Get all sides of every story at groundnews.com/Ve and read the news with a data-driven approach. Subscribe through my link for 50% off unlimited access -their biggest discount of the year.
you can try to prove it as much as you want with curves and graphs and stats... it won't stop this squirrel drama from making Trump to win the presidential election 🐿
It's probably a big difference if the smart people (or whatever group performs especially badly as long as they recognize themselves) have been warned that this is a "setup" question for them to fail.
@@HaukeLagingI still don't understand how anyone who has considered the question for more than a few seconds can get this wrong, like.. just double the non-cream numbers and you have the answer, I guess I'm just dumb
i got it right for the wrong reason lmao, I went "Experimental and Control are not same size thats dupid, im picking Controll. cuz they had less people :3"
There is a problem with every study. IF it is shown that smart people are selective in their choice of facts, then it means that scientists are also selective in their choice of facts - and this is a fact as data analysis is not only based on numerical analysis, as not ALL variables are observed, scientists choose variable to observe and sometimes even select data to use based on these beliefs. Now as long as this is some "apolitical" topic it works usually fine, although bias is still a huge issues. But when they are paid based on the results or topic is political, then that research is affected even to a level of fraud. Sabine Hossenfelder recently had some nice videos about current state of science. All this casues that people do not trust ALL research as new myths are created on daily basis, or people are served with ideas based on a tiny fraction of truth. Like: if EU limits emissions of CO2 then the world will be saved, even though rest of the world makes so much more.
@@OldSkullSoldier But I just worry about the systematic problems about financing really. Otherwise, it is well known that each single study might be biased. At least in natural sciences that is not a big problem. Even if an model of little use is popular, eventually a more useful model (AKA "the thruth") will win out.
I used to do a dumb statistics joke back in 2006 based on official UK Department of Transport car accident figures from 2000. It went something like this. 20% of car accidents involved excessive speed. Which means that 80% of car accidents had appropriate speed. 20% of car accidents involved somebody who was over the blood alcohol limit. Therefore, 80% of car accidents were caused by sober people. My conclusion? Drive at 100mph while drunk is safest.
That is something I am missing here. How many people/cities were involved in the fictious rash/gun study? Because the neutral participants for whom nothing changed are also important for the conclusion. In your analysis it would be the ratio of sober/drunk people who drove without an accident.
Rationality is not a character trait, it's a process. If you fool yourself into believing that you're rational by default, you open yourself up to the most irrational thinking.
And, on the flip side, if you assume the "scientist" is smarter than you, you are also more likely to fall for the "Appeal to expert" fallacy and go, "But so and so in a white lab coat said it so it must be true!"
@@commentinglife6175that's not a flip side, it's the same thing. scientific reasoning like that is why peer review and replicability are core to the scientific method. most scientists acknowledge that scientists are fallible as well
@@plwadodveeefdv If I have learned anything, Scientist are as likely, if not more likely to make really terrible assertions based on bias. They also often let bias affect their ability to see the big picture when looking at statistics. Gun crime for example: Almost all of it is from gang violence or s**cide but lab coats will continuously preach that removing guns fixes the problem but ignoring the fact the tools of the trade just shift.
I think a big part of why people don't change their views when presented with evidence is that evidence needs a level of trust, most things we disagree about aren't things we can easily see ourselves, we have to trust whoever is collecting the data. And we easily distrust anyone that says something that goes against our beliefs.
I'd also add that it is easier than ever to mistrust the data simply because the capabilities of the researcher are so expanded today. An economics professor I listen to makes this point very clearly when he talks about how easy it is to slice and dice the data and run multiple regressions over and over. In a way, having a more limited tool set provided a sense of "comfort" simply because it was so time consuming to re-run the calculations, we could assume someone was reporting their initial results rather than their fifth or sixth time now that the results matched what they wanted to report.
For the gun control example, I had a discussion with a European buddy who was talking about how Europe has low gun crime rates. Like, what a surprise. It doesn't matter that acid attacks, stabbings, assault, etc. are on the rise, they don't have gun crime! So if you see a statistic about that it's important to see trends before the change happened and what actually is being measured, if you're saying gun control reduced gun crime because there's zero guns, that's not actually measuring total crime trends that could otherwise be rising.
"most things we disagree about aren't things we can easily see ourselves" I don't think so. Take climate change for example. Within my own life span, I myself could clearly see the massive shift in weather pattern, and I am a city guy. Those like farmers whose work heavily depends on weather, they'd be even more likely to notice the shift. Or like with g^n control, I don't believe anyone would actually have a hard time seeing the ridiculousness in things like "without g^ns, a criminal would just be as dea_dly with a knaife." More likely, it starts from what gains or loses people have with such conclusion. A guy who's fond of g^ns would more likely not want to believe in the benefits of g^n control, and then he'd just fuel that narrative with "reasons" that fit, to the point it becomes a "belief".
I think this is definitely an aspect of it. But also changing your position in a political matter means changing your identity. And most aren’t willing to make that trade.
What struck me the most about about listening to people defend their answer to the gun control version of the question was that they were not engaging with the numbers at all. All of the arguments they made were about whether it would 'make sense' for gun control to work/not work. Numeracy doesn't matter if you're not looking at the numbers. This is actually the biggest problem I have with my students. Not in terms of political bias but they way they don't engage with the facts presented in a problem and instead reason based on what they would like to be true. It's incredibly frustrating to deal with.
That's very true actually, in light of hard evidence, we tend to rationalize against it if doesn't align with our political views or readily accept it if it does! A good reason is given about why this happens in the video but I'd like to add that these political beliefs become a part of our identity and questioning it is like working against yourself and most people don't like doing that!
The Problem in this example results from presenting data that goes against any reasonable view one could take. Gun control _works._ That is clearly evident by just looking at all other culturally western democracies, where _none_ have the same issues as the US, and in _none_ guns are as easy to access. This specific example therefore imo doesn't work - if you present a proponent of gun control data that lacks any context as to why a ridiculous case has resulted from gun control, as in gun crime going up, then they will rightly find issue with this. Especially if you frame it in a general, specific manner ("gun control") but only specifically mean one measure ("concealed carry"). The hypothetical should be somewhat believable if you want people to intellectually engage with it.
Your experience seems to be represented in this video pretty accurately. The woman who saw the apolitical version and asked what kind of skin cream it was also wasn’t engaging with the numbers
That's the thing, these kinds of student gets so focused on their hypothesis, and when they conduct a study/paper then discovered the data, they tend to skew the result to support their hypothesis.
@@l4nd3rThe thing is if your approaching random people in a political context, presumable a rally, and asking them these questions, they aren't going to interpret that as you asking them a math question, they're rationally going to interpret that as you asking about their beliefs on the issue
Very subtle and clever wording in the title. I filled in the gap myself with "Smarter people do worse than dumber people", when in reality it was "Smarter people do worse than they are expected to"
Well, if you write a statement like "Group A does worse", is that not objectively equivalent with "Group A does worse than Group {Not A}"? Certainly "Group A does worse than it is expected to" is not the most plausible interpretation.
It's not "clever", it's just deliberately misleading. Same as the rest of the video, where the questions shown on screen don't match the ones they showed to people (which should be an immediate red flag about the honesty of this "study"), and where they count "the number of cities" instead of the number of _crimes._
Since I have a background in clinical trials and experimental design, I got fixated on why the groups weren't equal size (since the people or cities were supposed to be assigned randomly, why would you end up with two different sized groups?). I would have just ended up questioning why the research problem was trying to fool me. Does that make me more or less rational?
Yeah, also one of these is a controlled randomized experiment and one of these is a study. People are well within their rights to question the methodology and impose their own interpretation of the correlation on the study group as the correlation doesn't necessarily imply causation.
With cities it would be natural to have different sample sizes - the number of cities matching criteria is limited and there's a lot of "noise". You're not conducting an experiment here - you're trying to gather observations and you grab whatever you can. With cream, however, you're definitely right - the obvious thing to do would be to get equal sized groups. You COULD end up with some differences due to people dropping out of the experiment for some reason (it did last some time and stuff happens) but it shouldn't make one group several times larger than the other.
I work as a strategy analyst for a major retailer and on a weekly occurrence I have a suspicion or belief that something may save money, or be a better solution for the business. As I research and analyze data I am no longer surprised at how often I prove myself wrong and have quickly learned that data supporting for, against, or even no correlation for either are all equally valid answers. It’s not about getting the results I want but rather getting results period. I’ve applied this same practice to other aspects of my life and it has been very freeing (and humbling). Thanks for the video.
I agree. Most often we want our ideas to be right, but verifying it against real data is necessary. I would rather not disappoint people than lead them on with a fabrication.
One of the most important trait, and the one thats more difficult to learn as a Data Analyst it’s being impartial. Don’t putting one’s belief into an investigation, dashboard, or report it’s very important to learn what’s actually behind the data, not supporting your own idea of it and don’t manipulate the results to skew the outcome you want to have
I'm an ethics teacher and I have used ideas from your videos in the past for my own lessons. I think you do such a great job with these videos. Always love to see them!
I feel that a lot of comments are missing that a meaningful point of the video is that the title is missing the words "...than usual" or "...than expected." The "high numeracy" people still typically do better than the rest, just not as well as they are EXPECTED to do. When the video says "worse" it's in comparison to that demographic's EXPECTED outcome, not worse than the "lower numeracy" groups. The surprise is that by following intuition, "higher numeracy" respondents tend to perform comparably similarly to the "lower numeracy" respondents' results, in certain cases where their beliefs disagree with the presented information).
the title is kinda clickbait. High numeracy people do better because it is literally a numeracy question. It should say, "political beliefs can overshadow numeracy when polling people outside a political rally."
actually I don't think that's the case as you can see in the graph the high level numeracy people had the same level of accuracy as the 0 level numeracy group. Which indicates they are the exact same, regardless of how they are expected to do.
Yes! The data basically shows that people willingly turn off their own numeracy and choose to go with intuitive thinking when the numeracy would lead them to conclusions they dislike.
I think the problem is people are not relying on the data in the fictional scenario and are relying on facts they believe they know. If you showed them a scenario that showed gravity was driven by air pressure, you would have a hard time getting people to believe that even though the paper showed that.
I think this points toward a deeper question that I'm not sure was answered in the study: how rational is it to accept new information and abandon prior information? If the quality and quantity of previous evidence is high, asking a person to suspend that understanding for a guy with a cardboard sign so you can play with numbers is a factor that's hard to ignore.
When you change it to political, the distrust of the data skyrockets. All sides are used to being lied to. What this says about "science communication" is significant.
This was my intuition as well. If I was showed a chart with 4 numbers about gun control and crime I would be like I am not drawing any conclusions from these 4 numbers, please leave.
@@TheMustyrusty Unfortunately for many people there is no such thing as a trustworthy source if it contradicts their current understanding. 4 or 400 numbers, the results wouldnt be too different.
@@agafabaWhich is why we have millions of people who firmly believe that Greenland is facing a threat from unprecedented high temperatures not seen in 10000 years.
Either low numeracy skills because you don't understand it on any level or high numeracy skill because you understand it's underspecified. Sample size of the control and test weren't specified and its not necessarily reasonable to assume that control and test sample were different sizes which is required to reach the 'correct' answer.
man, i wish he would have paused the video a bit right after he asked the initial question with the skin cream. I immediately was told the answer by the first guy and had no time to make my own. Now i'll never know if me dumb or smarts
Why did you label it 'crime improved' and 'crime worsened'? That seems so much more ambiguous than 'increased' and 'decreased' wtf all of you in XAI760K ??
@@michajozwiak7650 What do you mean? It made all people of all levels more biased based on their politics, their actual intelligence wasn't actually related to the trends when it was a partisan issue.
No. They are saying the smarter you are the better you are at manipulating the numbers to see what you believe. It’s a surprising conclusion that I’ve heard before- smarter people are more ideological because they are better at manipulating data to “prove” their ideology.
Bro, my intuition went straight to comparing the proportions. Picking the number just because its the biggest one isn't intuitive at all. Ct's called guessing without thinking. Cuz bigger is generally better
The "higher positives" winning the argument (ie Poker maths) is kinda cultural thing I think that if they did this exact same experiment somewhere else like in China or Japan for example - I'd expect a different result
Me: 'I'm surprised Derek isn't making a political science video so close to elec.....ooooooh I see what you did there.' how long is XAI760K ready for us?
It seems like tribalism explains a lot in the current political moment, but I think it's one step removed from the primary cause. The problem is lack of trust. In that context, my "tribe" is just the people I trust. That said, tribalism isn't the only engine for trust - it's just the easiest. The way out of this is to restore people's trust in more than just their political cohort. Rabid individualism and self-serving institutions have done much to create the current crisis of trust, and building reliable institutions populated by people willing to speak & act in the best interest of those institutions is the best way out of this mess. Unfortunately, I don't think we'll reverse this political/intellectual culture of self-harm before we hit rock bottom.
For the real world, I agree with you that tribalism is the strongest factor - BUT some of that is that most people simply don't have the time or trained analysis skills to examine data and figure it out. So, in that sense, some part of tribalism is just an optimization of time usage (in a strange sense - sort of like thinking slow / thinking fast situations).
Check out social identity theory and the minimum group paradigm. It isn’t really a lack of trust. It’s tribalism, and that tribalism causes us to lack trust in people who aren’t part of our group.
Tribalism is good, actually, and the people who argue otherwise are inevitably the ones who want to erode their competitor's tribe to the advantage of their own. No one who complains about tribalism will ever go out to find common ground with their political enemies, or seek to find understanding of why they believe what they believe.
@@SomeGuy-ty7kr People tend to recognize tribalism in other people but not themselves. We humans are very good at rationalizing our behavior in terms that make us feel virtuous and smart, so we're often quick to think in terms of " that team bad" than we are to figure out why we believe what we believe, and why people who believe something different, or even incompatible, need not be stupid, wrong, or evil. It's easy to blame their tribe. That said, I'm not sure tribalism is either good or bad. Many people think religion is bad tribalism, and point to the millions killed in religious wars as proof. But Catholic Charities and the Salvation Army are also religious tribes. Tribalism is inevitable. Politics has been, and will always be tribal. That's why it's important to recognize that our problem isn't tribalism, per se, but that we've lost so much trust in each other and civic institutions that trust in our tribe is all we have left. That's not good.
@@livemusicfannc It's always been true that most people lack the time and/or background to verify most facts. That's why trust is important. When we can't decide for ourselves, who are we going to trust to decide for us? If your answer is, "That radio talk show host on the MAGA right." or "That social justice warrior on the woke left." you're probably going to be wrong about many things that have nothing to do with populism or social justice.
Just some thoughts that I think help/hurt why people would entrench or fail to engage with the numbers: - A skin care study is assessing the entire sample of people effected by the cream/non-cream (100% of the people in the study were measured) while ~500 samples in a gun control study would barely scratch the surface of all of the potential variables and implications of different crimes (boiled down to one number) - Political topics are inherently more difficult because its harder to quantify things like "control" and "safety" as these are relative terms that heavily change based on the population - There's an inherently emotional response about gun control/violence because it directly impacts life and death, along with people being passionate about whichever side they feel best serves the problem Overall, I think reducing a study about rashes down to a number is reductionist, but logical for people. Compared to something that they feel is inherently more nuanced (due to the emotional side of their reasoning framing it as logic) they would likely substitute (probably accurate) flaws with data collection processes in general, but ones that just so happen to agree with their beliefs. Confirmation bias at its finest haha.
This is similar to a video I've seen by After Skool, on "Why smart people believe dumb things." Apparently, stronger reasoning abilities can help you justify your views, instead of rationalizing your way to the correct conclusion. You start with a belief, and work on justifying it, instead of starting by looking at the evidence, and working towards the conclusion. Also something called the Nobel Prize Effect, I think? Where Nobel Laureates believe in pseudoscience and all sorts of mystical nonsense, because they are outside of their area of expertise. The solution that comes to mind is constantly questioning your beliefs and assumptions, I guess, which is easier said than done.
I think your conclusion "that the solution could be constantly questioning your beliefs and assumptions" is what means to be "interested in science", since "falsifying your hypothesis or symptoms, is the key of science"... people that reject to analyze they can be wrong and only take into account what supports their view, have "confirmation bias".... which is, in my opinion, the key of pseudoscience... The more knowledge one have the more resources to justify or defend your bias.... Knowledge (to comprehend how things work) does not equal to Intelligence (how knowledge is used).... So it's possible to have "stupid knowledgeable" people and " uneducated smart" people as well. :) Know i would have loved to learn which variables were considered as evidence of people "being interested in science" versus people "having knowledge in science". :)
Hearing the people talk about how they got to their wrong answer reminds me so much of listening to people add in all the caveats in their answers to simple philosphical questions (such as the trolly problem) to try to escape the point of the question. Adding in "well I'd have to look into the studies and the sources that they cite" is just missing the point of what is being asked. It is a completely fictional result that they are being asked to assess.
Well, I work in science. And if there is something that people don't realize, it's that intelligence comes in a lot of different flavors... I know many scientists that are brilliant but totally stubborn and dumb in the face of truth. Having an open mind, a critical mind, and a curious mind, are not necessarily because of intelligence. Many people with high IQ have the most blown up ego... Meanwhile you can find the most humble farmer in the most remote region, who can't read and can't do math, but has the most progressive views on human rights and the environment... Great men aren't all smart men, and a lot of smart men are just a bunch of cunts...
Something I think hinders the reasoning process it that people think logic is the highest form of reasoning. It's really not, it's just the building blocks of reasoning: the fact that you can stick 1,000 bricks together doesn't mean you've built a sound house. The true path to sound reasoning is objective rationality: where you continuously try to prove your own conclusions wrong, while searching for valid points within wrong arguments.
Things like this are why I strongly dislike people who say things like the solution to the problem is education or critical thinking. Here is data that shows social and emotional factors will cause those skills to be applied to things like creating confirmation bias.
The great problem with data in politics is that it's SO EASY to manipulate studies. Each side has a block of 'Institutes' that conduct studies where the study design virtually guarantees your outcome. Minimum wage for example. I saw a meta paper examining the state of economic studies on the minimum wage. Sure enough they ALMOST divided perfectly into halves. The ones that chose one methodology for trying to establish controls showed no effect of minimum wage on employment. Those using a different method to control for other factors showed a substantial negative impact on hours worked. I say ALMOST because there were two outliers that used the methodology that did not show effects, but those studies DID show small negative effects. But generally, which methodology you chose ABSOLUTELY DICTATED your outcome. Then, politicians from each side go on talking head shows and talk like the only studies that exist are the ones that agree with them. And we've seen how this can happen even in hard science. How many studies showed that [insert anything here] causes cancer? And then you read the particulars and find out that they stuffed the mice full of 10x their body weight of that substance every day for months. And that's before you get into p-hacking or outright fraud. The entire anti-vax movement came out of a single study that was a dude just straight up lying about his data because he wanted to be able to sue the drug companies for big money. He's ADMITTED in court that he lied about his data. But the antivax movement remains strong. So the term 'Studies show...' is the most overused starter phrase in politics. And the most worthless. And people know it, so they're RATIONALLY skeptical of the 'evidence'. What's needed more than anything is better agreement BEFORE a 'study' is done on what are and are not valid methodologies. There needs to be tighter peer review. And any paper that doesn't go through peer review shouldn't be reported on. No pre-print reporting. Don't waste our time telling us what the National Institute of Pre-Determind Biased Outcomes concluded about the effect of abortion on crime or whatever. But that doesn't sell ad impressions, does it?
Which is why people have to actually be trained on how to read studies (including unfortunately a number of people that do meta-studies). And observational studies are even more prone to this problem. Sloppy selection of subjects or situations, poor controls, etc.. All too often meta-studies just throw a bunch of stuff together without any deep analysis of which studies are even good basic science
And this is why I bristle when liberals (they do this more often) will get mad and call names because they have studies done by thinktanks but conservatives won't believe in them. Its because conservatives distrust since they have seen so many biased stories. Its not they are too stupid to read (always) just cautious in trusting everything. This is why liberals are like press = listen to everything they say
A Rice professor and the former mayor of Houston got caught red handed. The professor was hired to do the study. It didn’t give the desired answer. A new check was cut, a new study designed, and voila, it supported the mayor’s position. 🤣😂🤣
You bring up good points. But the autism thing is interesting. If we're talking about a proper scientific RCT experiment there is literally no proof they work, because they were never done. And people who then claim the polio vaccine cured polio are proven wrong by history, it already went way down before it got introduced. I too thought these people were nuts 4 years ago. But then again, we saw what happened in 2020-2022. So when circumstances changed I reexamined the claims and now I changed my belief. But it's definitely difficult to tell if it's one way or another for most vaccines.
He quickly calculated that death by firearm is 11x more likely if you're black, skewing the crime rate affectable by policy for him to be near zero, and thus dismissing those heavily diluted numbers.
Since your "fits my rule" video (on confirmation bias, can't remember the title) I try to take an evening, every now and then, to prove myself wrong on something I feel strongly about. It has helped a lot! I either get a better understanding of why I am right (if I failed) or I learn something new, and change my mind, which is just an awesome "aha!" moment 😁 I keep a list of things I changed my mind on (and sometimes even changed back) to remind myself. I hope it helps. Thanks for teaching me that!
@@itstruce. Here's one: I thought I completely understood why and how the Keto diet worked. Even advocated for it among friends. Turns out its main hypothesis is disproven, and (in short) it doesn't work in any special way unless you have epilepsy, but may be easier to follow for some. Edit: It was an example of what some call "Mechanism/Mechanistic bias". When the entire explanation seems logical, but isn't actually backed by good science. Usually means: "It's more complicated" or plain wrong.
@@Wrackey The one that did it for me is "market efficiency". In theory, products will get cheaper, while getting better. But when markets make the measure of profits as a target, it ceases to be a good measure. This perfectly explains the disconnect of quality, price and volume production.
The problem with this experiment is when you ask people to consider a study on a nameless skin-care product that they have no prior knowledge of, you're testing their ability to come to the correct answer based on the numbers. But when you show people a study on a matter they probably have a lot of prior familiarity with, like climate change or gun control, you're not testing their ability to find the empirically right answer, you're asking them to rationalize the study itself based on all of the knowledge they have on the subject matter. The questions shouldn't be does "X lower gun violence", or "does Y affect climate change", the same way you would ask "does Z reduce the rash". You would have to ask them does THIS specific study show a reduction in gun violence, climate change, etc.
I was going to comment about it, on the question about gun control, there's no two sides over here, the data is there on an international scale, having hypothetical data and asking people to ignore all of their previous experiences is unreasonable.
That format also calls for people to take a public position on camera, not knowing if they're being tricked by an activist's video. Plus, while gun control supporters consider it a question of lower vs higher crime rates, many/most opponents of gun control consider it a civil liberties question. That is, even if crime goes up, they'd still support it because it's a civil right, rather than a purely functionalist policy question. That said, confirmation bias is absolutely a thing and this was a good video.
@@crafty1098 Even in the case of a pure numbers question that graph only states total crimes being worsened/improved. It doesn't state whether the thing which was just made illegal was included in those 'crimes'. Imagine if you made cars illegal, there'd be tens of millions of criminals overnight, crime is skyrocketing! But it doesn't state what 'type' of crime is occurring/not occurring, only that it is/isn't. Further it doesn't explain the *other* side of the equation, and that is the numbers of people saved via guns, which is as far as I am aware, like... inordinately disproportional, like 10x more people protected with them than harmed by them because there are just far more law abiding people with guns than criminals with guns, by default. It's one of the reasons that data like that is inherently biased. Even if the data showed that gun crime went down due to a complete and total lack of guns, if you don't also look into if other forms of crime worsened or improved because of that decision you're being disingenuous at best, outright manipulating people at worst.
This is a critical point. It's kind of like like asking someone what's 2+2 and then asking them if the data showed that 2+2=5 would that change their previous belief. If people have reasons for believing that 2+2=4 they are not going to immediately submit to new data without vetting it (that would be completely irrational). In my experience, most people are just not rigorous in vetting data, which is why emotional arguments are so common in politics (they are often more effective than trying to have productive debates).
Your point is valid, but i think the video is still fair. I would also predict that the rewording of "did THIS study show a reduction in Y?" wouldn't change much. Here's why: in Psychology there is a term called "flexible thinking". It's brought up in decision making research. A person who can think flexibly will reduce or block out the influence of their prior knowledge when making a decision from information provided. However, most of the time, when a decision is difficult or information is hard to process, we get cognitively lazy. So instead, we rely on a belief we already have to make our decision. What the video showed me is that people with higher numeracy are less likely to think flexibly and only consider the evidence if it contradicts their beliefs. That's a pretty interesting result
Aren't people with higher numeracy more likely to have opinions formed from previously investigating data on the most hot-button topics? How do you control for that in this experiment?
They should still correctly interpret the numbers from the single study, as oppose to relying on previous information to assume what the numbers state. Now, should one study change a person's opinion on a topic they've previously researched? Probably not, but numbers are numbers.
They explicitly told people the data was made up, and to interpret the numbers. Any prior data should have been irrelevant because all they needed to know was in front of them.
Title should be something more like, "People with the capacity to analytically consider data may take shortcuts to their preferred conclusion." But... I'm guessing that would suck for SEO purposes.
@@wax2562 I mean... is it? Are people with higher numeracy scores "smarter" than everyone else? And when the title says "these questions," it pretty clearly suggests specific questions which disproportionately confound "smart people," rather than abstractions designed to elicit an emotional/tribal response, where "smart people" answer the question with accuracy similar to that of not-"smart people."
I think it's very important how you phrase the question. "Based on these numbers did gun control increase crime or decrease crime?" is very different from just showing the numbers and then asking if gun control is effective. You can agree that the data leads you to a certain conclusion but still disagree with the methods with which the data was gathered or presented.
"Increased crime" is indeed ambiguous in these sorts of things because, when you make something illegal, there will of course be more crime, because something was just made illegal
yes, it did not seem that he asked in a very standardized way watching the video. And the one person did not even seem to consider his data sheet really. But then again, I doubt they really wanted to reproduce the result as they said (they did not show their own statistics). They just needed some people say weird stuff to fill the gaps in the video and make this more entertaining. I am sure in the paper they were more careful.
And if you make gun ownership illegal, you are going to get people committing the crime of owning a gun, because it is now illegal. In the UK for example, there are about 6000-7000 recorded offences every year related to gun ownership which would be eliminated if gun ownership were made legal. But the 29 gun homicides per year would likely increase, to somewhere nearer 10,000 if it ends up being anything like the USA.
This. When I realized that the study was about politics, the wording became far more apparent that it was aimed at leading than observing. At that point it had nothing to do with "intelligent people do worse than unintelligent people" and had everything to do with "political bias taints how we interpret data", which includes the way that the data was obtained for this study. It's worthless when you lead people towards or against their biases instead of asking them to make a rational observation based on the data and numbers in front of them irrespective of their political biases. I'm pretty sure there is a rule about not doing this in your data collecting methods when it comes to peer reviewable studies.
I'd say it's implied in this hipotetical situation that gun control is the only factor in the increase/decrease of crimes. The fact that you need to talk about phrasing and justifying the answer for the "gun control" case IS the point of the video, because if they asked you about the "skin cream" case, you would just accept that it's a simplified problem and directly answer if it helped or not. And a perfectly rational person would answer both questions the same way.
@11:18 the problem is that is purely reactive policy/action, which is the most expensive kind of action. To be proactive on something, you have to have agreement on causes and work to fix those. We've forgotten the old sayings of "a stitch in time saves nine" and "an ounce of prevention is worth a pound of cure". We see the same thing these days with the Y2K issues we had. Since a bunch of companies spend massive amounts of time and money *proactively* fixing things so that "nothing happened", today we have a bunch of people thinking Y2K was all just a big hoax. And that kind of thinking has continued to everything else. We won't be proactive on anything, we'll just have to reactively deal with the symptoms rather than proactively try to address the causes.
The other side of this issue is that to come up with the solution to the cause, i.e., the thing you do to fix the problem, you have to be sure of the cause itself. Part of that is understanding that if A causes B, then we can use the existence of A to predict B. Unfortunately, when it comes to some issues (like climate change), the predictions are faulty - illustrating that we actually do not know the cause. Why would I spend money to fix this problem you identified here if you can't show me it will actually solve the issue you claim it is solving? To put this in a non-political example, if I say my car won't start, telling me to put air in the tires won't do me a lot of good - even if the tires are flat!
@@commentinglife6175 I suggest you look at climate change prediction more. Even the predictions that Exxon made (and hid) back in the 1970s are reasonably accurate.
We need a combination of short term and long term problem solving. The issue is that long term work requires trust (which we're lacking in the U.S.). But successful cooperation on short term problems is a good way to foster the trust needed to tackle the long term stuff.
There are reasons why Iceland is the country with lowest crime rate and Venezuela is the country with the highest crime rate. Hint: Guns don't crack even the top ten of these factors.
Sorry, but this video is pretty bad. In most of the interviews you can see people criticizing the validity of the model, or modelling. This makes perfect sense. It happens less for the skin cream because most of the respondents do not have a concern about skin care.
@@peterfox7663 Not 100% sure what you mean, but... what I see is people refusing to do something they consider bad just because a random strange asks them... and I mean it in both direction here.
I love that this took an anthropological spin. Acknowledging your bias is a very difficult task, but important to keep at the forefront of your mind when analyzing a new information. Experiments like this demonstrate it beautifully. It never hurts to ask yourself, how does my culture/society shape my worldview?
@@BrandonWong55nah, research would include taking these numbers into account. She is rejecting research because she doesn't understand it. She wants a simpler opinion that is not founded on data.
They were similarly presented to the people being interviewed, during the video demonstration for us, it didn't go in depth in the second case to avoid exact repetition
Who do you believe more: "Acshually 🤓"-guy with fake numbers, or yourself? Smart people know that if gut feeling tells you against data, something wrong with data because it made up by much more stupid people. Misleading data is everywhere. In this particular case they again was right, because they spotted that data was fake, which is true.
Great video man! I totally see how this would be the case. I've been convinced of something that went against my beliefs based on data, but it's rare. The difficult part of trusting "studies" is that they're often made up with conclusions only drawn using p-hacking (which you have an excellent video on).
There’s a few different meanings of the word “rational”. One is using logic and facts to come to a consistent conclusion based on truth. This is the usual definition we think of. This could be called philosophical rationality. The other comes from economic literature. A rational human in the economic sense is one who pursues their self interest and doesn’t sabotage themself. I think there are many more people who are rational economically than are rational philosophically. I think it can be important to remember that even if someone is supporting something that isn’t 100% true, they’re just trying to act in their best self interest.
And it's worth noting that both types of rational thinking are vulnerable to Confirmation Bias, because the human brain is very, very good at discarding information that goes against our pre-established world views.
There's a problem with how you formulated the rationality in the economic sense, because with this formulation everyone will be considered rational, because everyone pursues what they believe is their self interest. At that point, the term loses usefulness. What it should be about is being rational in the economic sense meaning that one is pursuing that which is objectively in their self interest, rather than subjectively. Because that's where most people fall short. That's the hard part because it requires perfectly aligning oneself with actual objective reality and actual truth, as this in practice is not even possible to perform perfectly as the world is too chaotic and applying precise bayesian reasoning to every little thing is computationally explosive. I find it useful to split the meaning of the word rationality the way Yudkowsky does: 1. Epistemic rationality: systematically improving the accuracy of your beliefs. (truth) 2. Instrumental rationality: systematically achieving your values. (winning) And I love the way Vervaeke puts it: In practice, rationality is knowing when to use logic.
@@griffinshorts785 Chat GPT response: The Maratha Reservation Protest is a movement in the Indian state of Maharashtra, where members of the Maratha community are pushing for reserved seats in education and government jobs to improve their socio-economic conditions. Historically influential, many Marathas now face economic hardship and limited job opportunities, which has led them to seek inclusion in India’s affirmative action system for disadvantaged groups. The government attempted to provide this reservation, but India’s Supreme Court struck it down, citing legal limits on quotas. This has fueled protests as Marathas demand a solution, making it a significant issue in Maharashtra’s politics and raising broader questions about balancing affirmative action in a diverse society.
As social creatures, it is easy to become misled by those around us for sure. Logical thinking is so hard to come by now, but these videos definitely help in spreading the word.
I think a big difference between the cream example and the gun control example is how abstract the question is framed. For the cream it is just a maths problem, but for gun control it is a question where there are preconceived notions that prevent the person from interpreting the data, or even really looking at it, expecting the data to reflect their personal beliefs.
I was looking for this. This seems like another poorly designed study. Participants answering the skin cream question interpreted it as a math question, and were evaluated as if it was a math question. Participants answering the gun control question interpreted it as a social policy question, and were evaluated as if it was a math question. I read through the study. There is minimal effort to inform participants that they're evaluating just the data presented, and the choices they're given are phrased as "cities that enacted a ban on carrying concealed handguns were more likely to have a decrease in crime" rather than "this data supports..." The cream "control" question is also a very poor control. Rather than being simply apolitical, it's presented as entirely hypothetical, in expected math test fashion, rather than being a real world question like the gun control question.
"Personal beliefs" here also can just be "correctly being aware of scientific consensus". It doesn't matter how many fake studies you show me data from to explain that climate change isn't real. My answer will be that climate change is real, because that's just a fact.
this is the exact point where "isolated study" becomes something that is NOT taken as "isolated study". you can agree that the study says something, and disagree that the study is correct or was done correctly, which is what I think had happened here. especially with the method of the study changing from lab experiments to crime statistics, the former being a highly controlled environment with low margins of error while the other is marred with statistical error margins that often eclipse the sample size by a sizable factor. with that, people are often biased to pool from intuition and prior experience instead of the currently provided information as it is deemed "useless", which is what likely happened here.(the actual information provided to the people hearing about the study was useless as analysis of crime statistics require a LOT more than "got better" and "got worse" graph)
The two being treated as equivalent is such SOP in the soft sciences. The overall conclusion that politics (or any strongly held belief) can cause people to irrationally reject data was fine. Talk to someone about their favorite sports team/star. And, yes, today politics is a team sport with equally stupid fans.
@@barongerhardt It's not a fine conclusion at all, because people with real data about a real thing are being presented with fake data in the most unreliable way possible, and the study is assuming that will not have an effect on the way people answer. People don't know any data about skin cremes and those are presented as controlled laboratory studies. People DO know things about political topics, and they're being presented data as uncontrolled, unvetted statistics. It's not difficult to see how that could completely skew the data that they are attempting to treat as a fixed variable. Completely different types of data, completely different contexts, and then a causation is being provided based on the faulty correlation. All huge red flags of a bad study with bad conclusions.
@barongerhardt but is this study actually showing those results? I'd argue that this study shows no real data because the hand cream example is not a proper control group for this experiment due to the factors I mentioned in my original comment.
Whenever I see poll data of any kind, I am most wondering about the questions asked, how they're asked, etc. I see so many studies where I can predict "unexpected" outcomes based on poorly worded questions, multiple interpretations of choices, whether it was multiple choice or open-ended, etc. There's one other thing I want to know specifically for the political version of the question. As the skin cream is made up, no one could have read a previous study on that exact skin cream. Some people could have read existing studies on this exact political question, and I don't mean "read a random article online", I want to know if having read a peer reviewed paper on this topic has an any difference on the outcome.
A big one is where they compare countries. I saw one where they measured agreement with the statement "I trust most people, in general" and the results were completely messed up by the translation, because other languages express it using words that are stronger/weaker
But previous data wouldn't have changed the data presented here, so it's just another cause for bias. The participants were presented numbers in a vacuum and couldn't keep it that way before answering, that's the point.
@@GreenOnionBrother at least for the video that's not really the point, since its second half implies this matters because it is representative of the way people answer things outside of a vacuum. Also, just because you say "please consider in a vacuum", doesn't mean people can simply shut down their beliefs. And who knows, one might also argue that the difference between cream and politics is actually not down to "bias" so much as it is about having or not having context.
@@user-sl6gn1ss8p I don't know what to tell you. They were presented numbers that only allow one conclusion, but due to bias (and this includes previous studies and statistics they've read, regardless of their validity) failed to answer a question they would otherwise have less of a problem with. That is the point. How bias affects or rational thinking.
@@GreenOnionBrother my point is just that it was not clear to me, from the video, how well this could disentangle "bias" from "rational thinking", as in, how much can this actually show that people might make worse decisions in this sense due to their bias - the "tangle" being the fact that this bias may include, as you said, perfectly valid information. To be clear, I'm not disputing that the effect exists - I'm just not sold that the study shows that much, going by the video.
Maybe I misunderstood, but the study showed more that if you present the data in a bad way, people don't look into it much unless it conflicts with their worldview.
Yes I doubt if you gave the question in a math test, there would be any difference from the worldview. In this case most people just assumed that the data you show me is not worth a lot. Context matters, on the street clearly people assume you are asking about a topic, usually you want to make a point.
Great episode! Thank you for it! Also, congrats on the courage of speaking about it when this topic might upset peoples. Please make more that kind of content.
10:40 want to end the partisanship!! Then ensure media is not this biased and vitriolic. Media now is driving this division far more then any political rhetoric. Demand they go back to journalism, not opinions.
I understand your sentiment, but this naive. This is an incentive alignment issue; we're not in a position to "demand" anything. All we can do is seek out sources that are consistent with our values. And incidentally: good journalism IS opinionated, just not in defiance of fact.
The data used in this imaginary experiment was deliberately chosen to trigger spontaneous answers. Firstly, very different sample sizes were chosen for the experimental and control groups, making a direct comparison difficult. Secondly, a very strong difference between the two results (improved/deteriorated) was chosen in both groups, and prime numbers (223 and 107) were even used on the left-hand side to discourage probands from calculating the ratios in their heads accurately. They also only give two possible answers (whether the experiment showed that the skin condition of people treated with the cream was more likely to “get better” or “get worse.”) which may lead probands to think that the table must show a clear difference between the two groups. I also found no information on how much time the probands had to answer this question… However, a look at the statistics shows that it is debatable whether there is a significant difference between the groups at all, as a normal chi-square test gives a p-value of p = 0.0472 and using the Yates correction, p > 0.05. Also the entire experimental setup is vague, controls missing, etc. So if I had to answer this question I would say that the experiment and data is not good enough to make a statement about the creams effect on skin condition (even if you look at the results without a statistical test). As a side note, please never use the word ‘significant’ when you don't show results of statistical tests like at 1:42. I'd like to see what the trial results look like when you use more ‘normal/scientific’ data.
@@meneldal Although it is not good scientific practice, it does occur quite frequently. However, the use of suitable statistical tests and careful interpretation are particularly important in cases like that.
My first thought was about the p-value as well (before I realized it was American propaganda) I would have answered "I don't know, the numbers don't mean anything" and I think actual scientist that work with data would give the same answer, I don't know who these "high numeracy" people are, but the correct answer is idk.
I've participated in some form of online discussion for many decades now... since before the internet became widely available. And from this, I learned a long time ago that one cannot change another person's beliefs with rational discussion. It stands to reason that my own beliefs don't change through rational discussion. You can sure foment a lot of negative emotions, though, by playing with another's beliefs. If you wanted to cause another country's citizens to polarize to destabilizing levels, imagine how effective your campaign would be to use social media to heighten the negative emotions of folks on all sides of the political spectrum.
The questions are not equal. The problem is whether the questions exist in a vacuum or not. We have no prior knowledge of the skin cream and do not expect any bias. The gun control question has already been researched by both sides with their particular biases. So this just becomes one more study in a sea where bias is rampant. If it contradicts what you already 'know', then you are going to question the study, the sources, the funding, etc. You expect the study to be biased.
which does not make it impossible to analyze data objectively. after that, you can either change for mistakes in the experiment that gave you the numbers or you just learned something new. Science works, but only if you can stomach being wrong.
@@erumaaro6060 sometimes it's just hard to know you are wrong. like the video shows, we all have subconscious biases that are hard to keep check. I always try to do this. when i am looking unto the data that's connected something i am emotionally attached to. i try to look at it twice or three times again but in a different context of what if i was not me.
That said, I do think that confirmation bias is a real thing, and if people see data that is consistent with their ideology, they are less likely to question it. do you take part in XAI760K ? great work by them!
7:14 I think the difference between an observational study on crime and gun control with the potential for many uncontrolled variables is not fundamentally the same as a controlled experimental study on skin cream. The former leaves a lot more room for the interpretation of the results and theorisation about the nature of the relationship. Assuming that there is a "correct" interpretation of those figures when they are applied to a variable as complicated as crime is debatable.
But there are partially obvious human behavior aspects, such as the more likely a thief is to get shot, the less likely they will try it. This is proven by drive-up muggings being popular in California but literally never going to occur in Texas. Once there is a near zero chance the victim is armed and even less chance the police are around and watching, why not? It's the "easy pickins" theorem.
The questions measure what you think these results indicate, not what you think the overall effect is. You could say, "THIS study says my beliefs don't accomplish the goal measured, however I doubt that this is a valid study." Then it's back to being a math question. But I think you have to be careful. Opponents of gun control don't see it primarily as a crime prevention measure at all. Many issues are like this, a collision of differing goals and priorities rather than arguments about a technical point.
@@alexoolau How do you make training data, reward functions, and/or ensure a lack of bias when any human examples you could draw from come from biased entities? This isn't hypothetical either. For automated processing of applications they have to try and avoid bias against marginalized groups as they build them today. Like, if in your training data women are less likely to be accepted than men the program will learn to associate more stereotypically feminine names with being less qualified. Even if in theory there wasn't any bias in the origional grading (perhaps women just had less qualifications on average in historical data because of lack of access) well, now you've created a bias that the human evaluators didn't have.
10:12 The scary thing is that people don’t realize this as often as they should. Of course we’re all the same. If you’ve met 10 people you’ve met pretty much all of them, same as if you’ve met 10 cats you’ve met pretty much all of them.
10:38 I think what makes this issue particularly bad in the USA is the two-party system... In my country we have around 5 or 6 major political parties that get voted into our version of congress, two of them are somewhat similar to democrats and republicans (though tbh democrats would be considered center-right and republicans far-right in my country), but we have lots of other parties that are somewhat in between or focus on certain topics. Only having two parties only produces two options for every question because it's expected that a party has one viewpoint on a subject which always leads to a "we vs. them" where no in between options are really discussed because that would be closer to the enemy and the enemy is always wrong because otherwise they would be on our side.
@polvoazul I assume you mean 50 parties that can be voted for on the ballot, not 50 parties that get into your countries equivalent of congress. There are also lots and lots of parties in my country, probably even more than 50, but there are only 5-6 major parties that have politicians with a seat in "congress".
In all domains that is probably impossible to ever be. We can only medicate by constantly reminding ourselves that sometimes, if not most of the times, we are as dumb as a rock.
There's a few factors in play: 1. There is an emotional cost to changing your mind. Sometimes that cost is pretty heavy, so you will reject contradictory information. 2. Those of us with high numeracy have seen so many tilted/misleading studies that we automatically want to look deeper when it contradicts our held beliefs but we don't as often apply the same scrutiny to those which support them.
Don't biases arise from applying ethics to what we encounter in the world? Why would it be desirable to remain unbiased, always in the middle of whatever biases the outside world confronts one with?
Big issue: When you gave people the chart saying 'Crime improved' and 'Crime Worsened'; They both can be misinterpreted. It should've said 'Crime increased' and 'Crime decreased'. obviously XAI760K is in a great spot
Ironically the guy who answered the question with "well it depends on how guns were used and the resulting outcomes" was the most accurate, because he's saying that "the results of a study cannot always be determined based upon the statistics alone, there may be more variables that determine causal factors". It's nuance and ability to question science that is the most useful, and also highly problematic in science communication. We cite studies, we quote outcomes and take things as hard truth, yet we hear another explanation, and another perspective, and suddenly fact seems dubious. This is the problem of human reasoning--we want to be right and be right quickly. Truth takes time and nuance and a lot of work to build certainty. If there's a takeaway: be wary of fast answers and easy reasoning, if something seems obvious, ask more questions. Further--it's okay to accept things, but it's more important to be open to change and to listen to new evidence and ideas.
Very true indeed, I believe it's important also we recognise how people read(or don't properly read) scientific papers. Instead of putting the hard work of going through a whole paper, also making sure if everything is done correctly or not is not done by most people, they just read few lines and run with it! A thorough rigorous study of a scientific paper is the way to go about it, knowing about control groups, sampling biases, statistical or data analysis, the methodology used etc etc are important.
How critical I am to a claim in a paper, is proportional to how much I disagree with the claim. I readily accept something that sounds reasonable to save time. And this action seems justifiable to me rationally… it’s very time consuming and boring to “review paper as unbiased expert”
I just leave it up to the experts to critically read through and review everything, but yeah doesn’t quite happen nowadays….This is why unbiased experts are important, maybe AIs are much better at this
Can't help but notice your diatribe is not about clinical testing of topical creams 😅 It's a piece of paper with 4 small unjustified numbers in a grid, it's not that deep
It's so interesting that they have successfully divided us almost perfectly 50-50 and convinced each group that if the other party wins, our country is over. We are all stuck in this frame of mind and all closed off to outside ideas. I once experienced bias in myself. It was related to equipment in a niche field of extreme sports. It took me 2 years to open myself to the idea that I was wrong about a product and the manufacturer, which I had worked with for a while. When I finally overcame that mental block it honestly rocked my world and made me question what other areas I had these same biases.
From a game theory perspective it does make a lot of sense that it would be so 50-50 and polarizing, ideas that we disagree on are more likely to inflame into bigger issues, and each side reworks their beliefs to be more palatable to the masses only when they're behind in the polls. We're unfortunately stuck in a system that basically guarantees that there will be polarization between two roughly equally powerful parties. The issues may change, but the polarization itself is eternal
Considering one party has already made one attempt to violently overthrow the American government, it seems like this time its actually true that if one side gets elected, that will destroy american democracy. Not everything is rhetoric.
Who is "they". There is only one rational side left, there are still irrational people on it but the side itself is much more rational compared to the other side. And its not the one that has 75% of respondents denying human involvement in causing climate change.
Did anyone notice how Derek referenced at 11:58 that these things are more obvious now because we can post, Tweet... Well, you don't tweet anymore now that Twitter is X, but the cultural inertia of Twitter is so great that even in a video about science communication, you can't shake the notion of tweeting as a useful way to spread information. This cultural inertia is why even after the platform was essentially destroyed as a productive medium, it will continue to survive for at least several years to come, and why none of its competitors had a chance of stealing the market.
On X, Community Notes pointed out that Kahan didn't replicate. The premise of the video is either unproven or false. I find that amazing. We live in wild times.
I think the major issue here is that the more intelligent the individual, the better they will be at convincing others _and themselves_ of what they believe in, especially on a matter that is polarized based on a person's values which are only justified by logic post-hoc. It's already incredibly difficult to train yourself to approach a question as if you were an unbiased scientist attempting to interpret the data, so throwing in factors that are heavily influenced by emotion and cultural osmosis would tilt the desire to the default belief you've convinced yourself of many times in the past. Knowing this does indeed help you to understand that we are all humans and utilize the same tools in order to think, but I can't be optimistic that simply recognizing that will resolve polarization issues that arise from adherence to values. A person does not rationally think their way into the values that the believe in, so polarization that results from disagreements about the inherently emotional are going to be inevitable and actions taken by polarized individuals can get very nasty. I believe the best we can do is mediate where possible, but also recognize that disagreements on fundamental values are going to happen, and as such people will consider their values or cultures to be superior in one way or another to a different set of values or cultures.
![[LIVE] : ONE ลุมพินี 85 | คู่เอก "ยอดเหล็กเพชร vs ผึ้งหลวง"](http://i.ytimg.com/vi/Qq_19bnnxEo/mqdefault.jpg)
Get all sides of every story at groundnews.com/Ve and read the news with a data-driven approach. Subscribe through my link for 50% off unlimited access -their biggest discount of the year.
.
Hiii
first
you can try to prove it as much as you want with curves and graphs and stats... it won't stop this squirrel drama from making Trump to win the presidential election 🐿
it depends if its gonna help me with general knowledge for my life
Me, getting the question right: "... oh no."
It's probably a big difference if the smart people (or whatever group performs especially badly as long as they recognize themselves) have been warned that this is a "setup" question for them to fail.
That was my thought as well 😅
@@HaukeLagingI still don't understand how anyone who has considered the question for more than a few seconds can get this wrong, like.. just double the non-cream numbers and you have the answer, I guess I'm just dumb
i got it right for the wrong reason lmao, I went "Experimental and Control are not same size thats dupid, im picking Controll. cuz they had less people :3"
You might want to watch the video then.
"You are *not* immune to propaganda"
- garfield
But that one guy from twitter said I am!
@@TheOneWhoSometimesSaysOk its called X, mkaay?
no u
Garfield is the reason why I hate Mondays and love lasagna
@@mariocuric6690 Nobody calls it X, lmfao.
Veritasium: A study on skin cream is “apolitical”
USA: hold my beer.
There is a problem with every study. IF it is shown that smart people are selective in their choice of facts, then it means that scientists are also selective in their choice of facts - and this is a fact as data analysis is not only based on numerical analysis, as not ALL variables are observed, scientists choose variable to observe and sometimes even select data to use based on these beliefs.
Now as long as this is some "apolitical" topic it works usually fine, although bias is still a huge issues. But when they are paid based on the results or topic is political, then that research is affected even to a level of fraud. Sabine Hossenfelder recently had some nice videos about current state of science.
All this casues that people do not trust ALL research as new myths are created on daily basis, or people are served with ideas based on a tiny fraction of truth. Like: if EU limits emissions of CO2 then the world will be saved, even though rest of the world makes so much more.
@@OldSkullSoldier I don't think you get the joke buddy, he means that America is political and the quotes say 'apolitical'
@@OldSkullSoldier But I just worry about the systematic problems about financing really. Otherwise, it is well known that each single study might be biased. At least in natural sciences that is not a big problem. Even if an model of little use is popular, eventually a more useful model (AKA "the thruth") will win out.
"hold ny gun"
Studyon skin cream is sooooo racist OMG! How can you not see it? You are not woke enough. Go take the green pill.
I used to do a dumb statistics joke back in 2006 based on official UK Department of Transport car accident figures from 2000. It went something like this.
20% of car accidents involved excessive speed. Which means that 80% of car accidents had appropriate speed.
20% of car accidents involved somebody who was over the blood alcohol limit. Therefore, 80% of car accidents were caused by sober people.
My conclusion? Drive at 100mph while drunk is safest.
That is something I am missing here. How many people/cities were involved in the fictious rash/gun study? Because the neutral participants for whom nothing changed are also important for the conclusion. In your analysis it would be the ratio of sober/drunk people who drove without an accident.
Let’s be real, this is a correlation vs. causation type of scenario. Looking at ratios in cases like these are useless
We should test it, for science.
Nice joke😂!
michael bayes has entered the chat
Rationality is not a character trait, it's a process. If you fool yourself into believing that you're rational by default, you open yourself up to the most irrational thinking.
And, on the flip side, if you assume the "scientist" is smarter than you, you are also more likely to fall for the "Appeal to expert" fallacy and go, "But so and so in a white lab coat said it so it must be true!"
@@commentinglife6175that's not a flip side, it's the same thing. scientific reasoning like that is why peer review and replicability are core to the scientific method. most scientists acknowledge that scientists are fallible as well
This is only the case if you're wrong about being rational by default.
@@plwadodveeefdv It all comes back to confirmation bias yeah
@@plwadodveeefdv If I have learned anything, Scientist are as likely, if not more likely to make really terrible assertions based on bias. They also often let bias affect their ability to see the big picture when looking at statistics. Gun crime for example: Almost all of it is from gang violence or s**cide but lab coats will continuously preach that removing guns fixes the problem but ignoring the fact the tools of the trade just shift.
I think a big part of why people don't change their views when presented with evidence is that evidence needs a level of trust, most things we disagree about aren't things we can easily see ourselves, we have to trust whoever is collecting the data.
And we easily distrust anyone that says something that goes against our beliefs.
I'd also add that it is easier than ever to mistrust the data simply because the capabilities of the researcher are so expanded today. An economics professor I listen to makes this point very clearly when he talks about how easy it is to slice and dice the data and run multiple regressions over and over. In a way, having a more limited tool set provided a sense of "comfort" simply because it was so time consuming to re-run the calculations, we could assume someone was reporting their initial results rather than their fifth or sixth time now that the results matched what they wanted to report.
For the gun control example, I had a discussion with a European buddy who was talking about how Europe has low gun crime rates.
Like, what a surprise. It doesn't matter that acid attacks, stabbings, assault, etc. are on the rise, they don't have gun crime!
So if you see a statistic about that it's important to see trends before the change happened and what actually is being measured, if you're saying gun control reduced gun crime because there's zero guns, that's not actually measuring total crime trends that could otherwise be rising.
"most things we disagree about aren't things we can easily see ourselves"
I don't think so. Take climate change for example. Within my own life span, I myself could clearly see the massive shift in weather pattern, and I am a city guy. Those like farmers whose work heavily depends on weather, they'd be even more likely to notice the shift. Or like with g^n control, I don't believe anyone would actually have a hard time seeing the ridiculousness in things like "without g^ns, a criminal would just be as dea_dly with a knaife."
More likely, it starts from what gains or loses people have with such conclusion. A guy who's fond of g^ns would more likely not want to believe in the benefits of g^n control, and then he'd just fuel that narrative with "reasons" that fit, to the point it becomes a "belief".
I think this is definitely an aspect of it. But also changing your position in a political matter means changing your identity. And most aren’t willing to make that trade.
Everything that's wrong with the world right now in one paragraph
What struck me the most about about listening to people defend their answer to the gun control version of the question was that they were not engaging with the numbers at all. All of the arguments they made were about whether it would 'make sense' for gun control to work/not work. Numeracy doesn't matter if you're not looking at the numbers.
This is actually the biggest problem I have with my students. Not in terms of political bias but they way they don't engage with the facts presented in a problem and instead reason based on what they would like to be true. It's incredibly frustrating to deal with.
Or the guy defending gun control as "you cant make it illegal and then take it away.... like drugs" 🤣🤣🤣
That's very true actually, in light of hard evidence, we tend to rationalize against it if doesn't align with our political views or readily accept it if it does!
A good reason is given about why this happens in the video but I'd like to add that these political beliefs become a part of our identity and questioning it is like working against yourself and most people don't like doing that!
The Problem in this example results from presenting data that goes against any reasonable view one could take. Gun control _works._ That is clearly evident by just looking at all other culturally western democracies, where _none_ have the same issues as the US, and in _none_ guns are as easy to access. This specific example therefore imo doesn't work - if you present a proponent of gun control data that lacks any context as to why a ridiculous case has resulted from gun control, as in gun crime going up, then they will rightly find issue with this. Especially if you frame it in a general, specific manner ("gun control") but only specifically mean one measure ("concealed carry").
The hypothetical should be somewhat believable if you want people to intellectually engage with it.
Your experience seems to be represented in this video pretty accurately. The woman who saw the apolitical version and asked what kind of skin cream it was also wasn’t engaging with the numbers
That's the thing, these kinds of student gets so focused on their hypothesis, and when they conduct a study/paper then discovered the data, they tend to skew the result to support their hypothesis.
“The first step in solving any problem is recognizing there is one”
- Jeff Daniels (The Newsroom)
8:30 "The numbers are actually not real whatsoever" is so hilarious 🤣🤣
So it’s even
Lol ,I didn't understand that conversation btw them .
@@prajwalshivgan2847 that person missed the point of the question completely because the presented 'results' were against his personal beliefs.
@@l4nd3rThe thing is if your approaching random people in a political context, presumable a rally, and asking them these questions, they aren't going to interpret that as you asking them a math question, they're rationally going to interpret that as you asking about their beliefs on the issue
@@hadhamalnam the difference being? Lol.
Smart people doing worse? Finally, my moment to shine!
Political people are not smart
Smart comment!
Made me laugh, thanks friend.
😂
I can say you are a boy just by your name and pic
Very subtle and clever wording in the title. I filled in the gap myself with "Smarter people do worse than dumber people", when in reality it was "Smarter people do worse than they are expected to"
crazy i did too, probably an ego driven reaction
Well, if you write a statement like "Group A does worse", is that not objectively equivalent with "Group A does worse than Group {Not A}"? Certainly "Group A does worse than it is expected to" is not the most plausible interpretation.
Yes, very clever, like a pop science magazine (or channel) wanting to get more clicks
It's not "clever", it's just deliberately misleading. Same as the rest of the video, where the questions shown on screen don't match the ones they showed to people (which should be an immediate red flag about the honesty of this "study"), and where they count "the number of cities" instead of the number of _crimes._
Same, it's misleading
Since I have a background in clinical trials and experimental design, I got fixated on why the groups weren't equal size (since the people or cities were supposed to be assigned randomly, why would you end up with two different sized groups?). I would have just ended up questioning why the research problem was trying to fool me. Does that make me more or less rational?
less rational since group sizes do not matter if they are great enough
Yeah, also one of these is a controlled randomized experiment and one of these is a study. People are well within their rights to question the methodology and impose their own interpretation of the correlation on the study group as the correlation doesn't necessarily imply causation.
With cities it would be natural to have different sample sizes - the number of cities matching criteria is limited and there's a lot of "noise". You're not conducting an experiment here - you're trying to gather observations and you grab whatever you can.
With cream, however, you're definitely right - the obvious thing to do would be to get equal sized groups. You COULD end up with some differences due to people dropping out of the experiment for some reason (it did last some time and stuff happens) but it shouldn't make one group several times larger than the other.
6:05 - "crime improved" and "crime worsened" is a strange wording
Improved means decreased
@@LooperEpicwhat if I improved my criminal methods
@@LooperEpic yes, that's how it's usually worded, to avoid confusion.
@@LooperEpic it also means "people are getting better at doing crime"
"Turn up the air conditioner" / "Turn down the air conditioner"
6:04 Why did you label it "crime improved" and "crime worsened"? That seems so much more ambiguous than "increased" and "decreased"
To make it work for this silly video
Because study participants are famously so dumb that they don't know whether crime increasing is worse or better.
Only if you're a dork enough to think that more crime is an improvement
exactly what i was feeling
I no longer understand the table with those labels. You’re saying almost 300 cities outlawed guns? I don’t think there are even that many cities
I work as a strategy analyst for a major retailer and on a weekly occurrence I have a suspicion or belief that something may save money, or be a better solution for the business. As I research and analyze data I am no longer surprised at how often I prove myself wrong and have quickly learned that data supporting for, against, or even no correlation for either are all equally valid answers. It’s not about getting the results I want but rather getting results period. I’ve applied this same practice to other aspects of my life and it has been very freeing (and humbling). Thanks for the video.
The ability to be humble and see our misconceptions when they are presented to us is really rare and really what we need more of right now
What was the most surprising thing that saved money or cost money?
I agree. Most often we want our ideas to be right, but verifying it against real data is necessary.
I would rather not disappoint people than lead them on with a fabrication.
One of the most important trait, and the one thats more difficult to learn as a Data Analyst it’s being impartial. Don’t putting one’s belief into an investigation, dashboard, or report it’s very important to learn what’s actually behind the data, not supporting your own idea of it and don’t manipulate the results to skew the outcome you want to have
Derek, that solution you were talking about not having is right here.
I'm an ethics teacher and I have used ideas from your videos in the past for my own lessons. I think you do such a great job with these videos. Always love to see them!
I feel that a lot of comments are missing that a meaningful point of the video is that the title is missing the words "...than usual" or "...than expected."
The "high numeracy" people still typically do better than the rest, just not as well as they are EXPECTED to do.
When the video says "worse" it's in comparison to that demographic's EXPECTED outcome, not worse than the "lower numeracy" groups. The surprise is that by following intuition, "higher numeracy" respondents tend to perform comparably similarly to the "lower numeracy" respondents' results, in certain cases where their beliefs disagree with the presented information).
the title is kinda clickbait. High numeracy people do better because it is literally a numeracy question. It should say, "political beliefs can overshadow numeracy when polling people outside a political rally."
actually I don't think that's the case as you can see in the graph the high level numeracy people had the same level of accuracy as the 0 level numeracy group. Which indicates they are the exact same, regardless of how they are expected to do.
Yes, and in the end it's basically the same conclusion that Conscientiousness and IQ aren't correlated.
AI is only hope for correcting human tribal bias.
Yes! The data basically shows that people willingly turn off their own numeracy and choose to go with intuitive thinking when the numeracy would lead them to conclusions they dislike.
I think the problem is people are not relying on the data in the fictional scenario and are relying on facts they believe they know. If you showed them a scenario that showed gravity was driven by air pressure, you would have a hard time getting people to believe that even though the paper showed that.
AI is only hope for correcting human tribal bias.
@@alexoolau More like the hope for tribal people to propagate their tribal bias under the guise of unbiased facts...
@@alexoolauabsolutely not lol
@@alexoolau if you feed ai your biases, it will regurgitate back your biases
I think this points toward a deeper question that I'm not sure was answered in the study: how rational is it to accept new information and abandon prior information? If the quality and quantity of previous evidence is high, asking a person to suspend that understanding for a guy with a cardboard sign so you can play with numbers is a factor that's hard to ignore.
When you change it to political, the distrust of the data skyrockets. All sides are used to being lied to. What this says about "science communication" is significant.
This was my intuition as well. If I was showed a chart with 4 numbers about gun control and crime I would be like I am not drawing any conclusions from these 4 numbers, please leave.
@@TheMustyrusty Unfortunately for many people there is no such thing as a trustworthy source if it contradicts their current understanding. 4 or 400 numbers, the results wouldnt be too different.
Science communicators lie, too. PIs on studies with big funding wildly misrepresent their studies all the time.
@@agafabaWhich is why we have millions of people who firmly believe that Greenland is facing a threat from unprecedented high temperatures not seen in 10000 years.
@@toomanymarys7355 This comment would be hilarious if it wasn't so sad...
I don't know where you get the idea you get for your videos, but damn... I appreciate you, love having videos like this.
I don’t even understand the question. Where does that put me
Einstein level. 😂
Either low numeracy skills because you don't understand it on any level or high numeracy skill because you understand it's underspecified. Sample size of the control and test weren't specified and its not necessarily reasonable to assume that control and test sample were different sizes which is required to reach the 'correct' answer.
man, i wish he would have paused the video a bit right after he asked the initial question with the skin cream. I immediately was told the answer by the first guy and had no time to make my own. Now i'll never know if me dumb or smarts
In the hood lil bro 👊 😎
That would be the Democrats xd
7:50 "illegal crime... I think."
That's hilarious
Why did you label it 'crime improved' and 'crime worsened'? That seems so much more ambiguous than 'increased' and 'decreased' wtf all of you in XAI760K ??
I think it is probably an improvement if crime decreases
Because not all crime is bad? Idk i agree that it is intentionally ambiguous.
"That seems so much more ambiguous"
so what? it does not affect the numbers which is what he is asking about
@@bighillraft just realized we are talking to a bot. Anything with XAI760K is pushing some crypto nonsense.
Waait you're right... does crime improved mean that the crime got more sophisticated?🤣
Teaching how to get immune to propaganda and not spreading it while a lot of people are doing so, this channel is the best.
So you’re saying politics makes people biased?! The hell you say!
Point is - it makes more numerous people more biased.
Politics make smarter people even more biased - not something you'd expect
@@michajozwiak7650 What do you mean? It made all people of all levels more biased based on their politics, their actual intelligence wasn't actually related to the trends when it was a partisan issue.
No. They are saying the smarter you are the better you are at manipulating the numbers to see what you believe. It’s a surprising conclusion that I’ve heard before- smarter people are more ideological because they are better at manipulating data to “prove” their ideology.
@@swparsons true, the smarter you are the better you are at manipulating yourself and others
I really like how humbly you try and deal with a topic like this. Well done on this thought provoking video
Bro, my intuition went straight to comparing the proportions. Picking the number just because its the biggest one isn't intuitive at all. Ct's called guessing without thinking. Cuz bigger is generally better
Yeah, literally any smart person thinks that way
I think the "smarter people do worse" is meant to apply to the politically "charged" version of the question, not the skincare one.
Same here. Actually, my first thought was why didn't the same number didn't get the cream as not.
The "higher positives" winning the argument (ie Poker maths) is kinda cultural thing
I think that if they did this exact same experiment somewhere else like in China or Japan for example - I'd expect a different result
Even "stupid" people will look at how many sample is getting worse, and its much higher
All this tells me is that people who can't do proper data visualization shouldn't post papers.
Only Veritasium could make a political video that increases my faith in humanity
"increases my faith in humanity"? Every time I watch a video where he does a survey, I lose faith in humanity. Are people REALLY this dumb?
It's not a political video though
Gaytasium vitamins for harris
One spoonful for you all
to me it decreased.
but it also made me hope that things can change.
Me: 'I'm surprised Derek isn't making a political science video so close to elec.....ooooooh I see what you did there.' how long is XAI760K ready for us?
It seems like tribalism explains a lot in the current political moment, but I think it's one step removed from the primary cause. The problem is lack of trust. In that context, my "tribe" is just the people I trust. That said, tribalism isn't the only engine for trust - it's just the easiest. The way out of this is to restore people's trust in more than just their political cohort. Rabid individualism and self-serving institutions have done much to create the current crisis of trust, and building reliable institutions populated by people willing to speak & act in the best interest of those institutions is the best way out of this mess. Unfortunately, I don't think we'll reverse this political/intellectual culture of self-harm before we hit rock bottom.
For the real world, I agree with you that tribalism is the strongest factor - BUT some of that is that most people simply don't have the time or trained analysis skills to examine data and figure it out. So, in that sense, some part of tribalism is just an optimization of time usage (in a strange sense - sort of like thinking slow / thinking fast situations).
Check out social identity theory and the minimum group paradigm. It isn’t really a lack of trust. It’s tribalism, and that tribalism causes us to lack trust in people who aren’t part of our group.
Tribalism is good, actually, and the people who argue otherwise are inevitably the ones who want to erode their competitor's tribe to the advantage of their own. No one who complains about tribalism will ever go out to find common ground with their political enemies, or seek to find understanding of why they believe what they believe.
@@SomeGuy-ty7kr People tend to recognize tribalism in other people but not themselves. We humans are very good at rationalizing our behavior in terms that make us feel virtuous and smart, so we're often quick to think in terms of " that team bad" than we are to figure out why we believe what we believe, and why people who believe something different, or even incompatible, need not be stupid, wrong, or evil. It's easy to blame their tribe.
That said, I'm not sure tribalism is either good or bad. Many people think religion is bad tribalism, and point to the millions killed in religious wars as proof. But Catholic Charities and the Salvation Army are also religious tribes. Tribalism is inevitable. Politics has been, and will always be tribal. That's why it's important to recognize that our problem isn't tribalism, per se, but that we've lost so much trust in each other and civic institutions that trust in our tribe is all we have left. That's not good.
@@livemusicfannc It's always been true that most people lack the time and/or background to verify most facts. That's why trust is important. When we can't decide for ourselves, who are we going to trust to decide for us? If your answer is, "That radio talk show host on the MAGA right." or "That social justice warrior on the woke left." you're probably going to be wrong about many things that have nothing to do with populism or social justice.
Just some thoughts that I think help/hurt why people would entrench or fail to engage with the numbers:
- A skin care study is assessing the entire sample of people effected by the cream/non-cream (100% of the people in the study were measured) while ~500 samples in a gun control study would barely scratch the surface of all of the potential variables and implications of different crimes (boiled down to one number)
- Political topics are inherently more difficult because its harder to quantify things like "control" and "safety" as these are relative terms that heavily change based on the population
- There's an inherently emotional response about gun control/violence because it directly impacts life and death, along with people being passionate about whichever side they feel best serves the problem
Overall, I think reducing a study about rashes down to a number is reductionist, but logical for people. Compared to something that they feel is inherently more nuanced (due to the emotional side of their reasoning framing it as logic) they would likely substitute (probably accurate) flaws with data collection processes in general, but ones that just so happen to agree with their beliefs. Confirmation bias at its finest haha.
This is similar to a video I've seen by After Skool, on "Why smart people believe dumb things." Apparently, stronger reasoning abilities can help you justify your views, instead of rationalizing your way to the correct conclusion. You start with a belief, and work on justifying it, instead of starting by looking at the evidence, and working towards the conclusion.
Also something called the Nobel Prize Effect, I think? Where Nobel Laureates believe in pseudoscience and all sorts of mystical nonsense, because they are outside of their area of expertise.
The solution that comes to mind is constantly questioning your beliefs and assumptions, I guess, which is easier said than done.
I think your conclusion "that the solution could be constantly questioning your beliefs and assumptions" is what means to be "interested in science", since "falsifying your hypothesis or symptoms, is the key of science"... people that reject to analyze they can be wrong and only take into account what supports their view, have "confirmation bias".... which is, in my opinion, the key of pseudoscience... The more knowledge one have the more resources to justify or defend your bias.... Knowledge (to comprehend how things work) does not equal to Intelligence (how knowledge is used).... So it's possible to have "stupid knowledgeable" people and " uneducated smart" people as well. :)
Know i would have loved to learn which variables were considered as evidence of people "being interested in science" versus people "having knowledge in science". :)
Hearing the people talk about how they got to their wrong answer reminds me so much of listening to people add in all the caveats in their answers to simple philosphical questions (such as the trolly problem) to try to escape the point of the question. Adding in "well I'd have to look into the studies and the sources that they cite" is just missing the point of what is being asked. It is a completely fictional result that they are being asked to assess.
Well, I work in science. And if there is something that people don't realize, it's that intelligence comes in a lot of different flavors... I know many scientists that are brilliant but totally stubborn and dumb in the face of truth. Having an open mind, a critical mind, and a curious mind, are not necessarily because of intelligence. Many people with high IQ have the most blown up ego... Meanwhile you can find the most humble farmer in the most remote region, who can't read and can't do math, but has the most progressive views on human rights and the environment... Great men aren't all smart men, and a lot of smart men are just a bunch of cunts...
Something I think hinders the reasoning process it that people think logic is the highest form of reasoning. It's really not, it's just the building blocks of reasoning: the fact that you can stick 1,000 bricks together doesn't mean you've built a sound house. The true path to sound reasoning is objective rationality: where you continuously try to prove your own conclusions wrong, while searching for valid points within wrong arguments.
Things like this are why I strongly dislike people who say things like the solution to the problem is education or critical thinking. Here is data that shows social and emotional factors will cause those skills to be applied to things like creating confirmation bias.
The great problem with data in politics is that it's SO EASY to manipulate studies. Each side has a block of 'Institutes' that conduct studies where the study design virtually guarantees your outcome. Minimum wage for example. I saw a meta paper examining the state of economic studies on the minimum wage. Sure enough they ALMOST divided perfectly into halves. The ones that chose one methodology for trying to establish controls showed no effect of minimum wage on employment. Those using a different method to control for other factors showed a substantial negative impact on hours worked. I say ALMOST because there were two outliers that used the methodology that did not show effects, but those studies DID show small negative effects. But generally, which methodology you chose ABSOLUTELY DICTATED your outcome. Then, politicians from each side go on talking head shows and talk like the only studies that exist are the ones that agree with them.
And we've seen how this can happen even in hard science. How many studies showed that [insert anything here] causes cancer? And then you read the particulars and find out that they stuffed the mice full of 10x their body weight of that substance every day for months. And that's before you get into p-hacking or outright fraud. The entire anti-vax movement came out of a single study that was a dude just straight up lying about his data because he wanted to be able to sue the drug companies for big money. He's ADMITTED in court that he lied about his data. But the antivax movement remains strong.
So the term 'Studies show...' is the most overused starter phrase in politics. And the most worthless. And people know it, so they're RATIONALLY skeptical of the 'evidence'. What's needed more than anything is better agreement BEFORE a 'study' is done on what are and are not valid methodologies. There needs to be tighter peer review. And any paper that doesn't go through peer review shouldn't be reported on. No pre-print reporting. Don't waste our time telling us what the National Institute of Pre-Determind Biased Outcomes concluded about the effect of abortion on crime or whatever. But that doesn't sell ad impressions, does it?
Which is why people have to actually be trained on how to read studies (including unfortunately a number of people that do meta-studies). And observational studies are even more prone to this problem. Sloppy selection of subjects or situations, poor controls, etc.. All too often meta-studies just throw a bunch of stuff together without any deep analysis of which studies are even good basic science
And this is why I bristle when liberals (they do this more often) will get mad and call names because they have studies done by thinktanks but conservatives won't believe in them. Its because conservatives distrust since they have seen so many biased stories. Its not they are too stupid to read (always) just cautious in trusting everything. This is why liberals are like press = listen to everything they say
A Rice professor and the former mayor of Houston got caught red handed. The professor was hired to do the study. It didn’t give the desired answer. A new check was cut, a new study designed, and voila, it supported the mayor’s position. 🤣😂🤣
You bring up good points. But the autism thing is interesting. If we're talking about a proper scientific RCT experiment there is literally no proof they work, because they were never done. And people who then claim the polio vaccine cured polio are proven wrong by history, it already went way down before it got introduced. I too thought these people were nuts 4 years ago. But then again, we saw what happened in 2020-2022. So when circumstances changed I reexamined the claims and now I changed my belief. But it's definitely difficult to tell if it's one way or another for most vaccines.
@@nunyabidness3075 These days given the bias/manipulation/fraud it's about a 50/50 whether a study is right or wrong. You might as well flip a coin.
"See mom? I scored bad on that test cuz im too smart for it."
"See mom, i failed thr test cuz guns were banned!" 😂
see mom, banned thr failed were test cuz i guns
thr
He quickly calculated that death by firearm is 11x more likely if you're black, skewing the crime rate affectable by policy for him to be near zero, and thus dismissing those heavily diluted numbers.
@sugy747 I need the stuff you are taking , looks really effective.🤣🤣🤣
Since your "fits my rule" video (on confirmation bias, can't remember the title) I try to take an evening, every now and then, to prove myself wrong on something I feel strongly about. It has helped a lot! I either get a better understanding of why I am right (if I failed) or I learn something new, and change my mind, which is just an awesome "aha!" moment 😁 I keep a list of things I changed my mind on (and sometimes even changed back) to remind myself. I hope it helps. Thanks for teaching me that!
Can you share examples from your list?
How do you attempt to do that?
@@itstruce. Here's one: I thought I completely understood why and how the Keto diet worked. Even advocated for it among friends. Turns out its main hypothesis is disproven, and (in short) it doesn't work in any special way unless you have epilepsy, but may be easier to follow for some. Edit: It was an example of what some call "Mechanism/Mechanistic bias". When the entire explanation seems logical, but isn't actually backed by good science. Usually means: "It's more complicated" or plain wrong.
@@Wrackey thanks
@@Wrackey The one that did it for me is "market efficiency". In theory, products will get cheaper, while getting better. But when markets make the measure of profits as a target, it ceases to be a good measure.
This perfectly explains the disconnect of quality, price and volume production.
The problem with this experiment is when you ask people to consider a study on a nameless skin-care product that they have no prior knowledge of, you're testing their ability to come to the correct answer based on the numbers. But when you show people a study on a matter they probably have a lot of prior familiarity with, like climate change or gun control, you're not testing their ability to find the empirically right answer, you're asking them to rationalize the study itself based on all of the knowledge they have on the subject matter.
The questions shouldn't be does "X lower gun violence", or "does Y affect climate change", the same way you would ask "does Z reduce the rash". You would have to ask them does THIS specific study show a reduction in gun violence, climate change, etc.
I was going to comment about it, on the question about gun control, there's no two sides over here, the data is there on an international scale, having hypothetical data and asking people to ignore all of their previous experiences is unreasonable.
That format also calls for people to take a public position on camera, not knowing if they're being tricked by an activist's video.
Plus, while gun control supporters consider it a question of lower vs higher crime rates, many/most opponents of gun control consider it a civil liberties question. That is, even if crime goes up, they'd still support it because it's a civil right, rather than a purely functionalist policy question.
That said, confirmation bias is absolutely a thing and this was a good video.
@@crafty1098 Even in the case of a pure numbers question that graph only states total crimes being worsened/improved. It doesn't state whether the thing which was just made illegal was included in those 'crimes'. Imagine if you made cars illegal, there'd be tens of millions of criminals overnight, crime is skyrocketing! But it doesn't state what 'type' of crime is occurring/not occurring, only that it is/isn't. Further it doesn't explain the *other* side of the equation, and that is the numbers of people saved via guns, which is as far as I am aware, like... inordinately disproportional, like 10x more people protected with them than harmed by them because there are just far more law abiding people with guns than criminals with guns, by default.
It's one of the reasons that data like that is inherently biased. Even if the data showed that gun crime went down due to a complete and total lack of guns, if you don't also look into if other forms of crime worsened or improved because of that decision you're being disingenuous at best, outright manipulating people at worst.
This is a critical point. It's kind of like like asking someone what's 2+2 and then asking them if the data showed that 2+2=5 would that change their previous belief. If people have reasons for believing that 2+2=4 they are not going to immediately submit to new data without vetting it (that would be completely irrational). In my experience, most people are just not rigorous in vetting data, which is why emotional arguments are so common in politics (they are often more effective than trying to have productive debates).
Your point is valid, but i think the video is still fair. I would also predict that the rewording of "did THIS study show a reduction in Y?" wouldn't change much.
Here's why: in Psychology there is a term called "flexible thinking". It's brought up in decision making research. A person who can think flexibly will reduce or block out the influence of their prior knowledge when making a decision from information provided.
However, most of the time, when a decision is difficult or information is hard to process, we get cognitively lazy. So instead, we rely on a belief we already have to make our decision.
What the video showed me is that people with higher numeracy are less likely to think flexibly and only consider the evidence if it contradicts their beliefs. That's a pretty interesting result
“The first step in solving any problem is recognizing there is one.” - Jeff Daniels (The Newsroom)
10:17 The use of the word 'scary' in this context is, in and of itself, scary.
It went from cream and rashes to politics extremely fast
[starts a rumor that the skin cream has 5G in it]
There's more crossover between those two topics than you might be comfortable knowing about.
Never trust Big Cream
It went from intuition to bias real fast
I
I don’t get the rash question because if more people are improving, then why isn’t that the correct answer?
Aren't people with higher numeracy more likely to have opinions formed from previously investigating data on the most hot-button topics? How do you control for that in this experiment?
They should still correctly interpret the numbers from the single study, as oppose to relying on previous information to assume what the numbers state. Now, should one study change a person's opinion on a topic they've previously researched? Probably not, but numbers are numbers.
because people with more education or higher social standing tend to have more confidence in the correctness of their world view.
^ this is the correct answer. People who think they are smart are rigid in their beliefs and are unwilling to change.
If people were "previously investigating data" you wouldn't have such polarized, opposite beliefs. You would be drawing the same conclusions.
They explicitly told people the data was made up, and to interpret the numbers. Any prior data should have been irrelevant because all they needed to know was in front of them.
This video is like the study in the Video. The Title has nothing to deal with the Numbers that are presented.😂
Correct title should be "On these questions, smart people do not do better" (which still requires ignoring the average joe and Einsteins in 6:33)
Title should be something more like, "People with the capacity to analytically consider data may take shortcuts to their preferred conclusion." But... I'm guessing that would suck for SEO purposes.
@@michaelkruse9818 the title is 100% correct, they in fact did do worse then when give the study with guns compared to the study the cream
I cancelled the video and disliked it
I really dislike clickbaitint titles in generell
@@wax2562 I mean... is it? Are people with higher numeracy scores "smarter" than everyone else? And when the title says "these questions," it pretty clearly suggests specific questions which disproportionately confound "smart people," rather than abstractions designed to elicit an emotional/tribal response, where "smart people" answer the question with accuracy similar to that of not-"smart people."
Great episode, great timing ❤️💚🤍
I think it's very important how you phrase the question. "Based on these numbers did gun control increase crime or decrease crime?" is very different from just showing the numbers and then asking if gun control is effective.
You can agree that the data leads you to a certain conclusion but still disagree with the methods with which the data was gathered or presented.
"Increased crime" is indeed ambiguous in these sorts of things because, when you make something illegal, there will of course be more crime, because something was just made illegal
yes, it did not seem that he asked in a very standardized way watching the video. And the one person did not even seem to consider his data sheet really. But then again, I doubt they really wanted to reproduce the result as they said (they did not show their own statistics). They just needed some people say weird stuff to fill the gaps in the video and make this more entertaining. I am sure in the paper they were more careful.
And if you make gun ownership illegal, you are going to get people committing the crime of owning a gun, because it is now illegal. In the UK for example, there are about 6000-7000 recorded offences every year related to gun ownership which would be eliminated if gun ownership were made legal. But the 29 gun homicides per year would likely increase, to somewhere nearer 10,000 if it ends up being anything like the USA.
This. When I realized that the study was about politics, the wording became far more apparent that it was aimed at leading than observing. At that point it had nothing to do with "intelligent people do worse than unintelligent people" and had everything to do with "political bias taints how we interpret data", which includes the way that the data was obtained for this study. It's worthless when you lead people towards or against their biases instead of asking them to make a rational observation based on the data and numbers in front of them irrespective of their political biases. I'm pretty sure there is a rule about not doing this in your data collecting methods when it comes to peer reviewable studies.
I'd say it's implied in this hipotetical situation that gun control is the only factor in the increase/decrease of crimes.
The fact that you need to talk about phrasing and justifying the answer for the "gun control" case IS the point of the video, because if they asked you about the "skin cream" case, you would just accept that it's a simplified problem and directly answer if it helped or not. And a perfectly rational person would answer both questions the same way.
@11:18 the problem is that is purely reactive policy/action, which is the most expensive kind of action. To be proactive on something, you have to have agreement on causes and work to fix those. We've forgotten the old sayings of "a stitch in time saves nine" and "an ounce of prevention is worth a pound of cure". We see the same thing these days with the Y2K issues we had. Since a bunch of companies spend massive amounts of time and money *proactively* fixing things so that "nothing happened", today we have a bunch of people thinking Y2K was all just a big hoax. And that kind of thinking has continued to everything else. We won't be proactive on anything, we'll just have to reactively deal with the symptoms rather than proactively try to address the causes.
The other side of this issue is that to come up with the solution to the cause, i.e., the thing you do to fix the problem, you have to be sure of the cause itself. Part of that is understanding that if A causes B, then we can use the existence of A to predict B. Unfortunately, when it comes to some issues (like climate change), the predictions are faulty - illustrating that we actually do not know the cause. Why would I spend money to fix this problem you identified here if you can't show me it will actually solve the issue you claim it is solving? To put this in a non-political example, if I say my car won't start, telling me to put air in the tires won't do me a lot of good - even if the tires are flat!
@@commentinglife6175 I suggest you look at climate change prediction more. Even the predictions that Exxon made (and hid) back in the 1970s are reasonably accurate.
We need a combination of short term and long term problem solving. The issue is that long term work requires trust (which we're lacking in the U.S.). But successful cooperation on short term problems is a good way to foster the trust needed to tackle the long term stuff.
@8:24 That's a man who believes he is way smarter than he is 😅
That’s you average American which identifies with freedom for you in a nutshell😂
@@Leonardo-hy1foAnd where are you from?
There are reasons why Iceland is the country with lowest crime rate and Venezuela is the country with the highest crime rate. Hint: Guns don't crack even the top ten of these factors.
12:12 me?! I would never
😂😂😂
8:35 That is pretty much what I expect from someone with a bottle of "Smart Water" in their hands :D
This video went in such a different direction than I expected and I love it. Great video! Very thought-provoking.
Damn, this video just brought a brand new unexplored part of my subconscious to my conscious mind, I'm SO grateful for this channel.
Sorry, but this video is pretty bad. In most of the interviews you can see people criticizing the validity of the model, or modelling. This makes perfect sense. It happens less for the skin cream because most of the respondents do not have a concern about skin care.
yeah but some of us DID care about the cream, and KNEW the stupid model was stupid...
But the exercise isn't to evaluate the validity of the results
@@peterfox7663 Not 100% sure what you mean, but... what I see is people refusing to do something they consider bad just because a random strange asks them... and I mean it in both direction here.
@@MarcoServetto The validity of the model is irrelevant, though.
@@peterfox7663 People are more than permitted to question the validity of the exercise, especially if they don't know what it is.
I love that this took an anthropological spin. Acknowledging your bias is a very difficult task, but important to keep at the forefront of your mind when analyzing a new information. Experiments like this demonstrate it beautifully. It never hurts to ask yourself, how does my culture/society shape my worldview?
3:03 I hate people like this...
Anecdotal Bias. When you only believe what you see.
Gramps only trying to figure out if she wants to use cortisone or eurax. god bless her soul
"I did my own research"
@@BrandonWong55nah, research would include taking these numbers into account. She is rejecting research because she doesn't understand it. She wants a simpler opinion that is not founded on data.
What an amazing video! Very insightful.
8:30 made me laugh so hard. it's not real bro
there's always that one person
There was SO MUCH MORE discussion about the actual numerical figures in the skin cream version than in the gun control version!
They were similarly presented to the people being interviewed, during the video demonstration for us, it didn't go in depth in the second case to avoid exact repetition
Who do you believe more: "Acshually 🤓"-guy with fake numbers, or yourself?
Smart people know that if gut feeling tells you against data, something wrong with data because it made up by much more stupid people. Misleading data is everywhere.
In this particular case they again was right, because they spotted that data was fake, which is true.
Uh, yeah, that's the point of the video
you (and so many other commenters) literally doing exactly what the video is trying to explain... and it's quite amusing
You saw a few people from a large number of people surveyed.
2:02 Why did I expect him to say "onet housand one hundred and onety one" 😂
Great video man! I totally see how this would be the case. I've been convinced of something that went against my beliefs based on data, but it's rare.
The difficult part of trusting "studies" is that they're often made up with conclusions only drawn using p-hacking (which you have an excellent video on).
There’s a few different meanings of the word “rational”.
One is using logic and facts to come to a consistent conclusion based on truth. This is the usual definition we think of. This could be called philosophical rationality.
The other comes from economic literature. A rational human in the economic sense is one who pursues their self interest and doesn’t sabotage themself.
I think there are many more people who are rational economically than are rational philosophically. I think it can be important to remember that even if someone is supporting something that isn’t 100% true, they’re just trying to act in their best self interest.
And it's worth noting that both types of rational thinking are vulnerable to Confirmation Bias, because the human brain is very, very good at discarding information that goes against our pre-established world views.
There's a problem with how you formulated the rationality in the economic sense, because with this formulation everyone will be considered rational, because everyone pursues what they believe is their self interest. At that point, the term loses usefulness. What it should be about is being rational in the economic sense meaning that one is pursuing that which is objectively in their self interest, rather than subjectively. Because that's where most people fall short. That's the hard part because it requires perfectly aligning oneself with actual objective reality and actual truth, as this in practice is not even possible to perform perfectly as the world is too chaotic and applying precise bayesian reasoning to every little thing is computationally explosive.
I find it useful to split the meaning of the word rationality the way Yudkowsky does:
1. Epistemic rationality: systematically improving the accuracy of your beliefs. (truth)
2. Instrumental rationality: systematically achieving your values. (winning)
And I love the way Vervaeke puts it: In practice, rationality is knowing when to use logic.
9:28 random Maratha reservation still haha
Thought I was the only one to notice haha 😂
Real
What is a Maratha Reservation?
@@griffinshorts785 just Google it
@@griffinshorts785
Chat GPT response:
The Maratha Reservation Protest is a movement in the Indian state of Maharashtra, where members of the Maratha community are pushing for reserved seats in education and government jobs to improve their socio-economic conditions. Historically influential, many Marathas now face economic hardship and limited job opportunities, which has led them to seek inclusion in India’s affirmative action system for disadvantaged groups. The government attempted to provide this reservation, but India’s Supreme Court struck it down, citing legal limits on quotas. This has fueled protests as Marathas demand a solution, making it a significant issue in Maharashtra’s politics and raising broader questions about balancing affirmative action in a diverse society.
As social creatures, it is easy to become misled by those around us for sure. Logical thinking is so hard to come by now, but these videos definitely help in spreading the word.
Thank you for all your work and this video in particular. Stellar job
1:59 "math doesn't help". Goes on to explain how to solve the problem with math.
Ngl I barely used math i just knew based off the amount that didn't work to did work
@@C.S.Argudo🐒
Cause the mathematician is the problem 😂
I think a big difference between the cream example and the gun control example is how abstract the question is framed.
For the cream it is just a maths problem, but for gun control it is a question where there are preconceived notions that prevent the person from interpreting the data, or even really looking at it, expecting the data to reflect their personal beliefs.
Yeah. Umm. That's what the video is about.
I was looking for this. This seems like another poorly designed study.
Participants answering the skin cream question interpreted it as a math question, and were evaluated as if it was a math question.
Participants answering the gun control question interpreted it as a social policy question, and were evaluated as if it was a math question.
I read through the study. There is minimal effort to inform participants that they're evaluating just the data presented, and the choices they're given are phrased as "cities that enacted a ban on carrying concealed handguns were more likely to have a decrease in crime" rather than "this data supports..."
The cream "control" question is also a very poor control. Rather than being simply apolitical, it's presented as entirely hypothetical, in expected math test fashion, rather than being a real world question like the gun control question.
And here again, we see how the debate about "how was this research done" can be as important, if not more important, than the research itself.
sounds like the same back peddling the people in the video were doing. you're doing great!
"Personal beliefs" here also can just be "correctly being aware of scientific consensus". It doesn't matter how many fake studies you show me data from to explain that climate change isn't real. My answer will be that climate change is real, because that's just a fact.
this is the exact point where "isolated study" becomes something that is NOT taken as "isolated study".
you can agree that the study says something, and disagree that the study is correct or was done correctly, which is what I think had happened here.
especially with the method of the study changing from lab experiments to crime statistics, the former being a highly controlled environment with low margins of error while the other is marred with statistical error margins that often eclipse the sample size by a sizable factor.
with that, people are often biased to pool from intuition and prior experience instead of the currently provided information as it is deemed "useless", which is what likely happened here.(the actual information provided to the people hearing about the study was useless as analysis of crime statistics require a LOT more than "got better" and "got worse" graph)
The two being treated as equivalent is such SOP in the soft sciences. The overall conclusion that politics (or any strongly held belief) can cause people to irrationally reject data was fine. Talk to someone about their favorite sports team/star. And, yes, today politics is a team sport with equally stupid fans.
@@barongerhardt It's not a fine conclusion at all, because people with real data about a real thing are being presented with fake data in the most unreliable way possible, and the study is assuming that will not have an effect on the way people answer. People don't know any data about skin cremes and those are presented as controlled laboratory studies. People DO know things about political topics, and they're being presented data as uncontrolled, unvetted statistics. It's not difficult to see how that could completely skew the data that they are attempting to treat as a fixed variable. Completely different types of data, completely different contexts, and then a causation is being provided based on the faulty correlation. All huge red flags of a bad study with bad conclusions.
@barongerhardt but is this study actually showing those results?
I'd argue that this study shows no real data because the hand cream example is not a proper control group for this experiment due to the factors I mentioned in my original comment.
Excellent work on this thank you
Whenever I see poll data of any kind, I am most wondering about the questions asked, how they're asked, etc. I see so many studies where I can predict "unexpected" outcomes based on poorly worded questions, multiple interpretations of choices, whether it was multiple choice or open-ended, etc.
There's one other thing I want to know specifically for the political version of the question. As the skin cream is made up, no one could have read a previous study on that exact skin cream. Some people could have read existing studies on this exact political question, and I don't mean "read a random article online", I want to know if having read a peer reviewed paper on this topic has an any difference on the outcome.
A big one is where they compare countries. I saw one where they measured agreement with the statement "I trust most people, in general" and the results were completely messed up by the translation, because other languages express it using words that are stronger/weaker
But previous data wouldn't have changed the data presented here, so it's just another cause for bias. The participants were presented numbers in a vacuum and couldn't keep it that way before answering, that's the point.
@@GreenOnionBrother at least for the video that's not really the point, since its second half implies this matters because it is representative of the way people answer things outside of a vacuum. Also, just because you say "please consider in a vacuum", doesn't mean people can simply shut down their beliefs.
And who knows, one might also argue that the difference between cream and politics is actually not down to "bias" so much as it is about having or not having context.
@@user-sl6gn1ss8p I don't know what to tell you. They were presented numbers that only allow one conclusion, but due to bias (and this includes previous studies and statistics they've read, regardless of their validity) failed to answer a question they would otherwise have less of a problem with. That is the point. How bias affects or rational thinking.
@@GreenOnionBrother my point is just that it was not clear to me, from the video, how well this could disentangle "bias" from "rational thinking", as in, how much can this actually show that people might make worse decisions in this sense due to their bias - the "tangle" being the fact that this bias may include, as you said, perfectly valid information.
To be clear, I'm not disputing that the effect exists - I'm just not sold that the study shows that much, going by the video.
Maybe I misunderstood, but the study showed more that if you present the data in a bad way, people don't look into it much unless it conflicts with their worldview.
This is exactly it! The study is more about confirmation bias and criticality of research methodology than it is about the getting the "answer" wrong.
This!😂
Yes I doubt if you gave the question in a math test, there would be any difference from the worldview. In this case most people just assumed that the data you show me is not worth a lot.
Context matters, on the street clearly people assume you are asking about a topic, usually you want to make a point.
@@jon1913 As soon as he switched to gun control I immediately thought, is the study coming from CNN or Fox News lol.
This
3:02 Her numeracy skill = √-1
Thats why she got mustache
Her numeracy skills = i ?
@@jakobschepers2122it's imaginary
what does that say about you, if she actually thought more about the answer than just what was given
@Blox117 yeah but what was given was everything you needed to know to answer the question.
Also it's a joke it's fine
Great episode! Thank you for it!
Also, congrats on the courage of speaking about it when this topic might upset peoples.
Please make more that kind of content.
Me: "I'm surprised Derek isn't making a political science video so close to elec.....ooooooh I see what you did there."
10:40 want to end the partisanship!! Then ensure media is not this biased and vitriolic. Media now is driving this division far more then any political rhetoric. Demand they go back to journalism, not opinions.
Couldn’t agree more; unfortunately this will never happen
Yup.
News media basically needs to be boring.
Remove the emotional hooks, and the sensationalism.
I understand your sentiment, but this naive. This is an incentive alignment issue; we're not in a position to "demand" anything. All we can do is seek out sources that are consistent with our values.
And incidentally: good journalism IS opinionated, just not in defiance of fact.
Problem with this is dumb people vote too @@DanKaschel
@@UltraTrev the problem with... What?
The data used in this imaginary experiment was deliberately chosen to trigger spontaneous answers.
Firstly, very different sample sizes were chosen for the experimental and control groups, making a direct comparison difficult. Secondly, a very strong difference between the two results (improved/deteriorated) was chosen in both groups, and prime numbers (223 and 107) were even used on the left-hand side to discourage probands from calculating the ratios in their heads accurately.
They also only give two possible answers (whether the experiment showed that the skin condition of people treated with the cream was more likely to “get better” or “get worse.”) which may lead probands to think that the table must show a clear difference between the two groups. I also found no information on how much time the probands had to answer this question…
However, a look at the statistics shows that it is debatable whether there is a significant difference between the groups at all, as a normal chi-square test gives a p-value of p = 0.0472 and using the Yates correction, p > 0.05. Also the entire experimental setup is vague, controls missing, etc. So if I had to answer this question I would say that the experiment and data is not good enough to make a statement about the creams effect on skin condition (even if you look at the results without a statistical test). As a side note, please never use the word ‘significant’ when you don't show results of statistical tests like at 1:42.
I'd like to see what the trial results look like when you use more ‘normal/scientific’ data.
Also the first question I'd be asking is why the 2 groups have such a different size
@@meneldal Although it is not good scientific practice, it does occur quite frequently. However, the use of suitable statistical tests and careful interpretation are particularly important in cases like that.
@@meneldal And where is the data for the outcomes where the rash neither worsened nor improved ?
My first thought was about the p-value as well (before I realized it was American propaganda) I would have answered "I don't know, the numbers don't mean anything" and I think actual scientist that work with data would give the same answer, I don't know who these "high numeracy" people are, but the correct answer is idk.
I've participated in some form of online discussion for many decades now... since before the internet became widely available. And from this, I learned a long time ago that one cannot change another person's beliefs with rational discussion. It stands to reason that my own beliefs don't change through rational discussion. You can sure foment a lot of negative emotions, though, by playing with another's beliefs. If you wanted to cause another country's citizens to polarize to destabilizing levels, imagine how effective your campaign would be to use social media to heighten the negative emotions of folks on all sides of the political spectrum.
ITS WYLIE FROM “TO SCALE” !!!!!!!! So happy you two are working together 🫶
The questions are not equal. The problem is whether the questions exist in a vacuum or not. We have no prior knowledge of the skin cream and do not expect any bias. The gun control question has already been researched by both sides with their particular biases. So this just becomes one more study in a sea where bias is rampant. If it contradicts what you already 'know', then you are going to question the study, the sources, the funding, etc. You expect the study to be biased.
in short, PROPAGANDA HITS
which does not make it impossible to analyze data objectively.
after that, you can either change for mistakes in the experiment that gave you the numbers or you just learned something new.
Science works, but only if you can stomach being wrong.
@@erumaaro6060 sometimes it's just hard to know you are wrong. like the video shows, we all have subconscious biases that are hard to keep check.
I always try to do this.
when i am looking unto the data that's connected something i am emotionally attached to.
i try to look at it twice or three times again but in a different context of what if i was not me.
Me looking at the rash numbers and wondering wanna know more about XAI760K too bro
AI without human emotion can give better judgment.
XAI760K seems like an error code for AI bot comments where they can't write something that they're trying to write about. Dead internet. bot broke
It's some crypto token spam
@@teavis2762Or is it to drive search results for some upcoming scam product?
That said, I do think that confirmation bias is a real thing, and if people see data that is consistent with their ideology, they are less likely to question it. do you take part in XAI760K ? great work by them!
Report this comment for crypto scamming
Amazing video, just a shame it didn't come out sooner.
7:14 I think the difference between an observational study on crime and gun control with the potential for many uncontrolled variables is not fundamentally the same as a controlled experimental study on skin cream. The former leaves a lot more room for the interpretation of the results and theorisation about the nature of the relationship. Assuming that there is a "correct" interpretation of those figures when they are applied to a variable as complicated as crime is debatable.
But there are partially obvious human behavior aspects, such as the more likely a thief is to get shot, the less likely they will try it. This is proven by drive-up muggings being popular in California but literally never going to occur in Texas. Once there is a near zero chance the victim is armed and even less chance the police are around and watching, why not? It's the "easy pickins" theorem.
The questions measure what you think these results indicate, not what you think the overall effect is. You could say, "THIS study says my beliefs don't accomplish the goal measured, however I doubt that this is a valid study." Then it's back to being a math question.
But I think you have to be careful. Opponents of gun control don't see it primarily as a crime prevention measure at all. Many issues are like this, a collision of differing goals and priorities rather than arguments about a technical point.
The question is if I should be proud of answering the questions correct or not 😂
There is no correct answer for political question.
Politics is about someone interests against someone else's interests.
AI is the solution. Without emotional influence, AI should give better judgment.
@@alexoolau
How do you make training data, reward functions, and/or ensure a lack of bias when any human examples you could draw from come from biased entities?
This isn't hypothetical either. For automated processing of applications they have to try and avoid bias against marginalized groups as they build them today. Like, if in your training data women are less likely to be accepted than men the program will learn to associate more stereotypically feminine names with being less qualified. Even if in theory there wasn't any bias in the origional grading (perhaps women just had less qualifications on average in historical data because of lack of access) well, now you've created a bias that the human evaluators didn't have.
10:12 The scary thing is that people don’t realize this as often as they should. Of course we’re all the same. If you’ve met 10 people you’ve met pretty much all of them, same as if you’ve met 10 cats you’ve met pretty much all of them.
Damn, I'll check my biases more closely now. Thanks.
Damn, we are eating well. 2 videos this week! 🎉🎉
6:36 This is peak stupidity, people like this guy would look at any statistic and twist it his way regardless of what the numbers say lmao
That's just it. Stats is but just tool, how u use it is completely in the hands of the person.
Skeptics Guide to the Universe opened my eyes to this. They’re amazing.
10:38 I think what makes this issue particularly bad in the USA is the two-party system... In my country we have around 5 or 6 major political parties that get voted into our version of congress, two of them are somewhat similar to democrats and republicans (though tbh democrats would be considered center-right and republicans far-right in my country), but we have lots of other parties that are somewhat in between or focus on certain topics. Only having two parties only produces two options for every question because it's expected that a party has one viewpoint on a subject which always leads to a "we vs. them" where no in between options are really discussed because that would be closer to the enemy and the enemy is always wrong because otherwise they would be on our side.
My country has 50 parties and it still sucks. Bipartisanship is not the cause.
Yes I agree @Blex_040
@polvoazul I assume you mean 50 parties that can be voted for on the ballot, not 50 parties that get into your countries equivalent of congress. There are also lots and lots of parties in my country, probably even more than 50, but there are only 5-6 major parties that have politicians with a seat in "congress".
The average thanksgiving dining table discussion: 4:17
The hardest thing in life is to remain unbiased
In all domains that is probably impossible to ever be. We can only medicate by constantly reminding ourselves that sometimes, if not most of the times, we are as dumb as a rock.
There's a few factors in play:
1. There is an emotional cost to changing your mind. Sometimes that cost is pretty heavy, so you will reject contradictory information.
2. Those of us with high numeracy have seen so many tilted/misleading studies that we automatically want to look deeper when it contradicts our held beliefs but we don't as often apply the same scrutiny to those which support them.
why the hell should i be unbiased? i'm not an estimator
Don't biases arise from applying ethics to what we encounter in the world? Why would it be desirable to remain unbiased, always in the middle of whatever biases the outside world confronts one with?
@@orangejuice7964 you should try to be unbiased if you want your beliefs to be closer to the truth
super awesome, thank you for making this
Big issue: When you gave people the chart saying 'Crime improved' and 'Crime Worsened'; They both can be misinterpreted. It should've said 'Crime increased' and 'Crime decreased'. obviously XAI760K is in a great spot
Bot
@@a_plant-_- no
@@ShoukoTakuda1059 lmao?
Bot
Ironically the guy who answered the question with "well it depends on how guns were used and the resulting outcomes" was the most accurate, because he's saying that "the results of a study cannot always be determined based upon the statistics alone, there may be more variables that determine causal factors". It's nuance and ability to question science that is the most useful, and also highly problematic in science communication. We cite studies, we quote outcomes and take things as hard truth, yet we hear another explanation, and another perspective, and suddenly fact seems dubious.
This is the problem of human reasoning--we want to be right and be right quickly. Truth takes time and nuance and a lot of work to build certainty. If there's a takeaway: be wary of fast answers and easy reasoning, if something seems obvious, ask more questions. Further--it's okay to accept things, but it's more important to be open to change and to listen to new evidence and ideas.
Very true indeed, I believe it's important also we recognise how people read(or don't properly read) scientific papers.
Instead of putting the hard work of going through a whole paper, also making sure if everything is done correctly or not is not done by most people, they just read few lines and run with it!
A thorough rigorous study of a scientific paper is the way to go about it, knowing about control groups, sampling biases, statistical or data analysis, the methodology used etc etc are important.
How critical I am to a claim in a paper, is proportional to how much I disagree with the claim. I readily accept something that sounds reasonable to save time. And this action seems justifiable to me rationally… it’s very time consuming and boring to “review paper as unbiased expert”
I just leave it up to the experts to critically read through and review everything, but yeah doesn’t quite happen nowadays….This is why unbiased experts are important, maybe AIs are much better at this
@@arunjosephshadrach9539The core of the scientific method is trying to prove *yourself* wrong.
Can't help but notice your diatribe is not about clinical testing of topical creams 😅
It's a piece of paper with 4 small unjustified numbers in a grid, it's not that deep
I was told there would be smart people in this video.
I feel lied to
politics
Indeed, where are the non-Americans?
Brilliant - this is one of your best ever videos!! Regardless of which party you vote for - this information is valuable.
It's so interesting that they have successfully divided us almost perfectly 50-50 and convinced each group that if the other party wins, our country is over. We are all stuck in this frame of mind and all closed off to outside ideas.
I once experienced bias in myself. It was related to equipment in a niche field of extreme sports. It took me 2 years to open myself to the idea that I was wrong about a product and the manufacturer, which I had worked with for a while. When I finally overcame that mental block it honestly rocked my world and made me question what other areas I had these same biases.
From a game theory perspective it does make a lot of sense that it would be so 50-50 and polarizing, ideas that we disagree on are more likely to inflame into bigger issues, and each side reworks their beliefs to be more palatable to the masses only when they're behind in the polls. We're unfortunately stuck in a system that basically guarantees that there will be polarization between two roughly equally powerful parties. The issues may change, but the polarization itself is eternal
Considering one party has already made one attempt to violently overthrow the American government, it seems like this time its actually true that if one side gets elected, that will destroy american democracy.
Not everything is rhetoric.
Who is "they". There is only one rational side left, there are still irrational people on it but the side itself is much more rational compared to the other side. And its not the one that has 75% of respondents denying human involvement in causing climate change.
In a sense, the prediction is the prophecy.
They're not sports teams. Neutrality is not objectivity.
3:09 jus answer the question bro
Did anyone notice how Derek referenced at 11:58 that these things are more obvious now because we can post, Tweet... Well, you don't tweet anymore now that Twitter is X, but the cultural inertia of Twitter is so great that even in a video about science communication, you can't shake the notion of tweeting as a useful way to spread information. This cultural inertia is why even after the platform was essentially destroyed as a productive medium, it will continue to survive for at least several years to come, and why none of its competitors had a chance of stealing the market.
On X, Community Notes pointed out that Kahan didn't replicate.
The premise of the video is either unproven or false.
I find that amazing. We live in wild times.
I think the major issue here is that the more intelligent the individual, the better they will be at convincing others _and themselves_ of what they believe in, especially on a matter that is polarized based on a person's values which are only justified by logic post-hoc. It's already incredibly difficult to train yourself to approach a question as if you were an unbiased scientist attempting to interpret the data, so throwing in factors that are heavily influenced by emotion and cultural osmosis would tilt the desire to the default belief you've convinced yourself of many times in the past.
Knowing this does indeed help you to understand that we are all humans and utilize the same tools in order to think, but I can't be optimistic that simply recognizing that will resolve polarization issues that arise from adherence to values. A person does not rationally think their way into the values that the believe in, so polarization that results from disagreements about the inherently emotional are going to be inevitable and actions taken by polarized individuals can get very nasty. I believe the best we can do is mediate where possible, but also recognize that disagreements on fundamental values are going to happen, and as such people will consider their values or cultures to be superior in one way or another to a different set of values or cultures.