Skip to main content
Category: Effects
Type: A belief that one’s private attitude differs from other people’s, even when public behavior matches. In the counted studies, Princeton students rated their own comfort with campus drinking below the comfort they gave the average student. The score is that rating, not a drink tally, a vote, or a rescue
Origin: Deborah A. Prentice and Dale T. Miller, “Pluralistic Ignorance and Alcohol Use on Campus: Some Consequences of Misperceiving the Social Norm,” Journal of Personality and Social Psychology, 1993, volume 64, number 2, pages 243–256. The PDF was read. Received May 1, 1992; revision received August 19, 1992; accepted September 16, 1992. Miller and McFarland (1987, 1991), Allport (1924), Latané and Darley (1970), Schanck (1932), O’Gorman (1975), Breed and Ktsanes (1961), and Packard and Willower (1972) were not opened
Also known as: This PDF does not print a second title. It is not the False Consensus Effect. The same students can show both a mean gap and a positive correlation
Quick Answer — Pluralistic Ignorance, as this page uses the 1993 sentence, is the belief that one’s private attitudes differ from other people’s even though one’s public behavior is the same. In Study 1, 132 Princeton undergraduates rated their own comfort with campus drinking at 5.33 and the average student’s at 7.04, on a scale from 1 to 11. The file records those ratings. It does not record an office, a household, or a drink count as the definition.

​
What is Pluralistic Ignorance?

Pluralistic Ignorance, on this page, is the 1993 name for a psychological state in which a person believes that his or her private attitudes and judgments differ from those of others, even though his or her public behavior is identical.
Pluralistic ignorance is a psychological state characterized by the belief that one’s private attitudes and judgments are different from those of others, even though one’s public behavior is identical.
Prentice and Miller print that sentence and cite Miller and McFarland (1991) for it. That chapter was not opened, so this page does not treat the citation as a second reading. Their own score is comfort with “the alcohol drinking habits of students at Princeton.” A useful picture is a seminar in which nobody raises a hand. This paper retells that classroom: each student can explain his or her own silence as fear of looking foolish, and can read the same silence in classmates as proof that the material is clear. No hand count is in this PDF. The same sheet can also show a False Consensus Effect. This paper’s split is a mean gap versus a positive correlation between self and other. Study 1 had both. Groupthink is concurrence pressing a group toward one decision. These questionnaires did not ask a group to settle a plan.

​
Pluralistic Ignorance in 3 Depths

  • Beginner: Own comfort 5.33. The average student, as these 132 people pictured that student, 7.04. Same scale, from 1 to 11. The two numbers are ratings of comfort, not drinks.
  • Practitioner: Write three lines before using the name. Study 2 still found own comfort below friends, and friends below or beside the average student. The gap survived both question orders.
  • Advanced: A mean gap and a positive correlation can sit on one sheet. Study 1’s self–other correlation was r = .37. In Study 3 the men’s comfort moved toward the figure they had given the average student. That Target × Time test is p = .06. The women’s gap stayed. Reported drinks did not change with the calendar.

​
Origin

This 1993 article uses the name and does not print the year the name first appeared. The definition sentence cites Miller and McFarland (1991). The introduction cites Allport (1924) for a different claim: people assume that all members of a group endorse that group’s social norms. Neither book was opened. No page from either one is copied here. The article lists earlier demonstrations and does not reprint their cells: Schanck (1932), Breed and Ktsanes (1961), O’Gorman (1975), Packard and Willower (1972), and Miller and McFarland (1987). Those were not opened. The authors place the studies at Princeton because they say private feelings about alcohol were in transition while public ease still looked like the norm. They describe alcohol as central to campus social life and cite Clitherow (1991) for that setting. That article was not opened. No ranking of a reunion, and no hours-a-day claim, is copied here. Preparation of the article was helped by National Institute of Mental Health Grant MH44069 to Dale T. Miller. They also read the Bystander Effect, through Latané and Darley (1970), as a case in which inaction is taken for confidence that nothing serious is happening. That book was not opened. No helping rate from it is in this PDF. A later paragraph about inebriated students on this campus is the authors’ surmise. It is not a count of incidents.

​
Key Points

Four splits keep a comfort gap, a friend line, a semester, and a signature sheet from becoming one verdict.
1

The name is the private gap, not the drink tally

Pluralistic Ignorance, in the sentence above, is a belief that private attitudes differ while public behavior matches. Study 1 asked 132 undergraduates, 69 women and 63 men, how comfortable they felt and how comfortable the average Princeton undergraduate felt. The scale ran from 1, not at all comfortable, to 11, very comfortable. Own comfort averaged 5.33 (SD 2.73). The average student averaged 7.04 (SD 1.63). Women were 4.68 against 7.07. Men were 6.03 against 7.00. The target effect was F(1, 130) = 55.52, p < .0001. The sex-by-target interaction was F(1, 130) = 9.96, p < .005. The gap was significant for men, F(1, 62) = 10.35, p < .005, and for women, F(1, 68) = 51.95, p < .0001.
2

The estimated norm was tighter than the private ratings

Own ratings spanned the whole 11-point scale in a relatively uniform distribution. Estimates of the average student gathered near 7. The variance ratio was F(131, 131) = 2.99, p < .0001. Students also bracketed the range they thought held 50 percent of attitudes. The median of those brackets was a width of 4, from 5 to 9. The actual ratings had a mean of 5.33 and an interquartile range from 3 to 8. The paper says the means of these bracket summaries sat close to the medians. Two features are the point: the estimated center sits above the private center, and the estimated spread is the tighter picture.
3

Friends sit between the self and the average student

Study 2 used 242 undergraduates, 145 women and 97 men, and added a line for friends. Own comfort stayed below both friends and the average student, Tukey tests at p < .05, in both question orders. The target effect was F(2, 476) = 54.52, p < .0001. Order changed the size of the gaps, F(2, 476) = 3.45, p < .05, and the gaps were larger when the average-student question came first. Friends versus the average student reached p < .05 only in that order. When the self question came first, the totals were self 5.91, friends 6.61, and the average student 7.01. When the average student came first, the totals were self 5.41, friends 6.49, and the average student 7.30. Estimates of the average student were less variable than self ratings, F(241, 241) = 3.14, and less variable than friend ratings, F(241, 241) = 2.46, both p < .0001.
4

Feeling more out of step tracked fewer protest signatures

Study 4 is the fall 1991 keg ban, a policy the paper says the president of Princeton imposed. The president is not named. Of 94 undergraduates, 52 women and 42 men, 40 women and 29 men said the average student was more negative about the ban than they were. Eleven women and 11 men said the average student felt about the same. Two men and one woman said the average student was more positive, and those three were dropped. The “more negative” group was itself less opposed to the ban: women 4.70 and men 4.97, against 0.64 and 1.09, on a scale from 0, totally opposed, to 10, totally in favor, F(1, 90) = 48.26, p < .0001. After that attitude was controlled, the “more negative” group was less willing to collect protest signatures, F(1, 90) = 18.94, p < .0001. In the more-negative group, women averaged 6.05 signatures and men 3.14, against 49.09 and 30.00 in the same-feeling group. Hours followed, F(1, 90) = 10.99, p < .005. Expected reunions followed, F(1, 89) = 8.10, p < .01. Likelihood of donating money did not, F(1, 90) = 1.08, p > .10.

​
Applications

These four moves follow splits the 1993 paper prints. They are not a claim that a class, an office, a household, or another campus copies 5.33, 7.04, 7.08, or a signature mean.

Put the private mean beside the estimated mean

Study 1’s pair is 5.33 and 7.04 on one scale. In a class, a lab group, or a household plan, write the private rating and the rating given to “most people” before the room is called comfortable, opposed, or agreed. A single slogan has dropped the pair.

Add the friends line before the campus line

Study 2’s friends line sat between own comfort and the average student. The friends-versus-average contrast was significant only when the average-student question came first. A meeting that has only a guess about “the average employee” has not yet asked about the people in the room.

Split the calendar by sex before saying the gap closed

In Study 3, on the shifted scale, men’s own comfort moved from 5.84 to 7.08. The interaction is p = .06. Women’s own comfort stayed near 6, and their estimated average stayed higher. Reported drinks did not change over the same weeks. A sentence that says “the campus got used to it” has merged those three results.

Keep a private sheet off an unrun group conversation

The closing section says a campaign aimed at one person may move a private attitude and leave the perceived norm in place. It suggests saying the private attitudes out loud in the group. These four studies did not run that session, and they did not print a before-and-after for it. Bandwagon Effect is movement with a crowd that can be seen. Here the missing piece is the private rating behind the public ease.

​
Case Study

The counted case is Study 3, not the keg ban and not a rescue. Fifty second-year Princeton students, 25 women and 25 men, were drawn at random from the student telephone directory and interviewed twice by telephone. The first call came in the second or third week of the fall term. More than 90 percent of the students contacted agreed. About eight weeks later every one of the 50 agreed again. The paper says the authors assumed eight to nine weeks would be long enough to see internalization if it was going to appear. The interviewer asked each student to tell roommates that he or she was the survey person, so the second call would find the same student. Responses were described as anonymous. The article was received on May 1, 1992. The comfort questions used a 0-to-10 scale, from not at all comfortable to very comfortable. Table 3 adds 1 to each answer so the printed means sit beside Studies 1 and 2. The means below are those printed means. In September, women rated themselves 6.08 and the average student 7.16. Men rated themselves 5.84 and the average student 7.48. In December, women rated themselves 5.94 and the average student 7.74. Men rated themselves 7.08 and the average student 7.58.
Male students shifted their attitudes over time in the direction of what they mistakenly believed to be the norm, whereas female students showed no such attitude change.
That sentence is the abstract. The tests are narrower. A sex-by-target-by-time interaction was F(1, 48) = 3.92, p = .05. For women, the self–other gap held across both calls, F(1, 24) = 11.94, p < .005, and the target-by-time interaction was F < 1. For men, the September gap was F(1, 24) = 8.24, p < .01. In December the gap was F(1, 24) = 1.69, p > .10. The men’s target-by-time interaction was F(1, 24) = 3.92, p = .06. A result at p = .06 is the test the paper prints. It is not a second test that cleared .05. The December means, 7.08 and 7.58, are closer than 5.84 and 7.48. They are not a printed zero. The attitude–norm correlation moved with that split. For men it went from .34 in September, marked p < .05 one-tailed, to .76 in December, p < .001 two-tailed. For women it went from .60, p < .01, to −.08. The paper also averaged drinks in the past week with drinks in a typical week. That index correlated .78 and .93 across the two questions at the two interviews. An analysis of variance found a sex difference, means 2.69 drinks for women and 6.16 for men, F(1, 46) = 5.17, p < .05, and no change over time, F < 1. Table 5 prints the four cells: women 3.60 then 1.79, men 5.74 then 6.44. Those cells are not the collapsed 2.69 and 6.16, and this page does not average them into a new mean. The lesson inside this file is narrow. On these two calls, the estimated comfort of the average student stayed high. Men’s own ratings moved toward it. Women’s own ratings did not. The drink index did not show a time effect. Abilene Paradox is a group that agrees to a trip. This file has no trip. The authors say the gender split was not predicted, and they offer drinking’s place in men’s social life, and Princeton’s history as a men’s college that had admitted women for more than two decades, as speculation. Those sentences are not a further test.

​
Boundaries and Failure Modes

Pluralistic Ignorance, as this PDF lets us count it, is four studies of Princeton undergraduates. Studies 1, 2, and 4 recruited volunteers at a paid questionnaire session. The department’s note says those sessions were advertised in the student newspaper and in introductory courses, paid 6 dollars, and asked only sex and class. The authors call Studies 1 and 2 self-selected, and they call Study 3 the check against that limit. Study 3 is still 50 sophomores at one university, reached by telephone, in one fall. Two uses overshoot the paper. One exports 5.33, 7.04, 7.08, 6.05, or 49.09 onto a class, an office, a household, or another year. The other treats the abstract’s “shifted” as a test that cleared .05. The December result is easy to over-read. The men’s interaction is p = .06. The December self–other test is a failure to find a difference, p > .10, not a measurement that the gap was zero. Women’s means did not move toward the average student, and the paper does not claim a significant widening either, because that interaction was F < 1. Several drinking reports were extreme: two men and one woman in September, and one man and one woman in December. The paper says the pattern was the same when those points were replaced or dropped. The replacement means are not printed here. Study 4 does not show that the ban caused people to skip a reunion or a gift. The comparisons are between students who already differed in how deviant they felt, with attitude controlled by analysis of covariance. The paper calls that control reasonable and not perfect, and cites Huitema (1980). That book was not opened. Donations were not significant. The design is correlational. The authors also cannot separate two accounts of why the public ease and the private rating diverge: students may display more support than they feel and fail to assume the same of peers, or they may think their own discomfort is already obvious. The second account is where this PDF uses “illusion of transparency,” citing Miller and McFarland (1991). Illusion of Transparency on this site is a later page about how much an inner state shows. This PDF did not run that speech study. The authors say they have no direct evidence that picks one account.

​
Common Misconceptions

Three readings turn a correlation, a drink index, or an anonymous sheet into a different study.
This paper separates them. A mean gap is the private rating minus the estimated rating. False consensus, as this paper uses the name, is a positive correlation between the two ratings. Study 1 had r = .37, p < .001, beside means of 5.33 and 7.04. Study 2 had r = .27, p < .01. Students anchored the average student on themselves and still placed that student above themselves. The False Consensus Effect page is that correlation question. It is not this mean gap.
Comfort is the 1-to-11 rating, or the 0-to-10 interview later shifted by 1. The drink index is a separate question. Men reported more drinks than women, 6.16 against 2.69 in the collapsed test, and Table 5 still shows drinks at both calls. Men’s comfort moved toward the estimated norm. The drink analysis found no time effect. Own comfort in Study 1 also ran across the whole scale. The gap is not a count of abstainers.
The paper considers that impression-management reading and gives three reasons to doubt it. The sheets in the mass sessions were anonymous. In Study 4, feeling more out of step still tracked fewer signatures, fewer hours, and fewer expected reunions after own attitude was controlled. In Study 3, men moved their ratings toward the norm they had described as more comfortable. The authors say that move is a poor fit for a plan to keep looking more progressive than one’s peers. The closing suggestion, that a group be shown the private ratings, was not one of the four studies.
These six pages sit beside Pluralistic Ignorance. Each one asks a different question.

False Consensus Effect

A positive correlation between self and other. This PDF also found that correlation, and still found a mean gap of 5.33 against 7.04.

Bystander Effect

The 1970 book is their analogy for reading other people’s inaction as confidence. No helping count from that book is in this PDF.

Illusion of Transparency

This PDF uses the phrase for a student who may think private discomfort is already visible. The linked page is a later speech study. The two were not the same sample.

Groupthink

Pressure toward one group decision. These instruments asked for comfort ratings, not for a shared plan.

Bandwagon Effect

Movement with a crowd that can be seen. The miss here is the private rating behind a public display of ease.

Abilene Paradox

A group that agrees to a trip. This PDF scores comfort and a keg ban. It does not score a trip.

​
One-Line Takeaway

Before a quiet campus is called Pluralistic Ignorance, put 5.33 beside 7.04, and keep the men’s p = .06 shift apart from a drink index that showed no time effect.