1 00:00:15,370 --> 00:00:24,210 Speaker 1: Pushkin. Hi, Tim Harford here with two pieces of news 2 00:00:24,490 --> 00:00:28,690 Speaker 1: and a brand new mini cautionary tale for you. News 3 00:00:28,690 --> 00:00:32,210 Speaker 1: headline number one. We're back in the studio recording a 4 00:00:32,330 --> 00:00:37,890 Speaker 1: new Monster season of fourteen episodes of Cautionary Tales. There 5 00:00:37,930 --> 00:00:43,890 Speaker 1: will be wonderful actors, hilarious calamities and tragic catastrophes Cautionary 6 00:00:43,890 --> 00:00:50,250 Speaker 1: Tales Season two. Stay tuned. News headline number two. I 7 00:00:50,330 --> 00:00:53,170 Speaker 1: have a brand new book coming. If you're listening in 8 00:00:53,210 --> 00:00:56,250 Speaker 1: the US or Canada, it's out early next year and 9 00:00:56,330 --> 00:01:00,210 Speaker 1: it's called The Data Detective. It's never too early to 10 00:01:00,250 --> 00:01:03,610 Speaker 1: pre order, so please consider doing that. But if you're 11 00:01:03,650 --> 00:01:06,890 Speaker 1: listening anywhere else in the world, the book is out 12 00:01:07,210 --> 00:01:10,610 Speaker 1: right now and it's called How to Make the World 13 00:01:10,690 --> 00:01:13,930 Speaker 1: Add Up. On the surface, it's a book about statistics, 14 00:01:14,210 --> 00:01:17,250 Speaker 1: but really it's a book about how to think clearly 15 00:01:17,370 --> 00:01:19,650 Speaker 1: about the world. And if you want to know what 16 00:01:19,650 --> 00:01:24,890 Speaker 1: the book is really about, gather close. I'm Tim Harford 17 00:01:25,450 --> 00:01:30,610 Speaker 1: and this is a special mini episode of Cautionary Tales. 18 00:01:37,970 --> 00:01:40,370 Speaker 1: You can make an estimate of how many children have 19 00:01:40,450 --> 00:01:43,450 Speaker 1: been born into a Dutch or Danish family by counting 20 00:01:43,450 --> 00:01:46,250 Speaker 1: the storks nests on the roof of their house. In 21 00:01:46,330 --> 00:01:50,130 Speaker 1: statistical terminology, it would be said that a positive correlation 22 00:01:50,210 --> 00:01:56,090 Speaker 1: has been found to exist between these two things. These 23 00:01:56,130 --> 00:01:59,490 Speaker 1: are the words of Daryl Huff, the author of the 24 00:01:59,570 --> 00:02:04,210 Speaker 1: most popular book about statistics ever written, How to Lie 25 00:02:04,410 --> 00:02:08,690 Speaker 1: with Statistics. But of course, storks do not actually deliver 26 00:02:08,810 --> 00:02:11,890 Speaker 1: baby is, no matter how strong the correlation may be. 27 00:02:12,770 --> 00:02:17,210 Speaker 1: Half explains, big houses attract big and potentially big families, 28 00:02:17,450 --> 00:02:20,690 Speaker 1: and big houses have more chimney pots on which storks 29 00:02:20,730 --> 00:02:25,970 Speaker 1: may nest. Of course, sometimes it seems you can prove 30 00:02:26,450 --> 00:02:31,530 Speaker 1: anything with statistics, especially things that aren't true. Darryl Huff's 31 00:02:31,530 --> 00:02:36,250 Speaker 1: little book is a master class in skepticism. He exposes mistakes, 32 00:02:36,410 --> 00:02:42,090 Speaker 1: half truths, and outright lies. Why, he explains, the crooks 33 00:02:42,090 --> 00:02:45,330 Speaker 1: already know these tricks, honest men must learn them in 34 00:02:45,370 --> 00:02:50,010 Speaker 1: self defense. Daryl Hulf made statistics seem like a game, 35 00:02:50,570 --> 00:02:54,290 Speaker 1: or perhaps a stage magician's trick, all could fun, but 36 00:02:54,570 --> 00:02:58,410 Speaker 1: never to be trusted. And I worry that we're starting 37 00:02:58,450 --> 00:03:02,330 Speaker 1: to trust nobody. We're starting to believe that lying with 38 00:03:02,450 --> 00:03:07,210 Speaker 1: statistics is all anyone ever does. But skepticism can easily 39 00:03:07,290 --> 00:03:13,930 Speaker 1: curdle into cynicism, and cynicism can be corrosive after all. 40 00:03:14,170 --> 00:03:17,210 Speaker 1: In nineteen fifty four, the very same year that Half 41 00:03:17,290 --> 00:03:22,250 Speaker 1: published How To Lie With Statistics, two researchers, Richard Doll 42 00:03:22,330 --> 00:03:26,810 Speaker 1: and Austin Bradford Hill, produced one of the first convincing 43 00:03:26,890 --> 00:03:32,290 Speaker 1: demonstrations that smoking cigarettes caused lung cancer. They couldn't have 44 00:03:32,330 --> 00:03:36,610 Speaker 1: figured that out without statistics. Even to notice the dramatic 45 00:03:36,650 --> 00:03:39,650 Speaker 1: increase in lung cancer rates in the mid twentieth century 46 00:03:40,050 --> 00:03:44,890 Speaker 1: required a statistical view of the world. So what Darrell 47 00:03:44,970 --> 00:03:48,810 Speaker 1: Half saw as a trick, Richard Doll and Austin Bradford 48 00:03:48,890 --> 00:03:52,410 Speaker 1: Hill saw as a tool, and a tool that has 49 00:03:52,490 --> 00:03:58,290 Speaker 1: saved many millions of lives. Two visions of statistics had 50 00:03:58,330 --> 00:04:01,850 Speaker 1: emerged at the same time, and they very quickly came 51 00:04:01,890 --> 00:04:07,250 Speaker 1: into conflict, because as the evidence on smoking and cancer emerged, 52 00:04:07,810 --> 00:04:13,130 Speaker 1: the tobacco companies fought back. When their product was so dangerous, 53 00:04:13,370 --> 00:04:18,010 Speaker 1: so demonstrably dangerous, how on earth could they persuade people 54 00:04:18,050 --> 00:04:24,010 Speaker 1: to keep smoking? In a stroke of twisted genius, they 55 00:04:24,050 --> 00:04:28,970 Speaker 1: realized the answer scientific evidence was the enemy, so they 56 00:04:29,010 --> 00:04:32,690 Speaker 1: had to give people reason to doubt that evidence without 57 00:04:32,810 --> 00:04:37,770 Speaker 1: attacking it head on. They chose to emphasize uncertainty, the 58 00:04:37,810 --> 00:04:41,650 Speaker 1: fact that experts disagree and to call for more research. 59 00:04:41,770 --> 00:04:48,690 Speaker 1: Before anyone did anything hasty, they manufactured doubt, and it 60 00:04:48,730 --> 00:04:52,330 Speaker 1: turns out that doubt is a really easy product to make. 61 00:04:53,970 --> 00:04:57,850 Speaker 1: A couple of decades ago, two psychologists carry Edwards and 62 00:04:58,050 --> 00:05:02,290 Speaker 1: Edward Smith, conducted an experiment to test how people felt 63 00:05:02,290 --> 00:05:07,530 Speaker 1: about all sorts of politically fraught questions, including abortion rights, 64 00:05:07,890 --> 00:05:12,530 Speaker 1: smacking children, and the death penalty for under sixteens. They 65 00:05:12,610 --> 00:05:15,690 Speaker 1: invited people to produce as many arguments as they could 66 00:05:15,850 --> 00:05:20,330 Speaker 1: in favor of or against the conclusion, given three minutes 67 00:05:20,410 --> 00:05:24,490 Speaker 1: per topic. Of course, people found it easier to produce 68 00:05:24,610 --> 00:05:27,890 Speaker 1: arguments that backed up what they already believed. Lots of 69 00:05:27,930 --> 00:05:32,050 Speaker 1: researchers have found that, but more strikingly, Edwards and Smith 70 00:05:32,130 --> 00:05:36,770 Speaker 1: found that disbelief seemed to flow more fluidly than belief. 71 00:05:37,530 --> 00:05:41,770 Speaker 1: It's much easier to argue against positions you dislike than 72 00:05:41,890 --> 00:05:46,370 Speaker 1: in favor of positions you support. Doubt has a special 73 00:05:46,450 --> 00:05:49,410 Speaker 1: kind of power, and it is the easiest thing in 74 00:05:49,410 --> 00:05:52,170 Speaker 1: the world to suggest that you should doubt what the 75 00:05:52,210 --> 00:05:56,050 Speaker 1: statistics are telling you. This is why the simplest move 76 00:05:56,090 --> 00:06:00,450 Speaker 1: in any argument today is to encourage people to believe 77 00:06:00,970 --> 00:06:08,650 Speaker 1: nothing doubt is a powerful weapon. Now, statistics are a 78 00:06:08,770 --> 00:06:13,010 Speaker 1: natural rural target, but honest statistics need to be defended 79 00:06:13,410 --> 00:06:17,410 Speaker 1: because while it's easy to lie with statistics, it's even 80 00:06:17,410 --> 00:06:20,450 Speaker 1: easier to lie without them. And there are plenty of 81 00:06:20,530 --> 00:06:23,130 Speaker 1: things in this world, from the link between smoking and 82 00:06:23,210 --> 00:06:27,090 Speaker 1: cancer to how to deal with the coronavirus epidemic, that 83 00:06:27,170 --> 00:06:31,370 Speaker 1: we simply can't begin to understand without statistics. In a 84 00:06:31,450 --> 00:06:37,290 Speaker 1: complex world, they are an essential tool. In the spring 85 00:06:37,290 --> 00:06:41,570 Speaker 1: of nineteen sixty five, a US Senate committee was pondering 86 00:06:41,610 --> 00:06:44,490 Speaker 1: the life or death matter of whether to put a 87 00:06:44,530 --> 00:06:49,410 Speaker 1: health warning on packets of cigarettes. An expert witness appeared, 88 00:06:49,810 --> 00:06:52,850 Speaker 1: and he wanted to throw doubt on the scientific evidence, 89 00:06:53,730 --> 00:06:58,370 Speaker 1: so he turned to the topic of storks and babies. 90 00:06:58,970 --> 00:07:02,170 Speaker 1: It turns out he said that households with more children 91 00:07:02,530 --> 00:07:06,290 Speaker 1: also have more storks on their roofs. That old story 92 00:07:06,330 --> 00:07:10,490 Speaker 1: about babies being delivered by storks was true, the expert 93 00:07:10,530 --> 00:07:15,330 Speaker 1: went on, of course it wasn't. Correlation is not causation, 94 00:07:16,290 --> 00:07:20,890 Speaker 1: and similarly, just because smoking was correlated with lung cancer 95 00:07:21,450 --> 00:07:26,490 Speaker 1: didn't mean, not for a moment, that smoking caused cancer. 96 00:07:27,730 --> 00:07:30,970 Speaker 1: Do you honestly think there is as casual a relationship 97 00:07:31,010 --> 00:07:35,450 Speaker 1: between statistics linking smoking with disease as there is about storks, 98 00:07:35,890 --> 00:07:40,170 Speaker 1: asked the committee chair. The expert witness replied that the 99 00:07:40,250 --> 00:07:46,050 Speaker 1: two seemed to me the same. The witness's name was 100 00:07:46,130 --> 00:07:49,690 Speaker 1: Darryl Huff. He had been paid by the tobacco lobby 101 00:07:49,690 --> 00:07:53,410 Speaker 1: to do what he did best, weave together witty examples, 102 00:07:53,610 --> 00:07:57,370 Speaker 1: some statistical savvy, and a certain amount of cynicism to 103 00:07:57,530 --> 00:08:01,650 Speaker 1: cast doubt on the idea that cigarettes were dangerous. He 104 00:08:01,690 --> 00:08:04,570 Speaker 1: was even working on a sequel to his masterpiece, although 105 00:08:04,650 --> 00:08:08,610 Speaker 1: it was never published. The sequel's name was How to 106 00:08:08,770 --> 00:08:16,210 Speaker 1: Lie with Smoking Statistics. I understand that statistics can seem intimidating, 107 00:08:16,490 --> 00:08:20,490 Speaker 1: and they are often used deceitfully, but we mustn't give 108 00:08:20,570 --> 00:08:24,810 Speaker 1: way to cynicism or feel helpless under the deluge of numbers. 109 00:08:25,410 --> 00:08:30,690 Speaker 1: Understanding statistical claims doesn't require a deep technical background. If 110 00:08:30,690 --> 00:08:34,410 Speaker 1: you're able to think, then you're able to think sensibly 111 00:08:34,490 --> 00:08:39,530 Speaker 1: about statistical claims, which is why I've written a book 112 00:08:39,570 --> 00:08:42,770 Speaker 1: about how to think more clearly about numbers and about 113 00:08:42,770 --> 00:08:45,370 Speaker 1: the world. If you're listening in the US or Canada, 114 00:08:45,530 --> 00:08:49,130 Speaker 1: that book is called The Data Detective, and it's out 115 00:08:49,130 --> 00:08:52,090 Speaker 1: early next year, and if you're listening anywhere else. The 116 00:08:52,090 --> 00:08:54,610 Speaker 1: book is called How to Make the World Add Up, 117 00:08:55,290 --> 00:08:57,530 Speaker 1: and it's out now as a book, an e book, 118 00:08:57,690 --> 00:09:01,410 Speaker 1: and as an audiobook read by me. If you like 119 00:09:01,490 --> 00:09:04,290 Speaker 1: what I do here on Cautionary Tales, I really think 120 00:09:04,290 --> 00:09:07,730 Speaker 1: you're going to love the book. The world is an 121 00:09:07,730 --> 00:09:11,930 Speaker 1: amazing place, and these days, whatever you're interested in, the 122 00:09:12,050 --> 00:09:15,290 Speaker 1: chances are you'll understand it much better through the lens 123 00:09:15,450 --> 00:09:19,890 Speaker 1: of statistics. Don't be cynical, don't assume it's all a 124 00:09:19,970 --> 00:09:23,370 Speaker 1: lie or a trick. Don't be afraid to pick up 125 00:09:23,370 --> 00:09:34,130 Speaker 1: this statistical telescope and gaze around. This mini episode of 126 00:09:34,210 --> 00:09:38,210 Speaker 1: Cautionary Tales was written and read by me Tim Harford. 127 00:09:38,650 --> 00:09:41,610 Speaker 1: The producer was Ryan Dilley, and the music and sound 128 00:09:41,650 --> 00:09:47,650 Speaker 1: design by Pascal Wise. Cautionary Tales is a Pushkin Industry's production. 129 00:10:09,570 --> 00:10:11,530 Speaker 1: I don't don't be