1 00:00:05,680 --> 00:00:08,360 Speaker 1: Hey, welcome to Stuff to Blow Your Mind. My name 2 00:00:08,400 --> 00:00:11,440 Speaker 1: is Robert Lamb and I'm Joe McCormick. And it's Saturday. 3 00:00:11,520 --> 00:00:15,520 Speaker 1: Time to go into the vault. This episode originally published 4 00:00:15,560 --> 00:00:19,919 Speaker 1: on August three, and it's called Regression to the Mean, 5 00:00:20,120 --> 00:00:23,440 Speaker 1: which is about regression to the mean. Uh, it's an 6 00:00:23,480 --> 00:00:27,400 Speaker 1: episode about a statistical phenomenon. But don't don't don't run away. 7 00:00:27,600 --> 00:00:31,440 Speaker 1: It's actually, I think super interesting and having this idea 8 00:00:31,560 --> 00:00:35,800 Speaker 1: in your toolkit really helps you understand the information that 9 00:00:35,880 --> 00:00:39,839 Speaker 1: you encounter in the world, uh much better. Absolutely, So 10 00:00:39,960 --> 00:00:45,599 Speaker 1: let's dive right in. Welcome to stot to Blow Your 11 00:00:45,600 --> 00:00:55,320 Speaker 1: Mind production of My Heart Radio. Hey, welcome to Stuff 12 00:00:55,360 --> 00:00:57,960 Speaker 1: to Blow Your Mind. My name is Robert Lamb and 13 00:00:58,040 --> 00:01:01,319 Speaker 1: I'm Joe McCormick. And in today this episode, we are 14 00:01:01,360 --> 00:01:04,240 Speaker 1: going to be focusing on a topic that is already 15 00:01:04,280 --> 00:01:07,600 Speaker 1: something that's very well known to people who are familiar 16 00:01:07,640 --> 00:01:11,600 Speaker 1: with quantitative research and statistics, but less known to the 17 00:01:11,640 --> 00:01:14,640 Speaker 1: general public. And uh, and I think that's a tragedy 18 00:01:14,680 --> 00:01:17,360 Speaker 1: because it's an idea that should really be part of 19 00:01:17,520 --> 00:01:21,160 Speaker 1: everybody's basic critical thinking tool kit, no matter what your 20 00:01:21,240 --> 00:01:25,080 Speaker 1: job is. And so in order to introduce this concept. 21 00:01:25,120 --> 00:01:27,039 Speaker 1: I thought it would be best to start with a 22 00:01:27,160 --> 00:01:30,720 Speaker 1: with a direct illustration from the real world of people 23 00:01:30,800 --> 00:01:35,959 Speaker 1: reaching incorrect conclusions by not understanding the subject of today's episode. 24 00:01:36,280 --> 00:01:38,080 Speaker 1: And so the illustration I want to start with is 25 00:01:38,319 --> 00:01:43,080 Speaker 1: an interesting story told by the psychologist Daniel Kaneman that's 26 00:01:43,120 --> 00:01:47,960 Speaker 1: about the illusory power of screaming at pilots. Uh So, 27 00:01:48,000 --> 00:01:50,800 Speaker 1: the context of the story is that Knemon says he 28 00:01:50,880 --> 00:01:54,800 Speaker 1: was giving a lecture about positive reinforcement to a group 29 00:01:54,880 --> 00:01:58,360 Speaker 1: of flight instructures. I think this was in the nineteen sixties, 30 00:01:59,120 --> 00:02:03,320 Speaker 1: and Kneman was trying to inform them about what he 31 00:02:03,320 --> 00:02:06,160 Speaker 1: believed at the time was the best consensus of scientific 32 00:02:06,200 --> 00:02:10,240 Speaker 1: research on learning and reinforcement, which was at the time 33 00:02:10,320 --> 00:02:13,160 Speaker 1: that if these flight instructors wanted their students to have 34 00:02:13,200 --> 00:02:17,120 Speaker 1: the best possible outcomes, they should focus more on praising 35 00:02:17,200 --> 00:02:20,040 Speaker 1: the students when they did well, then on chewing them 36 00:02:20,080 --> 00:02:23,520 Speaker 1: out when they did something wrong. And Knomon says that 37 00:02:23,680 --> 00:02:27,000 Speaker 1: when he finished his talk, one of the flight instructors 38 00:02:27,000 --> 00:02:28,880 Speaker 1: that he had been giving this lecture two got up 39 00:02:29,200 --> 00:02:31,720 Speaker 1: and tried to dispute him. He said, no, you're wrong. 40 00:02:32,080 --> 00:02:34,480 Speaker 1: And so the direct quote economy and gives from the 41 00:02:34,520 --> 00:02:38,400 Speaker 1: instructor here is, on many occasions I have praised flight 42 00:02:38,440 --> 00:02:43,160 Speaker 1: cadets for clean execution of some aerobatic maneuver, and in general, 43 00:02:43,200 --> 00:02:46,000 Speaker 1: when they try it again, they do worse. On the 44 00:02:46,000 --> 00:02:50,040 Speaker 1: other hand, I've often screamed at cadets for bad execution, 45 00:02:50,360 --> 00:02:53,520 Speaker 1: and in general they do better the next time. So 46 00:02:53,600 --> 00:02:57,240 Speaker 1: please don't tell us that reinforcement works and punishment does not, 47 00:02:57,560 --> 00:03:01,120 Speaker 1: because the opposite is the case. So you might think 48 00:03:01,160 --> 00:03:03,160 Speaker 1: he has a good point here. If you accept that 49 00:03:03,240 --> 00:03:06,520 Speaker 1: this flight instructor has had a lot of direct experience 50 00:03:06,639 --> 00:03:10,360 Speaker 1: working with students, and you trust him to remember the 51 00:03:10,400 --> 00:03:14,320 Speaker 1: relative frequency of these events pretty well, you might assume 52 00:03:14,520 --> 00:03:18,160 Speaker 1: that he has a meaningful rebuke to konom In. Here again, 53 00:03:18,240 --> 00:03:20,560 Speaker 1: he says that most of the time, after a cadet 54 00:03:20,600 --> 00:03:23,440 Speaker 1: does something bad and he screams at them, they do 55 00:03:23,520 --> 00:03:26,359 Speaker 1: better the next time, And after a cadet does something 56 00:03:26,400 --> 00:03:29,240 Speaker 1: good and he praises them, they actually do worse the 57 00:03:29,320 --> 00:03:33,240 Speaker 1: next time. So if he's remembering these experiences correctly, and 58 00:03:33,280 --> 00:03:35,680 Speaker 1: he's had a lot of them, it would really seem 59 00:03:35,760 --> 00:03:39,480 Speaker 1: like evidence that praise has a negative effect on learning, 60 00:03:40,000 --> 00:03:43,880 Speaker 1: maybe by making the student pilots soft and overconfident. Or something, 61 00:03:44,200 --> 00:03:47,800 Speaker 1: and getting chewed out is good for skill development. I 62 00:03:47,800 --> 00:03:50,120 Speaker 1: think it's quite easy to see the allure of this, 63 00:03:50,120 --> 00:03:53,680 Speaker 1: this false conclusion right right, And it's and you can 64 00:03:53,720 --> 00:03:56,120 Speaker 1: also easily imagine how you kind of build upon this 65 00:03:56,280 --> 00:04:01,320 Speaker 1: with certain loosely backed up you know, folk ideas about 66 00:04:01,360 --> 00:04:03,880 Speaker 1: how you encourage people and how people learn, and you 67 00:04:03,960 --> 00:04:05,520 Speaker 1: got to stay on and if they if you tell 68 00:04:05,560 --> 00:04:07,760 Speaker 1: them they're doing a good job, they'll get lazy, right, 69 00:04:07,840 --> 00:04:11,840 Speaker 1: folk wisdom, tough guy mentality. But Koneman saw something different 70 00:04:11,880 --> 00:04:15,040 Speaker 1: in this response, and he says that he immediately set 71 00:04:15,120 --> 00:04:19,080 Speaker 1: up an experiment on the spot to demonstrate the flaw 72 00:04:19,279 --> 00:04:22,120 Speaker 1: in the flight instructors thinking here, so I want to 73 00:04:22,160 --> 00:04:25,960 Speaker 1: read from Knomen's description, He says, I immediately arranged a 74 00:04:26,040 --> 00:04:30,039 Speaker 1: demonstration in which each participant tossed two coins at a 75 00:04:30,120 --> 00:04:34,520 Speaker 1: target behind his back without any feedback. We measured the 76 00:04:34,560 --> 00:04:37,440 Speaker 1: distances from the target and could see that those who 77 00:04:37,440 --> 00:04:41,359 Speaker 1: had done best the first time had mostly deteriorated on 78 00:04:41,400 --> 00:04:44,719 Speaker 1: their second try, and vice versa. But I knew that 79 00:04:44,800 --> 00:04:48,560 Speaker 1: this demonstration would not undo the effects of lifelong exposure 80 00:04:48,800 --> 00:04:53,760 Speaker 1: to a perverse contingency. So to explain this, this experiment 81 00:04:53,800 --> 00:04:56,000 Speaker 1: a little bit better. Right. He has people stand with 82 00:04:56,080 --> 00:04:58,560 Speaker 1: their backs to a target so they couldn't see it, 83 00:04:58,960 --> 00:05:01,040 Speaker 1: and they would take two a attempts to throw a 84 00:05:01,080 --> 00:05:04,400 Speaker 1: coin and hit the target without any feedback of any kind. 85 00:05:04,400 --> 00:05:08,240 Speaker 1: So they're not getting praised, they're not getting chewed out, nothing. Uh. 86 00:05:08,240 --> 00:05:10,880 Speaker 1: And after staging a number of these, he found again 87 00:05:10,920 --> 00:05:13,960 Speaker 1: what he suspected, that the people who were the closest 88 00:05:14,040 --> 00:05:17,040 Speaker 1: on the first throw did worse on their second throw, 89 00:05:17,440 --> 00:05:20,120 Speaker 1: and the people who were farthest away on their first 90 00:05:20,120 --> 00:05:23,680 Speaker 1: throw tended to do better on the second throw. So 91 00:05:24,040 --> 00:05:27,520 Speaker 1: what condiment is actually demonstrating here is something that doesn't 92 00:05:27,600 --> 00:05:31,560 Speaker 1: really have anything to do with learning or reinforcement, or 93 00:05:31,600 --> 00:05:36,320 Speaker 1: really skills or even human psychology. Instead, this demonstration is 94 00:05:36,400 --> 00:05:41,320 Speaker 1: showing the effects of chance, luck, and statistics. What he 95 00:05:41,360 --> 00:05:44,880 Speaker 1: was showing is the subject we're talking about today, regression 96 00:05:45,000 --> 00:05:48,000 Speaker 1: to the mean. Uh. You'll you'll see that phrase a 97 00:05:48,000 --> 00:05:50,800 Speaker 1: lot in in scientific literature and in statistics. But if 98 00:05:50,800 --> 00:05:53,960 Speaker 1: it helps to put it in more everyday terms, anytime 99 00:05:53,960 --> 00:05:57,000 Speaker 1: you see regression to the mean, you can translate it 100 00:05:57,000 --> 00:06:01,039 Speaker 1: in your head as trending toward the average, trending toward 101 00:06:01,160 --> 00:06:05,919 Speaker 1: the average. So to make the coin tossing illustration even clearer. 102 00:06:06,520 --> 00:06:09,400 Speaker 1: Imagine you throw the coin not twice, but that you 103 00:06:09,440 --> 00:06:12,720 Speaker 1: throw the coin a hundred times. So you stand there 104 00:06:12,720 --> 00:06:14,800 Speaker 1: throwing the coin a hundred times. And then let's say 105 00:06:14,839 --> 00:06:19,960 Speaker 1: afterwards you average together the distance from the target across 106 00:06:20,040 --> 00:06:22,600 Speaker 1: all a hundred throws, and you'll come up with some 107 00:06:22,720 --> 00:06:26,160 Speaker 1: kind of average distance from target. Uh, just to make 108 00:06:26,240 --> 00:06:28,440 Speaker 1: up a number for the sake of argument. Common doesn't 109 00:06:28,440 --> 00:06:30,760 Speaker 1: give this. But let's say the average distance from the 110 00:06:30,760 --> 00:06:34,880 Speaker 1: target across all your throws is nine centimeters. And remember 111 00:06:34,920 --> 00:06:37,320 Speaker 1: that you're getting no feedback at all here, so it's 112 00:06:37,400 --> 00:06:39,599 Speaker 1: unlikely that you will be getting much better as you 113 00:06:39,640 --> 00:06:43,040 Speaker 1: go on. So, given that the average distance from the 114 00:06:43,080 --> 00:06:46,279 Speaker 1: target is nine cimeters, if you throw a coin once 115 00:06:46,279 --> 00:06:49,120 Speaker 1: and it happens to be two centimeters from the target 116 00:06:49,200 --> 00:06:52,640 Speaker 1: so really close, is your next throw likely to be 117 00:06:52,720 --> 00:06:56,680 Speaker 1: about the same as that one, better or worse. Obviously, 118 00:06:56,720 --> 00:07:00,480 Speaker 1: it is overwhelmingly likely that your next throw will be worse, 119 00:07:00,680 --> 00:07:04,400 Speaker 1: just due to chance, probably closer to the average of 120 00:07:04,520 --> 00:07:08,159 Speaker 1: nine cimeters away. And the same goes for throws that 121 00:07:08,160 --> 00:07:11,280 Speaker 1: are really far off. You throw something three hundred centimeters 122 00:07:11,280 --> 00:07:14,800 Speaker 1: off your next random toss just by chance is likely 123 00:07:14,840 --> 00:07:18,320 Speaker 1: to be much better, much closer. So simply put, most 124 00:07:18,360 --> 00:07:22,200 Speaker 1: of the time, if you're sampling something in a series 125 00:07:22,440 --> 00:07:27,400 Speaker 1: over time, if one sample produces an extreme value, the 126 00:07:27,480 --> 00:07:30,600 Speaker 1: next one in the series is more likely to be 127 00:07:30,840 --> 00:07:34,320 Speaker 1: closer to the average instead of extreme in the same way. 128 00:07:34,880 --> 00:07:37,440 Speaker 1: In my experience. Uh, this is this is why it 129 00:07:37,480 --> 00:07:40,760 Speaker 1: can sometimes be liberating to start off a game of 130 00:07:40,800 --> 00:07:44,760 Speaker 1: bowling with just a disastrous gutter ball, because because I 131 00:07:44,840 --> 00:07:48,080 Speaker 1: know that I'm good enough that that's probably not going 132 00:07:48,120 --> 00:07:50,280 Speaker 1: to happen twice in a row, but it's definitely going 133 00:07:50,320 --> 00:07:52,400 Speaker 1: to happen at some point in the game because I'm 134 00:07:52,440 --> 00:07:56,120 Speaker 1: not that good, you know. I put like playing you know, 135 00:07:56,280 --> 00:07:58,880 Speaker 1: once a year or even with less frequency these days. 136 00:07:59,320 --> 00:08:01,720 Speaker 1: Oh yeah. And also like why I think a lot 137 00:08:01,760 --> 00:08:04,160 Speaker 1: of us have intuitions that when you try something for 138 00:08:04,200 --> 00:08:06,360 Speaker 1: the first time and you do really good on the 139 00:08:06,440 --> 00:08:09,440 Speaker 1: first attempt, that makes you kind of nervous because you 140 00:08:09,520 --> 00:08:11,320 Speaker 1: just know you're probably not going to live up to 141 00:08:11,400 --> 00:08:13,680 Speaker 1: that repeatedly. Yeah, like if you get if you get 142 00:08:13,680 --> 00:08:16,840 Speaker 1: a strike that first time, then that that first um 143 00:08:17,120 --> 00:08:18,760 Speaker 1: what is it round? I can't even remember. This is 144 00:08:18,760 --> 00:08:22,240 Speaker 1: how one frequently I bowl, Um, the first role. So 145 00:08:22,400 --> 00:08:26,200 Speaker 1: the first role first, the first column. You know, So, 146 00:08:26,280 --> 00:08:29,800 Speaker 1: the tendency of regression to the mean or or trending 147 00:08:29,840 --> 00:08:32,760 Speaker 1: towards the average is pretty obvious when you're dealing with 148 00:08:32,840 --> 00:08:36,719 Speaker 1: something like lots of random coin tosses with no feedback, 149 00:08:37,400 --> 00:08:40,800 Speaker 1: But it becomes much more obscure when you're dealing with, say, 150 00:08:40,960 --> 00:08:44,800 Speaker 1: a more more limited numbers of outcomes. In the series, 151 00:08:44,840 --> 00:08:50,319 Speaker 1: you're looking at and introducing possibly influential variables like pilot 152 00:08:50,480 --> 00:08:54,280 Speaker 1: skill and instructor feedback. After all, we would expect that 153 00:08:54,760 --> 00:08:58,400 Speaker 1: some variables having to do with instructor feedback should have 154 00:08:58,520 --> 00:09:01,280 Speaker 1: an effect on pilots ill, right, That's the point of 155 00:09:01,280 --> 00:09:04,440 Speaker 1: teaching is to have an effect over time, and after all, 156 00:09:04,480 --> 00:09:07,880 Speaker 1: in this one scenario, the conomen describes the the instructor 157 00:09:08,000 --> 00:09:11,920 Speaker 1: believed that his verbal abuse of the students was so 158 00:09:12,080 --> 00:09:15,000 Speaker 1: motivating that it made them instantly better on the stick. 159 00:09:15,480 --> 00:09:19,640 Speaker 1: And you can't necessarily rule that out, but it's unlikely. 160 00:09:19,840 --> 00:09:22,880 Speaker 1: I think I'm convinced that regression to the mean could 161 00:09:22,920 --> 00:09:27,200 Speaker 1: more easily explain this flight instructor's belief that screaming at 162 00:09:27,240 --> 00:09:30,839 Speaker 1: pilots for screw ups made them better at planes, because, again, 163 00:09:30,920 --> 00:09:34,280 Speaker 1: on average, even in the absence of any feedback at all. 164 00:09:34,840 --> 00:09:38,800 Speaker 1: If a pilot in training executes a maneuver perfectly, the 165 00:09:39,000 --> 00:09:42,800 Speaker 1: random fluctuation from one execution to the next will tend 166 00:09:42,840 --> 00:09:45,760 Speaker 1: to mean that their next attempt probably won't be as 167 00:09:45,800 --> 00:09:48,880 Speaker 1: good as that really good when the last time. And likewise, 168 00:09:48,920 --> 00:09:51,720 Speaker 1: if they make a major error totally botch a maneuver, 169 00:09:51,960 --> 00:09:54,320 Speaker 1: they're more likely to do better the next time just 170 00:09:54,400 --> 00:09:57,920 Speaker 1: by chance. Both of these tendencies are regression towards the mean. 171 00:09:58,679 --> 00:10:01,839 Speaker 1: But then Conomon actually draw is a really interesting observation 172 00:10:01,920 --> 00:10:05,920 Speaker 1: about about about our psychology and about culture from this fact, 173 00:10:06,520 --> 00:10:10,040 Speaker 1: so to quote him directly, this was a joyous moment 174 00:10:10,080 --> 00:10:13,000 Speaker 1: in which I understood an important truth about the world. 175 00:10:13,559 --> 00:10:16,600 Speaker 1: Because we tend to reward others when they do well 176 00:10:17,080 --> 00:10:20,160 Speaker 1: and punish them when they do badly, and because there 177 00:10:20,320 --> 00:10:23,080 Speaker 1: is regression to the mean, it is part of the 178 00:10:23,200 --> 00:10:27,760 Speaker 1: human condition that we are statistically punished for rewarding others 179 00:10:28,000 --> 00:10:31,280 Speaker 1: and rewarded for punishing them. And that was one of 180 00:10:31,280 --> 00:10:33,160 Speaker 1: those things that when I read it, I was just like, 181 00:10:33,200 --> 00:10:37,720 Speaker 1: oh my god, that's so true. Um, yeah, yeah, And 182 00:10:37,760 --> 00:10:40,640 Speaker 1: in this specific instance, it makes me think about the 183 00:10:40,720 --> 00:10:44,640 Speaker 1: special effect of reversion to the mean, fallacies on motivating 184 00:10:44,679 --> 00:10:47,720 Speaker 1: belief in the effectiveness of of not just screaming at 185 00:10:47,760 --> 00:10:51,320 Speaker 1: pilots in this one case, but all kinds of punishment behaviors, 186 00:10:51,920 --> 00:10:55,680 Speaker 1: for example, corporal punishment. Thankfully you hear this less often 187 00:10:55,760 --> 00:10:58,400 Speaker 1: these days, but I remember when I was younger, I 188 00:10:58,520 --> 00:11:01,120 Speaker 1: used to hear people who would defice end the parental 189 00:11:01,160 --> 00:11:04,280 Speaker 1: practice of spanking children by saying, you know, I don't 190 00:11:04,320 --> 00:11:06,439 Speaker 1: I don't care what the site scientists say. I don't 191 00:11:06,480 --> 00:11:09,200 Speaker 1: care what the research says. I know from experience that 192 00:11:09,320 --> 00:11:12,760 Speaker 1: it works. To the extent that comments like this were 193 00:11:12,880 --> 00:11:15,880 Speaker 1: based on any real experience and observation and not just 194 00:11:15,920 --> 00:11:18,760 Speaker 1: sort of a free form, self justifying statement that had 195 00:11:18,800 --> 00:11:21,480 Speaker 1: nothing to do with experience. I bet a lot of 196 00:11:21,480 --> 00:11:26,240 Speaker 1: it was fallacious inference of causation actually based on regression 197 00:11:26,280 --> 00:11:29,480 Speaker 1: to the mean, just like in this condiment example. But anyway, 198 00:11:29,640 --> 00:11:31,600 Speaker 1: I thought it would be interesting to talk a bit 199 00:11:31,679 --> 00:11:34,800 Speaker 1: more about regression to the mean today, because it's one 200 00:11:34,800 --> 00:11:37,079 Speaker 1: of those things that, again, once you see it, it's 201 00:11:37,080 --> 00:11:41,000 Speaker 1: it's pretty simple, it's actually actually pretty clear, but understanding 202 00:11:41,000 --> 00:11:43,440 Speaker 1: it can help you have a better sense of how 203 00:11:43,480 --> 00:11:47,480 Speaker 1: good science works and help keep you from drawing hasty 204 00:11:47,559 --> 00:11:51,200 Speaker 1: inferences in everyday life. Yeah, because it is it is 205 00:11:51,240 --> 00:11:55,120 Speaker 1: interesting how kind of an insidious the results can be 206 00:11:55,200 --> 00:11:59,920 Speaker 1: the idea that that again, praise is ultimately punished because 207 00:12:00,120 --> 00:12:01,960 Speaker 1: is there's going to be a regression to the mean, 208 00:12:02,400 --> 00:12:05,679 Speaker 1: to to to to the mean, and then likewise there 209 00:12:05,720 --> 00:12:10,240 Speaker 1: can be this illusion, uh that uh that's screaming at 210 00:12:10,240 --> 00:12:12,640 Speaker 1: pilots and so forth is going to be the successful 211 00:12:12,640 --> 00:12:15,440 Speaker 1: way to go about things. Um. So yeah, this is 212 00:12:15,480 --> 00:12:17,760 Speaker 1: I think this is an important episode to cover because 213 00:12:17,760 --> 00:12:19,320 Speaker 1: it's the kind of thing that it's the kind of 214 00:12:19,320 --> 00:12:21,640 Speaker 1: tool you kind of need tucked in your back pocket, 215 00:12:22,000 --> 00:12:25,200 Speaker 1: even if you're just doing something like like scanning science 216 00:12:25,240 --> 00:12:28,240 Speaker 1: headlines on a you know, a news server or social 217 00:12:28,280 --> 00:12:32,600 Speaker 1: media message board. Yeah, because of course, understanding regression to 218 00:12:32,640 --> 00:12:36,120 Speaker 1: the mean is extremely important in what scientists do when 219 00:12:36,120 --> 00:12:40,559 Speaker 1: they design good experiments. If you don't take into account 220 00:12:40,559 --> 00:12:43,480 Speaker 1: regression to the mean, you can incorrectly believe you have 221 00:12:43,559 --> 00:12:46,520 Speaker 1: discovered some kind of tiger repellent or something. Uh. This 222 00:12:46,559 --> 00:12:49,319 Speaker 1: concern plays a huge role in the history of medicine. 223 00:12:49,720 --> 00:12:52,080 Speaker 1: It's part of the design of good medical research, or 224 00:12:52,120 --> 00:12:56,360 Speaker 1: really any field that seeks to find remedies for problems. 225 00:12:56,960 --> 00:13:01,080 Speaker 1: So consider a very basic hypothetical, uh path medicines, say 226 00:13:01,120 --> 00:13:03,160 Speaker 1: from a hundred years ago. So you know, you have 227 00:13:03,320 --> 00:13:06,880 Speaker 1: you have a foot pain that you've never really had before. Uh, 228 00:13:07,440 --> 00:13:08,920 Speaker 1: you know, you want it to go away. So you 229 00:13:08,960 --> 00:13:11,280 Speaker 1: go to the store and you buy a bottle of 230 00:13:11,400 --> 00:13:15,280 Speaker 1: doctor Field Grades No Fail Pantasy for tumors, ulcers, cramps, 231 00:13:15,280 --> 00:13:18,640 Speaker 1: and rooms, and you you pull the cork out, you 232 00:13:18,720 --> 00:13:21,440 Speaker 1: chug it, and then the next day your foot feels better. 233 00:13:22,000 --> 00:13:25,319 Speaker 1: Now you can conclude from this that the doctor Field 234 00:13:25,320 --> 00:13:28,679 Speaker 1: Grades cured you. But how do you know actually that 235 00:13:28,720 --> 00:13:31,520 Speaker 1: the feelings in your foot didn't just regress to the 236 00:13:31,559 --> 00:13:34,599 Speaker 1: mean because the average is a low amount or no 237 00:13:34,720 --> 00:13:37,200 Speaker 1: amount of foot pain. And if you don't have a 238 00:13:37,240 --> 00:13:41,240 Speaker 1: medication that's tested with control groups and and randomized allocation 239 00:13:41,280 --> 00:13:44,480 Speaker 1: into the groups, then how do you know that that 240 00:13:44,559 --> 00:13:47,920 Speaker 1: the medicine actually did anything at all? Yeah? Yeah, So 241 00:13:47,960 --> 00:13:50,479 Speaker 1: many of the examples you see for this and the applications, 242 00:13:50,520 --> 00:13:53,600 Speaker 1: you're dealing with some sort of situation in the world, 243 00:13:53,600 --> 00:13:58,439 Speaker 1: whether there is fluctuation and or change happening, often separately 244 00:13:58,520 --> 00:14:00,560 Speaker 1: from whatever is being tested. So in this case, yeah, 245 00:14:00,559 --> 00:14:03,679 Speaker 1: the doctor Field grads could have just been like just water. 246 00:14:03,960 --> 00:14:06,400 Speaker 1: It just just you know, but there is the illusion 247 00:14:06,440 --> 00:14:09,360 Speaker 1: that it worked because things got better. But if you 248 00:14:09,360 --> 00:14:11,520 Speaker 1: don't have a control group and to you know, to 249 00:14:11,600 --> 00:14:13,120 Speaker 1: drive home what that is. That would be like if 250 00:14:13,160 --> 00:14:15,680 Speaker 1: you had a had like three different groups and a 251 00:14:15,720 --> 00:14:19,440 Speaker 1: study of doctor Field Greats elixir. Here, one group was 252 00:14:19,480 --> 00:14:23,720 Speaker 1: taking doctor Field graades elixer, another group was taking I 253 00:14:23,760 --> 00:14:25,840 Speaker 1: don't know, let's say a half dose of Feel Grade 254 00:14:25,960 --> 00:14:29,640 Speaker 1: or maybe a competitor's tonic. And in one group the 255 00:14:29,680 --> 00:14:34,120 Speaker 1: control group was taking nothing was or was taking you know, 256 00:14:34,160 --> 00:14:37,960 Speaker 1: just water or something to that effect, something completely innate. Uh. 257 00:14:38,000 --> 00:14:42,080 Speaker 1: And that would be that would be a group that 258 00:14:42,080 --> 00:14:45,360 Speaker 1: you would judge the results of the other categories by, right, 259 00:14:45,360 --> 00:14:47,880 Speaker 1: and you would need to randomly sort the people into 260 00:14:47,880 --> 00:14:50,280 Speaker 1: those groups. So it wasn't just that, you know, the 261 00:14:50,680 --> 00:14:54,000 Speaker 1: only the people with real severe foot pain we're taking 262 00:14:54,040 --> 00:14:57,080 Speaker 1: the doctor Field grades because the more extreme their pain 263 00:14:57,160 --> 00:15:00,160 Speaker 1: to begin with, probably the more likely they are are 264 00:15:00,200 --> 00:15:03,560 Speaker 1: to have that pain be lessened or go away over time, 265 00:15:03,640 --> 00:15:06,520 Speaker 1: just naturally. Right. And uh. And I'm going to have 266 00:15:06,560 --> 00:15:08,880 Speaker 1: a more specific example of this a little later in 267 00:15:08,880 --> 00:15:10,800 Speaker 1: the podcast. So if you if you still don't get it, 268 00:15:10,840 --> 00:15:13,040 Speaker 1: just hang on we'll we'll have another example in a 269 00:15:13,080 --> 00:15:21,120 Speaker 1: bit thank. I was looking at an article in the 270 00:15:21,120 --> 00:15:24,960 Speaker 1: British Medical Journal from nineteen that was just a collection 271 00:15:25,040 --> 00:15:28,360 Speaker 1: of different examples of regression to the mean in real 272 00:15:28,400 --> 00:15:32,040 Speaker 1: life medical research. This was by J. Martin Bland and 273 00:15:32,120 --> 00:15:36,480 Speaker 1: Douglas J. Altman called statistics notes some examples of regression 274 00:15:36,480 --> 00:15:39,520 Speaker 1: towards the mean, and they point out a very common 275 00:15:39,560 --> 00:15:42,200 Speaker 1: type of example. So this will be similar to what 276 00:15:42,240 --> 00:15:45,600 Speaker 1: we just talked about. The author's right. In clinical practice, 277 00:15:45,640 --> 00:15:49,960 Speaker 1: there are many measurements such as weight, serum, cholesterol concentration, 278 00:15:50,120 --> 00:15:54,480 Speaker 1: or blood pressure, for which particularly high or low values 279 00:15:54,520 --> 00:15:58,240 Speaker 1: are signs of underlying disease or risk factors for disease. 280 00:15:58,840 --> 00:16:01,800 Speaker 1: People with extreme values of the measurements such as high 281 00:16:01,840 --> 00:16:05,280 Speaker 1: blood pressure may be treated to bring their values closer 282 00:16:05,320 --> 00:16:07,960 Speaker 1: to the mean. If they are measured again, we will 283 00:16:08,000 --> 00:16:10,640 Speaker 1: observe that the mean of the extreme group is now 284 00:16:10,720 --> 00:16:13,480 Speaker 1: closer to the mean of the whole population. That is, 285 00:16:13,680 --> 00:16:17,160 Speaker 1: it is reduced. This should not be interpreted as showing 286 00:16:17,200 --> 00:16:20,680 Speaker 1: the effect of the treatment. Even if subjects are not treated, 287 00:16:20,760 --> 00:16:24,120 Speaker 1: the mean blood pressure will go down owing to regression 288 00:16:24,160 --> 00:16:27,320 Speaker 1: towards the means. So again something starts with an extreme 289 00:16:27,480 --> 00:16:30,520 Speaker 1: value in certain types of cases, you would just expect 290 00:16:30,560 --> 00:16:33,800 Speaker 1: it to have a less extreme value the next time 291 00:16:33,920 --> 00:16:38,040 Speaker 1: due to random fluctuation. Uh So again, you know this 292 00:16:38,080 --> 00:16:41,000 Speaker 1: could fill you with despair because you might wonder, well, 293 00:16:41,040 --> 00:16:42,880 Speaker 1: then how could you ever know if a treatment was 294 00:16:42,920 --> 00:16:45,640 Speaker 1: effective or not. But again, this is where the standard 295 00:16:45,680 --> 00:16:49,080 Speaker 1: practices of science based medicine come to play. Instead of 296 00:16:49,160 --> 00:16:52,120 Speaker 1: just taking people with some extreme measurement and giving them 297 00:16:52,120 --> 00:16:56,680 Speaker 1: a treatment, you randomize them into test groups and control groups, 298 00:16:56,680 --> 00:16:58,280 Speaker 1: like we were just talking about. So if you have 299 00:16:58,320 --> 00:17:01,720 Speaker 1: a large enough sample, you really randomize the groups. People 300 00:17:01,760 --> 00:17:05,600 Speaker 1: with the extreme starting conditions will somewhat regress towards the mean, 301 00:17:05,760 --> 00:17:08,280 Speaker 1: but they will all regress toward the mean on average 302 00:17:08,400 --> 00:17:11,640 Speaker 1: the same rate, whether they're receiving a real potential treatment 303 00:17:12,040 --> 00:17:14,440 Speaker 1: or they're in the placebo group. But if the treatment 304 00:17:14,480 --> 00:17:17,600 Speaker 1: actually does something helpful, this effect will manifest as the 305 00:17:17,640 --> 00:17:21,880 Speaker 1: difference between the two groups. So good scientific research, good 306 00:17:21,920 --> 00:17:25,440 Speaker 1: medical research has methods for excluding the effects of reversion 307 00:17:25,480 --> 00:17:28,480 Speaker 1: to the mean on their findings. We have the tools, 308 00:17:28,520 --> 00:17:32,480 Speaker 1: but we can still fall into the trap of regression 309 00:17:32,480 --> 00:17:35,679 Speaker 1: to the mean fallacies, especially in our day to day lives. 310 00:17:35,840 --> 00:17:39,320 Speaker 1: Drawing inferences the way that that the pilot and Inconomens 311 00:17:39,320 --> 00:17:42,159 Speaker 1: story did, or or even in science if we're not 312 00:17:42,240 --> 00:17:45,960 Speaker 1: careful and deliberate about designing experiments. And in addition to 313 00:17:46,119 --> 00:17:49,879 Speaker 1: just a methodology design that has you know, a randomized 314 00:17:49,920 --> 00:17:52,960 Speaker 1: groups and control groups, there are also ways of trying 315 00:17:53,000 --> 00:17:56,320 Speaker 1: to counteract regression to the mean, just through statistical methods 316 00:17:56,400 --> 00:17:59,560 Speaker 1: that are maybe less reliable, but there are statistical methods 317 00:17:59,560 --> 00:18:03,680 Speaker 1: people can used to try to apply sort of modifiers 318 00:18:03,760 --> 00:18:06,800 Speaker 1: to data in order to estimate regression to the mean 319 00:18:07,000 --> 00:18:10,600 Speaker 1: and uh and counteract its effects. So again, we have 320 00:18:10,720 --> 00:18:13,679 Speaker 1: tools within scientific research to to figure this out, and 321 00:18:13,720 --> 00:18:16,760 Speaker 1: it's a lot of what science does is trying to 322 00:18:16,800 --> 00:18:19,560 Speaker 1: sort out the difference between regression to the mean and 323 00:18:19,680 --> 00:18:23,240 Speaker 1: actual effects of interventions. But in our day to day lives, 324 00:18:23,320 --> 00:18:26,080 Speaker 1: we still fall for regression to the mean fallacies all 325 00:18:26,080 --> 00:18:29,240 Speaker 1: the time. Yeah, and it's important to realize too that 326 00:18:29,320 --> 00:18:31,680 Speaker 1: it's not just a situation where regression towards the mean 327 00:18:32,080 --> 00:18:36,000 Speaker 1: could create an illusion of something working when it doesn't. Uh. 328 00:18:36,040 --> 00:18:40,800 Speaker 1: You know, sometimes it can just potentially overstate um the 329 00:18:40,840 --> 00:18:43,879 Speaker 1: effects of something. For an example of that that I 330 00:18:43,920 --> 00:18:46,960 Speaker 1: was looking at was that regression towards the mean, or 331 00:18:46,960 --> 00:18:49,600 Speaker 1: the failure to account for it can also overstate the 332 00:18:49,600 --> 00:18:53,520 Speaker 1: effectiveness of something like traffic light cameras. Is it making 333 00:18:53,560 --> 00:18:57,680 Speaker 1: a difference and cutting down on accidents? Perhaps, but any 334 00:18:57,760 --> 00:19:02,520 Speaker 1: actual effectiveness could potentially be overstated by failure to account 335 00:19:02,680 --> 00:19:05,680 Speaker 1: for just regression towards the mean. Oh yeah, so where 336 00:19:05,680 --> 00:19:09,240 Speaker 1: do you tend to install things like that? High acts 337 00:19:09,400 --> 00:19:12,440 Speaker 1: like problem areas? Right, So, if there's like a stretch 338 00:19:12,520 --> 00:19:15,800 Speaker 1: of road that has a lot of problems on people 339 00:19:15,920 --> 00:19:18,520 Speaker 1: really speeding a lot there or crashing a lot there, 340 00:19:18,920 --> 00:19:22,160 Speaker 1: that might be where you stage the intervention. It's possible 341 00:19:22,200 --> 00:19:26,480 Speaker 1: some things like that fluctuate naturally over time in different locations, 342 00:19:27,200 --> 00:19:29,280 Speaker 1: and you put the cameras in place, and it could 343 00:19:29,280 --> 00:19:31,359 Speaker 1: have an effect, but maybe not as much of an 344 00:19:31,359 --> 00:19:35,159 Speaker 1: effect as it looks like it is taking place. Again, 345 00:19:35,240 --> 00:19:39,560 Speaker 1: if you don't factor regression towards the mean into the study. 346 00:19:40,040 --> 00:19:43,840 Speaker 1: Right now, While our TM is a very important phenomenon 347 00:19:43,880 --> 00:19:46,840 Speaker 1: to understand and take into account, it certainly doesn't apply 348 00:19:47,040 --> 00:19:51,320 Speaker 1: to every sequence of values you could repeatedly sample, so 349 00:19:51,400 --> 00:19:53,840 Speaker 1: you also have to be careful not to apply it 350 00:19:53,880 --> 00:19:57,720 Speaker 1: in situations where it isn't warranted. I was you know, 351 00:19:57,920 --> 00:19:59,879 Speaker 1: there are a million examples. You could cite one that 352 00:20:00,040 --> 00:20:02,919 Speaker 1: came to my mind as the orbital decay of a satellite. 353 00:20:03,240 --> 00:20:06,439 Speaker 1: Let's say you've got a communication satellite in lower orbit 354 00:20:06,880 --> 00:20:09,040 Speaker 1: and you get a reading on its altitude and the 355 00:20:09,119 --> 00:20:13,320 Speaker 1: reading is lower than the satellites average altitude. Uh. Now 356 00:20:14,000 --> 00:20:16,000 Speaker 1: you might say, hey, I think this means we need 357 00:20:16,040 --> 00:20:18,600 Speaker 1: to program a reboost to insert it back into the 358 00:20:19,160 --> 00:20:22,640 Speaker 1: orbit where it's supposed to be. And somebody could erroneously 359 00:20:22,920 --> 00:20:25,960 Speaker 1: apply regression to the mean here and say, nah, we 360 00:20:26,000 --> 00:20:28,240 Speaker 1: don't need to do that. The satellite might just return 361 00:20:28,320 --> 00:20:31,520 Speaker 1: to its average altitude. It doesn't apply in this scenario, 362 00:20:31,600 --> 00:20:34,439 Speaker 1: even though you are taking repeated measurements of a value 363 00:20:34,480 --> 00:20:39,119 Speaker 1: over time, because we know things about the physical characteristics 364 00:20:39,160 --> 00:20:42,840 Speaker 1: determining the orbit of satellites and in lower th orbit uh, 365 00:20:42,880 --> 00:20:46,440 Speaker 1: and that due to factors like atmospheric drag, their altitude 366 00:20:46,440 --> 00:20:50,359 Speaker 1: tends to trend steadily downward over time in a consistent 367 00:20:50,400 --> 00:20:54,320 Speaker 1: direction down down, down, So eventually you will need a 368 00:20:54,359 --> 00:20:56,600 Speaker 1: reboost in order to put it back up to the 369 00:20:56,640 --> 00:21:00,240 Speaker 1: correct distance. So regression to the mean apply is to 370 00:21:00,440 --> 00:21:04,440 Speaker 1: certain kinds of data that are repeatedly sampled data where 371 00:21:04,480 --> 00:21:09,320 Speaker 1: there is natural random fluctuation back and forth, not a 372 00:21:09,440 --> 00:21:12,320 Speaker 1: steady trend in the data in one direction on the 373 00:21:12,359 --> 00:21:15,720 Speaker 1: relevant time scale. The other thing that's important to understand 374 00:21:15,800 --> 00:21:18,800 Speaker 1: is that systems where you expect to find regression to 375 00:21:18,880 --> 00:21:22,639 Speaker 1: the mean are systems in which the repeated data values 376 00:21:22,680 --> 00:21:26,800 Speaker 1: you're sampling are to some degree determined by luck or chance. 377 00:21:27,359 --> 00:21:30,800 Speaker 1: If a series of values is influenced almost entirely by 378 00:21:31,080 --> 00:21:34,720 Speaker 1: deterministic influence, like in the satellite example, by like the 379 00:21:34,800 --> 00:21:39,199 Speaker 1: laws of physics, or by some extremely reliable skill with 380 00:21:39,320 --> 00:21:43,399 Speaker 1: little room for variation, values don't really regress towards the 381 00:21:43,440 --> 00:21:46,040 Speaker 1: mean in the same way because there's just less random 382 00:21:46,080 --> 00:21:49,760 Speaker 1: fluctuation back and forth to begin with. The more chance 383 00:21:49,880 --> 00:21:53,119 Speaker 1: and random variation plays a role in the outcome, the 384 00:21:53,240 --> 00:21:55,960 Speaker 1: more you will tend to observe regression towards the mean 385 00:21:56,040 --> 00:21:58,920 Speaker 1: after an extreme sample in in whatever it is you're 386 00:21:58,960 --> 00:22:02,879 Speaker 1: looking at, I've I've read that the progression towards the 387 00:22:02,880 --> 00:22:05,760 Speaker 1: mean is is not to be confused with the law 388 00:22:05,800 --> 00:22:09,119 Speaker 1: of large numbers. For example, uh. This is the the 389 00:22:09,200 --> 00:22:11,800 Speaker 1: law that that states, as a sample size becomes larger, 390 00:22:12,040 --> 00:22:15,600 Speaker 1: the sample mean gets closer to the expected value. So 391 00:22:15,920 --> 00:22:18,480 Speaker 1: a coin flipping example is key here. Flip a coin 392 00:22:18,840 --> 00:22:21,640 Speaker 1: and the random results are going to ultimately average out 393 00:22:21,960 --> 00:22:25,320 Speaker 1: to a point five proportion. But if you only flip 394 00:22:25,400 --> 00:22:29,000 Speaker 1: the coin ten times, you might not see this breakdown. Um. 395 00:22:29,040 --> 00:22:32,399 Speaker 1: And this also applies to say, even odds on the 396 00:22:32,480 --> 00:22:34,400 Speaker 1: rolling of a of a D six of a six 397 00:22:34,440 --> 00:22:38,520 Speaker 1: sided die. Uh So for example, two regular people, that's 398 00:22:38,560 --> 00:22:42,240 Speaker 1: just to die that nerves like us, it's a D six. Yeah. 399 00:22:42,480 --> 00:22:43,680 Speaker 1: D six is what I could get my hands on. 400 00:22:43,720 --> 00:22:45,040 Speaker 1: Because I was like, well, I'm gonna do an example. 401 00:22:45,040 --> 00:22:47,160 Speaker 1: I'm gonna try it myself. So while I was putting 402 00:22:47,160 --> 00:22:50,359 Speaker 1: together notes for this, I went ahead and rolled ten times, 403 00:22:50,800 --> 00:22:53,919 Speaker 1: and I got even even odd even odd even even 404 00:22:53,960 --> 00:22:57,480 Speaker 1: even even odd. So that's that's seven to three in 405 00:22:57,600 --> 00:23:00,200 Speaker 1: favor of even. So it might make you wonder, well, 406 00:23:00,320 --> 00:23:02,720 Speaker 1: is this die broken? Does this D six need to 407 00:23:02,760 --> 00:23:06,479 Speaker 1: go away? Because it can't be trusted to roll? Uh? 408 00:23:07,240 --> 00:23:10,960 Speaker 1: You know a balanced array of odd and even numbers. Well, no, 409 00:23:11,160 --> 00:23:13,520 Speaker 1: that's not the case. Uh. And if I were to 410 00:23:13,680 --> 00:23:17,919 Speaker 1: roll this, say another hundred times, another thousand times, I 411 00:23:17,920 --> 00:23:20,879 Speaker 1: would see things even out even more to where we 412 00:23:20,920 --> 00:23:24,720 Speaker 1: would see this, uh, this point five proportion of odd 413 00:23:24,840 --> 00:23:28,440 Speaker 1: versus even right. So these are not exactly the same thing, 414 00:23:28,480 --> 00:23:30,639 Speaker 1: regression to the mean and the law of large numbers, 415 00:23:30,640 --> 00:23:35,359 Speaker 1: but they are closely related. Both observations require you to 416 00:23:35,440 --> 00:23:38,840 Speaker 1: think about statistical tendencies over time, over a time period 417 00:23:38,840 --> 00:23:42,160 Speaker 1: of repeated sampling, and both are premised on the knowledge 418 00:23:42,160 --> 00:23:46,119 Speaker 1: that repeated samples will tend towards the average. But regression 419 00:23:46,160 --> 00:23:48,760 Speaker 1: to the mean has to do with the idea that 420 00:23:48,840 --> 00:23:51,960 Speaker 1: if you start with an extreme observation and there is 421 00:23:52,040 --> 00:23:55,439 Speaker 1: some role of chance or luck in determining the value 422 00:23:55,480 --> 00:23:58,160 Speaker 1: of this observation, the next time you sample it, it's 423 00:23:58,200 --> 00:24:01,080 Speaker 1: more likely to be closer to the average. The law 424 00:24:01,119 --> 00:24:03,959 Speaker 1: of large numbers is that if in the real world, 425 00:24:04,040 --> 00:24:07,520 Speaker 1: the more times you run something, the closer your outcomes 426 00:24:07,560 --> 00:24:09,640 Speaker 1: in the real world will will be to the sort 427 00:24:09,680 --> 00:24:13,040 Speaker 1: of perfect mathematical average that you would estimate just given 428 00:24:13,080 --> 00:24:15,800 Speaker 1: the chances to begin with. Now, I want to come 429 00:24:15,800 --> 00:24:18,720 Speaker 1: back to regression towards the mean in um in medical 430 00:24:18,760 --> 00:24:21,640 Speaker 1: studies because I found a really interesting one that came 431 00:24:21,680 --> 00:24:24,399 Speaker 1: out earlier this year. Uh So, a lot of a 432 00:24:24,440 --> 00:24:27,879 Speaker 1: lot of the examples you find involving regression to the 433 00:24:27,920 --> 00:24:30,840 Speaker 1: mean involved sports or economics, and I found. This one 434 00:24:30,880 --> 00:24:34,639 Speaker 1: discussed in a New York Times article again from earlier 435 00:24:34,680 --> 00:24:38,200 Speaker 1: this year titled Intense strength training does not ease knee pain, 436 00:24:38,320 --> 00:24:41,760 Speaker 1: study finds by Gina Colada. Uh, this is referring to 437 00:24:41,800 --> 00:24:45,640 Speaker 1: a study published in Jama that entailed an eighteen month 438 00:24:45,640 --> 00:24:50,000 Speaker 1: clinical trial involving three d and seventy seven participants. Okay, okay, 439 00:24:50,000 --> 00:24:52,399 Speaker 1: So the basic situation, the setup for this paper is 440 00:24:52,440 --> 00:24:57,000 Speaker 1: that a lot of people have knee osteoarthritis, and one 441 00:24:57,000 --> 00:25:00,080 Speaker 1: of the go to treatment recommendations has long been and 442 00:25:00,359 --> 00:25:04,840 Speaker 1: strength training. So in this study they decided to look 443 00:25:04,840 --> 00:25:08,520 Speaker 1: into it with three basic groups, one that received intense 444 00:25:08,600 --> 00:25:13,000 Speaker 1: strength training, another that received moderate strength training, and another 445 00:25:13,160 --> 00:25:17,240 Speaker 1: that received counseling on healthy living. So that third group, 446 00:25:17,400 --> 00:25:19,879 Speaker 1: that's the control group, they did not have any amount 447 00:25:19,920 --> 00:25:23,439 Speaker 1: of strength training, just uh, you know, some positive counseling 448 00:25:23,480 --> 00:25:27,479 Speaker 1: about healthy living. Sure, so the researchers here apparently actually 449 00:25:27,480 --> 00:25:30,320 Speaker 1: expected to see the intense strength training take the lead 450 00:25:30,400 --> 00:25:35,040 Speaker 1: that they were looking to identify what has been just 451 00:25:35,040 --> 00:25:38,960 Speaker 1: sort of accepted wisdom, um and and again this this 452 00:25:39,000 --> 00:25:42,440 Speaker 1: has been the predominant treatment idea. But instead they found 453 00:25:42,440 --> 00:25:46,240 Speaker 1: that the results were the same for all three groups quote, 454 00:25:46,320 --> 00:25:50,520 Speaker 1: everyone reported slightly less pain, including those who had received 455 00:25:50,600 --> 00:25:53,800 Speaker 1: only counseling. Now why is that? Well, as Colotta points out, 456 00:25:53,840 --> 00:25:56,600 Speaker 1: there's there's always room for other effects, especially say the 457 00:25:56,640 --> 00:26:01,199 Speaker 1: placebo effect. Uh but regression to the is also a 458 00:26:01,240 --> 00:26:04,560 Speaker 1: heavy consideration here and certainly could work in congress with 459 00:26:04,600 --> 00:26:07,639 Speaker 1: the placebo effect. Right, So, you don't necessarily have to 460 00:26:07,720 --> 00:26:11,160 Speaker 1: assume that the counseling actually helped to heal people's knees, 461 00:26:11,200 --> 00:26:13,040 Speaker 1: though it may have in in in some it may 462 00:26:13,040 --> 00:26:15,560 Speaker 1: have had some kind of mechanistic effect in some way, 463 00:26:15,760 --> 00:26:18,840 Speaker 1: a mind body kind of thing. But you would also 464 00:26:18,920 --> 00:26:22,520 Speaker 1: just expect over time, people who have an extreme starting position, 465 00:26:22,560 --> 00:26:24,720 Speaker 1: who were starting with a lot of knee pain, to 466 00:26:24,920 --> 00:26:28,840 Speaker 1: get gradually better over time. Yeah, so a Colatta rights 467 00:26:28,920 --> 00:26:31,960 Speaker 1: quote are the right As symptoms tend to surge and subside, 468 00:26:32,240 --> 00:26:35,040 Speaker 1: and people tend to seek out treatments when the pain 469 00:26:35,160 --> 00:26:37,880 Speaker 1: is at its peak, when it declines, as it would 470 00:26:37,920 --> 00:26:42,040 Speaker 1: have anyway, they ascribed the improvement to the treatment. Uh. 471 00:26:42,119 --> 00:26:44,280 Speaker 1: So you know this would this would roughly equate to 472 00:26:44,400 --> 00:26:46,720 Speaker 1: yelling at your knee when it's in pain, and it 473 00:26:46,800 --> 00:26:49,879 Speaker 1: really make it certainly relates to many other health scenarios 474 00:26:49,880 --> 00:26:53,320 Speaker 1: as well various medications and even things like prayer and 475 00:26:53,600 --> 00:26:58,640 Speaker 1: you know, supernatural um treatments and attempts to to deal 476 00:26:58,680 --> 00:27:00,960 Speaker 1: with pain, et cetera. Yeah, I mean it could apply 477 00:27:01,040 --> 00:27:05,360 Speaker 1: to any intervention that is aimed at influencing something that 478 00:27:05,520 --> 00:27:09,280 Speaker 1: is naturally variable on its own, right. Yeah, and you 479 00:27:09,320 --> 00:27:11,640 Speaker 1: know something that's again any kind of system in which 480 00:27:11,720 --> 00:27:14,840 Speaker 1: change occurs when fluctuation occurs. Uh, you know, you can 481 00:27:14,880 --> 00:27:18,040 Speaker 1: you can see this applying to not only physical pain, 482 00:27:18,160 --> 00:27:22,639 Speaker 1: but also uh, emotional distress, things of that nature, you know. 483 00:27:22,760 --> 00:27:25,399 Speaker 1: So again, I think this is an important tool to 484 00:27:25,480 --> 00:27:34,359 Speaker 1: have in our our logic tool kit. Now there are 485 00:27:34,400 --> 00:27:37,840 Speaker 1: even cases where I'm tempted to think about the application 486 00:27:38,080 --> 00:27:41,960 Speaker 1: of regression to the mean, but but where it's probably 487 00:27:41,960 --> 00:27:45,240 Speaker 1: a lot harder to quantify exactly what the effects are. 488 00:27:45,840 --> 00:27:49,640 Speaker 1: It's cases where it can be difficult to separate out, 489 00:27:49,720 --> 00:27:53,280 Speaker 1: say the effects of some kind of deterministic influence like 490 00:27:53,400 --> 00:27:56,720 Speaker 1: skill versus how how strong the effect of chance or 491 00:27:56,800 --> 00:27:59,080 Speaker 1: luck is. But I think about things even in the 492 00:27:59,119 --> 00:28:01,840 Speaker 1: world of the like I think about, you know, the 493 00:28:01,920 --> 00:28:05,239 Speaker 1: sophomore album by by a band that has like a 494 00:28:05,240 --> 00:28:09,280 Speaker 1: really stellar debut album. Uh, you know, often that is 495 00:28:09,280 --> 00:28:13,440 Speaker 1: perceived is disappointing, and you have to wonder, like, Okay, 496 00:28:13,560 --> 00:28:17,119 Speaker 1: is it is that often true? Because I don't know 497 00:28:17,200 --> 00:28:19,240 Speaker 1: if people get famous and it goes to their heads 498 00:28:19,280 --> 00:28:21,320 Speaker 1: and then they you know, they get full of themselves 499 00:28:21,320 --> 00:28:23,919 Speaker 1: and make something dumb, or is it because when somebody 500 00:28:23,920 --> 00:28:27,720 Speaker 1: has a debut album that's really well received to some extent, 501 00:28:27,920 --> 00:28:31,840 Speaker 1: it's so good partially because of luck or chance, and 502 00:28:31,880 --> 00:28:35,440 Speaker 1: that's an outlier that you're as you're starting sample yeah, yeah, 503 00:28:35,480 --> 00:28:37,640 Speaker 1: And certainly this is an area that's there's a lot 504 00:28:37,680 --> 00:28:40,560 Speaker 1: more subjectivity here and and so it's not the kind 505 00:28:40,560 --> 00:28:43,320 Speaker 1: of thing you can necessarily have a control group for anything. 506 00:28:44,000 --> 00:28:46,000 Speaker 1: But but I think it is quite interesting. And I 507 00:28:46,040 --> 00:28:48,520 Speaker 1: did find as I was looking around for some jazzy 508 00:28:48,640 --> 00:28:51,800 Speaker 1: or examples or possible examples of aggression to the mean, um, 509 00:28:51,920 --> 00:28:56,080 Speaker 1: I found one that that actually gets into a little 510 00:28:56,080 --> 00:28:58,360 Speaker 1: bit into the idea of you know, the first and 511 00:28:58,480 --> 00:29:02,000 Speaker 1: second album. But also uh, the idea of follow up 512 00:29:02,000 --> 00:29:06,240 Speaker 1: films and Hollywood sequels has pointed out both good Yeah. 513 00:29:06,480 --> 00:29:09,920 Speaker 1: Has pointed out by Joanna Deong in two thou eighteen 514 00:29:09,920 --> 00:29:14,320 Speaker 1: on the blogs scientifically sound movie sequels are potentially a 515 00:29:14,320 --> 00:29:18,120 Speaker 1: great example of aggression to the mean. Quote, Hollywood sequels 516 00:29:18,280 --> 00:29:20,920 Speaker 1: are only made if the original film is a quote 517 00:29:21,000 --> 00:29:25,080 Speaker 1: unquote high quality success. But the average quality of sequels 518 00:29:25,080 --> 00:29:28,160 Speaker 1: will be closer to the mean than average quality of 519 00:29:28,200 --> 00:29:31,560 Speaker 1: originals of sequels because of regression to the means, So 520 00:29:31,600 --> 00:29:34,760 Speaker 1: sequels tend to be of lower quality than the original. 521 00:29:35,040 --> 00:29:38,440 Speaker 1: Now I might somewhat dispute the premise here that Hollywood 522 00:29:38,480 --> 00:29:41,880 Speaker 1: sequels are only made to films that are high quality 523 00:29:41,920 --> 00:29:46,400 Speaker 1: to begin with. Um, But but I still think this 524 00:29:46,480 --> 00:29:49,600 Speaker 1: is onto something because there is a movie that gets 525 00:29:49,640 --> 00:29:53,160 Speaker 1: a sequel tends to have something about it, something that 526 00:29:53,200 --> 00:29:56,000 Speaker 1: people are responding to, whether it's a movie that I 527 00:29:56,080 --> 00:29:58,800 Speaker 1: would like or not. Right, I mean, so sometimes obviously 528 00:29:58,840 --> 00:30:01,200 Speaker 1: the situation is the film just made a lot of mine. 529 00:30:01,200 --> 00:30:03,160 Speaker 1: I mean, I guess that's the key thing. It didn't 530 00:30:03,200 --> 00:30:05,880 Speaker 1: make a lot of money. If so, producers are going 531 00:30:05,920 --> 00:30:08,040 Speaker 1: to be more inclined to say, let's do that again, 532 00:30:08,160 --> 00:30:11,560 Speaker 1: Let's have that experience again of all that money coming in. 533 00:30:12,000 --> 00:30:16,960 Speaker 1: And sometimes this this certainly matches up with a quality film. 534 00:30:17,040 --> 00:30:19,920 Speaker 1: You have something that really captures people's imagination and it 535 00:30:20,000 --> 00:30:22,640 Speaker 1: is of high quality and uh and you know, so 536 00:30:22,680 --> 00:30:26,280 Speaker 1: it's really firing on all cylinders. But you know, and yes, 537 00:30:26,360 --> 00:30:29,320 Speaker 1: certainly in some cases it's just the right film at 538 00:30:29,320 --> 00:30:31,760 Speaker 1: the right time. Or or maybe it has nothing to 539 00:30:31,800 --> 00:30:33,840 Speaker 1: do with the film itself. Maybe it's who's in it, 540 00:30:34,000 --> 00:30:36,000 Speaker 1: or I don't know what's going on in the zeitgeist 541 00:30:36,320 --> 00:30:39,000 Speaker 1: during that particular era. Well, the way I would think 542 00:30:39,000 --> 00:30:42,000 Speaker 1: about this is, and I think again, this is onto something. 543 00:30:42,040 --> 00:30:46,680 Speaker 1: It highlights that when we experience confusion where we say, like, wow, 544 00:30:46,760 --> 00:30:49,520 Speaker 1: you know, the Exorcist is such a great horror movie 545 00:30:49,520 --> 00:30:52,640 Speaker 1: and The Exorcist Too is so bad? How could that 546 00:30:52,680 --> 00:30:54,680 Speaker 1: be the case? You know, why is it? Why is 547 00:30:55,000 --> 00:30:58,160 Speaker 1: such a bad sequel to such a great movie? It's 548 00:30:58,200 --> 00:31:01,480 Speaker 1: because of the compare a son of the original to 549 00:31:01,560 --> 00:31:05,640 Speaker 1: the sequel that we're experiencing this confusion. Another way you 550 00:31:05,680 --> 00:31:08,640 Speaker 1: could just look at it is most horror movies are 551 00:31:08,720 --> 00:31:13,120 Speaker 1: direc most movies are bad, and it is only by 552 00:31:13,200 --> 00:31:17,360 Speaker 1: comparing the The Exorcist Too to The Exorcist that you 553 00:31:17,520 --> 00:31:20,480 Speaker 1: notice this steep drop off. Where another way of looking 554 00:31:20,480 --> 00:31:23,560 Speaker 1: at it is that The Exorcist Too is bad like 555 00:31:23,840 --> 00:31:27,040 Speaker 1: most horror movies are, and the first one was an outlier. 556 00:31:27,160 --> 00:31:29,320 Speaker 1: At the beginning, it was a first film in a 557 00:31:29,400 --> 00:31:34,600 Speaker 1: series that happened to be really good A cut above. Yeah, absolutely, like, yeah, 558 00:31:34,640 --> 00:31:36,080 Speaker 1: I think this is the correct way to look at it, 559 00:31:36,120 --> 00:31:38,840 Speaker 1: and also keeping in mind that just how amazing it 560 00:31:38,880 --> 00:31:41,960 Speaker 1: is that any film gets completed, like even a bad film, 561 00:31:42,040 --> 00:31:45,120 Speaker 1: Like a lot of people probably worked pretty hard to 562 00:31:45,240 --> 00:31:48,000 Speaker 1: make that happen, even if the end results don't really 563 00:31:48,000 --> 00:31:50,320 Speaker 1: please anyone at all. But but yeah, I think this 564 00:31:50,440 --> 00:31:53,360 Speaker 1: is also an interesting inversion of the opening example of 565 00:31:53,440 --> 00:31:56,320 Speaker 1: yelling at pilots as well, because most of the time, 566 00:31:56,560 --> 00:31:59,800 Speaker 1: if a flawed movie comes out, people are not clamoring 567 00:31:59,840 --> 00:32:05,080 Speaker 1: for the sequel. Um Sequels are rarely guaranteed, so you're 568 00:32:05,080 --> 00:32:07,400 Speaker 1: not often going to hear things like, oh, well, that 569 00:32:07,480 --> 00:32:09,800 Speaker 1: wasn't great. I hope the next one is an improvement. 570 00:32:09,840 --> 00:32:12,400 Speaker 1: I mean some people say that, some people I've said 571 00:32:12,400 --> 00:32:14,000 Speaker 1: things like that before, where it will be like, oh, 572 00:32:14,320 --> 00:32:16,680 Speaker 1: really flawed film, but maybe there's like a cool idea 573 00:32:16,920 --> 00:32:19,040 Speaker 1: I kind of wish it would they would remake it, 574 00:32:19,280 --> 00:32:22,560 Speaker 1: even though there's no like logical reason that there would 575 00:32:22,560 --> 00:32:26,160 Speaker 1: be like a there would be money behind that idea. Well, 576 00:32:26,200 --> 00:32:28,240 Speaker 1: I guess it's kind of different when you're talking about 577 00:32:28,240 --> 00:32:31,000 Speaker 1: a one off creative project versus something. I mean, we 578 00:32:31,040 --> 00:32:33,560 Speaker 1: live in a kind of different era now because we 579 00:32:33,760 --> 00:32:36,600 Speaker 1: were at the height of this you know, cinematic universe 580 00:32:36,680 --> 00:32:40,960 Speaker 1: thing with a huge number of the big budget movies 581 00:32:41,000 --> 00:32:44,320 Speaker 1: that come out. The big event movies are not one 582 00:32:44,360 --> 00:32:48,440 Speaker 1: off creative products, but they are a product that exists 583 00:32:48,520 --> 00:32:51,880 Speaker 1: within some kind of franchise or universe or something. So 584 00:32:51,920 --> 00:32:54,720 Speaker 1: you just know automatically that there's gonna be another one, 585 00:32:54,720 --> 00:32:57,280 Speaker 1: whether this one is good or not. Yeah, like either 586 00:32:57,320 --> 00:33:00,320 Speaker 1: it's an established film universe where like you know, they 587 00:33:00,320 --> 00:33:03,560 Speaker 1: put out another Marvel movie and it's just terrible, Well, 588 00:33:03,600 --> 00:33:06,080 Speaker 1: obviously there's enough momentum. They're not going to stop. They're 589 00:33:06,080 --> 00:33:08,360 Speaker 1: not gonna be like, oh, well, less and learned, Well 590 00:33:08,400 --> 00:33:11,440 Speaker 1: we'll stop then. No, No, there's gonna be another. Another 591 00:33:11,480 --> 00:33:14,800 Speaker 1: example of this might be a successful franchise in another medium, 592 00:33:14,880 --> 00:33:18,400 Speaker 1: say a book series, so like the Harry Potter books 593 00:33:18,440 --> 00:33:20,800 Speaker 1: for example, or I don't know, Lord of the Rings, 594 00:33:21,000 --> 00:33:23,400 Speaker 1: where you know that once they make the Fellowship of 595 00:33:23,440 --> 00:33:25,720 Speaker 1: the Rings, there's going to be a follow up. They're 596 00:33:25,720 --> 00:33:29,600 Speaker 1: gonna do another one. So in these ways, unless it's 597 00:33:29,640 --> 00:33:32,920 Speaker 1: the seventies and it's uh, Lord of the Rings movie 598 00:33:32,920 --> 00:33:36,080 Speaker 1: that that ends with Helm's Deep. Well, but they picked 599 00:33:36,080 --> 00:33:41,000 Speaker 1: that up eventually. But kay, but but yeah, probably the 600 00:33:41,000 --> 00:33:43,080 Speaker 1: Harry Potter films are a better example. And there may 601 00:33:43,080 --> 00:33:46,160 Speaker 1: be spe specific you know, things about how that wasn't 602 00:33:46,200 --> 00:33:49,520 Speaker 1: guaranteed either. Uh, you know, the economic reality can always 603 00:33:49,520 --> 00:33:51,840 Speaker 1: come into play. But for the most part, like those 604 00:33:51,880 --> 00:33:54,280 Speaker 1: were when when that started, you knew they were gonna 605 00:33:54,320 --> 00:33:56,040 Speaker 1: keep making these at least they were going to make 606 00:33:56,040 --> 00:33:58,520 Speaker 1: a follow up, so you could have comments like, well 607 00:33:59,120 --> 00:34:00,680 Speaker 1: there were that was just on of flawed in some 608 00:34:00,720 --> 00:34:02,720 Speaker 1: of the some of its execution. I hope that they 609 00:34:02,760 --> 00:34:05,400 Speaker 1: fix that in the next film. For the most part, yeah, 610 00:34:05,440 --> 00:34:07,680 Speaker 1: with one offs, this is not the case. It's like, 611 00:34:07,760 --> 00:34:11,440 Speaker 1: if if this film fizzles, then only you know a 612 00:34:11,480 --> 00:34:14,600 Speaker 1: few like rare people are going to be clamoring for 613 00:34:14,640 --> 00:34:17,480 Speaker 1: a sequel or dreaming about what the sequel would be. Yeah, 614 00:34:17,520 --> 00:34:19,640 Speaker 1: I think this observation, but regression to the mean and 615 00:34:19,680 --> 00:34:23,479 Speaker 1: movie sequels is actually very on point, but more so 616 00:34:23,640 --> 00:34:26,080 Speaker 1: for the films of yester Year, where the more the 617 00:34:26,120 --> 00:34:29,080 Speaker 1: more common thing was you'd have a an independent sort 618 00:34:29,080 --> 00:34:31,759 Speaker 1: of creative product that it's its own thing, and then 619 00:34:31,920 --> 00:34:34,680 Speaker 1: if it resonated with somebody, if it did well, there 620 00:34:34,680 --> 00:34:36,960 Speaker 1: would be sequels. I think it's a little it applies 621 00:34:37,000 --> 00:34:39,359 Speaker 1: a little bit less today when there's just you know, 622 00:34:39,560 --> 00:34:43,560 Speaker 1: we're in the world of franchises and extended universes and 623 00:34:43,600 --> 00:34:47,480 Speaker 1: there's just sort of like a guaranteed, ongoing uh conveyor 624 00:34:47,520 --> 00:34:50,279 Speaker 1: belt of of new stuff within the Marvel world or 625 00:34:50,320 --> 00:34:52,839 Speaker 1: the Star Wars world or whatever. Yeah, but I think 626 00:34:52,840 --> 00:34:56,320 Speaker 1: it it is a worthwhile way to think about creative 627 00:34:56,920 --> 00:35:00,200 Speaker 1: the creative process, and you know, as opposed to some 628 00:35:00,239 --> 00:35:02,439 Speaker 1: of these alternate sort of folk wisdomy ways of thinking 629 00:35:02,480 --> 00:35:05,440 Speaker 1: about it. For example, on Weird House Cinema, we recently 630 00:35:05,440 --> 00:35:07,920 Speaker 1: talked about Toby Hooper. Toby Hooper is one of those 631 00:35:07,920 --> 00:35:10,920 Speaker 1: directors who's often you'll often you'll see descriptions. I think 632 00:35:10,920 --> 00:35:12,960 Speaker 1: we've even read part of a review where they they 633 00:35:13,000 --> 00:35:14,759 Speaker 1: really they talk about, oh, well, you know he put 634 00:35:14,760 --> 00:35:19,160 Speaker 1: out Texas Chainsaw Masacre directed that film and this was great. 635 00:35:19,200 --> 00:35:22,800 Speaker 1: It was, you know, just a real lightning bolt um 636 00:35:23,160 --> 00:35:26,600 Speaker 1: to the cinematic world into horror itself as a genre. 637 00:35:27,000 --> 00:35:29,439 Speaker 1: And then the idea that well he was never able 638 00:35:29,480 --> 00:35:32,040 Speaker 1: to capture that magic again. You know that his his 639 00:35:32,120 --> 00:35:35,080 Speaker 1: career was just like one long slide after that, which 640 00:35:35,120 --> 00:35:38,120 Speaker 1: I don't think is a fair assessment, especially if you 641 00:35:38,560 --> 00:35:43,080 Speaker 1: employ regression to the mean, you know, the idea being that, yeah, 642 00:35:43,160 --> 00:35:44,960 Speaker 1: he did kind of get lightning in a bottle with that, 643 00:35:45,040 --> 00:35:48,359 Speaker 1: with that first big film, that that he was able 644 00:35:48,440 --> 00:35:53,080 Speaker 1: to to really bring something together that is an outlier, um, 645 00:35:53,120 --> 00:35:55,840 Speaker 1: but that that that's just going to happen. That's just 646 00:35:55,880 --> 00:35:59,000 Speaker 1: the way these things work, right, So most movies aren't 647 00:35:59,040 --> 00:36:01,600 Speaker 1: that good, So you of the random chance of like 648 00:36:01,840 --> 00:36:04,560 Speaker 1: how how good his ideas and execution are from one 649 00:36:04,640 --> 00:36:06,879 Speaker 1: year to the next is going to set in and 650 00:36:06,960 --> 00:36:09,759 Speaker 1: you might have a different idea about his career if 651 00:36:09,760 --> 00:36:14,359 Speaker 1: you were to say, like randomly chronologically reorder all his movies, right, 652 00:36:14,440 --> 00:36:15,879 Speaker 1: you know, like if you were to put the worst 653 00:36:15,960 --> 00:36:19,640 Speaker 1: ones earlier on or something, people might feel differently about it. Yeah, 654 00:36:19,640 --> 00:36:22,360 Speaker 1: well then they would talk about, well, okay, TCM was 655 00:36:22,480 --> 00:36:26,000 Speaker 1: peak Toby Hooper, like this was his peak output. Because 656 00:36:26,040 --> 00:36:28,760 Speaker 1: this is the kind of the kind of view of 657 00:36:28,880 --> 00:36:31,759 Speaker 1: an artist's you know, creative trajectory that we tend to 658 00:36:32,280 --> 00:36:35,799 Speaker 1: want to um to follow along, you know, because it's 659 00:36:35,800 --> 00:36:38,960 Speaker 1: more story shaped, the idea of assent and then eventually 660 00:36:39,040 --> 00:36:41,799 Speaker 1: decent that there's gonna be uh, there's gonna be a 661 00:36:41,840 --> 00:36:44,520 Speaker 1: period of high noon in their creative out output, and 662 00:36:44,560 --> 00:36:47,440 Speaker 1: sometimes that does match up with the reality. But I 663 00:36:47,480 --> 00:36:49,880 Speaker 1: don't know. Even then, we I think we tend to 664 00:36:49,960 --> 00:36:53,239 Speaker 1: overlook the dogs in the filmographies of people we love, 665 00:36:53,320 --> 00:36:55,960 Speaker 1: you know. Oh yeah, uh, But then again, I mean, 666 00:36:56,160 --> 00:36:59,480 Speaker 1: this is interesting because in talking about regression to the 667 00:36:59,480 --> 00:37:03,839 Speaker 1: mean applying to creative products like movies, we are acknowledging 668 00:37:04,040 --> 00:37:07,799 Speaker 1: that the creative process is not purely a product of 669 00:37:07,880 --> 00:37:10,920 Speaker 1: talent and skill, that there is a significant amount of 670 00:37:11,080 --> 00:37:14,040 Speaker 1: chance and luck involved in something like how good a 671 00:37:14,080 --> 00:37:16,880 Speaker 1: movie turns out to be? Um, And it's hard to 672 00:37:16,880 --> 00:37:20,080 Speaker 1: know exactly how to like how to picture that influence 673 00:37:20,080 --> 00:37:22,600 Speaker 1: of chance and luck, you know, like, what what is 674 00:37:22,680 --> 00:37:26,319 Speaker 1: that in the creative process? It's obviously true because there 675 00:37:26,360 --> 00:37:29,440 Speaker 1: are people who can be incredibly skilled in one instance 676 00:37:29,480 --> 00:37:31,759 Speaker 1: and then I don't know, things just don't go right 677 00:37:31,800 --> 00:37:34,120 Speaker 1: the next time, and to make something that nobody really likes. 678 00:37:34,280 --> 00:37:37,760 Speaker 1: But uh, but that's that's just not often how people 679 00:37:37,800 --> 00:37:40,120 Speaker 1: like to think about creative talents, and people like to 680 00:37:40,120 --> 00:37:42,799 Speaker 1: think about the creative process like it is much more 681 00:37:42,920 --> 00:37:47,480 Speaker 1: strictly deterministic. Yeah yeah, or or you look at things 682 00:37:47,520 --> 00:37:50,040 Speaker 1: like the Star Wars films, and you kind of like 683 00:37:50,080 --> 00:37:51,920 Speaker 1: fall into this idea of thinking this is stuff that 684 00:37:52,080 --> 00:37:55,880 Speaker 1: is mind out of the mythic earth, and you know, 685 00:37:55,920 --> 00:37:58,279 Speaker 1: it just makes sense that things would accumulate and get better. 686 00:37:58,400 --> 00:38:01,480 Speaker 1: So um, but really looking back on it, especially if 687 00:38:01,480 --> 00:38:04,080 Speaker 1: you actually like watch documentaries, and there's some great ones 688 00:38:04,120 --> 00:38:08,640 Speaker 1: about the production of those films, like it's it's amazing 689 00:38:08,680 --> 00:38:10,920 Speaker 1: that Star Wars, the first one in New Hope was 690 00:38:10,960 --> 00:38:13,760 Speaker 1: as good as it was, and then it's nothing short 691 00:38:13,800 --> 00:38:16,400 Speaker 1: of I mean, it's it's just a pure miracle that 692 00:38:16,480 --> 00:38:19,760 Speaker 1: the second one was so much better and like really 693 00:38:19,840 --> 00:38:23,040 Speaker 1: nailed it. Like if if the second film had had floundered, 694 00:38:24,040 --> 00:38:28,319 Speaker 1: I mean, just imagine how different the cinematical landscape would 695 00:38:28,320 --> 00:38:31,200 Speaker 1: have been for decades to come. Yeah, So it's it's 696 00:38:31,239 --> 00:38:34,440 Speaker 1: amazing if the first film in a series is good, 697 00:38:34,640 --> 00:38:37,200 Speaker 1: and it's super amazing if the second one is good. 698 00:38:37,520 --> 00:38:39,440 Speaker 1: And and this is why I think we often find 699 00:38:39,440 --> 00:38:42,520 Speaker 1: too that if if part one in part two of 700 00:38:42,640 --> 00:38:45,440 Speaker 1: something are of high quality, then you've got to look 701 00:38:45,440 --> 00:38:47,960 Speaker 1: out for that part three because that part three, that 702 00:38:48,040 --> 00:38:49,960 Speaker 1: part three may be coming to get you. But likewise, 703 00:38:50,160 --> 00:38:54,719 Speaker 1: if a part two is rubbish, um, you know, subjectively, 704 00:38:55,120 --> 00:38:57,680 Speaker 1: then then part three might pick it up and uh 705 00:38:57,760 --> 00:38:59,759 Speaker 1: and get things back on track. So you certainly see 706 00:38:59,800 --> 00:39:02,200 Speaker 1: that that kind of fluctuation as well. I have a 707 00:39:02,280 --> 00:39:04,400 Speaker 1: question I actually don't know the answer to, but this 708 00:39:04,440 --> 00:39:08,920 Speaker 1: would be interesting in terms of I don't know the 709 00:39:09,080 --> 00:39:12,040 Speaker 1: high performing output, whether that is in whether that is 710 00:39:12,160 --> 00:39:15,720 Speaker 1: a creative endeavor like you know, writing books or creating movies, 711 00:39:15,840 --> 00:39:19,200 Speaker 1: or whether that's something even like athletics, like athletic performance, 712 00:39:19,800 --> 00:39:22,960 Speaker 1: do you expect to see more random fluctuation in the 713 00:39:23,040 --> 00:39:29,640 Speaker 1: performance of collaborative output versus individual output? So say, um, 714 00:39:29,800 --> 00:39:33,239 Speaker 1: do you expect more influence of random chance and fluctuation 715 00:39:33,280 --> 00:39:36,600 Speaker 1: in the quality of uh books written by a single 716 00:39:36,719 --> 00:39:39,400 Speaker 1: author versus you know, movies that have the input of 717 00:39:39,480 --> 00:39:43,440 Speaker 1: hundreds of thousands of people? Uh? Or in in the 718 00:39:43,480 --> 00:39:46,120 Speaker 1: realm of say sports, like do you expect more random 719 00:39:46,200 --> 00:39:50,160 Speaker 1: variation in the output of an individual athletes like you know, 720 00:39:50,160 --> 00:39:55,120 Speaker 1: an individual gymnast or something, or in team sports? Yeah? 721 00:39:55,280 --> 00:39:57,640 Speaker 1: I could see it going both ways, because yeah, if 722 00:39:57,680 --> 00:39:59,520 Speaker 1: you think too hard to about even just like the 723 00:39:59,520 --> 00:40:02,239 Speaker 1: film and aology, you can easily get into discussions of 724 00:40:02,280 --> 00:40:04,040 Speaker 1: like okay, well is it the same cast and crew 725 00:40:04,600 --> 00:40:07,440 Speaker 1: that are producing the sequel. Uh, you know, what happens 726 00:40:07,440 --> 00:40:09,279 Speaker 1: when the budget is different, what happens when there are 727 00:40:09,280 --> 00:40:11,440 Speaker 1: other constraints, what happens when suddenly there are a whole 728 00:40:11,440 --> 00:40:14,719 Speaker 1: bunch of producers that have their ideas about what things 729 00:40:14,719 --> 00:40:16,400 Speaker 1: should be. I mean, there's so many different factors to 730 00:40:16,440 --> 00:40:18,880 Speaker 1: take into place. Uh. You know, with this example that 731 00:40:19,160 --> 00:40:22,759 Speaker 1: you know, perhaps doesn't bear too close of scrutiny, but 732 00:40:22,760 --> 00:40:24,920 Speaker 1: but but it's but it's still I think serves as 733 00:40:24,960 --> 00:40:28,520 Speaker 1: a nice um illustration of the overall trend that we're 734 00:40:28,520 --> 00:40:30,600 Speaker 1: talking about here. Well, it does bring up the fact 735 00:40:30,600 --> 00:40:33,040 Speaker 1: that since I mentioned athletes that you know, I don't 736 00:40:33,040 --> 00:40:35,160 Speaker 1: know a lot about sports. I'm not a big sports fan. 737 00:40:35,239 --> 00:40:37,640 Speaker 1: But but clearly, but regression to the mean is something 738 00:40:37,680 --> 00:40:41,080 Speaker 1: that has widely been applied to the world of sports. Uh. 739 00:40:41,080 --> 00:40:44,880 Speaker 1: For example, in the observation that often after having a 740 00:40:44,920 --> 00:40:48,400 Speaker 1: really stellar season, either an individual athlete or a sports 741 00:40:48,400 --> 00:40:54,080 Speaker 1: team will be perceived to underperform the next season. And again, 742 00:40:54,120 --> 00:40:56,320 Speaker 1: that very well could have something to do with regression 743 00:40:56,320 --> 00:40:58,680 Speaker 1: to the mean. Like, you know, the fact that they're 744 00:40:58,760 --> 00:41:02,040 Speaker 1: observed having in a using season is actually an outlier. 745 00:41:02,520 --> 00:41:06,160 Speaker 1: You're starting your expectations then and saying like, Okay, now 746 00:41:06,160 --> 00:41:08,760 Speaker 1: they're going to be the best forever. Just by random 747 00:41:08,800 --> 00:41:11,480 Speaker 1: fluctuation over time, you would expect their next season to 748 00:41:11,520 --> 00:41:14,799 Speaker 1: probably be not as good as the first. I wonder 749 00:41:14,840 --> 00:41:17,319 Speaker 1: to what an extent this can be applied to, say, 750 00:41:17,360 --> 00:41:20,239 Speaker 1: the world of the culinary arts, or even just like 751 00:41:20,640 --> 00:41:24,160 Speaker 1: various food crops, like say the selecting a cantalope at 752 00:41:24,160 --> 00:41:26,560 Speaker 1: the grocery store, that sort of thing. I mean, I 753 00:41:26,600 --> 00:41:28,560 Speaker 1: guess it would apply to pretty much anything where you're 754 00:41:28,560 --> 00:41:32,200 Speaker 1: sampling in a series over time, there's plenty of random 755 00:41:32,239 --> 00:41:36,520 Speaker 1: fluctuation in what you're sampling, and the first thing you 756 00:41:36,560 --> 00:41:39,040 Speaker 1: sample is an outlier in some way really good or 757 00:41:39,080 --> 00:41:42,440 Speaker 1: really bad. If those things hold true, then you can 758 00:41:42,480 --> 00:41:45,600 Speaker 1: probably expect you're going to see some regression one way 759 00:41:45,680 --> 00:41:48,360 Speaker 1: or the other. Yeah. Yeah. By the way, I was 760 00:41:48,400 --> 00:41:51,640 Speaker 1: looking around for like really stellar examples of a sequel 761 00:41:51,800 --> 00:41:56,440 Speaker 1: film that is widely believed to be uh rubbish, and 762 00:41:56,480 --> 00:41:59,279 Speaker 1: I think The Exorcist Too is the primary example. Like 763 00:41:59,480 --> 00:42:01,520 Speaker 1: you get into some of the other examples that pop up, 764 00:42:01,600 --> 00:42:05,320 Speaker 1: I feel like there's room for disagreement. Um. For instance, 765 00:42:05,440 --> 00:42:08,080 Speaker 1: Texas Chainsaw Masacre two is one which I saw popping 766 00:42:08,160 --> 00:42:11,080 Speaker 1: up on some of these lists for disappointing sequels. But 767 00:42:11,480 --> 00:42:13,360 Speaker 1: I think that's entirely based on who you ask. I 768 00:42:13,360 --> 00:42:16,319 Speaker 1: think if you ask us, we will agree that that 769 00:42:16,440 --> 00:42:19,640 Speaker 1: that t c M Two is is actually a great film. 770 00:42:19,840 --> 00:42:21,719 Speaker 1: It's different from the first one, perhaps if you go 771 00:42:21,760 --> 00:42:24,520 Speaker 1: into if you go into part two with the expectations 772 00:42:24,560 --> 00:42:26,839 Speaker 1: you had for part one, you may see it as 773 00:42:26,840 --> 00:42:30,160 Speaker 1: a dip in quality. But depending on what else you're 774 00:42:30,160 --> 00:42:31,879 Speaker 1: bringing to the table, you might see it as an 775 00:42:31,880 --> 00:42:34,400 Speaker 1: increase in in quality, or at least or something that 776 00:42:34,440 --> 00:42:37,080 Speaker 1: maybe is different but on par with the original. I mean, 777 00:42:37,120 --> 00:42:38,880 Speaker 1: it's certainly not for everybody. I mean, it is a 778 00:42:39,000 --> 00:42:41,839 Speaker 1: It is a gross, disgusting film in in a way 779 00:42:41,880 --> 00:42:44,400 Speaker 1: like the first one, probably even grosser, but also a 780 00:42:44,680 --> 00:42:48,919 Speaker 1: sort of satirical masterpiece. Um but I just had another 781 00:42:48,960 --> 00:42:51,320 Speaker 1: thought when you said that The Exorcist Too is regarded 782 00:42:51,320 --> 00:42:53,440 Speaker 1: as like one of the best examples of a sequel. 783 00:42:53,520 --> 00:42:55,759 Speaker 1: That's really rubbish. I mean, it makes me also wonder 784 00:42:55,840 --> 00:42:59,560 Speaker 1: about the pretty high estimation critics generally have of the 785 00:42:59,600 --> 00:43:02,839 Speaker 1: exer Is three. Makes me wonder if the effect of 786 00:43:02,880 --> 00:43:06,919 Speaker 1: The Exorcist to being so bad actually makes people sort 787 00:43:06,960 --> 00:43:09,640 Speaker 1: of over. You know, they're like they're ready to be 788 00:43:09,680 --> 00:43:12,920 Speaker 1: impressed by the Exorcist three. Yeah. Yeah, I wonder if 789 00:43:12,920 --> 00:43:14,960 Speaker 1: that's the case too with it with especially when you 790 00:43:15,000 --> 00:43:17,720 Speaker 1: have a situation with the part three coming back and 791 00:43:17,920 --> 00:43:22,040 Speaker 1: restoring uh some I don't know, some level of quality 792 00:43:22,080 --> 00:43:24,399 Speaker 1: to a franchise. I mean there's also like the Star 793 00:43:24,440 --> 00:43:27,799 Speaker 1: Trek example, right, I mean that was long the Long 794 00:43:27,840 --> 00:43:29,480 Speaker 1: held up as an example of like, okay, you have 795 00:43:29,520 --> 00:43:32,560 Speaker 1: you even Star Treks and your odd Star Treks, right, uh. 796 00:43:32,560 --> 00:43:35,400 Speaker 1: And I think you've made a similar case for the 797 00:43:36,320 --> 00:43:38,839 Speaker 1: Faster and Furious movies, right, I mean, once you get 798 00:43:38,880 --> 00:43:40,600 Speaker 1: to a certain point in the series, I think it's 799 00:43:40,680 --> 00:43:44,160 Speaker 1: pretty much all uh, you know, a nitrous boosted brain. 800 00:43:44,239 --> 00:43:47,320 Speaker 1: It's it gets you know, it's all like we're driving 801 00:43:47,400 --> 00:43:51,000 Speaker 1: cars in space now and flying and all that. But um, 802 00:43:51,080 --> 00:43:53,760 Speaker 1: but for the earlier ones, yeah, I'd say the odd 803 00:43:53,800 --> 00:43:57,000 Speaker 1: ones are better. Like, uh, three is the first one 804 00:43:57,040 --> 00:44:01,319 Speaker 1: where it really starts getting ludicrously weird. Four is kind 805 00:44:01,320 --> 00:44:04,759 Speaker 1: of a uh and then five starts. Five is when 806 00:44:04,760 --> 00:44:07,480 Speaker 1: the rock shows up, and then but by seven year 807 00:44:07,520 --> 00:44:11,080 Speaker 1: golden all right, well we're gonna go ahead and close 808 00:44:11,160 --> 00:44:12,799 Speaker 1: this one out here. But we'd obviously love to hear 809 00:44:12,840 --> 00:44:17,440 Speaker 1: from everyone about this about regression towards the mean, just 810 00:44:17,560 --> 00:44:22,640 Speaker 1: in our daily lives, in various scientific studies. Perhaps you 811 00:44:22,680 --> 00:44:25,040 Speaker 1: have thoughts about how this applies to something we've discussed 812 00:44:25,040 --> 00:44:27,319 Speaker 1: on the show in the past, because I know we've 813 00:44:27,600 --> 00:44:32,080 Speaker 1: we've mentioned regression to the mean in passing before, but 814 00:44:32,120 --> 00:44:34,759 Speaker 1: certainly we've never taken the opportunity to really dive into 815 00:44:34,800 --> 00:44:36,960 Speaker 1: it and explain it like we did today. Yeah, I 816 00:44:36,960 --> 00:44:39,680 Speaker 1: know it's come up in passing, just in us making 817 00:44:39,719 --> 00:44:42,359 Speaker 1: comments here and there about like the importance of of 818 00:44:42,440 --> 00:44:45,760 Speaker 1: randomized trials and control groups and all that. In the meantime, 819 00:44:45,800 --> 00:44:47,600 Speaker 1: if you would like to listen to other episodes of 820 00:44:47,600 --> 00:44:50,280 Speaker 1: Stuff to Blow Your Mind, you will find them wherever 821 00:44:50,360 --> 00:44:52,680 Speaker 1: you get your podcast. Just look for the Stuff to 822 00:44:52,680 --> 00:44:55,720 Speaker 1: Blow your Mind podcast feed. We have our core episodes 823 00:44:55,760 --> 00:44:59,839 Speaker 1: on Tuesdays and Thursdays, Artifact episodes on Wednesday, listener mail 824 00:44:59,880 --> 00:45:02,279 Speaker 1: on Monday's. On Fridays, we do a little bit of 825 00:45:02,320 --> 00:45:04,279 Speaker 1: a weird house cinema. That's our times. We just talk 826 00:45:04,320 --> 00:45:07,880 Speaker 1: about some sort of a strange film. Uh, and you know, 827 00:45:08,040 --> 00:45:11,840 Speaker 1: tease apart what makes it strange? Uh, let's see what 828 00:45:12,040 --> 00:45:13,560 Speaker 1: else so yeah, have you go to stuff to Blow 829 00:45:13,600 --> 00:45:15,279 Speaker 1: your Mind dot com that will send you to the 830 00:45:15,280 --> 00:45:19,000 Speaker 1: I Heart listing for our show, and there's a button 831 00:45:19,000 --> 00:45:20,600 Speaker 1: there for a store if you want to click on 832 00:45:20,640 --> 00:45:23,319 Speaker 1: that wealth then you can buy some merchandise that has 833 00:45:23,320 --> 00:45:26,160 Speaker 1: Stuff to Blow your Mind logos and whatnot on it, 834 00:45:26,320 --> 00:45:29,200 Speaker 1: or perhaps weird House cinema logos and whatnot on it, 835 00:45:29,239 --> 00:45:31,439 Speaker 1: and you can get mugs, t shirts, all that kind 836 00:45:31,440 --> 00:45:35,880 Speaker 1: of stuff. Uh, just for fun, you know, or but 837 00:45:36,080 --> 00:45:37,719 Speaker 1: if you want to be all business about it, you 838 00:45:37,760 --> 00:45:40,000 Speaker 1: can make it work to get our merch Now, now 839 00:45:40,239 --> 00:45:42,399 Speaker 1: this this is just for fun, Like, don't use these 840 00:45:42,719 --> 00:45:45,839 Speaker 1: these clothing items to clothe your nakedness. These are just 841 00:45:45,960 --> 00:45:49,239 Speaker 1: for fun. This is just purely extra clothing. Okay, this 842 00:45:49,280 --> 00:45:52,200 Speaker 1: should not be your core clothing. Huge thanks as always 843 00:45:52,200 --> 00:45:55,360 Speaker 1: to our excellent audio producer Seth Nicholas Johnson. If you 844 00:45:55,360 --> 00:45:57,480 Speaker 1: would like to get in touch with us with feedback 845 00:45:57,520 --> 00:45:59,680 Speaker 1: on this episode or any other to suggest a topic 846 00:45:59,719 --> 00:46:01,800 Speaker 1: for the future, just to say hello, you can email 847 00:46:01,880 --> 00:46:12,520 Speaker 1: us at contact at stuff to Blow your Mind dot com. 848 00:46:12,560 --> 00:46:15,040 Speaker 1: Stuff to Blow Your Mind is production of I Heart Radio. 849 00:46:15,400 --> 00:46:17,520 Speaker 1: For more podcasts for My Heart Radio, visit the I 850 00:46:17,560 --> 00:46:20,400 Speaker 1: Heart Radio app, Apple podcasts, or wherever you listening to 851 00:46:20,440 --> 00:46:32,360 Speaker 1: your favorite shows.