Cognitive Dissonance: The $1 Study and a 39-Lab Replication

Key Points

  • Festinger and Carlsmith's 1959 paper reported that students paid $1 to call a boring task enjoyable rated it +1.35 on a -5 to +5 scale, against -.45 for a control group and -.05 for students paid $20, with 20 students in each condition.
  • A 2024 multilab replication based on a different experiment, with 4,898 participants in 39 labs, reported no significant difference between high and low choice but a significant difference between counterattitudinal and neutral essays.
  • Its authors say the results call into question whether the induced-compliance paradigm gives robust evidence for dissonance.
  • A reanalysis by other researchers reports an effect only at higher reported choice, and its authors call it post-hoc and limited.

Published .

Looking out from the shaded stone arcade of Stanford's Main Quad, three rounded arches frame a sunlit plaza, palm trees and the mosaic-fronted Memorial Church under a blue evening sky.
The Main Quad arcade at Stanford University with Memorial Church beyond, photographed in 2014. Photo: Jawed, CC BY-SA 4.0, via Wikimedia Commons, resized, photo source.

In a 1959 experiment at Stanford, Leon Festinger and James Carlsmith paid students either $1 or $20 to tell another student that a boring task had been fun, and reported that the $1 group later rated the task as more enjoyable1. A 2024 study run in 39 labs, based on a different experiment in the same family, did not find a significant difference between the high-choice and low-choice conditions2. The sources below show what each reported and where the argument stands.

What Festinger and Carlsmith did

The paper tested a prediction from Festinger's theory of cognitive dissonance: if a person is induced to say something contrary to a private opinion, the opinion tends to move toward what was said, and the larger the pressure used to elicit the behavior beyond the minimum needed, the weaker that tendency1. The participants were 71 male students in the introductory psychology course at Stanford University, and the experiment was listed as a two-hour study of “Measures of Performance”1.

Each student spent half an hour putting 12 spools onto a tray, emptying it and refilling it, then another half hour turning each of 48 square pegs on a board a quarter turn at a time1. The experimenter then explained that a paid helper normally told the next participant the tasks were enjoyable, and asked the student to do that job for $1 in one condition or $20 in another, with a sheet of points such as that the tasks were very enjoyable and exciting1. The student then spoke to a waiting participant, who was an undergraduate hired for the role. A control group did the same tasks and was never asked to talk to her1.

Afterward an interviewer who did not know the condition asked four questions, each answered on a rating scale, the first being whether the tasks were interesting and enjoyable on a scale from -5 to +51. Data from 11 of the 71 students were discarded for reasons the paper lists, such as suspicion about the real purpose, leaving 20 in each of the three conditions1.

What the 1959 paper reported

On the enjoyment question the average rating was -.45 in the control condition, +1.35 in the $1 condition and -.05 in the $20 condition. The $1 group differed from the control group at the .02 level (t = 2.48), and the $1 and $20 groups differed at the .03 level (t = 2.22); the $20 group was not significantly different from the control group1. On a question about wanting to take part in a similar experiment the means ran in the same direction (+1.20 for $1, -.62 for control, -.25 for $20), but the differences reached only the .08 level and the .15 level1.

The authors also considered an alternative explanation, that the $1 students had improvised more or been more convincing when talking to the waiting student. They rated recordings of those conversations and concluded that the $1 students did not improvise more or act more convincingly1. Their summary says the results strongly corroborate the theory. The reprint lists the paper as received November 18, 19571.

A 2024 multilab replication

David Vaidis and colleagues ran what they call a multilab constructive replication of the induced-compliance paradigm, based on an experiment by Croyle and Cooper rather than on the $1 and $20 design, in 39 labs from 19 countries and 14 languages, with 4,898 participants2. Participants wrote a counterattitudinal essay under high choice, a counterattitudinal essay under low choice, or a neutral essay under high choice2.

The abstract reports that the primary analyses failed to support the core hypothesis: there was no significant difference in attitude after writing a counterattitudinal essay under high choice compared with low choice. It also reports a significant difference after writing a counterattitudinal essay compared with writing a neutral essay, a pattern that held up across data exclusions, lab variability and the way attitude was assessed, and says the results call into question whether the induced-compliance paradigm provides robust evidence for cognitive dissonance2. Only the abstract of that paper was opened for this article.

A reanalysis and its limits

Shiva Pauer, Roman Linne and Hans-Peter Erb argued that the lack of a choice effect could come from methodological factors such as the laboratory setting, and reanalyzed the replication data, which they put at 1,573 participants, using two self-report items about how much choice participants felt instead of the experimental manipulation3. They report a significant interaction between perceived choice and writing a counterattitudinal essay (F(2, 1098) = 7.07, p < .001): attitude change was not significant at low reported choice but was significant at the mean and at high levels of reported choice3.

The authors write that this is consistent with the original theorizing, and also that it is a post-hoc response to the replication and is limited by relying on two manipulation check items3. They add that one of the two items showed a large difference between the high and low choice conditions (d = 0.80) and the other only a small one (d = 0.23)3. The version read here is the authors' preprint, which names its journal as Advances in Methods and Practices in Psychological Science.

What is and is not established

The sources support these statements. The 1959 paper reported a higher enjoyment rating for the $1 group than for the control or $20 groups, with 20 participants in each condition. The 2024 abstract reports that a very large multilab study found a significant difference between counterattitudinal and neutral essays but no significant effect of high versus low choice, and that its authors question the paradigm's evidence for dissonance.

The sources do not settle whether the choice part of the theory holds. The replication was based on a different experiment than the $1 and $20 design, and the reanalysis is exploratory by its authors' own account. The full text of the 2024 replication was not opened, so only its abstract is described here.

Trivia

  1. How much were students paid in the lower-paid group in the 1959 experiment? $1 (the other group was paid $20)1.
  2. How many of the 71 students remained in each of the three conditions after data were discarded? 201.
  3. On the -5 to +5 enjoyment question, what was the average rating in the $1 condition? +1.351.
  4. In how many countries were the labs in the 2024 multilab replication? 19 (39 labs, 14 languages)2.
  5. What was the total number of participants reported in the 2024 replication abstract? 4,8982.

Sources

  1. Leon Festinger and James M. Carlsmith, Cognitive Consequences of Forced Compliance, Journal of Abnormal and Social Psychology, 58, 203 to 210, 1959 (reprint, Classics in the History of Psychology). Accessed November 18, 2026.
  2. David C. Vaidis and colleagues, A Multilab Replication of the Induced-Compliance Paradigm of Cognitive Dissonance, Advances in Methods and Practices in Psychological Science, 7(1), 2024 (abstract, university repository record). Accessed November 18, 2026.
  3. Shiva Pauer, Roman Linne and Hans-Peter Erb, From the Illusion of Choice to Actual Control: Reconsidering the Induced-Compliance Paradigm of Cognitive Dissonance (preprint, 2024). Accessed November 18, 2026.