Looking Forward

The Methodology Wars

by Iran Prism 9 min read

In 2020, Dr. Ammar Maleki faced a problem that most political scientists never encounter: how do you measure the opinions of 93 million people when the act of having an opinion can get you killed?

Maleki, an Iranian-born researcher at Tilburg University in the Netherlands, had co-founded GAMAAN — the Group for Analyzing and Measuring Attitudes in Iran — with a colleague named Dr. Pooyan Tamimi Arab. Their premise was simple and radical: every existing poll of Iran was measuring the wrong thing.

Traditional telephone polls — the kind conducted by IranPoll in partnership with the University of Maryland — called Iranians on traceable phones and asked them what they thought about the government. The results consistently showed moderate regime support, high election turnout intentions, and broad satisfaction with the country’s direction. These results were cited by Western policymakers, journalists, and academics as evidence that the Islamic Republic maintained genuine popular backing.

Maleki and Tamimi Arab believed those polls were measuring fear, not opinion.

Their alternative: send a survey through encrypted channels — Telegram, Instagram, WhatsApp — accessible via VPN networks like Psiphon. No phone numbers. No voice interaction. No way to trace a response to a person. Let 50,000 Iranians answer from behind the wall of anonymity, and see what changes.

What changed was everything.

The Public Self and the Private Self

The theoretical foundation for GAMAAN’s approach comes from economist Timur Kuran and his theory of preference falsification.1

In regimes with high coercive capacity — where the cost of dissent ranges from job loss to torture to execution — citizens engage in a rational act of survival. They develop what Kuran describes as a dual existence:

The Public Self complies with regime rituals. It votes in sham elections. It participates in Quds Day rallies. It wears the mandatory hijab. It tells telephone pollsters that the Supreme Leader is doing a fine job. To an outside observer — or a Western academic tabulating survey results — this compliance is indistinguishable from genuine support.

The Private Self harbors the real grievances. It shares forbidden music on Telegram. It drinks alcohol despite the ban and the 80 lashes. It watches Reza Pahlavi’s speeches via VPN. It tells an anonymous encrypted survey the truth.

This isn’t speculation. Methodological studies using “list experiments” — a technique that measures sensitive attitudes indirectly — have confirmed the gap in authoritarian states. In China and Russia, direct questioning inflates regime support by 15 to 40 percent.2 In Iran, where the theocratic intrusion into private life extends to what you wear, drink, name your children, and believe — the gap is almost certainly wider.

The implication is stark: the “legitimacy” that traditional polls record is, to a significant degree, an artifact of fear. It measures the state’s capacity to intimidate, not its capacity to inspire loyalty.

Three Ways to Listen

Three organizations have tried to hear what Iranians actually think. They used fundamentally different methods, and their results tell fundamentally different stories.

IranPoll / University of Maryland

Method: Computer-Assisted Telephone Interviewing via Random Digit Dialing to landlines and mobile phones inside Iran. Operators speak Farsi and local dialects, calling from outside the country.

The landline problem: In Iran, landlines are registered to specific physical addresses and national ID numbers. When a citizen receives a call from an unknown number asking about political views, the rational assumption is that the caller is affiliated with the Ministry of Intelligence or the Basij. The “interviewer effect” — the social pressure of a live human interaction — triggers social desirability bias. Respondents give answers that signal “good citizenship.”

The results: IranPoll has consistently reported relatively high regime support — 70 percent or higher turnout intentions, high favorability for figures like General Qasem Soleimani — numbers that often contradict the visual evidence of mass protests and election boycotts.

IranPoll’s defense: They argue their data converges with census demographics on non-political questions. Age, education, and rural/urban splits match official data. They also note that online surveys skew young and educated, excluding the rural poor and elderly who lack internet access.

The critique: Demographic convergence on non-political questions does not validate accuracy on political ones. A sample can perfectly represent Iran’s age distribution while still capturing the performance of compliance rather than genuine opinion.

Stasis Consulting (Washington, D.C.)

Method: Primarily telephone polling, calling from the United States — which may reduce some fear compared to domestic pollsters.

The innovation: Stasis treats “Don’t Know” and “Refuse to Answer” as significant data points, not statistical noise. In their 2024 polls, refusal rates on presidential approval questions hit 26 percent — triple the historical norm.3

Their analytical principle: “In an authoritarian context, the refusal to answer a political question is often a proxy for opposition. When a respondent hangs up on a pollster asking about the Supreme Leader, they are not expressing apathy; they are expressing a calculation of risk.”

The results: Stasis findings on social issues tend to track closer to GAMAAN than to IranPoll — suggesting a “convergence of dissent” across different methods.

GAMAAN (Netherlands)

Method: Non-probability “Multiple Chain-Referral Sampling” (snowball method) via encrypted channels. Surveys distributed through Telegram, Instagram, and WhatsApp over VPN networks. No phone numbers, no voice interaction.

Sample sizes: Raw samples reach 50,000 to 100,000 respondents per survey.

Bias correction: Raw data weighted against five Iranian census variables — age, sex, education, province, rural/urban distribution. Sometimes past voting behavior is added to prevent opposition-activist skew.

Core thesis: “Bias in demographics can be mathematically corrected, but bias in truthfulness cannot.” By prioritizing absolute anonymity, GAMAAN argues it is the only organization accessing the Private Self.

The Parallel Blind

GAMAAN’s strongest defense against methodological criticism is what might be called the “Parallel Blind” test.

When their surveys ask non-political questions — “Do you use the internet?” “Are you employed?” “What language do you speak at home?” — the weighted results match independent economic data and census figures with high accuracy.

The logic: if the sample is representative for employment and language, it is statistically probable that it is representative for politics as well.

Critics have raised the possibility of bot manipulation — automated responses designed to skew results. GAMAAN counters that carrying out a massive bot attack that perfectly mimics the demographic distribution of Iran across age, sex, education, province, and rural/urban splits — well enough to survive rigorous raking protocols — would be extraordinarily difficult. The weighting itself is the defense.

But the most persuasive evidence isn’t internal. It’s the convergence.

Where Two Methods Agree

GAMAAN uses encrypted online surveys. Stasis uses telephone calls from Washington. The methods share almost nothing in common. And yet:

On mandatory hijab opposition: GAMAAN finds 72 percent. Stasis finds 68 percent.4

On election boycott intent: GAMAAN predicted a 77 percent boycott in 2024. Stasis predicted a record-low 34 percent turnout. Official results showed historically low turnout — around 40 percent officially, with observers estimating 20 to 30 percent in major cities. Both captured the same direction; the official numbers captured the state’s capacity to coerce participation through ID-card stamps required for state employment.

On reformism: Both find support for “gradual reform within the Islamic Republic” collapsed below 10 percent.

Two entirely different methodologies arriving at nearly identical numbers for politically sensitive questions. The probability that both are wrong in the same direction, by the same magnitude, is vanishingly small.

The truth is in that vicinity.

The Election Test

The 2024 parliamentary election provided a natural experiment — a chance to test the polls against reality.

GAMAAN predicted: 77 percent boycott. Stasis predicted: Record-low turnout of 34 percent. Official result: Approximately 40 percent turnout. Observers estimated the real figure at 20 to 30 percent in major cities.

Even among the roughly 40 percent who voted, many were coerced. State employees — teachers, civil servants, military personnel — need a stamp on their national ID card proving they voted. Without it, they risk losing their jobs, pensions, and access to government services. This is not hypothetical; it is a documented, systematic practice.

The polls accurately captured the intent of the electorate — rejection. The election results captured the capacity of the state to coerce participation.

When Western analysts point to Iran’s 40 percent turnout as evidence of residual legitimacy, they are measuring the same thing the traditional phone polls measure: the Public Self performing under duress.

The Spiral of Silence — Broken

Elisabeth Noelle-Neumann’s “Spiral of Silence” theory describes how people suppress opinions they perceive to be in the minority, creating a feedback loop that makes the dominant narrative appear more popular than it is.5 In Iran, the state’s monopoly on broadcast media — combined with the terrifying consequences of dissent — created a spiral so tight that millions of Iranians believed they were alone in their opposition.

Digital platforms broke the spiral.

When 50,000 Iranians filled out a GAMAAN survey and the results showed that 80 percent opposed the Islamic Republic, the data itself became a political event. The “Silent Majority” discovered it was, in fact, a majority. The encrypted channels didn’t just collect opinions — they allowed Iranians to see that their private discontent was the majoritarian view.

This is what Kuran’s theory predicts: authoritarian regimes appear robust and popular right up until the moment of collapse. The Private Self remains unmeasured until a triggering event — military defeat, economic crisis, or simply the knowledge that others feel the same — allows it to cascade into the public sphere.

The January 2026 uprising, in which hundreds of thousands chanted slogans that perfectly matched the polling data from two years earlier, was that cascade.

What This Means

The data from GAMAAN and Stasis carries a policy implication that Western governments have been reluctant to absorb.

For two decades, Western engagement with Iran has been filtered through the assumption that the regime maintains a meaningful base of popular support — that “moderates” and “reformists” within the system represent a constituency worth empowering. The polling data shows that this constituency has collapsed below 10 percent. The West has been engaging with a political faction that represents a fraction of the population.

Meanwhile, 78 percent of Iranians blame their own government — not sanctions — for the economic crisis. This means that targeted sanctions do not alienate the population, provided they are clearly distinguished from humanitarian goods. The common fear that “sanctions hurt ordinary Iranians and push them toward the regime” is contradicted by the people the sanctions are supposed to protect.

The data doesn’t prescribe a policy. But it demolishes the information environment in which current policy was formed.

The report’s conclusion is explicit: “The data suggests that the Islamic Republic is in a pre-revolutionary phase. The collapse of religious identity and the rejection of the political system are irreversible trends.”6

The question for the world is whether it will engage with the Iran that exists in the data — or continue talking to the mask.

This article is a companion to What Iranians Actually Want. For the ethnic and regional dimensions of this data, see The Map Problem.

Footnotes

  1. Timur Kuran, Private Truths, Public Lies: The Social Consequences of Preference Falsification, Harvard University Press, 1995; see also “Sparks and Prairie Fires: A Theory of Unanticipated Political Revolution,” Duke University, 1989 ↩

  2. Cambridge University Press, “Do Chinese Citizens Conceal Opposition to the CCP in Surveys? Evidence from Two Experiments,” The China Quarterly, 2019; ResearchGate, “Lying or Believing? Measuring Preference Falsification From a Political Purge in China,” 2016 ↩

  3. Middle East Institute, “New Polling Highlights Iranians’ Views on Iran’s Foreign Policy and Regional Role,” citing Stasis Consulting data, 2024 ↩

  4. GAMAAN, “Iranians’ Attitudes Toward the 2024 Elections,” Tilburg University, February 2024; Stasis Consulting telephone polling convergent findings ↩

  5. Elisabeth Noelle-Neumann, The Spiral of Silence: Public Opinion, Our Social Skin, University of Chicago Press, 1984 ↩

  6. GAMAAN, “Iranians’ Political Preferences in 2024: An Analytical Report on GAMAAN’s Survey Findings,” Tilburg University, August 2025 ↩