Acquiescence bias
Also known as: Yea-saying, Agreement bias
The tendency to agree with whatever a survey statement asserts.
What it means
A response style in which people tend to agree with assertions presented to them regardless of content, saying 'yes' or 'true' more readily than 'no' or 'false.' It threatens measurement because it inflates agreement with any item and can manufacture spurious correlations among questions worded in the same direction. Proposed drivers include deference to the perceived authority of the questioner, a desire to be agreeable, satisficing under cognitive load, and cultural norms favoring politeness. It is more pronounced among respondents with lower education or motivation and varies across cultures. Survey designers counter it by balancing positively and negatively keyed items, avoiding agree–disagree formats, and using forced-choice or item-specific response scales.
How it distorts a dataset
For an acquiescent respondent, every agree-disagree item picks up a small upward push unrelated to its content. Because it varies between people rather than within a questionnaire, that push acts like a hidden trait behind every item at once. Items keyed in opposite directions then correlate less than their content warrants, sometimes turning negative where they should not. Factor-analyze the result and one construct can split into two factors tracking wording rather than meaning. The damage is not noise, which averages out. It is structure, which survives aggregation and moves coefficients.
What the evidence shows
Saris and colleagues compared agree-disagree items against item-specific response options across many matched pairs and found the agree-disagree versions carried consistently lower measurement quality. Hill and Roberts made the point sharply by asking conspiracy and misperception questions in both keyings: on a true factual item — that the FBI director notified Congress about new Clinton emails in October 2016 — 64 percent agreed with the positive wording while 22 percent agreed with its negation. The largest shift in the paper, 43 points, came on an item about Obama's birthplace fielded in China; across 53 questions the shifts ranged down to -12.5 points. Acquiescence corrupts factual estimates, then, not merely conspiratorial ones. The bias also fell hardest on ideological respondents, inflating the apparent link between conservatism and conspiracy endorsement. Across 160 criterion-validity tests, Clifford and Engelhardt found acquiescence distorted the magnitude of relationships between constructs and could reverse their sign.
How big the effect really is
Effect size depends heavily on where the data come from. Cruz, Bouyamourn and Ornstein re-examined the Hill and Roberts data, found a coding error in the education variable, and ran two preregistered replications; Hill and Roberts promptly issued a corrigendum. On a low-cost panel the corrected pattern reappeared — a positive association between postgraduate education and acquiescence that reverses the original paper's own published finding and contradicts sixty years of survey research. On a larger, higher-quality panel that association vanished. The effect itself survived: Cruz and colleagues reproduce the main finding, and roughly five percentage points of bias remain on the clean panel, two to three times smaller than the cheap-panel estimate. What did not survive was the education moderator, much of which was fraud and inattention that attention checks alone did not remove. Note also that Cruz and colleagues re-tested only the education moderator; the ideological one rests on the same panel and has not been re-examined. Treat the bias as sample-dependent, and do not assume the textbook moderators hold.
Why balanced scales are not a clean fix
Balancing item keying is the standard remedy, and it is not free. Weijters, Baumgartner and Schillewaert show that reversed and negated items carry their own method bias, drawing on careless responding and confirmation as well as acquiescence, and that they lower reliability and produce the very wording-driven factor structures analysts must then explain away. The alternative fares unevenly too. Lelkes and Weiss compared agree-disagree and construct-specific versions of political efficacy items in the American National Election Study and found the construct-specific versions no more reliable or valid, even among respondents most likely to acquiesce. Rewriting response options is not the cure it is sold as.
Using it in practice
The standard advice — avoid the format, balance the keying — is still worth following, though as above the evidence on item-specific formats is split; where a question has a natural scale, using it costs nothing: ask how satisfied someone is rather than asking them to agree that they are satisfied. What that advice leaves out is why balance matters. Keying has to be balanced across at least two unrelated constructs, which is what makes an acquiescence factor identifiable rather than merely assumed, and it should be modelled as a common intercept instead of averaged away. A short block of heterogeneous, unrelated items yields a per-respondent estimate of agreement tendency. Screen the sample first, because on cheap panels fraud and inattention masquerade as response style. And check whether reverse-keyed items load on a wording factor before folding them into a scale mean.
Examples
A respondent rushing through a questionnaire checks 'agree' for both an item and its near-opposite a few questions later.
A patient form asks whether staff explained things clearly and, later, whether they kept you informed. Agreement runs high on both, and the two look tightly linked — mostly because they are worded alike.
A tenant surveyed by his landlord's own agent agrees that the building is well maintained. Asked in his own words what needs fixing, he lists a broken lift and damp in the bedroom.
Two random halves of a sample get opposite wordings of the same question. Each half's headline number is reported as the public's belief; the two numbers disagree by thirty points, and neither is the belief.
A product team fields an agree-disagree survey in eight markets. Scores run highest where politeness norms are strongest, and the roadmap chases a ranking that tracks response style rather than demand.
First described in Cronbach (1946); Couch & Keniston (1960).
Key references
- Clifford, S., & Engelhardt, A. M. (2026). Acquiescence bias and criterion validity: Problems and potential solutions for agree-disagree scales. Political Behavior. Advance online publication. doi.org/10.1007/s11109-026-10124-z
- Cruz, A., Bouyamourn, A., & Ornstein, J. T. (2026). Survey quality and acquiescence bias: A cautionary tale. Political Analysis, 34(3), 471-478. doi.org/10.1017/pan.2025.10030
- Hill, S. J., & Roberts, M. E. (2023). Acquiescence bias inflates estimates of conspiratorial beliefs and political misperceptions. Political Analysis, 31(4), 575-590. doi.org/10.1017/pan.2022.28
- Hill, S. J., & Roberts, M. E. (2025). Acquiescence bias inflates estimates of conspiratorial beliefs and political misperceptions - CORRIGENDUM. Political Analysis, 33(2), 178-180. doi.org/10.1017/pan.2024.27
- Lelkes, Y., & Weiss, R. (2015). Much ado about acquiescence: The relative validity and reliability of construct-specific and agree-disagree questions. Research & Politics, 2(3). doi.org/10.1177/2053168015604173
- Weijters, B., Baumgartner, H., & Schillewaert, N. (2013). Reversed item bias: An integrative model. Psychological Methods, 18(3), 320-334. doi.org/10.1037/a0032121
- Saris, W. E., Revilla, M., Krosnick, J. A., & Shaeffer, E. M. (2010). Comparing questions with agree/disagree response options to questions with item-specific response options. Survey Research Methods, 4(1), 61-79. doi.org/10.18148/srm/2010.v4i1.2682
Where this comes up
- Response Bias in Surveys: A Behavioral Science PerspectiveSurvey responses aren't always accurate. Learn about the most common types of response bias, why they occur, and practi…