> ## Documentation Index
> Fetch the complete documentation index at: https://meta.niceshare.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Romantic Uncertainty
> Romantic Uncertainty is higher attraction when liking is unclear between a lot and average. See the 2011 ratings, the thought path, and the limits.
**Category**: Effects
**Type**: A higher attraction rating when another person's liking is said to be either strong or only average, compared with a clear statement that the liking is strong. Dislike, silence, and a lottery priced below its worst prize are different claims
**Origin**: Erin R. Whitchurch, Timothy D. Wilson, and Daniel T. Gilbert, *Psychological Science* 22(2), February 2011, pages 172–175. Online 17 December 2010. DOI 10.1177/0956797610393745. Received 7 July 2010. Revision accepted 30 September 2010
**Also known as**: The paper's claim is that uncertainty can increase romantic attraction. Chinese popular writing sometimes files a mixed romantic signal under 蔡氏效应, a short name for the [Zeigarnik Effect](/effects/zeigarnik-effect). The published name "uncertainty effect" belongs to a different result: a risky prospect valued below its worst outcome (Gneezy, List, and Wu, 2006). Neither name is a second title here
**Quick Answer** — Romantic Uncertainty, in the 2011 study, is a higher attraction rating when college women were told that four men liked them either a lot or only an average amount, compared with women told that the men liked them a lot. The uncertain group also reported thinking about the men more than the clearly-liked group. The path from those thoughts to attraction, with the message held constant, is p = .10. The unknown was never "he might dislike you."
## What is Romantic Uncertainty?
Romantic Uncertainty is a higher romantic-attraction rating when a person is told that another's liking sits in a pair of good outcomes, "a lot" or "only average," than when the person is told the liking is "a lot."
> Women were more attracted to the men when there was only a 50 percent chance that the men liked them the best than when there was a 100 percent chance that the men liked them the best.
That sentence is Whitchurch, Wilson, and Gilbert in the 2011 discussion. Whitchurch and Wilson were at the University of Virginia. Gilbert was at Harvard. The comparison that matches [Reciprocity Principle](/principles/reciprocity-principle) is still in the same table. Women told that the men liked them a lot rated the men higher than women told that the men liked them only an average amount. The uncertain message sat above both.
The extra pull is a rating of fictitious profiles in one session, as news accounts of the study describe the profiles. It is not a count of dates, messages, or months in a relationship. [Zeigarnik Effect](/effects/zeigarnik-effect) is better memory for an interrupted task. Essays that shorten that name to 蔡氏效应 and then explain an unfinished romance are using a different study. Gneezy, List, and Wu's uncertainty effect is a price: people value a risky prospect less than its worst possible outcome. A gift card and a coin-flip between two gift cards are not a question about who likes whom.
### Romantic Uncertainty in 3 Depths
* **Beginner**: The lab's unknown had two pleasant answers, "likes me a lot" and "likes me an average amount." The unclear pair drew a higher attraction rating than the clear "a lot." A silence that might mean dislike was not in the design.
* **Practitioner**: Before another evening goes to a maybe, write the two poles you are actually unsure about. If one pole is rejection, harm, or no reply, this table does not cover it. If both poles are positive, the extra rating is still one lab's score, not a reason to wait the person out.
* **Advanced**: Reported thoughts and attraction do not share one order across the three messages. Average liking drew more reported thought than clear liking, and much less attraction. The path from thought to attraction, with the message held constant, is p = .10. "Be vague so they will think about you" overreads that path.
## Origin
Romantic Uncertainty, as this page uses the name, is the rating result in Whitchurch, Wilson, and Gilbert, *Psychological Science* 22(2), February 2011, pages 172–175. The paper qualifies what it calls a social psychological truism: people like others who like them. College women viewed the Facebook profiles of four male students who, they were told, had already seen their profiles. One group was told these were the four men who had liked them the best. The liked-best script in the paper includes the line that, of all the people who saw the profile, these four thought they would like her the best. A second group was told the men liked them only an average amount. A third group was told the men liked them either a lot or an average amount.
The authors tie the third cell to earlier work on the pleasures of uncertainty. Timothy D. Wilson, David B. Centerbar, Deborah A. Kermer, and Daniel T. Gilbert, *Journal of Personality and Social Psychology* 88(1), January 2005, pages 5–21, was not re-read past the abstract. The abstract's example is an unexpected gift of a dollar coin on an index card. The card's text was easy or hard to make sense of. Positive moods lasted longer when the text was hard to settle. Forecasters preferred the certain version and tended to predict a better mood there. That paper is about how long a good mood lasts after a positive event. It is not a rating of a person.
The reciprocity half of the 2011 table is the older claim. The paper cites Elliot Aronson and Philip Worchel (1966) among its sources for people liking those who like them. That 1966 paper was not re-read. This page does not move its conversation ratings onto the Facebook means.
A different published name has to stay off this page. Uri Gneezy, John A. List, and George Wu, *The Quarterly Journal of Economics* 121(4), November 2006, pages 1283–1309, call the uncertainty effect a case in which people value a risky prospect less than its worst possible realization. The PDF was not re-read. No gift-card price from it is printed here. Ondřej Rydval, Andreas Ortmann, Sasha Prokosheva, and Ralph Hertwig, *Experimental Economics* 12, 2009, pages 473–487, was read as the abstract. With a physical lottery and fuller information about the goods, willingness to pay for the lottery sat above willingness to pay for the worse prize in the three pricing tasks they repeated. That is a boundary on the 2006 price result. It is not a dating study.
Whitchurch's dissertation, *When Honesty Isn't the Best Policy: The Positive Effects of Uncertainty on Romantic Attraction* (University of Virginia, May 2009; advisor Timothy Wilson), was not re-read. The archived abstract says three studies found more attraction under uncertainty, that thoughts partly carried the attraction, and that an alternative explanation for the thoughts removed the extra attraction. No cell from those studies is on this page.
## Key Points
Four readings keep the message, the thought row, the mood row, and the two colliding names apart. A vague text message is not yet any of them.
The uncertain women were not told that the men might dislike them. The paper's own restatement is a 50 percent chance the men liked them the best, against a 100 percent chance. News accounts of the study, including Pamela Paul's *New York Times* note of 10 March 2011, say the design left out the possibility that the men found the women repulsive. A maybe that includes "not at all" is a different sentence. This table does not test it.
Liked-best attraction sits above average-liking attraction. The paper calls that contrast a replication of reciprocity. [Reciprocity Principle](/principles/reciprocity-principle) is the tendency to like people who like us. The uncertain cell is an exception the authors report inside a narrow band, not a repeal of the best-versus-average gap. A clear "I like you" was not the losing message against average liking.
The uncertain group reported the most thought, then average liking, then liked-best. Attraction does not follow that order. Average liking is the low attraction cell and the middle thought cell. In the contrast of uncertain against liked-best, the message predicted thought, b = 1.51 (SE = 0.70), p = .04. Thought predicted attraction with the message held constant only at b = 0.09 (SE = 0.05), p = .10. The abstract says the extra thought increased attraction. The printed path is marginal. Both stay.
[Zeigarnik Effect](/effects/zeigarnik-effect) is memory for a task that was interrupted. A popular short name for that memory claim is 蔡氏效应. An essay that uses the short name to explain why an unfinished romance lingers is not this Facebook contrast. Gneezy, List, and Wu's uncertainty effect is a lottery valued below its worst prize. [Variable Ratio Reinforcement](/effects/variable-ratio-reinforcement) is a reinforcer after a varying count of tries. Kind acts and cool acts were not counted here.
## Applications
Write the two poles, the person who is unsure, and what was actually rated. Then decide whether the note in front of you is this table, a different uncertainty, or a tactic the study does not recommend.
"Do they like me?" is not yet Romantic Uncertainty. Write the answers still in play. The 2011 pair was "likes me the best" and "likes me an average amount." If one answer is dislike, danger, or a silence with no rating at all, stop using this page as the explanation. Ask, or leave, on the facts you have.
In the same table, "liked you a lot" beat "liked you an average amount." A plain statement of interest is the reciprocity cell. It is not a mistake the uncertain cell corrects for every later week. One session of profile ratings does not say that clarity spoils a relationship.
The study told women a cover story about men who had already rated them. It did not test a person who hides interest in order to be thought about. Whitchurch's 2009 abstract says an alternative explanation for the thoughts removed the extra attraction. That abstract was not re-read as a PDF, and it is not a license to keep someone unsure. Leaving someone without a clear answer, so they will invest more, is pressure. This page does not recommend it.
A grade, a job reply, a medical result, or a parent's unexplained mood can be hard to settle. Wilson and coauthors' 2005 abstract is about a good mood lasting longer when a positive event is hard to explain. That dollar-coin card is not a partner. [Sunk Cost Fallacy](/effects/sunk-cost-fallacy) is staying because of time or money already spent. Extra thought about a profile is a third row. Write which one you mean.
## Case Study
The counted case is the 2011 experiment itself. UPI on 7 February 2011 and a HealthDay account archived from the same period say 47 female undergraduates at the University of Virginia took part. Those accounts say the women were told the study asked whether Facebook could work as a dating site, that male students at collaborating universities had viewed each woman's profile along with those of about 20 other women, and that the four profiles each woman then saw were fictitious and were presented as real. The paper excerpts read here do not reprint that cover story. They do say each woman saw four men's profiles and was told which kind of rating those men had given her. The attraction contrast is t(44) = 2.07. For three groups, error degrees of freedom of 44 match a sample of 47. The thought tests print degrees of freedom 43. This page does not invent which row is missing, and it does not split 47 into group sizes.
Table 1 prints attraction, reported thoughts, and two mood rows. Standard deviations are in parentheses. Attraction: uncertain 0.57 (0.44), liked best 0.12 (0.60), average liking −0.62 (0.71). The uncertain mean is above liked-best, t(44) = 2.07, p = .04. The paper says the liked-best mean is above average liking and calls that reciprocity. It does not print a second t for that pair in the excerpts read here. The questionnaire wording behind the attraction composite was not in those excerpts, so this page does not invent a 1-to-7 scale. The 0.57 is a table mean, not a chance of a date.
Reported thoughts: uncertain 5.07 (2.17), liked best 3.56 (1.67), average liking 4.63 (2.34). The overall effect of the message is F(2, 43) = 2.14, p = .13, eta-squared = .09. That overall test is not significant at 5 percent. The planned contrast of uncertain against liked-best is t(43) = 1.99, p = .05. The authors call that contrast significant, as predicted. Average liking sits between the other two thought means and at the bottom of attraction. The mediation drops that middle cell. It compares only uncertain (coded 1) with liked-best (coded 0). The message predicts thought at b = 1.51, which is the gap between 5.07 and 3.56. Thought predicts attraction, with the message held constant, at p = .10. A bootstrap interval is mentioned in the paper. The excerpt read here starts that 95 percent interval at −.03 and does not show the upper end, so the interval is not printed. Do not treat the indirect path as a settled test.
Mood, Time 1: uncertain 16.64 (3.62), liked best 16.09 (2.04), average liking 13.55 (4.25). Mood, Time 2: 16.89 (2.97), 15.24 (3.11), 13.40 (4.41). The excerpts do not state the clock time between the two rows, the top of the mood scale, or a test. Uncertain and liked-best are close at Time 1. Average liking is lower. The extra attraction is not a large mood gap against the women who were told they were liked best.
The lesson is a rating gap in this session. Women who were told the men's liking might be the best or only average rated those men higher than women who were told the liking was the best, and they reported more thought than that second group. The profiles were not dating partners. The men were not a sample of men. No follow-up message was counted.
## Boundaries and Failure Modes
Romantic Uncertainty is the wrong summary in three retellings. Every mixed signal is called this result. The thought row is said to explain the attraction row. The name is swapped for an unfinished task or for a lottery priced below its worst prize.
The first boundary is the pair of outcomes. Both poles in the uncertain cell were positive. The paper's 50 percent is the chance of being in the liked-best set, not the chance of being liked at all. A person who might be hostile, unsafe, or simply gone is outside the table. [Forbidden Fruit Effect](/effects/forbidden-fruit-effect) is a higher wish to see a violent description that adds advice, against a description that only names the content. A warning is not a liking rating. News accounts say the profiles were chosen to seem likeable and attractive. The result is not a test of uncertainty about someone the viewer already dislikes.
The second boundary is the thought story. The three-group thought test is p = .13. The order of thoughts is uncertain, then average, then liked-best. The order of attraction is uncertain, then liked-best, then average. Those orders disagree. The remaining path, p = .10, is the authors' mediation contrast of two cells. The dissertation abstract says that giving people another explanation for their thoughts removed the extra attraction. That sentence is a boundary on "I must like them because I keep thinking about them." It is not a cell mean, and the dissertation PDF was not re-read. HealthDay quotes Whitchurch attributing the rating to increased thought rather than to the thrill of a challenge, and comparing the thoughts to the difficulty of not thinking about a medical exam. That interview is her account of the mechanism. It is not a second sample.
The third boundary is the name. 蔡氏效应, in Chinese popular writing, shortens Bluma Zeigarnik's memory result. The existing Zeigarnik page owns that claim. Do not import its "about twice" memory line, or any unfinished-romance essay, as a second result beside 0.57 and 0.12. Do not import Gneezy, List, and Wu's gift-card prices. Rydval and coauthors' abstract already refuses the pricing pattern in the tasks they repeated. Do not export 0.57, 0.12, −0.62, 5.07, 3.56, 4.63, 16.64, 16.09, t(44) = 2.07, or p = .10 onto a dating app, a household, a class, a feed, or a later month. The sample is women students at one university, rating men, in one session, on profiles news accounts call fictitious.
## Common Misconceptions
Three retellings travel with the story. One equates any mixed signal with a deeper love. One treats the thoughts as a finished explanation. One pastes on a name from a memory study or from a pricing study.
The extra rating is uncertain against liked-best, inside a pair of positive ratings, in this session. Liked-best still beat average liking. The design, on the paper's wording and on the *New York Times* summary, did not include dislike. A partner who sometimes is warm and sometimes is cruel is not the uncertain cell. One hour with four profiles is not a relationship.
Uncertain women did report more thought than liked-best women, t(43) = 1.99, p = .05. The three-group test is p = .13. Average liking produced more reported thought than liked-best and less attraction. With the message held constant, thought predicts attraction at p = .10. The abstract's smoother sentence and that p value both belong on the page. A bootstrap interval that the excerpt starts at −.03 is not printed, because the upper end was not in the excerpt.
Zeigarnik's result is memory for interrupted work. The short name 蔡氏效应 belongs with that page. Gneezy, List, and Wu named the uncertainty effect for a risky prospect valued below its worst outcome. Rydval and coauthors' abstract does not find that pricing pattern in their three repeated tasks. A dollar coin on a confusing card, in the 2005 abstract, is a longer good mood. None of those three is a Facebook liking rating.
## Related Concepts
Romantic Uncertainty sits next to liking people who like us, memory for unfinished work, repeated contact, a warned pleasure, money already spent, and a prize after a changing count of tries.
People tend to like those who like them. In this table, liked-best still beat average liking. The uncertain cell is the exception inside that pair.
Unfinished tasks are remembered better than finished ones. The popular short name 蔡氏效应 points here. It is not the 2011 liking rating.
Repeated contact can raise liking without a message about how much the other person likes you. A familiar face is not the uncertain cell.
A warning can raise the wish to watch a described program. Advice on a fictitious film is not a rating of who likes whom.
Staying because of time or money already spent is a different reason to continue. Extra thought about a profile is not that bill.
A reinforcer after a varying number of tries is a schedule. No one in this study counted kind acts and cool acts against an average.
## One-Line Takeaway
When a liking feels unclear, write the two answers still in play. The 2011 rise sat between "a lot" and "average," and it does not license keeping someone unsure.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.