Small subgroups of the population seem much larger to many Americans | Hacker News Reader

Small subgroups of the population seem much larger to many Americans

today.yougov.com

171 pointsduck127 comments

Note: YouGov is a platform where internet users are paid small amounts to complete surveys. Their primary product is market research and focus group stuff. These type of posts seem to me like (advanced!) marketing content to drive clicks/eyeballs to their platform. It is not exactly an academic study and there may be a selection bias not truly representing a random set of Americans. If you are interested they do disclose methodology at the end of the post.

I do like their more entertaining content - such as this: "You’re doing Italian food all wrong, say Italians." [1] I'd like to see more data binned as "Acceptable/Not Acceptable/Divisive to Italians".

[1]: https://yougov.co.uk/topics/international/articles-reports/2...

Their elections forecasts aren't stellar, but their errors are still far smaller than the magnitude needed to meaningfully change these results.

These results are consistent with a group of people answering accurately and honestly mixed in with a large group of people giving random answers to complete the paid survey as quickly as possible (which will naturally drag the averages of each toward 50%, as they report). Wouldn’t wipe my ass with these findings.

Academic researchers use these same types of panels for their studies. Sometimes they use even lower quality sample sources since their budgets typically aren’t as high as corporate market researchers.

That said, research panels have huge issues across the board. I’d bet a significant majority of ‘Americans’ responding to this study are using vpns to participate from other countries.

Leaving aside the methodological issues, this is a general phenomenon not isolated to Americans. In my corner of the world (Vancouver, Canada) many students recently expressed outrage that, over the course of a 4 year degree, they hadn't even had a single Black instructor.

Based on the census-reported prevalence of Black individuals in the community, they would need to be significantly overrepresented in University teaching in order for most students to take courses taught by them!

That sort of level of critical thinking really should be the outcome of all degrees too, tbh!

Alas, it's not just the students who struggle in that respect. Over half of the faculty supported them.

You’re lucky they didn’t call for the firing of the people who read the census numbers, too.

I'm an alumni representative on the University's academic governing body (Senate) and was the only person in the room who receives neither grades nor a paycheque from the University.

In light of that, I have a standing offer to speak on behalf of any Senator who sends me written remarks and does not feel safe (either physically or career-wise) being publicly identified with their opinions. I'm sad to say that this offer has been utilized.

That's a sad state of affairs, thank you for doing the right thing and speaking up for those who have been chilled into silence.

That tells me how to value a degree from there when looking at resumes.

The good news is that it varies tremendously across the University. Most of the Mathematics faculty are quite numerate, for example.

If you're considering hiring someone who did a degree in Political Science, on the other hand...

A few of my co-workers brought this up at a previous workplace and the end result was they stopped allowing employees to ask questions at all-hands.

The questions were along the lines of: Why is D&I openly telling us they are prioritizing the hiring of engineers from Sub Group $A instead of Sub Group $B, when $B is also a minority but is much less fairly represented by our company in relation to the population. The fact of the matter is, they don't want to admit that some minorities in tech are more equal than others.

If they keep pushing this crap it's going to make people bigoted. Sometimes I wonder if that's their goal.

Nassim Taleb talks about a similar phenomenon in “Skin in the Game.” His proposition is that there are cases where a minority’s needs or values might be small relatively speaking but the cost of not accommodating them is more than that of not doing so so the niche becomes mainstream for something akin to economic forces. I listened to the audiobook so don’t have access to his sources but it might be a better place to start looking for more credible research into this idea of anyone is interested.

https://medium.com/incerto/the-most-intolerant-wins-the-dict...

You are talking about his idea of "the intolerant minority", but it has nothing to do with over- or under-representation of people in certain demographics.

If anything, it is almost the opposite. The idea is that intolerant minorities can get to have their preferences dominate if they are more-or-less evenly dispersed among the larger population.

Your numbers seem slightly off. While Vancouver is 1.7% African, it’s unlikely for a university to only recruit from a single city. Canada as a whole was 3.5% Black in 2016. I would expect unbiased selection should result in a number between them.

Assuming students had 45 courses the majority of students should eventually have a black instructor at 1.7% as 0.983^45 < 50%. At 3.5% and 45 courses 80% of students should have had a black instructor at some point.

Of course blacks are likely to be under represented in applicants for such jobs, but that’s a different issue.

Census data I saw was 1.2% Black for Metro Vancouver. Students take 30-40 courses (120 credits for a degree; almost all courses are 3 or 4 credits), and typically have 20-25 unique instructors since they'll take multiple courses with the same instructors in their programs.

98.8%^25 = 74%.

So your assuming every instructor was both recruited and lives in the Metro area for Vancouver, that’s simply not how collages actually work.

Further your 30 credit lower bound requires every class to be 4 credit hours which is clearly ridiculous, especially when you mention 1 credit hour courses.

Garbage in Garbage out.

PS: “Repeating same instructors” how tiny is your school? I repeated the instructors 4 times before graduation and I went to a small school.

Not Canadian but you're assuming that instructors only teach one course each which was definitely not the case during my degree. Most of the senior academics ran multiple courses across different years and the TA's taught on multiple courses even in the same term. I doubt I had more than 10 instructors across 4 years and three departments and subjects. Relatively small university of about 3500 though.

That sounds terrible. I think we had ~4k undergraduates but I very rarely repeated a professor. 1 time by choice, 3 times because they taught two required classes, but I never had TA’s.

Many professors would teach multiple courses, but they where split across undergrad programs and graduate classes plus general education requirements. Toss in some part time adjuncts who actually did what the course was for professionally or taught at another school and overlap was extremely rare.

A lot of comments are coming down on the result as “the methodology must be wrong”, and I feel they are expecting people to give mostly reasonable answers. Except the target of the survey is perception, so just a mental image disconnected with any reality.

Perception is usually wildly off, in particular when affecting political positions, and as noted in the article that’s not the first survey to come up with similar results. The Ipsos one [0] has French guessing 30% of population as Muslims (vs 7.5 for reality) for instance.

[0] https://www.ipsos.com/en-uk/perceptions-are-not-reality-what...

Yeah, kinda weird that everybody is so focused on inconsistencies. These questions directly relate to underlying assumptions for support for various public policies. Does HN think people have coherent views on public policies?

We are all biased toward thinking we personally hold coherent views.

We’re also biased toward thinking we’re average in our capabilities (amongst our in-groups, at least).

So it would stand to reason that those biases compose to believing our entire in-group, on average, holds coherent views.

We’re rarely wise enough to estimate our own margins of error, nor self-aware enough to know our own actual capabilities relative to others.

Expecting 30% of the US to live in NYC does strike me as particularly wild hah

The interesting thing is that the equivalent is actually true for Japan and South Korea. The metro populations of Tokyo and Seoul are enormous relative to national population.

Yes. This trend is also true on a global scale.

Looking at the stats[0] for 240 countries, 70+ have 20% of their population in the capital city, 41 of them having above 30% of the population.

[0] https://en.wikipedia.org/wiki/List_of_national_capitals_by_p...

New York City isn’t even the capital city of New York State.

Capitals in the US are often not the most populous city.

Or that one in FIVE people are transgender.

In their minds nearly every single person who is not christian is transgender. Must be a wild world for them.

I doubt that is the case. But rather what the media has been pushing. When half of the cast of every show created today is minority members it is easy for the public to assume that is representative of the nation.

This is too facile a response, I know people like this who are not Christian, and they are many times immigrants.

While these people don't meet very many transgender people, the media gives them the impression that this is because they live in a bubble, which for immigrants is often true. They estimate based on the assumption that their experience is very far from representative. They have the same impression of the number of homosexuals, and they usually know several, but they also get the impression that average Americans know far more than they do. I've lived in immigrant communities for a long time and this has been an amusing point of confusion for decades.

Overestimating their prevalence is generally considered more polite than underestimating their prevalence, so people sandbag on the overestimation side. It is similar to if you asked people how old you are or how much you weigh; they will always sandbag to the more socially flattering side of their true estimate.

I come from a Muslim country, and I never had any Christian think I was transgender. In fact I suspect Muslim Americans probably have the same misconceptions as Christians about how many transgender people there are.

In my experience, the liberal white non-Christians I know are just as ignorant about minorities, they just have different priors. A friend of mine thought Vietnamese people supported Trump to get back at America for the Vietnam War. This is a highly educated person. It’s like COVID—conservatives vastly underestimate the effectiveness of vaccines, while liberals vastly overestimate their risk of hospitalization and death. https://www.nytimes.com/2021/03/18/briefing/atlanta-shooting....

I think Izzydata’s take is close to reality, and the majority of people probably haven’t thought at all about what transgenders are.

They see them in the news, but don’t care enough to mentally separate them from LGBQ for instance.

Given the ratios in the communities I participate in, I'd also overestimate how many people in the general population are transgender.

And another 30% in Texas, and 32% in California. So 8% spread across the rest of the nation... It would seem we drastically oberestimate the number of Jewish asian Texans.

I suspect there is also a race component to the numbers being so wildly off. It’s something that I actually have noticed a lot over the years.

About 15 years ago when I was in high school, I asked a classmate “without looking around, how many Asians do you think are in this class right now” (he had just ranted about how the Chinese were taking over everything). He estimated 70%. There were 4 out of around 30 or so. a lady who I later found out was extremely racist once told me quite confidently that NYC was >90% black.

> The Ipsos one [0] has French guessing 30% of population as Muslims (vs 7.5 for reality) for instance.

Looking at prison population, 30% is actually significantly under the real figure (close to 50%).

[0] https://www.nytimes.com/2015/01/26/opinion/the-mill-of-musli...

Notable that these hugely incorrect percentages are probably the correct percentages for the representation of minority groups on TV commercials and shows/movies.

This would imply that minority groups make up 126% of TV actors. Even if that number weren’t mathematically impossible, reasonable people would agree that we’re not even close to even say 90% representation of minority groups on US TV.

Not every show, but lots of shows over represent minorities in main cast in what appears to be a conscious DEI effort.

For example, Zoey's Extraordinary Playlist gives the impression of high representation of minorities in tech (Sprqpoint is extremely DEI friendly being a woke silicon valley tech company, but even the main character's non-techy neighbor is a black genderfluid woman dating a gay black man).

The Dragon Prince gives the impression of high representation of minorities as well (with 1 main character and his kingly father black, 1 main character having gay parents, the rulers of the neighboring country being lesbian, etc).

I'm not saying that's wrong, just that minority groups don't need to make up 126% of TV actors in order to be overrepresented in main cast.

Heck, even outside of media, look at the Supreme Court. Soon 2/9 justices will be black, which is 22% of the court - an over representation of the black population. Again, not that I have a problem with that, just that DEI efforts can cause high visibility groups (such as a TV show's main cast) to be overrepressented by minorities.

The Dragon Prince is set in a fantasy world, so not really a fair comparison. It's possible that in their universe the chances of being black or homosexual is different than ours, and that the onscreen representation of those minorities is perfectly proportional. So trying to compare the two is a moot point.

Aside from what my sibling comment said about a fantasy world, why'd you pick Zoey's Extraordinary Playlist and The Dragon Prince? Surely shows like Abbott Elementary and Euphoria would be more culturally relevant? Or recent animated movies Encanto and Turning Red?

And even as far as the Dragon Prince is concerned, it's interesting how you mention it as overrepresentation when much of the diversity is relegated to secondary characters. Callum, Rayla, Claudia, Soren and Viren are all white (elves are in some ways minority coded, but the British accent makes that remain firmly in the realm of coding), and seem to be straight and cisgender.

I personally love stories with "overrepresentation" - and not only just to make up for a massive history of erasure and underrepresentation. It makes me so happy to hear people saying "I usually don't feel seen, but THIS character, they look/sound/act like me!" Honestly, as a gay man I still don't feel especially seen - and as far as gay characters I feel connected to beyond just that, it's pretty limited. I can think of maybe... 4 main characters? And they're all in the past few years.

I don’t get the “feeling seen” thing at all. There were few brown characters on TV when I was growing up, and I’m not sure I even noticed. And I think it’s even worse these days, because you have these white and white adjacent people writing the characters. So you have “representation” but they’re completely unrelateable projections of what white writers think brown people are like/what white audiences want brown people to be like.

I totally agree with you on the writers room issue! It's a bit less easy to see, but it's definitely important for representation to go beyond shallow stereotype, and actually touch on real experiences. For an example from my background, the "gay best friend" is a pretty tired trope, and while it technically counts as gay representation, I don't feel connected to the character since it doesn't reflect my reality. So for me feeling seen extends beyond to just visibility, but to accurate representation of someone like me.

Another reason why this "feeling seen" thing can matter to people is that stories can give people a template to imagine their own life. If I see a gay couple I can connect to, it becomes easier to imagine my own future - what my family may be like in the future, and what kind of person I might be in that relationship. And while it's absolutely possible to think of those sorts of things via straight representation, or just brainstorm separately, for me there's something extra to seeing it in action.

> Aside from what my sibling comment said about a fantasy world, why'd you pick Zoey's Extraordinary Playlist and The Dragon Prince? Surely shows like Abbott Elementary and Euphoria would be more culturally relevant?

For what it’s worth I don’t watch shows/TV, yet I’ve heard of both umvi’s examples but neither of yours. I suspect you might be overestimating their influence.

I'm not a twitter user, but I would think these stats speak to some measure of pop-cultural relevancy?

"‘Euphoria’ Is the Most-Tweeted TV Show of the Decade (So Far)"

https://variety.com/2022/digital/news/euphoria-most-tweeted-...

and

"‘Abbott Elementary’ Is No. 1 Most-Tweeted TV Comedy of the Year to Date"

https://news.yahoo.com/abbott-elementary-no-1-most-171433284...

Also this https://www.imdb.com/chart/tvmeter/?ref_=tt_ov_pop list has Euphoria as #2 (formerly #1), although to be fair Abbott is not on the list, and I'm personally not familiar with Pieces of Her. So fair enough, might be overestimating for Abbott - but it is on ABC so perhaps relevant to average people. And of all of these shows mentioned, I've only seen The Dragon Prince - I'm only aware of the other ones via discussions.

When the Grammys eliminated the "urban" award categories there were complaints from some that certain groups weren't half of the nominees.

Not all minorities classifications are mutually exclusive! Intersectionality!

Except popular criminal procedurals.

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2856766

Some people think 2 out of 10 people are transgender or 2 out of 10 make over 1 million per household? I'm struck by either the lack of awareness of the average person or the methodology of this survey.

Yeah, let's take the obvious choice and say this survey's methods might not stand up to scrutiny. Pew, a reputable outfit, says that most Americans report having never met someone who is transgender. That obviously can't be reconciled in any sensible way with a belief that 21% of Americans are transgender.

You could reconcile it by believing that people just don’t understand percentages.

Perception does not equal reality.

The obvious reconciliation is that "they" all live "over there".

This is probably true. I'd also add though, how would most Americans even know if they've met someone who's transgender? By definition, the transition phase is transitory.

We’re talking methodology so I think it’s fair to dig deeper.

The most obvious point is, how do you tell someone you met is transgender ? The question itself is pretty weird IMHO.

It would also be interesting to see how many gay people the Americans responding to the pew survey reported to have met.

All in all, these survey are about what people think, it seems pretty hard to come up with control numbers, or validate a methodology as much better than another in meaningful ways. At most we could have different surveys from different groups converging toward the same numbers.

A lot of the answers don’t make any sense, especially the percentage of people living in NYC. I’d love to see both the methodology, know who and how many were asked, and know more about some of these actual numbers (does # people who graduated high school include kids under 18?)

Do you mean, the actual value, or the value obtained in the survey for NYC population?

I assume the survey. After all the population of NYC would have to be about 100 million for the suggested percentage to be true.

Exactly

Maybe it's the amount of media dedicated to the phenomenon.

This and a poor understanding of math.

I wonder how these responses would have differed if they had simplified the questions to something like: “If you were in a room with 100 people in it, how many would you expect to be X, excluding yourself?”

It could have been asked with a local bias, 100 people from your town...

May have better reflected actual experience rather than general 'perception'

the lack of awareness of the average person

During Michael Bloomberg's brief presidential campaign, multiple reporters thought it was a plausible claim that he had spent over a million dollars per American (https://www.politifact.com/factchecks/2020/mar/06/msnbc/bad-...). This requires both impressive innumeracy and a total absence of common sense.

we can safely assume at least 150% of americans are innumerate

By those estimates, 30% live in California, 30% in Texas, and 30% in New York City. That leaves 10% for the entire rest of the US. Maybe the respondents of this poll just didn't care to think for 5 seconds for each answer, rather than revealing some deep-set misperceptions?

There probably were relatively few people who responded with those exact figures.

My guess is someone who said 30% of the country lived in CA is unlikely to say 30% of the country lives in Texas. Probably there are outliers for each far above 30% that average out in each category. Would love to see a deeper analysis on how the responses break down across questions.

Medians do improve things -- but not by that much.

It's not that they don't think (though they don't). It's that they don't have any kind of a realistic feel for how big various groups are. (Which means they would have to think, instead of just responding by intuition - and they don't think.)

This is not surprising? If I survey how much an appetizer costs, how much an entree costs, how much dessert costs, and how much a dinner costs, the answer to the fourth is not the sum of the first three, or even close. Or ask people how much time they spend sleeping, eating, working, driving to work, etc., and the answer is nowhere close to 24 hours.

People simply don't approach these questions from starting with a whole, then considering a portion.

Ask people if the government should help group X, yes. Sum up the costs of all their answers, hell no.

> This is not surprising?

I had assumed most Americans knew the US consisted of 50 states, so their estimates for the population of a single state should have started at 1/50 = 2%. Clearly I was grossly mistaken.

Maybe each person didn't individually answer each question?

> 30% live in California, 30% in Texas, and 30% in New York City

That’s also what most Europeans think.

IMO, a factor in why people overestimate minority groups is that the news media (and media in general) are almost always discussing one minority group or the other these days. It's in your face all the time. I'm in one minority group myself, and I'm kinda sick of hearing about it!

I wonder if they stuck the default position of the slider at 50% and that caused such high estimates.

There was a time during early covid where I did a bunch of research surveys to kill time, and there were quite a few where the I was questioning if the intent was to bias the result or if it was accidental with how they set up the survey form.

I believe that people overestimate the percentage of the population for many small minority groups. I do not, however, believe that the mean estimate for transgender population is 21%. I don't believe they think 30% live in NYC, 30% in Texas, and 32% in California, leaving 8% of the population living anywhere else. The lowest estimate is 20%, and the highest estimate is 65% (even though several categories actually included well over 80%).

It looks rather like a bunch of people said 50% to everything, skewing the means towards the middle.

I'm not surprised because I learned this myself. It's not just the uneducated 'can't find America on a map' people that you are picturing. My highly educated, policymaking friends, when asked responded with an estimated '30%' of people who are gay and protested that 'it must be' that high because that was their experience of the demographics in their offices at government and NGO sectors.

The implicit CDF of income is hilarious. Most people are broke ... but of the people who aren't broke, most of them are making a million a year!

I wonder if some misread that as wealth rather than income. I know I did initially.

I would posit that most people believe that wealth == income. People rarely know the difference.

One "real number" is actually rather far off - survey respondents estimated that 49% of people "own a home" but the article implies the actual number is 65%.

65% is the oft-cited "home ownership rate", however that is defined as the fraction of occupied homes with an owner in them. Move back in with your parents after college? Rent from a live-in landlord? Congrats, you "own a home" according to this absurd statistic.

A much more reasonable number would be something like the percent of adults who either own a home or are married to someone who does.

I think a lot of the inaccuracy is just innumeracy. People have a hard time conceptualising 1% or 99%.

The estimations range from 20% on the low end to 60% on the high end, which are numbers that easily come to mind and have relevance in reality (like you have probably cut something into 5 pieces before, but not into 100 or 1000).

You get the same effect if you ask people to estimate the likelihood of dying in very improbable events like school shootings or lightning strikes.

Not sure what to make of this:

"Black Americans estimate that, on average, Black people make up 52% of the U.S. adult population; non-Black Americans estimate the proportion is roughly 39%, closer to the real figure of 12%. First-generation immigrants we surveyed estimate that first-generation immigrants account for 40% of U.S. adults, while non-immigrants guess it is around 31%, closer to the actual figure of 14%."

It seems to me that if BA thought they were more than 50% of the population, then of course they would be outraged if they aren't getting at least 50% representation in executive positions, TV & movie actors, government officials, etc. Even if they got 14% representation (doubt they do, but I don't know), they would still be outraged. This seems like an unbelievably huge issue that needs to be solved.

I noticed YouGov data is wrong in at least 1 area, the percentage of Native Americans is much higher than 1%, they seem to have wrongly excluded a large % of the black population that is native to America but were part of a systematic paper genocide carried out by the US government.

If there is one obvious mistake, I wonder what else they got wrong or used the wrong criteria for.

If you haven't read the article yet you can test yourself. (I've copied and pasted the questions in random order. I don't know how they were presented to the original participants.)

"If you had to guess, what percentage of American adults..."

  * Are left-handed
  * Have at least a high school degree
  * Are Hispanic
  * Are Catholic
  * Have an advanced degree
  * Have a household income over $500,000
  * Are vegan or vegetarian
  * Are first-generation immigrants
  * Have a passport
  * Have a household income over $100,000
  * Live in Texas
  * Are fully vaccinated against COVID-19
  * Live in New York City
  * Own a smartphone
  * Are Jewish
  * Are Muslim
  * Have flown on a plane
  * Are a military veteran
  * Own a gun
  * Are Asian
  * Have a driver's license
  * Own a house
  * Are Native American
  * Have a household income over $1 million
  * Are married
  * Own a car
  * Are members of a union
  * Are atheists
  * Have at least one child
  * Have a household income over $50,000
  * Are Democrats
  * Are transgender
  * Are gay or lesbian
  * Are obese
  * Are Black
  * Are bisexual
  * Have a pet
  * Have read a book in the past year
  * Are white
  * Live in California
  * Voted in the 2020 election
  * Are Republicans
  * Have a household income over $25,000
  * Are Christian
  * Have at least a college degree

I saw polls during Covid where Americans estimated that 10% of the population had died of Covid.

It’s hard to know what’s going on with these polls. Perhaps it’s simply mathematical illiteracy.

All you’re seeing here is a bias toward 50% caused by random clicking.

And afterward the bias against the null hypothesis.

The "all you see is all there is" fallacy is probably in play here; if respondents were asked about each minority individually, simply the fact that other minorities not named didn't come readily to mind could account for the obviously poor math here. If all the minorities were listed and the respondents were asked to estimate all of them at the same time, you'd see a dramatically different result. I suspect that similar over estimations would happen if you exposed a test subject to a chart showing the relative proportion of ten different quantities (eg, a chart partitioned into different sized regions, distinctly colored) then asked them to estimate one of the quantities without referring back to the chart. It's not so much an estimation problem as it is a limited working memory problem (the old "seven, plus or minus two" rule).

I strongly suspect methodological error until poll results are replicated with other methodologies. The results trend toward the center which seems an unlikely result in many categories which are exclusive and widely known to be 'minority' (e.g. religion and sexual orientation).

Not American, but i was surprised by atheism at only 3%. Im aware that USA is strongly religious, but wow.

However, 29% of Americans are non-religious: https://www.pewforum.org/2021/12/14/about-three-in-ten-u-s-a...

I'm not even sure why atheism has a name. It doesn't feel like a category that anyone would identify with. Lots of people are simply not religious which is more like being null in the religion category while atheism is like -1.

As an American, I (roughly correctly) estimated Christians to be about 65%, and my wife and I both estimated athiests to be about 20%... I guess "athiest" is too strong of a term here, and most people (self included) would more accurately desribe themselves as non-religious or unaffiliated.

Atheism is a dirty word in America. Way more people have no particular religious beliefs, but they’re turned off by the word.

I wonder how much of this is innumeracy rather than mis estimation. What I mean is that people may actually intuitively understand more accurate proportions, but due to lack of education they don’t understand how to convert that to percentages. That may not even be due to not understanding the math itself (ie 1/5 is 20% and 1/100 is 1%) but perhaps not linking the two concepts, so they don’t even think in terms of their intuition when asked to give a percentage. Part of the reason why I suspect this is that even the smallest groups are measured at close to 20%.

But I bet the average person knows the groups are rare or very rare, just not what 20% actually means in terms of rarity

I think this is a lot of the explanation. What always surprises me is just how innumerate many people can be with little obvious impact on their day to day life. Also they are unable to do basic sanity checks like when I go in public how many of these types do I see etc. Also they must not have any other knowledge of demographic statistics to rely on.

There are many factors here, for sure. But I think a big one is just that most people don't have a good grasp of small percentages.

I think humans are more naturally inclined to ratios and proportions, and not percentages. I've found that people who aren't accustomed to how a logarithmic scale works likely won't have a good feel for what the extremities of the percentage range actually represent.

Many comments here are criticizing folks for not using critical thinking and logic to reason about answers, but they're missing the point.

The point isn't that people are so badly informed about the state of their own country, it's why are they so badly informed, and what is going to be done to improve it.

if i were yougov, i’d take this result as an indication that my methodology for aggregating public opinion is bunk.

It depends on what you're looking for. For instance, if you want to know what inflation is like is for financial modeling purposes, I certainly wouldn't trust surveys to provide that figure. However, if you're political operative trying to figure out whether "inflation" is going to be an election issue or not, I'd definitely use polls. If true inflation is 3% but everyone thinks it's 20% or whatever, then dismissing it as an issue would be a massive political blunder.

Disappointed they didn't ask people to estimate number of people who speak a non-english native tongue. I've noticed people way over-estimate the number of Spanish speakers, when in reality the US is quite overwhelmingly monolingual. Curious if this would be reflected in the data.

As I happen to have the new 2020 Census ACS data open in another tab, I'm interested in your estimate of the number of Americans who speak only English.

Some context on the pollster: if you watch the poll tracking charts on FiveThirtyEight, YouGov is the one that generates lots of insane outliers.

I’d be interested to know how the estimates compare given aspects of the estimators.

How on earth do you wind up at 20% of the population being transgender, 30% being gay/lesbian, and 30% being bisexual? Even if you combine all those into one group, it just doesn't make any sense on the face of it. The medians are a bit better, at 12, 24, and 24, but either this poll is ridiculously, incredibly badly done, or I've absolutely no idea how to think about my fellow person anymore.

I suspect the same way you arrive at all Trump supporters being racist, the problems of the country being the fault of conspiracy theorists, etc: heuristics, combined with a cultural aversion to understanding how the human mind works.

By guessing, and being asked a bunch of questions without inter-question context.

It’s just people going “probably a small number”, being asked to quantify that, and throwing out 20-30% because that feels like one.

Look how tightly the data is clustered around 50%. There are almost no questions for which people did not guess closer to 50% than the correct answer. Did they give people sliders with 50% as the default value? There's a maximum amount I'm willing to fuck with your ui to answer 50 questions. 1 percent is pretty difficult to hit on a slider.

If correct answers take more work to enter, then I expect there to be fewer correct answers.

There was a deepfake experiment on the site recently asking people to use a slider to estimate their confidence that a particular video, audio or text clip was faked. That used a slider between -100 confidence and 100 confidence, and I really sincerely doubt that anyone spend any amount of time thinking about what number they entered. It was just a way to introduce noise.

This seems both credible and reassuring

I actually thought about the difference between my percentage ratings :(

I'm sorry to have libeled you then.

TFA offers one reason:

uncertainty-based rescaling — leads people to systematically overestimate the size of small values and underestimate the size of large values.

Basically, when you're not sure what the actual number is, you tend to anchor your guess on 50%, not 0% or 100%.

Not far off from how much airtime / mindshare their ideas get.

Even the medians seem outrageously high.... just some basic life experience would suggest they're all low % across the board. Heck my personal guess was sub-1% up to maybe 1% for each category you mentioned.

Wow...these small loud groups make the other majority of americans live by their rules, e.g jewish demanding removal of christmas trees from public places, gays/lesbians changing education programs

Reply on news.ycombinator.com