By Maya Chen
The booking page asks how many seats. You have three thumbs-ups and a strong feeling about Friday. Then Jay checks the chat on his break. He works Friday. One friend who reacted meant “saw this.” The thumbs-ups haven’t changed. Your head count has.
The short answer
Write down how many people you expect can come, then ask each person the same private question. Decide beforehand whose attendance the plan needs. Count the answers you have and leave a blank beside anyone you still need to hear from.
Definitions and scope
Friday works for you. Jay can’t make it. Would you also guess different things about everybody else? In false-consensus research, people who pick an option tend to give it more support in their forecasts than people who pick the alternative. To judge the forecasts, you need answers from the same group the guesses were about. Ross, Greene and House, 1977.
Ross and colleagues asked 80 Stanford undergraduates to wear a sign. The students who said yes guessed 62.2% of peers would say yes. Those who declined guessed 33.0%. The reported estimates average the two sign conditions equally. For an observed comparison, 48 of those 80 students agreed. The share was 28 of 40 for one sign and 20 of 40 for the other. The guesses concerned peers. The actual counts cover the students in the study. Study 4, Table 4.
How it works
You know your own Friday. Social projection means expecting similarities between yourself and others. A 2005 synthesis found that people’s own judgments correlated more closely with estimates for groups they belonged to than with estimates for other groups. The association for one’s own group was stronger in lab groups than in existing groups. The older studies show that the pattern can appear in laboratory settings, and a particular person’s reason for guessing remains open. Robbins and Krueger.
Maybe the first three replies came from friends on the same shift. You’ve heard from three people with similar schedules. In one U.S. survey, estimates of flu vaccination across the country tracked what people thought their contacts did, even after the researchers accounted for whether respondents said they’d been vaccinated. These were impressions of contacts. The contacts’ behavior was never independently checked. Bruine de Bruin and colleagues, 2020.
What the evidence supports
Many Labs 2 found the gap again in two hypothetical decisions across online and lab samples. For the supermarket choice, people who picked the first option guessed 69.19% of others would also pick it. The other group guessed 43.35%. The replication confirms a pattern in the estimates. It leaves the cause of that pattern open. Klein and colleagues, 2018.
Misconceptions and uncertainty
The U.S. survey also found a reversal. Among respondents who saw less similarity in their contacts, those who reported getting vaccinated estimated lower coverage than those who reported skipping it. The authors called that pattern false uniqueness. At dinner, Jay’s actual answer gives you something concrete to plan around regardless of what you guessed about him. Bruine de Bruin and colleagues, Figure 1A.
Practical meaning and sources
Try it once with an optional plan.
- Write the prediction. Your list has six people, including you. You’re free Friday and guess three of the others are too. Write “four of six can come Friday,” including anyone who can also do Saturday.
- Ask the other five privately. Send the same options to each person. Friday only, Saturday only, either day, or neither. Add your own answer to the same list.
- Do the count. Imagine all six people respond. Two say Friday only, two Saturday only, one either and one neither. Three can come on each day. Your Friday guess was high by one. If you need all six, ask about another date. Leave skipped questions blank. Count someone under “neither” only if they chose it. Compare the guess with the final count once everyone answers.
Follow up once if you still need replies. Then make the plan provisional or pause. The private count checks one forecast. Whether this exercise improves group planning remains untested in these studies.
Use this for choices people can decline freely. Anyone can skip the question or turn down the plan. A group tally leaves that individual decision with the person affected. A poll cannot diagnose anyone. If worry about others’ views often interferes with daily life, a qualified mental-health professional can assess that distress.
Myth or Real?
Claim
A downvote count shows that a take is wrong. A Reddit commenter recalled someone using this argument against their take. Public comment.
Verdict
False. A displayed score reflects how a platform counts votes. The factual claim still needs evidence.
Facts
Researchers randomly gave comments a starting upvote, a starting downvote or no added vote on one social news site. An added upvote drew more upvotes from first viewers. Added downvotes prompted corrective voting. The experiment measured votes on that site. The accuracy of a comment and the size of any effect on Reddit remain separate questions. Muchnik, Aral and Taylor, 2013.
Takeaway
Ask what evidence would settle the specific factual claim. For a preference, ask who voted and which options were available.
Sources
Public comment · Original experiment
Sources
- Ross, Greene and House (1977). Eighty undergraduates at one campus. The estimates describe peers. The observed denominator is the study sample. Study 4 and Table 4.
- Klein and colleagues (2018). The supermarket analysis used 7,205 responses and the traffic-ticket analysis used 7,827. The gaps in standard-deviation units were 1.18 (95% confidence interval 1.13 to 1.23) and 0.95 (0.90 to 1.00), respectively. Replication results, pp. 459–460.
- Robbins and Krueger (2005). English-language studies available by June 2002. Pooled correlations were .46 (interval .42 to .50) for one’s own groups and .13 (.08 to .18) for other groups. Within one’s own groups, the means were .57 in lab groups and .40 in existing groups. Meta-analysis.
- Bruine de Bruin and colleagues (2020). The analysis used 351 people from a probability-recruited U.S. panel who answered in September 2011. Participants who said they had been vaccinated estimated 47.47% coverage, compared with 41.16% among those who said they hadn’t. The survey analyzed perceptions and self-reports. Survey results, Table 1.
- Lee and colleagues (2019). A separate survey of 300 adults from selected U.S., German and South Korean online panels also examined how contacts relate to wider estimates. Mirta Galesic co-authored both papers. Survey and network model.
- Muchnik, Aral and Taylor (2013). The study assigned 101,281 comments. Compared with controls, adding an upvote increased the first viewer’s chance of upvoting by 32% and the final mean score by 25%. The site displayed upvotes minus downvotes. Original experiment.
Topic hubs
Tags



