In this episode we explore the idea that all things eventually regress back to the mean. We apply this concept to instances of daily life, as well as expanding on this principle from an evolutionary perspective. We also think about how statistics can illustrate principles and guide thought, whether correctly or incorrectly.
Key discussion points in this episode:
- System 1 vs. system 2 thinking
- Regression vs Correlation
- What is regression to the mean?
- Sporting examples of regression to the mean
- Regression in relation to Individual averages vs population averages
- Luck isn’t given enough credit in our successes. Do we give luck enough credit?
- The role of regression to the mean in clinical trials
- What does a placebo or control group actually allow you to interpret?
- The arrogance in believing a particular view in science is ‘correct’
- The complexity of interpreting statistics, especially with the data surrounding COVID-19.
Links and references:
- Thinking, Fast and Slow by Daniel Kahneman
- The Art of Statistics by David Speigelhalter
- The Transmission of Epidemic Influenza
Enjoyed the episode?
Please consider leaving us a rating and review on Apple Podcasts/iTunes. It helps a tonne in helping new people discover the podcast!
Have a thought or insight about something we discussed? Or have a question/episode/new topic for us to delve into? Then drop us a message at [email protected]
Connect and stay up-to-date
We want to interact with our listeners as much as we can, so follow us on Instagram and/or Twitter for snippets from our episodes, random thoughts, and general discussion: @gettingit_pod.
Prefer to have a visual element to your podcast consumption? – Then listen and subscribe to us on YouTube.
What is Getting It?: In a Nutshell
A conversation where we explore topics both familiar and unfamiliar to us to find out what makes them interesting, so that we can expand our horizons and further our understanding of the world and people around us.
From science to lifestyle design, languages to religion, plus everything in between – anything can be interesting if exposed to you through the right lens. We hope to spark your curiosity through open-minded and thoughtful discussion, as well as a healthy dose of overthinking.
Subaan is a 4th year medical student, motion designer, and an avid rabbit hole explorer. He has keen interests in lifestyle design, technology, investing, and metabolic health. Follow him on Instagram and Twitter.
Dan is a 5th year medical student, pianist, and random fact connoisseur. He spends most of his time learning about languages, playing sports, music, and geopolitics. Follow him on Instagram and Twitter.
Note: This transcript was generated using Otter.ai. Therefore the transcript will not be 100% accurate in some parts.
Subaan Qasim 00:10
And in this episode, we explore the idea that all things eventually regress back to the mean, we apply this concept to instances of daily life, as well as expanding on this principle from an evolutionary perspective. We also think about how statistics can illustrate principles and guide thought, whether correctly or incorrectly. Hello, Dan.
Daniel Redfearn 00:27
Hello, good week to you. How are you?
Subaan Qasim 00:29
I’m very good as a greeting I’ve never really heard
Daniel Redfearn 00:33
full blown over, it just came to me. Good week to you a good week. How’s it going?
Subaan Qasim 00:38
Yeah, not bad.
Daniel Redfearn 00:39
Okay. You’re going to drop me some knowledge today.
Subaan Qasim 00:43
I hope so I don’t know how much knowledge it will be. But I’ve had a few thoughts on my mind, whether this ends up being a shorter episode, or just ends up graduating like we did last time or a while ago to full episode, we’ll see how much we can milk, this topic. And the topic is basically regression to the mean. And it’s because I’ve been reading stuff around it, I’ve been reading a lot in terms of statistics and the way our mind works about it. And interpreting statistics over the past year, year and a half.
Daniel Redfearn 01:11
I’ve always had an interest in it. But COVID really took it off. Because there was so much coming at us, I was like, let me, you know, make myself accountable for myself and learn about this. So I can probably make interpretations for myself and understand what’s going on. And I am also really glad that you’re doing this, because nowadays, I can really appreciate the importance of actually understanding statistics, not because obviously it can be manipulated, so easily, can’t it. But yeah, being able to fully understand a set of data and appreciate what’s going on, I don’t know, I can really see the value in that. And growing up, I was never very good at it. I remember we did statistics at school and a level, for example, and I just never really took the time to fully get my head around it. So even now, now that we’re getting into research more, or we’re getting into an age where research is more important. I really feel like a little bit not good enough at statistics. And I’m really hoping that I can sponsor a few basically. So I’m very glad that you’re doing this.
Subaan Qasim 02:08
Yeah, I mean, so it’s what I kind of started off with was started off with was reading the art of statistics by David Spiegel, I’ve mentioned that a couple of times, it’s a really good book to just kind of get a general understanding. First off might be a bit slow, if you do have experience with it. But if you don’t, then it’ll be a really good introduction. But the second half really does get interesting, and starts really, you know, breaking down certain intuitions that you’d have. And the other book that I’m reading right now is Thinking Fast and Slow by Daniel Kahneman, I think it’s how you say his last name. And that’s more worth looking out the way we use intuition. And the way we use critical thinking and the way they interact, which equals system one, which is your fast like, you don’t realize it’s your subconscious doing all the work in your system to which is your critical thinking and stuff. And your critical thinking is able to take in figures and statistics and kind of work, work things out and create a proper solution or proper judgment about a particular thing. When it comes to statistics, but our system, one which are fast, just intuitive thinking isn’t so good at that. And the thing is, is that 90% of the time when we’re interpreting things like statistics, we are using system one. And even, you know, psychologists or scientists or statisticians who are trained in this and spend all their day looking at these statistics, and the psychologists who end up studying how we think about statistics, even they struggle when they’re just in their day to day life, or when they get given surveys as part of experiments. Even they struggle to even over come their system one intuitions which think about things by trying to always create causality, or just kind of get rid of base rate, statistics or base rates in terms of how things occur.
Daniel Redfearn 03:51
So just really quickly, because I don’t want to get lost when because I’ve never heard of system one and system two. Yeah, before we started speaking. So you said system two is critical thinking?
Subaan Qasim 04:01
Yeah, it’s, you know, when you sit down and actually think about problem, that’s your system, too, right? You’re actively thinking about it. But when someone if I just taught you, what’s two plus two, four, yeah, he’s like that you didn’t really think about that. It was just you just know it. And you just know it’s correct. But then if I said, what, what 17 times 236?
Daniel Redfearn 04:19
That one is, I’m joking.
Subaan Qasim 04:22
But yeah, you see, it takes a bit longer, you have to actually think about it, you take some pause, you’re slow with it, that’s your system to actually having to come in to get and system one is just there as your frontline as your defense mechanism, right to make these assumptions and just put things together to make sure we don’t get information overload but can still function in the world around us. But then when there is something that’s too much, then it kind of offloads to system two, or it will offload a conclusion to system two. It will be like system two,
Daniel Redfearn 04:48
Is this correct? And these are just theories, right? These are there’s no like load. They’ve not located a part of the brain. Well, no,
Subaan Qasim 04:53
no, it’s just how our brain and mind and intuition and thinking works. Okay, yeah, yeah, that makes sense. And obviously in the book, he is basically a account of the field of research that he was in. And just like relations for it and different discoveries and how it all kind of transitioned. So I feel like system two is the one we want to really hone in on and get. Yeah, but the thing is, is that most of is always is usually about system one, because that’s what we’re always using. And that’s what leads to these rash conclusions that we usually get, or polarizing conclusions. Or we see a statistic, and we think it means one thing, but actually means another thing, we don’t really see deeper into what it’s saying, right? And the not to kind of bash it. Science, like scientific research and stuff. But the best lawyers are usually those who are smart, because they know how to get around the topics and over. If you’re very well versed in the topic, it’s hard to get lied to, like manipulate the information. Exactly. Yes. Yeah, I guess manipulation is a better word. So you know, there are some times some mental gymnastics or statistical play work that is that happens, or you know, the statistics will be fine. But the way you interpret it is just incorrect. And that wasn’t their intention. They didn’t put it out there in that way. But as a reader, it’s your responsibility at that point, to make sure you understand what it’s actually saying. So that’s why I kind of went reading into statistics to make sure I understand it from the fundamentals, really. So those two books, I haven’t finished Thinking Fast and Slow. But I yeah, so far, these are pretty interesting. And Thinking Fast and Slow, isn’t all about statistics is more about heuristics and stuff. But that’s a topic for another time. Really interesting, though, as well. Okay, so it was chapter 17, of Thinking Fast and Slow, that just kind of spurred all of this. And it’s about regression to the mean. And it was obviously spoken about in the art of statistics as well. And so luckily, I found understanding this chapter a lot easier, because I had that prior understanding, or at least prior exposure to the concept. And is one thing that even now I find in medical school, at least a lot of medical students find, find it hard to explain the difference between correlation and regression, and what the different analyses mean and stuff. Even I find it hard to explain, even though I’ve been reading a lot about it. But I think intuitively, even though I’m against this whole intuitions being what he’s talking against it, I seem to understand where they’re coming from. And the whole thing of regression to the mean, is where it all kind of starts. And yeah, like I said, it’s hard to wrap our head around what regression to the mean actually means. So it’s related to our system, one always wanting to make causal relationship between factors, or at least this is the side I’m attacking it from in terms of how we interpret statistics. Because a lot of the time when things happen is actually, you know, randomness, just kind of supposedly creating patterns. There are lots of theories like this, and there are distributions you can use to kind of predict how things are randomly distributed, which is weird to think about. And that’s an area of statistics, I just don’t understand. Or, you know, I’d have to spend a lot more time understanding it. And, yeah, it’s hard for us to take away this link where, oh, they I observed this, and it’s going to be because of this, rather than just accepting you as kind of random. Our system one just wants to create a causal relationship, and we’re emotionally attached to that causal relationship, because you know, it’d be useful in terms of survival. So that’s why that kind of happens.
Daniel Redfearn 08:18
So at this point, just quickly, I’m not even thinking about the fact we’re recording right now, but because I want to understand this better. Could I quickly ask the difference between correlation and regression? From my understanding so far? Okay. Is that right? Yeah. Okay. So the way I see correlation is that it’s a spectrum. So you haven’t had like, a correlation or no correlation? And you can just predict how likely it is that there is a correlation between two things? Is that kind of? No, that’s a pretty good explanation. So the way I see it is, whereas regression is like that is you can make a regression curve, because I’m just thinking back to like, anytime I’ve had data, there’s an objective regression curve that you can make. Whereas a car, you can just predict whether there is a correlation between two different things, right. And you can say, there’s almost there’s certainly, almost certainly a correlation are very unlikely to be a correlation. But a regression is just objective.
Subaan Qasim 09:13
Yeah, I can’t, I wouldn’t say objective, but I, I get what you mean, it’s more that regression kind of explains the relationship and defines if there is a relationship, like how much does x influence a y? Yeah, there’s correlation just kind of states that there is a relationship between x and y. Yes, yeah. Regression kind of quantifies it.
Daniel Redfearn 09:31
Yes. That’s, that’s how I see it. It’s probably very basic routing.
Subaan Qasim 09:35
Yeah. And that’s kind of the way I see it. And I think that’s a good general place to start, but obviously, is a lot deeper to a certain extent. But I think if you’re in that kind of area, you’re you’re going along the right lines.
Daniel Redfearn 09:45
Okay. Thank you. We can continue.
Subaan Qasim 09:48
Yeah, because otherwise we can just talk about for hours, it just get very confusing and to be fair, I I’m not a person to be able to explain them fully because I don’t understand them well enough myself, but the conversation isn’t really going to be too much about that Otherwise, I wouldn’t really have a basis to be talking from. Okay, so the example I was going to use is one with an issue that I’ve had with protein shakes and my complexion basically in terms of spot breakouts and stuff. So whenever I would drink protein shakes, I’d noticed that I’d sometimes get breakouts, which are normally I don’t usually get breakouts, at least over the past, you know, four years, I do breakouts, my skin’s fairly stable to not have many spots. But whenever I would drink a protein, or go through a phase where I start drinking protein shakes, because I’m doing a lot of exercises, I can need to, you know, add supplementary protein, I would, I would figure out that my skin would get worse during that period. And then eventually, I wouldn’t be exercising so intensely, and then I’ll you know, drop off, you know, having protein shakes, and I’d find out that my skin gets better or it maintains a better complexion. And I’ve noticed this cycle a few times. And I’ve always just kind of come to the conclusion. Well, I’ve come to the conclusion now that me drinking protein shakes leads to breakouts. And the thing is, I ended up doing a lot of, you know, n equals one experiments where I specifically started trialing it, and I would exclude other foods and other factors. So I was basically doing controlled experiments on myself and cycling it. And that’s probably the best way to figure it out for yourself. But I realized what I was just trying to solve is is is just regression to the mean that I’m that’s just being that’s influencing my perspective on how protein shakes affect my skin. Because my mean, my average state is to not really have any spots or breakouts. And, you know, unfortunate to be in that position. So I do appreciate that. But yeah, I’d have some protein shakes. And then I’ve noticed some breakouts. And what I didn’t really take into account, what I was trying to take into account actually was that it could just be pure coincidence, it could be another factor outside of it. But because our system wants just system one just wants to be like, you know, protein shake, a lot of people get reactions with it, it’s probably just that. But the thing is, is that when there is I was consuming whey protein, and I just generally stay with that. With all forms of dairy, I’m pretty fine with I consume quite a lot of dairy in terms of other we have so much milk, but I’ve more double cream and cream and cheese. So I’d consume a lot of that. And I don’t really have any problem God tissues, skin issues, or any kind of issues really. So then consuming whey protein, which I guess is the concentrated form of the protein and milk and stuff, or colostrum. It kind of makes sense. Okay, fine, maybe that’s part specifically might cause breakouts. But honestly, I don’t know, either, I don’t want to find the relationship there. So I never really wanted to put it to the protein shakes. So that was me trying to find an explanation, just thing is regression to the mean, because what could have happened is that I could have just had a good stretch of skin being fine, right. And my mean is actually slightly worse than having just perfect skin all the time. I don’t have perfect skin, but I’m just gonna say perfect skin for the sake of it. So because when my skin is perfect, I pay more attention. I’m not always you know, it’s in a good state. So I just kind of feel like that’s my main because I pay more attention. And then when it goes down to my normal state, because it’s normal, I don’t really pay any attention, even though that actually has slightly more breakouts. So I’m going for a good stretch, right. And then I started consuming protein shakes, and I get breakouts, but that’s just me randomly at that time regressing to the mean. Because in one thing in random events, it all just kind of reverts to the mean, essentially. And then so I did a lot of experiments to try and figure out is it actually causal? Is it this directly causing the sports. And the thing is, is that I did also notice, sometimes I’ll just get I wasn’t consuming the waypoint, I’ll just get sport breakouts, and then I’d go back to not having them. So I had to really figure this out. And it was all about this concept of regression to the mean. But I didn’t really know at the time.
Daniel Redfearn 13:51
I have a question, but I’m worried that it’s going to deviate. Can I ask if you think it’s going to give you a wound like bookmarklet? Okay. I’m struggling to think of like, exactly what we mean by mean, though, because the mean is surely every variable in your life put together. Like your the rest of your diet is a factor in that mean? Or like the amount of sleep you’re getting all the other stuff matter of showers, you tell you wash your face. Yeah. So So surely the eating or consuming the whey protein is another just another variable. So that is part of your main one, when you’re consuming it, if that makes sense. Yeah, yeah, that is true. I mean, it’s just your genetics.
Subaan Qasim 14:25
Yeah, well, okay, that’s really up to take into account your environment as well. And that arguably, arguably plays a bigger role in stuff like skin completely, or you spot breakouts and stuff. But I’m talking about the mean of my skin was the average state of my skin do I usually have a lot of sports or breakouts or not? Right? And for me, I think those using not really that many. And the whole point of the mean is that one of those variables is influencing where you what the state of my spots is. So specifically, I’m looking at whey protein. I might just be along my mean, but if I increase a whey protein is just going to It has a large effect on how my skin reacts. Whereas other things don’t really affect it. Well, you know, they’re just imperfect bones. So it doesn’t really affect it. But that’s the thing where all of these things are just kind of random how much sleep I get to a certain extent is going to be random, or the quality of my sleep is going to be random, or there are so many factors that I wouldn’t be able to put it all into account. It’s going to depend on my why eat, what time I eat, and then how much exercise I did in the day and the stress of work for you whether you’re on the Northern line,
Daniel Redfearn 15:28
sorry, no. One’s dusty.
Subaan Qasim 15:32
Yeah, yeah. So stuff like that. And there was this one factor that was just sticking out. And I’m like, is this right, but it could have just been some some other factor, right, maybe when I started doing lots of exercise, I don’t get enough sleep to them, that makes me break out or something or because I’m under more, you know, stress in my body that causes. But in the end, I have kind of figured out by not doing exercise and taking the protein shakes, but then keeping everything else as constant as possible, that there’s a pretty strong relationship between the two. So I’ve come to that conclusion, which is why I don’t really take whey protein anymore, I’ll just kind of make it when I do have to increase my protein intake via just normal food sources, or other protein sources. So you can talk about means in that way for a singular person about singular aspect, but you can also talk about means in terms of average height and stuff. So it was an example we were talking about earlier, where there’s a relationship between short parents always producing children that are taller than them and tall parents producing children that are smaller than them. Why? Because in the large population, this kind of stuff is actually somewhat random, as long as you take it within a similar population is going to be random. So the offspring of is going to revert to the mean height was going to start leading towards that, that’s what the regression line is, is the mean of all those points. So there’s a relationship, right? So taller parents will produce taller children. Right? And you know, there’s that relationship as well between short parents to all parents producing taller, taller children. But if you’re, the further you are away from the mean, the more likely you are, the more drastic the regression to the mean is,
Daniel Redfearn 17:06
what do you have to do then to go even further from the mean? So what would have to happen for at all, husband and wife to have an even taller,
Subaan Qasim 17:16
relatively even taller child? it it’s unlikely, it’s regression to them. That’s because it’s random. There’s a lot of random factors and in terms of genetics, and genes, and then also a lot of random environmental factors that is hard to take control of obviously, the main one, we have control over nutrition, but even then, that plays only a certain amount of the role. Other factors we can’t control, so it’s basically random at that point. So there’s not really much they can do to force a taller child, what’s most likely to happen is that they have a shortage. And then everyone would be surprised, like, You’re, you’re tall. So why is your child significantly shorter than you, you know, when they’re fully developed, but it’s just, that’s, that’s actually normal, whereas we see as abnormal, because it’s our intuition coming into play were taught all parents should have, you know, equally to a child, because the relationship is short parents, short children, to parents to children, but they are more likely to have children that are of the opposite height ratio, if that makes sense. So that’s one example. Another example is that of ski jumping. So this is an example in the book, ski ski jumping in the Winter Olympics, they have two attempts to basically go off that massive ramp and go as far as they can. And they do that weird pose where they like, corner member sites and like flying squirrel that they like, it’s pretty funny.
Daniel Redfearn 18:35
I’m just a side note. I used to watch that on Eurosport when I was about seven. And it’s the only sport channel we had. So I always watched the sport channel. And for some reason, there was always ski jumping on it. So I watched loads of ski jumping when I was about seven or eight years old, Oh, man. So
Subaan Qasim 18:49
you must have noticed or be an expert. And notice this relationship where if the person is competing, so they’d have two jobs, right? If they had a really good first job, their second job was just going to be worse. Statistically, it was most likely their second job was going to be worse, if they didn’t, it had an exceptionally good first jump, right. And if a person had a exceptionally bad first job, their second job is just going to be better. And the commentator in a particular scenario was like, our man, he had a really good first job, the nerves are gonna get to him to try and stay up to that and try and beat so he’s gonna end up doing badly. So now he was creating an explanation for it. And then when it was the other way around, was like, Oh, he had a bad first jump. So you know, he’s got nothing to lose. He’s relaxed, he’s more relaxed, and he’s just going to do well, right? That’s always trying to create the explanation for it a causal kind of explanation for it because our system when our brain wants that, right, but it’s just statistically you’re going to regress to the mean.
Daniel Redfearn 19:45
Oh, okay. So I’ve done a coin flip 50 times and I’ve got tails 50 times, what’s the chance of the 51st? One being tails?
Subaan Qasim 19:54
Okay, I don’t want to say you’d have to look at the site, look into the Poisson distribution. I don’t really know anything about it. Really. But Remember this being mentioned about how you can predict this is what I was saying where you can kind of predict random events on the way random events would occur. Right? So intuitively, it seems more likely heads is going to come up. But it’s actually still just wanting to,
Daniel Redfearn 20:15
because I have something to say about the ski jump.
Subaan Qasim 20:18
Okay, yeah, go ahead.
Daniel Redfearn 20:19
I obviously could be wrong. But the variables on that first ski jump, the variables are not the exact same as in the second ski jump, mental state conference, for example, those are big factors that determine how well you do it. And they change drastically after the outcome of your first one, those outcomes will undoubtedly affect your probabilities. And the second one, that surely is an explanation why so you can’t say, Oh, he’s like, got less to lose. Now he’s going to go for it. And as you know, that’s exactly why it’s happening. But I’m sure that is a factor. So each time you do it, because I think about that in sport. In tennis, for example, I often see after a bad loss, a player will come back to that same tournament the next year and do better in that tournament, or after winning it struggle to win it again, if that makes sense. And is because yeah, I guess the variables are changing what after you’ve done it again, whereas with the coin flip, the reason why I think it’s, that’s not the case is Yeah, because there’s only one, there’s literally such charts, if you’re doing the exact same, like you’re getting like a robot to do something.
Subaan Qasim 21:21
Yeah. So yes, things like this will play a difference. And I’m sure what, if you measured it, you’d be able to find some kind of correlation, but the magnitude of the correlation of that. So the the, you know, correlation basically, wouldn’t explain the magnitude of the effect you saw in terms of the drastic change from really good to really bad, or really good to just mediocre, or really bad to mediocre, right? It wouldn’t be able to explain that much. And I think this just comes from the fact that we as humans don’t like associating things with luck, or randomness, even though a lot of things that actually majority, majorly explained by randomness, or like, depending on how you see it. Let me just quickly try and find an example.
Daniel Redfearn 22:00
Because I do think there is a range applying this a lot to sport. Yeah, I do think there’s a range of like how well you can perform. And some of that is down to luck. But I think that’s actually part of a mentality, like an approach to failure sometimes. So I remember back to school. When sitting a class test, say I was, say, I was on the border of getting an A, and I get to be, I’ll be really disappointed. But there’s like a range of my capability, right? Like, based on how hard I studied and all those other factors. Say you need 80 for an A, you can’t 100% guarantee yourself a chance of getting at, but you’ll most likely fall in a range, right? Like within a number of standard deviations. Maybe for me, it would have been 76 to 85. That was if I, you know, say I didn’t get much sleep, I hadn’t eaten well, and stuff. I’ll go on the lower part of that scale. And I, you know, good factors on the day make me pass the 80. It’s my fault for not making sure the lower part of my range is not actually above is not above, making sure it’s above 80. because there’d be some people in our class who, even on their bad day when they didn’t sleep well, blah, blah. They’re their bad day was still a 90. And their good day was a 98. Do you know I mean?
Subaan Qasim 23:16
Yeah, but that’s their individual mean?
Daniel Redfearn 23:19
Yeah. So I’m saying each person, we will have a minimum and we all will, like deviate from that mean?
Subaan Qasim 23:24
Yeah. Yeah, I guess that’s one kind of getting out. I don’t understand it enough. Or I don’t have the vocabulary to explain it in a popular at a population level. Because I’ve just read these examples. And it makes a lot of logical sense when you go through those arguments. And at the time, I’d tried to argue against it in my head, and I just can’t, because I mean, the people writing the books are a lot smarter than me. So they’re probably thought about every single argument that I would think of. And a lot of the time, they’d say, like, make a statement. And I’m like, wait, what about this, and then the next point is, you probably think this. So they’re basically in my mind at that point. But I found this example from veritasium. You know, that channel on YouTube is a science communicator YouTuber, thing. He has a PhD in physics or something. And he made a video about success, and luck, and hard work and like the egocentric bias and stuff. And one example he created a simulation or maybe he’s talking about someone who did a simulation where he was looking at the astronaut program, like astronauts selection program, which is one of the most competitive programs in the world, there’s 18,000 applicants or something but only 11 gets selected. So insane numbers, and he made a model of the application and selection process, and participants were given two scores, a hard work or skill score that accounted for 95% of their overall score. And then the lock score was the last 5% but Within that lock score, they were given a score of zero to 100. Right? So it was still weighted in a ratio of 95 to five. Does that make sense?
Daniel Redfearn 25:10
What is the lock score involved? Sorry,
Subaan Qasim 25:12
the lock score just involved randomness, because it was a simulation, using like data points, it was just randomness at that point is just how lucky you are. Yeah, pure randomness. So to get a score of 100, overall, you’d have to have 95% work and 100%. luck, right? And then you’d have an overall score of 100. Whereas if you had we’d have a 95 and five, yeah. Yeah, yeah. But so the 5% of that lock is zero to 100. Yeah, yeah. Okay. So the ratio is 95 to five in terms of how it gets allocated, but
Daniel Redfearn 25:43
you never have 100% work.
Subaan Qasim 25:46
Yeah, but that only comes to 95%. return to school? Yeah. Okay. So if you had 100% work, which was 94 95% of the total score, and five and 100% lock, which is 5% of the score, that’s over 100. Right. So you get the ratios as 95 to five. And after running the simulation, 1000 times, the ones who are picked, had an average lock score of 95% 94.7%. Wow. So that shows how much randomness and lock is involved in something like that. And the moral of the simulation is basically, only 1.6 participants out of the 11 were selected based on skill. So that’s what 10% were based on their skill and hard work alone. The rest was just kind of luck at that point.
Daniel Redfearn 26:29
So you’ve maybe said you’d have so many applicants who are on a similar level. Is that right? Yeah, way to distinguish them is literally just a bit of luck. Really? Yeah. Okay, that makes sense. It’s really interesting.
Subaan Qasim 26:40
And so even though luckily contributed to 5% of their total score of their chance of getting selected for the program, if luck played no role, the nine or 10 of the selected applicants would have been different,
Daniel Redfearn 26:53
is really interesting. I mean, because that means doesn’t that mean, essentially, you’d have to eliminate all chance of luck being a factor to make it as fair as possible. Yeah. Because even a small amount of luck playing apart has a really big influence on the result. Yeah, is a result and an outcome the same thing? No, yeah.
Subaan Qasim 27:11
It depends how you define it initially. But yeah, that’s why when I say certain times, I like to define them. Because a lot with this technical talk and stuff, a lot of the time, you might see something, but it might mean some something out to someone else. Yeah. So I was reading a book called aspiration by Agnes kalon. And the first chapter was just explaining what she meant by aspiration, because in the typical way, we use aspiration was slightly different, what she was saying, what how she was defining it. So she spent a long time defining that
Daniel Redfearn 27:38
this was so interesting about linguistics, because each word has a slightly different meaning to different people, I like
Subaan Qasim 27:42
always match a link, link everything back to linguistic
Daniel Redfearn 27:45
language, linguistics, because just really quickly, my definition of result, and outcome would be a result is objective, you have a result, a number or something, an outcome, you can base it on results. And outcome could be something you take away from the result. Like, you know, this increases, that can also be a result, but the result is objective. You take from it. So now it comes is how you is can be how you interpret a result.
Subaan Qasim 28:10
Yeah, but it’s also like a lot of medical studies have the primary outcome, which might be death, right? Yeah. And secondary outcome might be hard to talk into the cardiovascular 30. Yeah, so those are just results in the way you were defining it. Those were? Because that’s not what you’re taking from it.
Daniel Redfearn 28:24
Yes, an outcome can be the same as
Subaan Qasim 28:26
always. Yeah. That’s why you should just define it. That’s why I was like, defining stuff like that. But yeah, I think you will just see people get what you mean, in the context, you’re speaking at that time. Okay, so to have explained the stuff about me having breakouts and stuff with whey protein. So yeah, it was actually quite a hard process for me to go through. And it was quite, I had to be really strict with myself to basically figure out if it is causally associated, I didn’t end up doing any statistical analysis. But with this kind of stuff, n equals one kind of experiments. I mean, if you’re really stringent and you end up, you know, expanding it over, you know, create proper hypothesis and theory and then study at least in proper samples. And also, you should probably do some kind of analysis on yourself as well. Yeah, I just kind of came to the conclusion where, okay, it’s very likely that is related to it. So when correlations are less than perfect, there’s always regression to the mean taking place. So unless a correlation is one or minus one, or, yeah, if it’s not one or minus one, there’s always some level of regression to the mean taking place. And it’s hard to take into account or people don’t like beat. So obviously, regression to the mean is a random process, right? People don’t like attributing a lot to that random process where, like I explained without luck simulation, there’s a lot that can depending on the situation, luck, and just randomness can play a very big role. And I think just kind of understanding and thinking about and just accepting it kind of puts you into a more humble state when you’re interpreting things and I’ll kind of get onto that later when it comes to more. Just Studies later on. So, yeah, I guess we see this a lot in sport to a certain extent. So another, as I said, the example of ski jumping, another example is golfing, where day one, someone might absolutely smash it in relation to the par value, or whatever the terminology is. Yeah, day one, they’re just going to smash it. And then you would expect for the rest of 20 Oh, man, they’re gonna win this one. They’re gonna smash it right. But the second day, they perform pretty badly. But they’re badly is relative to that really exceptionally good to them in though.
Daniel Redfearn 30:32
Because everyone’s means different
Subaan Qasim 30:34
things, because this happens to everyone. Yeah, in it.
Daniel Redfearn 30:38
But everyone’s got their own mean, right? Because the best players, because otherwise you’d expect any one random person to win every tournament. But the same is the same in a lot of sports work. Because like, I love golf, and yeah,
Subaan Qasim 30:50
and yeah, and those are the exception to the rules, I guess the best athletes, they are exceptions, right? They, they have been able to one raise their mean so high that they stick out and then stay so close and constantly keep improving or stay at that really high level. Right? Those are exceptions, you don’t get many of them in any sporting because Are you saying that everyone has a similar? Everyone has the same mean?
Daniel Redfearn 31:12
Subaan Qasim 31:14
but at a population level is normally distributed?
Daniel Redfearn 31:16
Yes. But if we because all of these guys are outliers, you know? Yeah. Okay. Yeah. 100, guys, or woman?
Subaan Qasim 31:23
Yeah, I guess it’s like the sport magazine effect where, you know, someone performed really well last season and other the front page of the sport magazine. And then it creates that in the interpretation where you expect them to do really well and stuff. And I guess then this comes into what the potential causal factors were, now they have so much pressure and stuff that they just perform worse, or maybe they didn’t really feel that pressure, or maybe that didn’t really have much, they might have had more pressure. But maybe the overall effect of that was probably not that much compared to just randomness that they were just not going to perform as well as they did previously.
Daniel Redfearn 31:56
This is, by the way, a super interesting topic. I love thinking about it in terms of sport, because the last few years, I’ve learned quite a lot of life lessons from watching sport. And this is something I’ve thought about how you get a top team, for example, I’m a Liverpool fan. So I’m not happy about this. But manchester united with, have you heard of Sir Alex Ferguson? Yes, yep. So he was the manager manager for a very long time. And what’s so incredible about him, and why he probably in my opinion, is the greatest ever manager in football, is because he made a premier league winning dynasty in the 90s. And then won multiple league titles, loads, he won 13 total. That’s incredible. That’s more than every other team in England by one on it like in his own career. But he had to build multiple teams to do that. So as in in the 90s, he had a team. And then he had to build another team again in the 2000s. Because all the 90s guys got old, she had to start again. So it’s not just that he was an he was a truly incredible manager, because that proves that his sort of average, his mean, team strength is very high. And he was constructed to construct that multiple times. So it’s not just about getting to the top, but it’s about staying at the stay at the top, you have to maintain that level of good. have that level of like ability. Yeah, so otherwise, you all regress, is that right? Is
Subaan Qasim 33:08
that my website regress to the mean? That’s where regression, they just call it regression. Right? So that’s where he’s such a blessing to where Yeah,
Daniel Redfearn 33:13
his movements are so high Yes, level was so high on any team where they are able to maintain that they’re having to maintain that really high level, they can’t just get to the level, and then be the same as everyone else. And then just on a higher level, they have to get to that level and stay at that level to stay at the top. I know that probably sounds quite obvious, but it’s really hard to do,
Subaan Qasim 33:32
I think Yeah, and most people don’t, most people do end up falling off at some point or have a randomly, you know, bad season. And that randomly bad season, people start saying try and make excuses for it to a certain extent where, okay, maybe it was severely injured, or he’s not going to perform so well. Maybe the next season. I’m talking about in general sports, not managers. So yeah, that might be a legit reason, right? He took the season out of the seasons gonna be rubbish because he didn’t even take place. But you’re looking at an individual level where someone’s you know, performing exceptionally for five seasons in a row in whatever sport and then they then the sixth season is pretty bad. People will say, Oh, he’s injured or he’s had these other stressors or whatever trauma kicks us but it could have actually just been random, where, you know, when it comes to individual people is really hard to assess this stuff because the level of change can be quite small because obviously you’re already in professional sporting players, they’re already compressed in a very small scope, right, because at the end of the distribution already, so small, very small variations can end up having quite large effects. But then the inter individual variability is still quite high. Whereas if you have someone like LeBron very consistent his his variability is probably quite low overall. I’m I think, I’m just gonna start mixing up terms and explanations here, because it’s, it’s hard to explain, hard for me to explain properly because again, This is stuff I’m learning more and more. So people will say, make these excuses or not excuses, but try and create these reasons as to why some performed badly. But it could have just literally been random with all the color. Okay, all of these effects that come into play, okay, this effect plus this effect plus this effect, because this effect may lead to a bad season. But all of those individual effects were basically random.
Daniel Redfearn 35:23
I, I agree. But what I would stipulate for sure, is that everyone, when we say, regressing to the mean, it’s all relative to their ability. And I’m saying Yeah, actually, different athletes can have quite significantly different means.
Subaan Qasim 35:36
Yeah, and it does also work on a population level, even within subset populations, like athletes, but I, I can’t articulate it. And obviously, there’s two two sides of it. And it’s always difficult to figure out if it was somewhat mostly random, or, actually was due to a legitimate reason,
Daniel Redfearn 35:54
because they say that the golfer say, on average, he or she shoots an average of 7171 shots per round. And that’s that the last 15 years, the average 71, and then they’re on a course where the power is 71. And they shoot 68. What you’re saying is that the next time they go out statistically, they’re more they’re not gonna, they’re more likely to actually shoot above 71 than 71. Is that what you’re saying? So then you’re saying it’s more likely there’ll be 71 or above? Above? 71? That? Yeah, then then just 71? Do you get what I’m saying? Yeah. As opposed to just being 71? Again,
Subaan Qasim 36:29
yeah. So I think this goes back to the system one thinking where, because we think someone had a really good day, the next day, that was going to be really good. And then when they don’t perform as good, we have exaggerated emotion in our head being like, while they’re playing really badly now. But what were the although the actual change wasn’t that much, and they’re still playing around the mean,
Daniel Redfearn 36:48
but what I’m struggling to fully understand is, surely, we have to factor in how important the variables are. So for example, the playing on the same course, they did really well yesterday, they’re feeling good, and they’re looser today, you can argue still, that they have a chance of doing a better score, or a really good score again, the next day, because, yeah, they’re compared to their normal career average, today, all the variables are just lined up very similarly to yesterday. So it’s actually quite likely they’ll shoot again, a very
Subaan Qasim 37:14
variables coming into places random the so then the overall effect was random.
Daniel Redfearn 37:19
But say it was their favorite weather, because you could you could literally explain it. That’s what I love. Like you could say, this guy loves playing in the wet for some reason, he’s a bit weird. He loves playing in the wet just shoots, it suits how he shoots because he doesn’t have much spin. So it just
Subaan Qasim 37:31
works. Okay, and then against wet with weather, and he has all the time played really well in wet weather. But now it’s wet weather again, and he performs badly, then people just make an another line of reasoning for it.
Daniel Redfearn 37:43
Oh, but what you’re saying is that’s just chance, like
Subaan Qasim 37:46
you I agree with, and it’s his favorite time. He’s feeling good. You just played bad.
Daniel Redfearn 37:50
So yeah, if we eliminate all other factors, so let’s take even wet weather away. Okay, let’s say every day he’s ever played in his whole career has always been the same weather, occasionally 26 degrees and the same humidity and everything. Yeah. And the same course. So that’s, that’s insane.
Subaan Qasim 38:05
This guy is just,
Daniel Redfearn 38:06
he’s just a science experiment. Yeah, literally. Yeah. So then if he shoots a 71, on average, throughout his career yesterday, he shoots a 68. And then today, he shoots a 74. That’s literally just because of chance, isn’t it? There’s no, there’s no explanation for it. Yeah. The think data is basically that the problem is the beauty of it is you can always speculate because there are so many variables you can blame on. Yeah,
Subaan Qasim 38:33
yeah. That’s the thing is speculation, though. And
Daniel Redfearn 38:36
I get it, I get it. We blame blaming it completely on the variables, but there’s also just an element of just chance.
Subaan Qasim 38:41
Yeah, because all of those variables are random in itself, produce a random effect. So like,
Daniel Redfearn 38:47
I’m with you now.
Subaan Qasim 38:48
Yeah. And so it’s the thing where, okay, each individual say if in a year, there were 100 murders or something on this topic is just because each and just each single, like homicide, right? Each individual one is going to be unrelated. What assume they’re all unrelated, you know, different person, different situation all random. So, like, one murder isn’t related to another murder is not the same person, same group or gang or whatever, which is, you know, individual murders, right? Each one is random. But each year, you can predict how it will you can. So this is where the, the example of using the Poisson distribution came in when I think this is in the art of statistics. And this really shook me and basically, they could predict how many murders there would be in each year. Even though each event is random, the total amount in the year would be random. They could still predict it with insane accuracy, because when all the individual random things come together, you can still predict it to a certain extent. So this starts getting really weird and I don’t understand it’d be really nice to actually get petition statistics, statisticians to come on and talk about So maybe explanation actually explained it properly because I don’t understand it fully. Okay, this is futures a bond speaking, the future editing. So bond speaking. So I just want to go through and make some clarifications on this whole thing with the Poisson distribution, what it is how it works and what it’s kind of used for, because at the time, we will, we will recording, I didn’t really expect to go onto this topic and mentioned this distribution, because I didn’t really know much about I’m not a statistician. And when I read, I read the book A while ago, and I couldn’t really remember much to do with it, I just kind of remember the vague idea about it. And yeah, I didn’t really want to say anything majorly wrong. So I just kind of left the example hanging, and didn’t really explain it further, since I didn’t want to see anything entirely wrong. So I now have Wikipedia and the book open in front of me. So I can kind of explain that example properly, just so you have some kind of closure on that entire topic on that example. So according to Wikipedia, the problem distribution is a discrete probability distribution that expresses the probability of a given number of events occurring in a fixed interval of time or space, if these events occur with a known constant mean rate. And independently of the time since the last event. That’s a bit jargony. But anyway, so the the Poisson distribution can also be used for the number of events in other specified intervals, such as distance, area or volume. So, for example, things that are independent in time and space, things like suicide rates, each individual suicide is unrelated, they are independent of each other. But then there’s still a fairly constant mean of the amount of suicides that occur in a year. So it seems like all these individual random events, obviously, those specific factors have their own factors going into it, but they’re independent of each other. So they’re random. And they kind of come together to have an average mean, that seems to happen a year on year. And the Poisson distribution kind of lets you predict how many suicides are going to happen in a particular year or something, or over a period of time. And so, so it’s not about really predicting how many will occur. But it’s more about predicting how many of those single events will occur in a specific amount of time. Let’s say on average, there are 10 suicides per day or something. But that doesn’t necessarily mean that say if you average it across 10 years worth of data, but that doesn’t mean, having 1000 suicides in one day is impossible, probably extremely unlikely, but not impossible. The Poisson distribution allows you to kind of calculate the probability that there is 1000 suicides in a single day, I don’t know why I’m using the example of suicides that just kind of came to mind. But it can be used in that kind of fashion. Or another example that I can see on Wikipedia right here is, for instance, a call center receives an average of 180 calls per hour, 24 hours a day, the calls are independent, ie receiving one does not change the probability of when the next one will arrive. The number of calls received during any minute has a Poisson probability distribution, ie the most likely numbers are two and three, for example, but two, one and four are also likely and there is a small probability of it being as low as zero, and a very small probability like could be 10. Another example is the number of decay events that occur from a radioactive source in a given observation period. So hopefully, that’s has to paint the picture of the use case of it. And the example or the thinking exercise that was given in the book of the art of statistics is, how often do we expect to see seven or more separate homicide incidents in England and Wales in a single day. And I’m just going to read out certain segments, and then just add any thoughts just to kind of make it clearer, because obviously, I can’t read the whole book and just give you miss on certain pieces of context. So to assess how Read Across of at least seven homicides in a day might be we can examine the data for three years. So which is 1095 days, between April 2013, and March 2016, in which there were 1545 homicide incidents in England and Wales. So that makes an average of 1.41 homicides per day. over this period, there were no days with seven or more incidents. But it would be very naive to therefore conclude that such an occurrence was impossible. So if we can build a reasonable probability distribution for the number of homicides per day, then we can answer the question of what’s the probability of there being seven homicides in a single day? Or how often do we expect that to happen? And this is where the Poisson distribution comes in. So the normal distribution requires two parameters, the population mean and the standard deviation. And the Poisson distribution depends on only its mean. So in the current example, the expected number of homicide incidents each day is 1.41. Because that’s the number. That’s the average or the mean, per day over that three year period. So using the Poisson distribution, you can calculate that there would be a probability of 0.01134 of exactly five homicides occurring in a day, which means that over 1095 days, we would expect 1095 times naught point naught 1134, which is 12.4 days on which there were precisely five homicide events. So over three years, you would expect 12.4 days on which there were precisely five homicide events. And that’s being calculated from the, or using the Poisson distribution on the mean number of 1.41 homicides per day. And the actual number of days on which there were five homicides in a three year period were 13. So is pretty accurate in that sense. And that kind of paints, the, the use case of the Poisson distribution and kind of how it works. So in answer to the question posed at the start of that section, we can calculate from the Poisson distribution, the probability of getting seven or more incidents in a day, which turns out to be naught point naught 7%. And means that we can expect such an event to happen on average, every 1535 days, or roughly once every four years, we can conclude that this event is fairly unlikely to happen in the normal run of things, but it’s not impossible. The fate of this mathematical probability distribution to the empirical date is almost disturbingly good. Even though there is a unique story behind every one of these tragic events, most of which are unpredictable, the data act as if they were actually generated by some known random mechanism. One possible view is to think that other people could have been murdered, but they weren’t. We’ve been so busy, we’ve observed one of many possible worlds that could have occurred, just as when we flip coins, we observe one of the many possible sequences. Yeah, so like I said, I couldn’t really remember much about it. And it just goes to show that I need to revise and understand it more. So I can just kind of go off the top of my head with it. But hopefully, that kind of gives you some closure on that example. And now you can go back to listening to the rest of the podcast and the other waffle that I probably gotten other things wrong on.
Daniel Redfearn 46:57
Subaan. Tom is oh my gosh, no, no, you’re Subaan? Wait. Can we cut that out?
Subaan Qasim 47:04
That’s funny, then there’s no way you think, any good
Daniel Redfearn 47:10
bit off topic, but I can’t really think of another time to fit it into an episode. So it was my friend say recently, who just mentioned something to me. He’s really good at stats. Okay. He Yeah, he did it fell Martin, you know, like, he loves maths. And, um, he said, obviously, in biology, we’re looking at like, quite large organisms, very complex things. And then we already know about this, like how chemistry is like, you go smaller, you go
Subaan Qasim 47:38
smaller, as far as biology is Applied Chemistry, chemistry is applied physics.
Daniel Redfearn 47:42
Like you go like, small into a cell, and then you end up doing chemistry. And then if you end up doing chemistry and looking at like, the individual atoms and stuff, then you end up like you’re doing physics this morning, you get Yeah. But then what you’re essentially doing ultimately, is just like with an electron, for example, it’s just probabilities, that is going to be in a given location that a given time you and you end up the smallest scale of everything is just statistics. Yep. Because there’s just a bunch of chance. Yeah. So it’s proof that everything is just kind of completely random. And you can just, you can just stratify and just assume, based off of probabilities that things are, yeah,
Subaan Qasim 48:20
yeah. And then so when things start working together in systems, then those random events become predictable. Yeah. And, you know, I probably gets into a lot of philosophy of that. So next smallest
Daniel Redfearn 48:31
Subaan Qasim 48:33
Yeah, it goes background is just applied. So chemistry is no biology is Applied Chemistry, chemistry is applied physics. Physics is applied maths, and maths is applied philosophy.
Daniel Redfearn 48:48
There we go. And the thing is, as well, like physics also goes, it goes the other way doesn’t, because physics is the largest scale possible as well.
Subaan Qasim 48:56
Daniel Redfearn 48:57
That’s what I was thinking. I was thinking, who is this? What? Who are the smartest people in the world? Like, what is that profession where he or she is the smartest? It’s probably
Subaan Qasim 49:05
physicists. Yeah. I’ve always had this thing with physics where I think everyone should like known physics to a decently high level. I don’t know how people can find physics interesting, because it starts to explain every single thing that we interact with and see and do in our lives. Even just okay, I really like technology, obviously, a lot of is based on physics, and electrons and stuff like that. So I like reading about that. But literally everything like why is glass seethrough. Most people probably can’t explain it. If you have, I don’t want to say basic because it’s obviously relative. But if you if you studied some physics, and the way waves interact with solid materials and stuff, and electrons, you’d be able to explain why. You know, you might not even initially come to mind even though you know the theory behind it. But then you just Google it and notice the first sentence a lot. Oh, yeah, it’s just because of this. I’m not gonna go into explain why. But yeah, those kind of questions you answer that without physics. So people like Elon Musk and try and think of other examples, but you know, all the big discoveries like Einstein, Newton, Tesla, like Nikola Tesla, Elon Musk’s Tesla company, they’re all physicists, and what they end up figuring out and producing in terms of theories and discoveries, and up changing the world, because they can be that profound. Obviously, it’s not going to be to the same scale, no, but the way, you know, Musk can just figure out all of this not figured out. But he always returns from first principles using physics, everything always comes down to it. And if you build up from the most fundamental principles, you can build everything on top of it. So I mean, that was a tangent. But yeah, examples when it comes to regression to the mean, and stuff comes up in medicine and health law, and I guess, all scientific studies, because scientific studies, the whole point is, the whole point of having a control or a placebo, is so that you can see if that changing of the intervention is bigger than what would be chance or the natural regression to the mean, or the natural state. Right. So one example is that of depression, because I was in the book, obviously, I noticed that, you know, obviously, a very sensitive matter. But this was the example. And I’m not saying that all your depression is just a statistical, you know, anomaly or something. But this is just an example that was given. So I’m just basically going to regurgitate it. So the example given in the book was depressed children treated with energy drinks improved significantly over a three month period. And, you know, yes, that could be true, and it probably does happen. But it’s also you could also say that children who are depressed, who do handstands for five hours a day, will also improve significantly, because the extreme end of the spectrum, in terms of emotional state, that they would just regress back to the mean, naturally. So this is essentially the placebo, you know, this is what they’re talking about when they say placebo effect, because someone will just naturally just heal from their disease, because they were just regressing back, it was just chance. And this is basically the basis of always having stringently controlled trials and placebos with or with placebos, or just, you know, well controlled groups. And in this case, the children just naturally gonna improve to a certain extent, because of natural regression, and natural states and fluctuations in emotional health. Because, you know, suddenly, something might happen in their situation in their livelihood. Maybe they weren’t in a good household, they change household, maybe foster home and stuff, that stuff that is random to a certain extent, you can’t predict that, and they’ve improved. So you’re gonna have that natural improvement, or even just naturally in your mental stage is going to improve, right? So most people kind of understand this, but it’s important to actually be aware of it. So this is where the interaction between system one and system two come in, where it’s like, when someone explains it to you, and you think about it, your system is like, yeah, duh. But when it’s said to you, people just take these conclusions about like, oh, x, a drug or x exercise reduces risk of x and stuff, right? And people like you, okay, it’s just sit on one just comes in, like, yeah, makes sense. So we’re always inclined to make causal explanation between factors. That’s our system one works. But yeah, it’s about just always being more aware of what’s going on, and how the world works in terms of statistics to see what the bigger picture actually is. I was just gonna say, so if we use the handstand example. And so if we have a spectrum from left to right, of like, minus one to one, depression, the level of depression,
Daniel Redfearn 53:37
and you measure in some way, like, measure a particular neurotransmitter or something, oh,
Subaan Qasim 53:41
yeah. Just proper clinical analysis. Yeah.
Daniel Redfearn 53:42
And zero is the average person, the mean, say that they are? point minus point eight, six. Yeah, right. There, you could list a huge number of variables that have that where their result has led them to minus point eight, six, introducing another variable, it would have to be less than minus point eight, six, to make them worse. And introducing that variable. Anything that won’t make it worse, it is more likely that that variable will make them better than worse is that you kind of get what I’m saying? Yeah, it would have to be an even worse than their already low level. So introducing any new variable is more likely to have them
Subaan Qasim 54:25
exactly. That’s the essence of randomness. And regression to the mean. Yeah, if you are minus one on that thing, anything, you can do anything, and you are more likely to go up and down. You can’t go down any further. So you are just going to go up in terms of emotional state or something. And this applies to essentially everything. So that’s where this whole point of was how strong is the relationship? How much does X variable, explain why variable or state? Very interesting. So this is what it means by getting down to the fundamentals of what does it mean to have a control group on what you’re trying to do? Eliminate when you’re trying to figure out or control for, you’re essentially trying to control for randomness. But what does that even mean? Because I guess it depends how you define randomness. There are different definitions of randomness. And those are important when you’re talking about randomness in different circumstances in different scenarios. But obviously, that’s just a whole nother topic. So, yeah, this is why post hoc hypotheses and explanations and dangerous to play with. And you, you seem to get this a lot at least. So with, I’ve done a lot of reading into the nutritions, for everything and lifestyle interventions and stuff. It’s just so hard to control for that there are always so many variables. And it becomes easy now, when you do see some kind of association to be like, Oh, yeah, this is because you want to just anchor towards a Association, because we want an explanation for something, just how humans work. But a lot of the time, it is random. So you see that, and then you hook on to it, and you get almost emotionally attached to it. So it’s hard to get away from it. So you’ll just meet, like, you know, post hoc explanation to explain your post hoc hypothesis. And you’ll just go down that rabbit hole, and then you’ll go and do experiments to prove your post hypo hoc hypothesis and explanation. And you don’t really get out with that, because you’re almost emotionally attached to it. And, you know, that’s been a big thing within the nutrition foods fair, because it’s really hard to control for stuff. So that’s why it’s really important to always go back to the fundamental principles of whatever you are researching or experimenting on in terms of how do certain metabolic functions work, what is the core principles and a lot, and this is where the problem of hyper specialization comes in, where if you only know about one particular metabolic pathway, unfortunately, everything interacts with each other. So if you’re an expert in that one pathway, but you anything you experiment on that you’re going to be, you’re almost chucking out a million different explanations, because the only explanations you can come up with for a particular result is that which is combined within your sphere of knowledge. And that’s why those who understand full body physiology to a really deep level, they can put together these different explanations that seem initially counterintuitive. And people will initially just be like, Oh, no, that’s counterintuitive. So it just doesn’t make sense. A common one is that of So okay, when you fast, or you go through a period of basically, when you go into ketosis, ketones have been shown to change mitochondrial bioenergetics, to lead to uncoupling, which is basically browning or Beijing of white adipose tissue, which is our normal store. Brown adipose tissue in adults are at low stores, but they’re basically I mean, for those who aren’t, okay, I’ll try and explain it fully for those who don’t have like a medical background or something. So we have white fat, which is our main storage one, which we’ve, you know, initially was assumed to not really do much, but obviously does a lot of brown fat. Children have a lot of it when they’re born, basically. And basically, you can basically burn the fat and it just gets generated into heat. So that’s called non shivering thermogenesis. So they can just produce heat or novelty for a child, it’s really important, because they have a high surface area to volume ratio, so they lose heat very quickly. That’s why small animals like rats also have very hot, fast metabolisms, and also have a lot of brown fat as well. larger animals tend to not really have as much brown fat, human adults don’t really have much brown fat, although there are areas of it. And the thing is, but there is this idea of browning or Beijing of orb browning of white tissue to beige tissue, which is kind of like an in between between white adipose and brown adipose. And where there’s certain amount of uncoupling of the mitochondria. uncoupling of the mitochondria just means that instead of producing ATP to produce energy, it just produces energy, ie heat, and it just dissipates that, which is usually wasteful, right, you don’t just want to get rid of your fat stores, basically all your glucose stores to just produce heat. That’s not what we want to do. Usually, however, when you go into ketosis, or state of ketogenesis, so you need to be fasting, usually, or not eat food, or at least not eat carbohydrates for a significant period of time. Let’s just say 24 hours and longer, depends how well metabolically flexible you are and how well adapted you are to fat oxidation. But if you stop eating, you ought to stop eating. Yeah, let’s just say you stop eating, you’re going to ketosis because you’re utilizing fats now, in terms of your body energy that promotes this browning of adipose tissue of your white adipose tissue. But then, if you think about it, you’re essentially wasting energy is used to generate heat. So if you’re starving, right, if you think from an evolutionary standpoint, whatever your views on evolution are, just Okay, let’s just think about the Paleolithic era or something. You you’re starving, and now your body is just burning energy, just producing heat, or at least your body’s gearing towards that. That makes no sense, right? But then, there’s why I guess to understand science, you need to understand that the history of that field of science And I guess when it comes to the body and stuff, anything related to evolution and just natural body processes, obviously, we’re very different now we adapt the environment for us rather than the other way around. So you’d have to think really deeply, it was really hard to people just want to kind of accept that. Or at least this is what I believe in think as well, that the reason that would happen, that you would start burning energy, food lighter producing heat, is when you’re starving, and you need to store and keep that energy for as long as possible, is because periods of lack of food tended to happen in the winter. So we know that we can survive with just have our body fat without any food for a significant period of time, depending on how well adapted you are and how much how fat you are, or how many swords you have. However, what’s going to kill you first, lack of energy from fat storage, or the cold, you can die instantly from the cold. And as you’re going to, at least from an evolutionary stance, you’re most likely to go through periods of famine, or just low food supply, you know, how to hunt and find animals, or there’s not even going to be any, you know, fruits or berries to find and pick this is winter. So your, your body can either just conserve the energy, but then you’ll freeze to death, which will happen pretty quickly. So it makes a compromise. So then it starts to paint the bigger picture, you know, okay, I’m understanding why this kind of work. But again, that requires a lot of knowledge and a lot of spread out knowledge, not just within one field, which is your specialization and hyper specialization is required. But you’d also have to spread out that rug made that rabbit hole wider, as well to actually take new ideas into account and formulate hypotheses that actually make
Daniel Redfearn 1:01:38
sense to just just so I can fully understand that because it’s quite interesting. So the route is an evolutionary thing, because so we brown the adipose tissue becomes Brown, because, yeah, historically, when there was a relationship, or the correlation between being in the cold, and starvation, so when you go into ketosis, your body is initiating that response to what is assuming is more likely that you’re in a cold situation, you’re trying to help you. We’re trying to help you create heat.
Subaan Qasim 1:02:10
Yeah. And I suppose these days, we don’t have that problem that much. Yeah, in the West. Yeah. And you know, he likes central heating and stuff, in other areas of the world is still going to be a problem. But I suppose in the West, where we’re very fortunate to have these kind of technologies to help us, you know, stay in thermal regulation, we don’t need that kind of stuff. But now they’re trying to exploit this software. If you go into ketosis, you’re essentially, is futile recycling of energy, you’re wasting energy to a certain extent, can we utilize this in some way? So that’s, anyway, that’s a side point.
Daniel Redfearn 1:02:40
But I wonder how we’re going to evolve now that we’ve, like, removed so many parameters, you know, physically,
Subaan Qasim 1:02:47
he Oh, man, oh, man, there’s a lot I have to say about this. Because, you know, we’ve already had certain aspects of this happen to us where, say, we’re dental health, our jaws are getting smaller. So we’re more people are getting more and more crooked teeth have trouble breathing, because the navel cavities and jaw and facial structures doesn’t develop properly. Why? Because we’re eating so much soft food these days. Whereas before, you’d have to chew on hard food, whether it’s like raw meat or you know, very under just you know, chewing meat and stuff like that, or just like chewing bark or something, obviously, not during bite, but something along those lines where, you know, like a banana used to be a very round thing, thick skin, and it was very fibrous, and you’d have to like, you need a jaw to be able to chew a banana, right. Whereas these days, even from childhood, we’re just giving them mashed up baby formula for like two or three years old, whereas they should begin to swallow food and eating solid food for a long time to actually let the jaw develop. But anyway, side point, I went on a massive tangent, but basically try and bring it back post hoc hypotheses and explanations can be dangerous to play with. And even psychologists and statisticians who understand the way our mind works with associations, regressions, and base rate statistics have been shown in studies to falter when it comes to reasoning with certain statistics, intuitive judgments, and relationships and assumptions and whatnot. And I guess it’s always a controversial topic, but I suppose with COVID these days, at least over the past year, and I guess this is kind of what initiated me into really reading trying to understand that much as much about interpreting statistics as I can. And just numbers in general. There are very, very polarizing right now where it’s either this or that or like there’s only one group saying, no, this doesn’t work or one group saying that it does work and there’s no in between or no even attempt of trying to explain the crossover between the two arguments and stuff, and trying to actually get to the truth because it’s likely that these extreme things probably aren’t correct to their entirety, entirety anyway. So, let’s take, you know, lockdowns, that’s one that’s of massive controversy. Whether which one’s right you know, over time or just it will become clearer as to what what is correct. Whether lockdowns were launched One correct to do or whether they weren’t? Well, I mean, logically they seem correct to do, because I knew locked down, you know, you can’t transmit it. So rates of transmission go down. So you know, hurry for lock downs. But then if rates don’t go down, then people were just inclined to say, Oh, we didn’t lock down hard enough or people were breaking the lockdown. So it just rendered the lockdown, you know, useless. And you can even go like the other way round as well. So yeah, it’s just extremely nuanced. And you have to take into account natural viral dynamics of respiratory transmitted viruses and other coronaviruses. And it’s what the rates are going to come down anyway, because of seasonal variability. I don’t know if you’ve read it’s an old book by Edgar Hopson, I think his name is called the transmission of influenza. And he’s been like 80 years studying influenza and stuff. And yes, coronaviruses influenza, but, you know, the transmitted in the same way it to a certain extent. And they share a lot of similarities in the way the viral dynamics in different areas of the world works and whatnot. So yeah, if you take into account certain borrowed dynamics, where we just going down anyway, if you’re going to assume that it was going to be modeled, again, like it was just gonna follow a gomperts curve, then log downs just weren’t going to do anything. Anyway, it’s especially the first one, right? So it depends, there’s a lot to take into account. So when you just get pure, like, physicists modeling, you know, how many deaths there are going to be, is a very close minded thing. And I’m sure they take into into account loads of variables and assumptions, but still very difficult. certain models were 12 to 15 times off, right. But then everyone said that was because of lockdown, it prevented all of those deaths, you’re making a post hoc explanation for something, whereas you didn’t directly test a specific hypothesis. So this is where it becomes really hard, you have to really think it’s really hard right? To think about this stuff and take everything into account. untested, especially with something like this, a small and random as a virus, and all the random variables that you have to take into account that can influence viruses, like, you know, the butterfly effect all over again, a butterfly, flapping its wings in, like southern Australia causes a tornado on the other side of the world. Because of all the random events, I just kind of came off from it so hard. But then people are very quick to jump to conclusions, especially early on when there was limited data and stuff. Even now, there’s still somewhat limited data, or at least we’re still taking in more data to understand everything, right? I’m not trying to say what was correct and what was not correct. Because, you know, we’ll find out later on, if we do ever find out later on, we don’t probably know. So yeah, you can get a lot of these circular arguments, I just kind of keep going round on itself. And so yeah, well, we will naturally want to place it on things. So we would naturally want to say Oh, the reason the deaths weren’t so high, according to that model was because of luck. And we just want to play some kind of explanation on it. However, if we did look down and the deaths were that high, we would, we would have said lockdowns don’t work, or would we have said, Look, we just didn’t lock down properly. Yeah, you see where I’m going with it basically is very hard to take into account everything. And you know, most people are in support lockdowns, and it makes a lot of logical sense that people tend to fall to that default, but and then people don’t stay open to the other other side. Right? You know, it’s all right to make mistakes and take to do to say that no, this is correct. And this is what works and just Oh, believe the science and whatnot. That is a very arrogant point of view to come from, because that’s assuming you managed to make all the assumptions correctly. And do everything correctly, take everything into account, create your data collection, data was good to collect, like, the data in itself was good as well. And you calculated everything perfectly right? So you are correct. The scientific method, by definition doesn’t let you prove anything directly. And he lets you to disprove things. So actually requires a collation of information and, you know, subsequent consilience of all the information. And there’s a decent body of literature that says, you know, locked down interventions didn’t really do too much in terms of impacting the transmission of the virus in the way we would have liked it to. But even trying to read some of those papers and understand the way it was analyzed, it’s just really complicated, at least for me, because I’m not an expert in statistics, or viral dynamics or immunology, right? I’m just trying to get as high as I can, as quickly as I can, and just try and understand it for myself to a certain extent. And just because something is against natural intuition and logic doesn’t mean it’s wrong. Like I was saying about ketosis and stuff as well. So it’s very likely that our intuition and thinking is flawed because of all the biases that we intrinsically have. Whether knowledge wise, we are not open to a certain piece of knowledge because we haven’t been exposed to it or just emotion. So it’s difficult to admit that and process that. But I think you really have to take away all of your ego and pride to just become humble. humble. I’m not saying that I’m, I’m really humble in that sense. But you know, it’s difficult to do. And you just have to take a step back and be like, okay, there’s really complicated, there’s a lot to take into account, we probably don’t know the full picture, and we probably can’t even take it into account. So I’ll leave at the regression to the mean, lock randomness. Do you have anything else to add
Daniel Redfearn 1:10:26
on? No, I mean, the only thing I can think of is that, in this discussion, there are probably things we’ve said that anyone who’s listened to might disagree with, in certain points, and they’re very welcome to like, you know, your messages, getting contacts, I’d be interested as well, to hear more thoughts on it something I I think it’s just a question of taking in more information, learning more about it, doing more reading, and then just constantly trying to improve our views. So yeah, we’ve said a few times during this conversation, like, oh, we’re not experts. We’re not experts. We don’t know. We’re just like speculating, we’re just trying to understand it better. But if there is someone who thinks they probably have a different viewpoint than if we
Subaan Qasim 1:11:01
explain something incorrectly, in DMS, on Twitter, or Instagram, or just email us votes at Getting it.co.uk please send us a follow No, please do not. Do be, you know, strict and criticize us if we do if we did say something wrong, and then we’ll you know, make the correction, not in the shownotes or in a future episode. So yeah, hopefully that was interesting to listen to. And I guess we’ll leave it there.
Daniel Redfearn 1:11:24
We’ll leave it there. Peace.
Subaan Qasim 1:11:27