications with her ethics committee (a two-person group she established after joining AAPOR's Transparency Initiative four years ago).

Her ethics committee's assessment: The survey is not clearly a push poll by AAPOR's technical definition — it does include genuine measurement items. But the "message testing" items include at minimum one materially misleading statement (the plural "accusations" implying a broader record) and one unverifiable negative claim (the job loss figure). If Clearwater conducts and releases this survey with its name attached, it is lending its credibility to messaging that it cannot verify and that it has reason to believe is misleading.

Discussion Questions

1. Is Margaret's proposed survey a push poll, a legitimate message-testing survey, or something in between? Use AAPOR's definition of push polls and the technical characteristics of the proposed survey to make your determination.

2. Evaluate Darren Walsh's responses to Margaret's two concerns. Are his responses legally accurate? Are they ethically satisfactory? What is the significance of the distinction?

3. Under the AAPOR Code of Professional Ethics, what are Margaret's obligations in this situation? Be specific about which provisions of the Code apply.

4. If Clearwater conducts the survey and releases results with Clearwater's name on them, what is the nature of the credibility transfer involved? How does this transfer of reputational credibility interact with voters' ability to evaluate the information they receive?

5. If Margaret declines this contract, what precedent does that set for Clearwater's business? What are the financial and reputational risks of the refusal? What are the professional costs of acceptance?

6. Suppose Margaret agrees to conduct only the genuine benchmark items (favorability, issue priorities, economic satisfaction) but refuses the "message testing" section. Apex says it will use the benchmark data from Clearwater but hire a separate vendor for the message testing, then release both sets of results together under a format that implies Clearwater endorsed the whole survey. Does Margaret have any obligation to prevent this outcome? If so, what options does she have?

7. What would you advise Margaret to do? Provide a specific recommendation and defend it by reference to at least two provisions of the AAPOR Code and at least one argument from the broader ethics framework developed in Chapter 38.

Aftermath (Hypothetical)

Margaret ultimately refuses to conduct the "message testing" items in their current form. She offers to conduct legitimately worded versions of the same questions — using the actual public record rather than the misleading framing — for the same fee. Apex declines and hires another firm.

Two weeks later, automated calls go out across the state using the misleading statement about the challenger's workplace record, explicitly citing a "recent poll" and the polling firm that Apex hired, a small operation with no AAPOR membership and no professional accountability. The calls are widely reported as a possible push-poll operation.

Margaret's firm is not implicated. But she wonders: did her refusal matter, given that the misleading information got out anyway through a different channel? This question — whether individual ethical choices matter in a field with inadequate systemic accountability — does not have an easy answer.