1
00:00:15,370 --> 00:00:31,570
Speaker 1: Pushkin. On the twenty first of August nineteen seventy four,

2
00:00:32,010 --> 00:00:35,450
Speaker 1: Elaine Oswald made a visit to the doctor's office in

3
00:00:35,490 --> 00:00:38,450
Speaker 1: a small town not far from Manchester in the north

4
00:00:38,450 --> 00:00:42,130
Speaker 1: of England. Oswald had a slight pain in her side,

5
00:00:42,410 --> 00:00:44,530
Speaker 1: but she was hoping to go into work later that day.

6
00:00:45,450 --> 00:00:48,970
Speaker 1: Oswald had never met this particular doctor before, and he

7
00:00:49,090 --> 00:00:51,330
Speaker 1: was only a few years older than she was. She

8
00:00:51,410 --> 00:00:54,730
Speaker 1: was only twenty five. He had spectacles and the kind

9
00:00:54,770 --> 00:00:57,530
Speaker 1: of big brown beard that was fashionable at the time.

10
00:00:58,130 --> 00:01:01,970
Speaker 1: The doctor couldn't have been friendlier or more accommodating. He

11
00:01:02,050 --> 00:01:04,290
Speaker 1: sat beside her, not over on the other side of

12
00:01:04,330 --> 00:01:06,850
Speaker 1: a big desk, and he told her she might have

13
00:01:06,970 --> 00:01:11,530
Speaker 1: kidney stones. He prescribed some strong painkillers, then suggested she

14
00:01:11,570 --> 00:01:15,010
Speaker 1: go home and rest. Leave your door unlocked. He said,

15
00:01:15,450 --> 00:01:18,090
Speaker 1: I'll come round after my morning clinic has finished and

16
00:01:18,170 --> 00:01:21,890
Speaker 1: do a blood test. Later that day he stopped by.

17
00:01:22,610 --> 00:01:25,130
Speaker 1: His wife and son were in the car outside. He said,

18
00:01:25,690 --> 00:01:27,970
Speaker 1: just a quick jab with a needle to draw the

19
00:01:27,970 --> 00:01:31,850
Speaker 1: blood and he'd be on his way. The needle slid

20
00:01:32,050 --> 00:01:37,650
Speaker 1: into her arm. The next thing Elaine Oswald remembers was

21
00:01:37,690 --> 00:01:40,490
Speaker 1: waking up on the floor with a doctor and two

22
00:01:40,530 --> 00:01:44,490
Speaker 1: paramedics trying to revive her. She was rushed to hospital,

23
00:01:44,970 --> 00:01:47,850
Speaker 1: where the staff, she recalled, treated her like the scum

24
00:01:47,890 --> 00:01:51,450
Speaker 1: of the earth. They assumed she'd overdosed on those painkillers.

25
00:01:52,210 --> 00:01:55,090
Speaker 1: The young doctor was much kinder. She must have had

26
00:01:55,130 --> 00:01:58,570
Speaker 1: an allergic reaction, he said, Thank goodness he'd been there

27
00:01:58,570 --> 00:02:01,570
Speaker 1: to administer the kiss of life. He promised to write

28
00:02:01,650 --> 00:02:05,210
Speaker 1: up the case for a medical journal. She was grateful,

29
00:02:05,250 --> 00:02:09,770
Speaker 1: of course, who wouldn't be. When she was discharged from hospital,

30
00:02:10,050 --> 00:02:13,130
Speaker 1: the kind doctor even invited her and her husband for dinner.

31
00:02:13,730 --> 00:02:16,410
Speaker 1: It was a pleasant evening. He gave her a medic

32
00:02:16,450 --> 00:02:20,890
Speaker 1: alert bracelet so that no future clinician would accidentally prescribe

33
00:02:20,930 --> 00:02:25,170
Speaker 1: similar drugs. She'd go on to have two children, toughing

34
00:02:25,210 --> 00:02:28,090
Speaker 1: it out through the agonies of labor without pain relief.

35
00:02:31,050 --> 00:02:35,450
Speaker 1: Elaine Oswald eventually moved to America, became a professor of English,

36
00:02:36,090 --> 00:02:39,330
Speaker 1: and for the next twenty five years assumed that the

37
00:02:39,450 --> 00:02:44,090
Speaker 1: kind doctor, a man called Harold Shipman, had saved her life.

38
00:02:45,250 --> 00:02:50,010
Speaker 1: She couldn't have been more wrong. I'm Tim Harford, and

39
00:02:50,170 --> 00:03:31,730
Speaker 1: you're listening to cautionary tales. It took twenty five years

40
00:03:31,770 --> 00:03:34,730
Speaker 1: for the world to realize that Harold Chipman was not

41
00:03:34,970 --> 00:03:38,290
Speaker 1: the kindly doctor who believed in his patients while others

42
00:03:38,330 --> 00:03:41,810
Speaker 1: treated them badly. Not the bold life saver who'd leap

43
00:03:41,890 --> 00:03:45,090
Speaker 1: to administer the kiss of life, Not the caring man

44
00:03:45,170 --> 00:03:47,890
Speaker 1: who'd go the extra mile to visit his patience at

45
00:03:47,930 --> 00:03:52,530
Speaker 1: home rather than drag them into the clinic. No, Harold

46
00:03:52,530 --> 00:03:57,810
Speaker 1: Shipman was a murderer, and not just any murderer. In

47
00:03:57,970 --> 00:04:03,250
Speaker 1: sheer numerical terms, he was the worst serial killer in history.

48
00:04:04,770 --> 00:04:09,410
Speaker 1: There are plenty of notorious killers, the charming Ted Bundy,

49
00:04:10,050 --> 00:04:14,090
Speaker 1: John Wayne Gaycy, the Killer Clown, the son of Sam

50
00:04:14,330 --> 00:04:19,570
Speaker 1: David Berkowitz. But while Bundy, Gaycey, and Berkowitz between them

51
00:04:19,730 --> 00:04:24,650
Speaker 1: killed more than seventy people, Shipman alone killed more than

52
00:04:24,690 --> 00:04:29,490
Speaker 1: two hundred. Often his victims would be people living by themselves,

53
00:04:30,050 --> 00:04:34,330
Speaker 1: elderly but in perfectly fine health. Shipman would come round,

54
00:04:34,810 --> 00:04:38,090
Speaker 1: inject them with an overdose of morphine, just as he

55
00:04:38,130 --> 00:04:41,770
Speaker 1: had injected young Elaine Oswald, and then sign a death

56
00:04:41,770 --> 00:04:47,170
Speaker 1: certificate saying that they had died of old age. When

57
00:04:47,210 --> 00:04:50,770
Speaker 1: I began researching this cautionary tale, I knew that Shipman

58
00:04:50,970 --> 00:04:55,330
Speaker 1: was a terrible man. I didn't understand quite how terrible

59
00:04:56,170 --> 00:04:59,050
Speaker 1: I vaguely had in mind a doctor who started down

60
00:04:59,090 --> 00:05:01,810
Speaker 1: this path by easing the deaths of patients who were

61
00:05:01,850 --> 00:05:05,450
Speaker 1: in pain and terminally ill, and then got carried away.

62
00:05:06,450 --> 00:05:10,610
Speaker 1: But the truth is so much more horrible than Shipman

63
00:05:10,650 --> 00:05:15,490
Speaker 1: would kill healthy people, then to explain their deaths, he'd

64
00:05:15,490 --> 00:05:19,170
Speaker 1: say they'd been drug addicts. He'd retrospectively tamper with their

65
00:05:19,210 --> 00:05:24,930
Speaker 1: medical records and leave the bereaved families bewildered. More than once,

66
00:05:25,210 --> 00:05:28,810
Speaker 1: he killed someone in his clinic, then would claim they'd

67
00:05:28,890 --> 00:05:32,570
Speaker 1: died of heart failure. He may have killed a severely

68
00:05:32,610 --> 00:05:36,970
Speaker 1: disabled four year old girl. When one middle aged man,

69
00:05:37,290 --> 00:05:43,170
Speaker 1: Jim King, was misdiagnosed with cancer, Shipman intercepted the letter

70
00:05:43,210 --> 00:05:45,370
Speaker 1: from the hospital with a good news that he was

71
00:05:45,450 --> 00:05:50,530
Speaker 1: cancer free after all. Shipman supplied Jim with morphine, got

72
00:05:50,610 --> 00:05:54,010
Speaker 1: him hooked, and watched as Jim lost his job and

73
00:05:54,210 --> 00:05:59,210
Speaker 1: pawned his possessions. Meanwhile, Shipman skimmed off some of that

74
00:05:59,290 --> 00:06:03,330
Speaker 1: morphine and used it to murder Jim King's own father.

75
00:06:04,770 --> 00:06:08,890
Speaker 1: Why did he do it? Plenty of people have speculated,

76
00:06:09,610 --> 00:06:14,450
Speaker 1: but nobody knows. One doctor, an expert witness at Shipman's

77
00:06:14,530 --> 00:06:18,410
Speaker 1: trial mused that while some doctors would relax from the

78
00:06:18,450 --> 00:06:21,450
Speaker 1: stresses of the profession by playing a round of golf,

79
00:06:22,250 --> 00:06:28,290
Speaker 1: Shipman seemed to relax by murdering his patience. Shipman, who

80
00:06:28,490 --> 00:06:32,610
Speaker 1: killed himself in prison, never offered an admission of guilt,

81
00:06:33,170 --> 00:06:37,530
Speaker 1: let alone an explanation. But this cautionary tale isn't going

82
00:06:37,570 --> 00:06:42,530
Speaker 1: to pick apart Harold Shipman's psychology. No, this tale, like

83
00:06:42,650 --> 00:06:47,170
Speaker 1: all our tales, is about the lessons we can learn here.

84
00:06:47,570 --> 00:06:50,090
Speaker 1: The lesson is that Harold Shipman could have been caught

85
00:06:50,370 --> 00:06:53,850
Speaker 1: much earlier. Maybe not as early as nineteen seventy four

86
00:06:53,930 --> 00:06:57,770
Speaker 1: when he injected the young Elaine Oswald with morphine, perhaps

87
00:06:57,810 --> 00:07:01,130
Speaker 1: intending to kill her, and perhaps with some other wickedness

88
00:07:01,170 --> 00:07:04,170
Speaker 1: in mind, but he could have been caught early enough

89
00:07:04,290 --> 00:07:11,050
Speaker 1: to have saved more than a hundred lives. In the

90
00:07:11,090 --> 00:07:15,010
Speaker 1: early hours of July the twenty ninth, nineteen seventy six,

91
00:07:15,570 --> 00:07:19,690
Speaker 1: in the Bronx, New York, Jody Valenti and her friend

92
00:07:19,810 --> 00:07:24,490
Speaker 1: Donna Lauria sat in their oldsmobile chatting. They were just

93
00:07:24,650 --> 00:07:29,490
Speaker 1: outside Donna's home, Donna's parents were inside. Both of them

94
00:07:29,530 --> 00:07:33,210
Speaker 1: were in medical training, Jody to be a nurse, Donna

95
00:07:33,490 --> 00:07:36,330
Speaker 1: to be a medic, but Donna would never get the

96
00:07:36,450 --> 00:07:40,890
Speaker 1: chance to finish her studies. Jody Valenti was a young

97
00:07:40,930 --> 00:07:45,210
Speaker 1: woman like Elaine Oswald, but her brush with death at

98
00:07:45,210 --> 00:07:48,650
Speaker 1: the hands of a serial killer was a complete contrast

99
00:07:48,730 --> 00:07:52,770
Speaker 1: with Elane's. As Jody and Donna were talking, a man

100
00:07:52,850 --> 00:07:56,650
Speaker 1: walked up, pulled out a pistol, and shot them both.

101
00:07:58,010 --> 00:08:02,690
Speaker 1: Donna Laurier died instantly. Jody Valenti took a bullet in

102
00:08:02,770 --> 00:08:08,090
Speaker 1: her leg and she survived. She was, of course traumatized,

103
00:08:08,410 --> 00:08:12,090
Speaker 1: and she obviously understood quite how close she had come

104
00:08:12,130 --> 00:08:16,370
Speaker 1: to death. Unlike Elaine Oswald, she didn't spend the next

105
00:08:16,530 --> 00:08:20,250
Speaker 1: twenty five years believing that her attacker was a hero

106
00:08:20,450 --> 00:08:24,570
Speaker 1: who had saved her life. The NYPD had a clearly

107
00:08:24,610 --> 00:08:29,330
Speaker 1: defined problem. Someone calling himself the Son of Sam was

108
00:08:29,370 --> 00:08:33,770
Speaker 1: wandering around New York City shooting young people, leaving some dead,

109
00:08:34,050 --> 00:08:37,810
Speaker 1: some disabled, and the whole community in a panic. He

110
00:08:37,890 --> 00:08:42,050
Speaker 1: had to be found, and eventually, in August the following year,

111
00:08:42,570 --> 00:08:50,410
Speaker 1: he was found. In the case of Harold Chipman, the

112
00:08:50,490 --> 00:08:55,450
Speaker 1: Greater Manchester Police faced a radically different problem. The problem,

113
00:08:55,530 --> 00:08:59,170
Speaker 1: in fact, was to realize that there was a problem,

114
00:08:59,970 --> 00:09:03,090
Speaker 1: because The police had no idea that people were being murdered.

115
00:09:03,890 --> 00:09:07,290
Speaker 1: People were dying, yes, but according to their doctor, they

116
00:09:07,290 --> 00:09:11,250
Speaker 1: were dying of natural causes. The deaths were a surprise

117
00:09:11,370 --> 00:09:14,490
Speaker 1: to friends and family. Most of the time, the victims

118
00:09:14,570 --> 00:09:18,450
Speaker 1: weren't seriously ill, just old and alone. In the morning,

119
00:09:18,530 --> 00:09:21,770
Speaker 1: there would be pottering around, catching the bus or dropping

120
00:09:21,810 --> 00:09:24,890
Speaker 1: in on a neighbor, in good shape and fine spirits.

121
00:09:25,770 --> 00:09:30,450
Speaker 1: And in the afternoon, dear kind doctor Shipman would come

122
00:09:30,490 --> 00:09:34,770
Speaker 1: by on a routine visit, and according to doctor Shipman,

123
00:09:35,650 --> 00:09:39,690
Speaker 1: well he'd find them dead, dead of old age. He

124
00:09:39,730 --> 00:09:43,210
Speaker 1: would often write on the death certificate, even though doctors

125
00:09:43,210 --> 00:09:47,410
Speaker 1: would normally be more specific. And while their friends and

126
00:09:47,490 --> 00:09:52,410
Speaker 1: relatives were shocked, they weren't shocked enough to call the police.

127
00:09:53,410 --> 00:09:56,330
Speaker 1: As far as the police were concerned, then there was

128
00:09:56,730 --> 00:10:05,170
Speaker 1: nothing to investigate. Sarah Marsland died on the seventh of

129
00:10:05,250 --> 00:10:09,210
Speaker 1: August nineteen seventy eight at her home in Hyde, a

130
00:10:09,250 --> 00:10:13,210
Speaker 1: small town near Manchester. Harold Shipman was a doctor. He

131
00:10:13,250 --> 00:10:18,010
Speaker 1: had moved to hide in nineteen seventy seven. By coincidence,

132
00:10:18,090 --> 00:10:20,330
Speaker 1: that was about the time my own family moved to

133
00:10:20,370 --> 00:10:24,090
Speaker 1: the area. I was four years old. I'm so glad

134
00:10:24,130 --> 00:10:27,450
Speaker 1: I never came any closer to his orbit. We can't

135
00:10:27,490 --> 00:10:30,530
Speaker 1: be absolutely sure that Sarah was one of Shipman's victims,

136
00:10:30,570 --> 00:10:34,130
Speaker 1: because nobody even suspected that a crime had been committed

137
00:10:34,410 --> 00:10:37,690
Speaker 1: until more than two decades after she had died. But

138
00:10:37,890 --> 00:10:42,810
Speaker 1: the circumstantial evidence is this. Although Sarah Marsland seems to

139
00:10:42,810 --> 00:10:46,570
Speaker 1: have had no particular health complaint, Harold Shipman came to

140
00:10:46,610 --> 00:10:51,890
Speaker 1: see her uninvited and unannounced. While he was there she died.

141
00:10:52,530 --> 00:10:55,770
Speaker 1: It wasn't unusual for Shipman to visit patients for no

142
00:10:55,890 --> 00:10:58,770
Speaker 1: particular reason. He did it all the time, and they

143
00:10:58,810 --> 00:11:02,690
Speaker 1: loved him for it. A good old fashioned, hard working doctor,

144
00:11:02,730 --> 00:11:06,130
Speaker 1: they said, someone with all the time in the world

145
00:11:06,170 --> 00:11:10,770
Speaker 1: for his patience. But even so, it is a striking

146
00:11:10,810 --> 00:11:14,090
Speaker 1: coincidence that some one would drop dead just when a

147
00:11:14,170 --> 00:11:16,730
Speaker 1: doctor happened to be popping in for a friendly visit.

148
00:11:18,090 --> 00:11:21,010
Speaker 1: One physician later testified that this was the sort of

149
00:11:21,050 --> 00:11:24,490
Speaker 1: coincidence there might be at once in a lifetime experience

150
00:11:24,530 --> 00:11:28,690
Speaker 1: for a family doctor, but for Harold Shipman, it seemed

151
00:11:28,690 --> 00:11:34,170
Speaker 1: to happen every few weeks. Nevertheless, nobody raised the alarm

152
00:11:34,250 --> 00:11:37,770
Speaker 1: about Sarah Marsland at the time, and why would they

153
00:11:37,850 --> 00:11:40,890
Speaker 1: raise the alarm. She was in her eighties and her

154
00:11:40,930 --> 00:11:44,650
Speaker 1: own doctor had declared that she had died of coronary thrombosis.

155
00:11:45,290 --> 00:11:50,650
Speaker 1: The situation didn't seem out of the ordinary. A few

156
00:11:50,770 --> 00:11:55,930
Speaker 1: years before Sarah Marsland's death, two psychologists, Daniel Carneman and

157
00:11:56,090 --> 00:12:01,050
Speaker 1: Amos Verski, began investigating patterns in the way we make judgments.

158
00:12:01,850 --> 00:12:05,570
Speaker 1: One pattern that they discovered helps to explain why nobody

159
00:12:05,650 --> 00:12:10,330
Speaker 1: suspected doctor Shipman for a very long time. That pattern

160
00:12:10,410 --> 00:12:15,050
Speaker 1: is known as the representativeness heuristic, a habit of mind

161
00:12:15,290 --> 00:12:20,090
Speaker 1: that leads us to sort a situation into strange or unremarkable,

162
00:12:20,410 --> 00:12:24,330
Speaker 1: depending not on the true likelihood, but whether it matches

163
00:12:24,370 --> 00:12:29,690
Speaker 1: our existing mental groupings. For an example of the representativeness heuristic,

164
00:12:30,050 --> 00:12:33,490
Speaker 1: consider the following description of a person. He's called Jeff.

165
00:12:34,050 --> 00:12:37,370
Speaker 1: Jeff is forty and very good looking. He works out,

166
00:12:37,690 --> 00:12:41,730
Speaker 1: practices yoga, and is a vegetarian. When he was a teenager,

167
00:12:41,970 --> 00:12:44,370
Speaker 1: Jeff was a movie buff, and he also took the

168
00:12:44,490 --> 00:12:48,050
Speaker 1: lead role in the school play. He's always been extroverted.

169
00:12:48,410 --> 00:12:52,050
Speaker 1: He's already gone through two divorces, and his current girlfriend

170
00:12:52,130 --> 00:12:55,890
Speaker 1: is fifteen years younger than him. Which of the following

171
00:12:55,930 --> 00:12:59,610
Speaker 1: do you think is more probable A. Jeff is now

172
00:12:59,610 --> 00:13:06,690
Speaker 1: a Hollywood movie star or B Jeff is now an accountant. Intuitively,

173
00:13:06,930 --> 00:13:10,650
Speaker 1: Jeff sounds like a movie star, but that's not right.

174
00:13:11,650 --> 00:13:14,970
Speaker 1: There are just a few dozen genuine movie stars in Hollywood,

175
00:13:15,330 --> 00:13:18,130
Speaker 1: while there are well over a million accountants in the

176
00:13:18,210 --> 00:13:22,410
Speaker 1: United States. Many thousands of them will, like Jeff, be

177
00:13:22,490 --> 00:13:26,770
Speaker 1: good looking, twice divorced vegetarians with a background in amateur dramatics.

178
00:13:27,490 --> 00:13:31,290
Speaker 1: The representativeness heuristic is a quick and easy way for

179
00:13:31,370 --> 00:13:34,530
Speaker 1: our subconscious mind to make decisions. We use it all

180
00:13:34,530 --> 00:13:38,570
Speaker 1: the time without knowing, and it often works, but it

181
00:13:38,650 --> 00:13:43,930
Speaker 1: can lead us astray. It led the community of Hide

182
00:13:43,930 --> 00:13:48,250
Speaker 1: astray too. In a subtly different way. Harold Chipman didn't

183
00:13:48,290 --> 00:13:51,450
Speaker 1: repeat his early mistake of drugging a twenty five year

184
00:13:51,490 --> 00:13:55,130
Speaker 1: old Elane Oswald. He began to target much older people,

185
00:13:55,530 --> 00:13:59,370
Speaker 1: people like the widow Sarah Marsland. Although Sarah was in

186
00:13:59,410 --> 00:14:03,330
Speaker 1: decent health, she fits the mental template of someone who

187
00:14:03,370 --> 00:14:06,810
Speaker 1: would die from natural causes. My point is not that

188
00:14:06,890 --> 00:14:10,490
Speaker 1: when an elderly person dies we should assume it was murder,

189
00:14:10,890 --> 00:14:14,090
Speaker 1: not even when the elderly person dies, just as her

190
00:14:14,170 --> 00:14:18,810
Speaker 1: doctor happens to call past unannounced. No. My point is

191
00:14:18,970 --> 00:14:23,130
Speaker 1: that when something fits neatly into our mental story, we

192
00:14:23,210 --> 00:14:26,490
Speaker 1: don't ask questions. We don't start to weigh up the

193
00:14:26,530 --> 00:14:31,810
Speaker 1: probabilities of murder versus natural causes. If we did, we'd

194
00:14:31,850 --> 00:14:35,130
Speaker 1: simply ask for an autopsy, wouldn't we But we don't,

195
00:14:36,050 --> 00:14:40,010
Speaker 1: and we don't because what we see fits naturally into

196
00:14:40,050 --> 00:14:44,170
Speaker 1: the story we expect. So the fundamental problem was not

197
00:14:44,330 --> 00:14:47,930
Speaker 1: only did people not realize that Shipman was a murderer,

198
00:14:48,370 --> 00:14:51,770
Speaker 1: they didn't even realize that there were any murders taking place.

199
00:14:52,570 --> 00:14:58,810
Speaker 1: The representativeness heuristic reassured them, nothing strange is happening. Move on,

200
00:14:59,450 --> 00:15:04,810
Speaker 1: there's nothing to see. Cautionary tales will be right back.

201
00:15:11,690 --> 00:15:16,890
Speaker 1: Shipmen murdered people with lethal doses of morphine, which left

202
00:15:16,970 --> 00:15:20,970
Speaker 1: no obvious trace unless there was an autopsy. But why

203
00:15:20,970 --> 00:15:24,650
Speaker 1: would there be one? The representativeness heuristic tells us nothing

204
00:15:24,690 --> 00:15:28,530
Speaker 1: stranger has happened. Shipman would sign the death certificate himself

205
00:15:28,650 --> 00:15:32,410
Speaker 1: to certify death from old age or heart failure. No

206
00:15:32,490 --> 00:15:36,410
Speaker 1: need to call the ambulance, he'd say, too late, no

207
00:15:36,570 --> 00:15:38,690
Speaker 1: need to call the police. He could deal with the

208
00:15:38,770 --> 00:15:45,050
Speaker 1: necessary paperwork himself again and again, Harold Shipman murdered people

209
00:15:45,330 --> 00:15:49,050
Speaker 1: in their own homes, and again and again. The friends

210
00:15:49,090 --> 00:15:52,130
Speaker 1: and the family of the victims did not realize that

211
00:15:52,210 --> 00:15:56,610
Speaker 1: a crime had been committed. Indeed, many people were grateful

212
00:15:56,650 --> 00:16:00,370
Speaker 1: to Shipmen, glad that in their final hours the patients

213
00:16:00,410 --> 00:16:03,130
Speaker 1: had had the close attention of the doctor they adored.

214
00:16:05,810 --> 00:16:09,690
Speaker 1: Not everyone felt that way. In nineteen ninety four, for example,

215
00:16:10,090 --> 00:16:13,770
Speaker 1: Alice Kitchen died suddenly at the age of seventy A

216
00:16:13,810 --> 00:16:16,530
Speaker 1: few hours after seeing her son and appearing to be

217
00:16:16,610 --> 00:16:19,890
Speaker 1: in good health. Doctor Shipman told her family that he

218
00:16:19,930 --> 00:16:22,330
Speaker 1: had called in to visit her, that she had clearly

219
00:16:22,330 --> 00:16:24,890
Speaker 1: suffered a stroke, but that she had refused to go

220
00:16:24,970 --> 00:16:28,170
Speaker 1: to hospital as he had suggested. It was a cruel

221
00:16:28,330 --> 00:16:33,650
Speaker 1: lie and an arrogant one. Alice Kitchen's family decided against

222
00:16:33,770 --> 00:16:37,530
Speaker 1: making a formal complaint, but they were angry. They thought

223
00:16:37,610 --> 00:16:42,690
Speaker 1: Shipman was guilty of negligence. In their book about Shipman's crimes,

224
00:16:43,170 --> 00:16:47,930
Speaker 1: Prescription for Murder, the journalists Brian Whittle and Jean Richie

225
00:16:48,330 --> 00:16:51,050
Speaker 1: muse on the nature of the murders and the people

226
00:16:51,090 --> 00:16:54,410
Speaker 1: who died. Their ages meant that the death would not

227
00:16:54,450 --> 00:16:58,810
Speaker 1: make any statistician raise an eyebrow seventy seventy four, sixty

228
00:16:58,850 --> 00:17:02,210
Speaker 1: nine eighty three, all within the range that death comes.

229
00:17:02,930 --> 00:17:06,890
Speaker 1: The deaths would not make any statistician raise an eyebrow.

230
00:17:07,730 --> 00:17:11,770
Speaker 1: It seems an uncontroversch or phrase. After all, elderly people

231
00:17:11,930 --> 00:17:15,130
Speaker 1: die all the time, don't they. But it's quite wrong.

232
00:17:15,810 --> 00:17:21,570
Speaker 1: The representativeness heuristic is soothing us into keeping our eyebrows unraised.

233
00:17:22,410 --> 00:17:28,850
Speaker 1: But statisticians don't use the representativeness heuristic. They use the data,

234
00:17:29,010 --> 00:17:32,650
Speaker 1: and any statistician given a look at the statistics behind

235
00:17:32,770 --> 00:17:37,370
Speaker 1: Harold Shipman's clinical practice would have raised more than an eyebrow.

236
00:17:38,370 --> 00:17:46,250
Speaker 1: They would have raised the alarm. Professor Sir David Spiegelhalter

237
00:17:46,530 --> 00:17:50,330
Speaker 1: is one of the UK's foremost statisticians. He's a brilliant

238
00:17:50,330 --> 00:17:53,850
Speaker 1: communicator of statistical ideas and the author of a great book,

239
00:17:54,130 --> 00:17:58,810
Speaker 1: The Art of Statistics. David was asked to provide advice

240
00:17:58,850 --> 00:18:02,410
Speaker 1: to the commissions set up after Shipman was jailed, invited

241
00:18:02,570 --> 00:18:07,130
Speaker 1: to answer the obvious question could Shipman have been stopped sooner?

242
00:18:07,850 --> 00:18:11,370
Speaker 1: And to David spiegel Alter and the other statisticians considering

243
00:18:11,370 --> 00:18:15,730
Speaker 1: the problem, that answer was, of course he could have

244
00:18:15,730 --> 00:18:19,050
Speaker 1: been stopped. All you had to do was look at

245
00:18:19,090 --> 00:18:24,010
Speaker 1: the numbers in the right way. Interrogating statistics to set

246
00:18:24,010 --> 00:18:27,210
Speaker 1: our alarm. Bells ringing was an idea developed by the

247
00:18:27,250 --> 00:18:31,370
Speaker 1: Allies during the Second World War. At Columbia University, New York,

248
00:18:31,650 --> 00:18:36,290
Speaker 1: the great Hungarian mathematician Abraham Vald was working on military

249
00:18:36,330 --> 00:18:41,610
Speaker 1: mathematics and he developed what he called sequential testing. Meanwhile,

250
00:18:41,930 --> 00:18:46,130
Speaker 1: the young mathematician named George Barnard was working in London

251
00:18:46,290 --> 00:18:50,770
Speaker 1: for the fabulously named Ministry of Supply. Because of the

252
00:18:50,810 --> 00:18:55,010
Speaker 1: wartime secrecy, Vald and Barnard weren't aware of each other's work,

253
00:18:55,410 --> 00:18:58,610
Speaker 1: but they were working on the same basic problem, which

254
00:18:58,650 --> 00:19:02,130
Speaker 1: is this. Let's say you have a process which produces

255
00:19:02,170 --> 00:19:06,330
Speaker 1: a random output. Say rolling a die, you are a one,

256
00:19:06,890 --> 00:19:15,530
Speaker 1: then a four, and another one, A two, one, five, three, one, six,

257
00:19:16,170 --> 00:19:19,530
Speaker 1: on you go. You can keep rolling as many times

258
00:19:19,570 --> 00:19:23,490
Speaker 1: as you like. So at what point do you say, Hey,

259
00:19:23,530 --> 00:19:28,010
Speaker 1: there's something strange about this dye. I'm rolling too many wands.

260
00:19:29,770 --> 00:19:32,330
Speaker 1: You can use the same idea to check products coming

261
00:19:32,370 --> 00:19:34,930
Speaker 1: off the production line. You don't want to stop the

262
00:19:34,930 --> 00:19:38,530
Speaker 1: conveyor belt just because of a single faulty product, but

263
00:19:38,810 --> 00:19:41,170
Speaker 1: neither do you want to keep the production line rolling

264
00:19:41,250 --> 00:19:45,690
Speaker 1: forever if there is a steady stream of problems. Rold

265
00:19:45,770 --> 00:19:49,210
Speaker 1: and Barnard would have been particularly focused on the manufacture

266
00:19:49,210 --> 00:19:52,730
Speaker 1: of ammunition and other armaments, but the maths can be

267
00:19:52,770 --> 00:19:56,490
Speaker 1: applied more widely. Sample some cookies to check whether they

268
00:19:56,490 --> 00:19:59,770
Speaker 1: have enough chocolate chips in them, or check the strength

269
00:19:59,770 --> 00:20:02,650
Speaker 1: of condoms by inflating them to see if they stand

270
00:20:02,690 --> 00:20:06,770
Speaker 1: up to the strain. Any product will have a failure rate.

271
00:20:07,530 --> 00:20:10,530
Speaker 1: But at what point do you say, hang on a minute,

272
00:20:11,410 --> 00:20:16,250
Speaker 1: something's wrong. David Spiegelholter and his colleagues told the Shipman

273
00:20:16,290 --> 00:20:20,610
Speaker 1: Inquiry that looking for suspicious patterns in medical records was

274
00:20:20,730 --> 00:20:24,770
Speaker 1: fundamentally similar to looking for suspicious patterns in dice rolls,

275
00:20:25,010 --> 00:20:28,370
Speaker 1: or cookies or condoms. You might want to make some

276
00:20:28,450 --> 00:20:31,850
Speaker 1: adjustments for the mix of cases. A doctor serving a

277
00:20:31,890 --> 00:20:34,570
Speaker 1: retirement community is going to have a very different case

278
00:20:34,610 --> 00:20:37,890
Speaker 1: mix from a doctor working on a military base, but

279
00:20:38,090 --> 00:20:42,090
Speaker 1: the principle is the same. Track deaths over time among

280
00:20:42,130 --> 00:20:45,610
Speaker 1: each doctor's patients, just as you might track faulty cookies

281
00:20:46,090 --> 00:20:50,930
Speaker 1: or faulty condoms. Spiegelholter and his colleagues concluded that the

282
00:20:51,010 --> 00:20:55,010
Speaker 1: kind of analysis developed by Vold and by Barnard could

283
00:20:55,010 --> 00:20:58,570
Speaker 1: have flagged Harold Shipman for close attention as early as

284
00:20:58,730 --> 00:21:03,850
Speaker 1: nineteen eighty four, fourteen years before he was eventually arrested.

285
00:21:04,330 --> 00:21:08,250
Speaker 1: More than a hundred murders could have been prevented, and

286
00:21:08,370 --> 00:21:12,570
Speaker 1: that's just statistical method. Other ways to slice the data

287
00:21:12,730 --> 00:21:16,130
Speaker 1: also raise questions. For example, there were a couple of

288
00:21:16,170 --> 00:21:19,850
Speaker 1: years in which Shipman went quiet, perhaps fearing that other

289
00:21:19,970 --> 00:21:23,050
Speaker 1: doctors in the clinic would notice what was happening. When

290
00:21:23,050 --> 00:21:25,650
Speaker 1: he left to set up shop as a lone practitioner

291
00:21:25,890 --> 00:21:30,370
Speaker 1: for the murders began again with hindsight. All this is

292
00:21:30,410 --> 00:21:33,930
Speaker 1: clear in the data. Even clearer is the fact that

293
00:21:33,970 --> 00:21:37,650
Speaker 1: so many of Shipman's patients died in the early afternoon,

294
00:21:38,250 --> 00:21:42,890
Speaker 1: a convenient time for Shipman's home visits. The pattern, says

295
00:21:42,930 --> 00:21:48,250
Speaker 1: Professor Spiegelhalter, requires no subtle statistical analysis. It is what

296
00:21:48,450 --> 00:21:54,490
Speaker 1: statisticians call interocular. Draw a graph and it hits you

297
00:21:54,570 --> 00:22:04,490
Speaker 1: between the eyes. Not all statistical anomalies result from foul play,

298
00:22:04,530 --> 00:22:07,930
Speaker 1: of course, David Spiegelhalter told me about one doctor who

299
00:22:08,010 --> 00:22:11,130
Speaker 1: had a truly extraordinary number of deaths on his watch,

300
00:22:11,530 --> 00:22:15,370
Speaker 1: even more than Harold Shipman. But there was an innocent explanation.

301
00:22:16,130 --> 00:22:19,930
Speaker 1: While Shipman's patients had often died suddenly, this doctor had

302
00:22:19,930 --> 00:22:23,410
Speaker 1: been treating terminally ill patients. He had gone to great

303
00:22:23,490 --> 00:22:26,210
Speaker 1: lengths to ensure they were able to die at home

304
00:22:26,850 --> 00:22:30,290
Speaker 1: rather than spending their final hours or days in hospital.

305
00:22:30,690 --> 00:22:33,810
Speaker 1: As a result, the doctor ended up signing a large

306
00:22:33,930 --> 00:22:38,490
Speaker 1: number of death certificates. But statistical analysis isn't designed to

307
00:22:38,570 --> 00:22:44,290
Speaker 1: prove guilt. It's designed to focus attention. Close inspection of

308
00:22:44,330 --> 00:22:48,330
Speaker 1: this doctor's work revealed a person who upheld the highest

309
00:22:48,450 --> 00:22:52,610
Speaker 1: standards of the medical profession. Close inspection of Harold Shipman's

310
00:22:52,610 --> 00:22:57,570
Speaker 1: practice would have revealed the appalling truth. A forensic analysis

311
00:22:57,570 --> 00:23:00,650
Speaker 1: of his medical record keeping, for example, would have shown

312
00:23:00,690 --> 00:23:04,810
Speaker 1: him back dating entries to invent medical problems after the fact,

313
00:23:05,290 --> 00:23:08,490
Speaker 1: and a single autopsy of one of the patients would

314
00:23:08,490 --> 00:23:12,730
Speaker 1: have revealed the lethal doses of morphine. All it would

315
00:23:12,730 --> 00:23:16,890
Speaker 1: have taken was someone to pay attention, and a simple

316
00:23:16,930 --> 00:23:20,370
Speaker 1: analysis of the numbers would have shown them which doctor

317
00:23:20,410 --> 00:23:24,690
Speaker 1: to pay attention to. But given just how simple this

318
00:23:24,810 --> 00:23:28,650
Speaker 1: statistical exercise would have been, given how many lives it

319
00:23:28,650 --> 00:23:31,450
Speaker 1: would have saved, and given the fact that we didn't

320
00:23:31,490 --> 00:23:35,770
Speaker 1: actually do it, I have a question, what else are

321
00:23:35,770 --> 00:23:45,970
Speaker 1: we missing? Cautionary tales will return in a minute. The

322
00:23:46,090 --> 00:23:49,930
Speaker 1: health authorities in the UK believe they now have statistical

323
00:23:49,930 --> 00:23:53,610
Speaker 1: alarm bells that would ring if another Harold Shipman comes along.

324
00:23:54,570 --> 00:23:58,930
Speaker 1: But what other stories are hiding in plain sight? In

325
00:23:59,010 --> 00:24:03,770
Speaker 1: twenty fourteen, and Case and Angus Deaton were spending the

326
00:24:03,810 --> 00:24:08,170
Speaker 1: summer together in a cabin in Montana. Case and Deaton

327
00:24:08,330 --> 00:24:12,450
Speaker 1: are married. Both are respected economists, and they had both

328
00:24:12,490 --> 00:24:16,490
Speaker 1: become deeply interested in the growing problem of suicide among

329
00:24:16,610 --> 00:24:21,370
Speaker 1: middle aged white Americans. To put that problem into context,

330
00:24:21,730 --> 00:24:25,250
Speaker 1: they decided to compare suicide to the more traditional forms

331
00:24:25,250 --> 00:24:29,570
Speaker 1: of death, such as heart disease and cancer. We went

332
00:24:29,610 --> 00:24:33,650
Speaker 1: to the Centers for Disease Control, downloaded the numbers, and

333
00:24:33,690 --> 00:24:37,250
Speaker 1: made the calculations. They write in their new book, Deaths

334
00:24:37,250 --> 00:24:40,970
Speaker 1: of Despair and the Future of Capitalism. To our astonishment,

335
00:24:41,330 --> 00:24:44,330
Speaker 1: it was not only suicide that was rising among middle

336
00:24:44,330 --> 00:24:49,410
Speaker 1: aged whites. It was all deaths. Not by much. But

337
00:24:49,570 --> 00:24:53,010
Speaker 1: death rates are supposed to fall year on year, so

338
00:24:53,050 --> 00:24:57,050
Speaker 1: even a pause was news, let alone an increase. We

339
00:24:57,170 --> 00:25:00,770
Speaker 1: thought we must have hit a wrong key. Constantly falling

340
00:25:00,810 --> 00:25:03,570
Speaker 1: death rates were one of the best and best established

341
00:25:03,610 --> 00:25:07,290
Speaker 1: features of the twentieth century. The finding was right there

342
00:25:07,330 --> 00:25:11,050
Speaker 1: in the data, but nobody, it seems, had thought to look.

343
00:25:11,810 --> 00:25:14,970
Speaker 1: We thought we must be wrong, because someone would know

344
00:25:15,050 --> 00:25:18,410
Speaker 1: about it. But they weren't wrong. They were just ignored.

345
00:25:19,290 --> 00:25:21,890
Speaker 1: The New England Journal of Medicine didn't want to publish

346
00:25:21,930 --> 00:25:26,090
Speaker 1: the results. The Journal of the American Medical Association rejected

347
00:25:26,130 --> 00:25:28,770
Speaker 1: us so quickly we thought it was an auto reply

348
00:25:28,890 --> 00:25:33,970
Speaker 1: because we'd used the wrong email address. Case and Dton

349
00:25:34,330 --> 00:25:38,850
Speaker 1: broadened and deepened their scrutiny of the numbers. Suicide was up,

350
00:25:39,410 --> 00:25:43,490
Speaker 1: so was chronic liver disease a sign of alcoholism. Even

351
00:25:43,530 --> 00:25:48,650
Speaker 1: more dramatically, deaths from poisoning were up. Poisoning sounds melodramatic,

352
00:25:48,850 --> 00:25:51,690
Speaker 1: like the cause of death in an Agatha Christie story,

353
00:25:52,330 --> 00:25:56,490
Speaker 1: but it usually means a fatal overdose of alcohol or drugs,

354
00:25:56,530 --> 00:26:02,210
Speaker 1: often opioids such as morphine or fentanyl. Once a rare problem,

355
00:26:02,610 --> 00:26:06,090
Speaker 1: drug overdoses have overtaken lung cancer as a cause of

356
00:26:06,130 --> 00:26:09,050
Speaker 1: death for forty five to fifty four year old white America.

357
00:26:09,090 --> 00:26:14,170
Speaker 1: And it all happened so quickly from barely being an

358
00:26:14,210 --> 00:26:17,490
Speaker 1: issue in the late nineteen nineties to making a major

359
00:26:17,570 --> 00:26:22,090
Speaker 1: dent in the mortality data just fifteen years later. It

360
00:26:22,370 --> 00:26:28,290
Speaker 1: is an ironic reversal of Harold Shipman's murderous career. Shipmen

361
00:26:28,530 --> 00:26:33,490
Speaker 1: killed vulnerable people with opioid overdoses. In the US, doctors

362
00:26:33,490 --> 00:26:39,730
Speaker 1: have simply been supplying ever more powerful opioids to vulnerable people. Misery,

363
00:26:40,090 --> 00:26:44,610
Speaker 1: pain or sheer accident have done the rest. Put these

364
00:26:44,650 --> 00:26:50,090
Speaker 1: three courses of death together, suicide, accidental overdoses, and livid disease,

365
00:26:50,690 --> 00:26:53,370
Speaker 1: and you have a category that Case and Dton named

366
00:26:53,890 --> 00:26:59,730
Speaker 1: deaths of despair. The toll dwarfs anything that one murderer

367
00:26:59,730 --> 00:27:03,570
Speaker 1: could achieve. Case and Dton found there were one hundred

368
00:27:03,610 --> 00:27:07,730
Speaker 1: and fifty eight thousand deaths of despair in twenty seventeen.

369
00:27:08,370 --> 00:27:11,170
Speaker 1: That is a similar scale to the first wave of

370
00:27:11,250 --> 00:27:15,930
Speaker 1: COVID nineteen deaths in the US. It was a catastrophe,

371
00:27:16,450 --> 00:27:19,650
Speaker 1: and it was a catastrophe that should have been plainly

372
00:27:19,770 --> 00:27:24,490
Speaker 1: visible in the statistics. Yet somehow nobody had taken the

373
00:27:24,530 --> 00:27:33,050
Speaker 1: effort to look. On the twenty fourth of June nineteen

374
00:27:33,170 --> 00:27:38,250
Speaker 1: ninety eight, Kathleen Grundy, the former Mayoress of Hyde, died

375
00:27:38,810 --> 00:27:43,450
Speaker 1: suddenly at the age of eighty one. It was a surprise.

376
00:27:44,010 --> 00:27:47,210
Speaker 1: She had been fit and socially active. On the same day,

377
00:27:47,690 --> 00:27:51,050
Speaker 1: a will arrived at a firm of local attorneys with

378
00:27:51,170 --> 00:27:54,650
Speaker 1: a covering letter. The will purported to be that of

379
00:27:54,890 --> 00:27:58,850
Speaker 1: Kathleen Grundy. It declared her intention to leave her house

380
00:27:58,930 --> 00:28:04,650
Speaker 1: to her dear family, doctor Harold Shipman, but the attorney

381
00:28:04,730 --> 00:28:08,250
Speaker 1: had never had any dealings with Kathleen Grundy and the

382
00:28:08,290 --> 00:28:12,650
Speaker 1: signatures looked odd. A few days later, a mysterious letter

383
00:28:12,730 --> 00:28:16,930
Speaker 1: from someone called Smith told the attorney that missus Grundy

384
00:28:17,010 --> 00:28:22,810
Speaker 1: had died. Puzzled, the attorney contacted Kathleen Grundy's daughter, who

385
00:28:22,850 --> 00:28:26,490
Speaker 1: was an attorney herself and well versed in the ins

386
00:28:26,610 --> 00:28:30,850
Speaker 1: and outs of making a will. Already stunned by her

387
00:28:30,850 --> 00:28:33,890
Speaker 1: mother's death, she was even more astonished to find herself

388
00:28:33,930 --> 00:28:39,370
Speaker 1: and her children abruptly disinherited. There had been no family argument,

389
00:28:39,890 --> 00:28:42,370
Speaker 1: no sign that had changed in the will was imminent,

390
00:28:43,130 --> 00:28:47,370
Speaker 1: and the new will was odd. Why send it to

391
00:28:47,410 --> 00:28:50,690
Speaker 1: an unknown firm of attorneys, Why was it riddled with

392
00:28:50,810 --> 00:28:54,970
Speaker 1: typos when her mother was a trained typist, and why

393
00:28:54,970 --> 00:28:58,210
Speaker 1: did it show no knowledge of the fact that Kathleen

394
00:28:58,250 --> 00:29:01,210
Speaker 1: Grundy owned a second house in Hyde and a holiday

395
00:29:01,250 --> 00:29:07,930
Speaker 1: cottage too. Kathleen Grundy's daughter called the police. It didn't

396
00:29:07,930 --> 00:29:10,370
Speaker 1: take long for the police to discover that the will

397
00:29:10,490 --> 00:29:13,250
Speaker 1: was a forgery, that the cover letter had been typed

398
00:29:13,290 --> 00:29:18,250
Speaker 1: on Harold Shipman's typewriter, and that missus Grundy's medical records

399
00:29:18,450 --> 00:29:23,530
Speaker 1: had been altered after her death. If Shipman hadn't made

400
00:29:23,610 --> 00:29:28,730
Speaker 1: such crass misjudgments, who knows, he might never have been caught.

401
00:29:30,010 --> 00:29:34,050
Speaker 1: As it was. Harold Shipman was arrested. Faced with the

402
00:29:34,170 --> 00:29:39,450
Speaker 1: need to conduct autopsy examinations, the police began the terrible

403
00:29:39,570 --> 00:29:43,770
Speaker 1: task of digging up the bodies all over the town

404
00:29:43,810 --> 00:29:49,050
Speaker 1: of Hyde. The slow process of uncovering Shipman's awful crimes

405
00:29:49,650 --> 00:29:56,210
Speaker 1: had begun. In the aftermath, some local people blamed themselves

406
00:29:56,250 --> 00:30:00,450
Speaker 1: for not having spoken up sooner. John Shaw, the gentle

407
00:30:00,570 --> 00:30:04,130
Speaker 1: taxi driver who spent his days driving elderly ladies around,

408
00:30:04,570 --> 00:30:07,650
Speaker 1: knew them well enough to attend funerals when they passed away,

409
00:30:08,730 --> 00:30:13,970
Speaker 1: but were simply too many funerals. Shaw told the journalists

410
00:30:13,970 --> 00:30:17,290
Speaker 1: Brian Whittle and Jean Ritchie. I noticed that all those

411
00:30:17,330 --> 00:30:21,970
Speaker 1: who were dying went to the same doctor, doctor Shipman. Eventually,

412
00:30:22,330 --> 00:30:25,370
Speaker 1: john Shaw went to the police to discover that they

413
00:30:25,370 --> 00:30:29,770
Speaker 1: were already investigating the death of Kathleen Grundy. Could he

414
00:30:29,810 --> 00:30:34,450
Speaker 1: have spoken up earlier, perhaps a year or two, But

415
00:30:34,570 --> 00:30:37,570
Speaker 1: the police admitted that he might well have been ignored

416
00:30:37,810 --> 00:30:41,730
Speaker 1: if he had. After all, he was just a taxi driver,

417
00:30:42,170 --> 00:30:46,530
Speaker 1: and Harold Shipman was a respected doctor. Other people had

418
00:30:46,570 --> 00:30:50,290
Speaker 1: also been growing concerned. There was Debbie Massey, a funeral

419
00:30:50,330 --> 00:30:53,890
Speaker 1: director who was responsible for burying or cremating many of

420
00:30:53,890 --> 00:30:59,010
Speaker 1: Shipman's victims. There was Linda Reynolds, another local doctor. Massey

421
00:30:59,090 --> 00:31:02,730
Speaker 1: and Reynolds raised the alarm in March of nineteen ninety eight.

422
00:31:03,250 --> 00:31:06,450
Speaker 1: Perhaps they could have spoken up in February or January.

423
00:31:06,690 --> 00:31:09,450
Speaker 1: Perhaps the police could have been more vigorous in responding.

424
00:31:09,970 --> 00:31:13,170
Speaker 1: But it's important to recognize we're talking about a matter

425
00:31:13,210 --> 00:31:18,010
Speaker 1: of weeks or months at best. If instead we had

426
00:31:18,050 --> 00:31:22,090
Speaker 1: collected the simplest of data sets, if we had run

427
00:31:22,210 --> 00:31:25,970
Speaker 1: the most basic analysis of that data, we would never

428
00:31:26,010 --> 00:31:28,770
Speaker 1: have needed to depend on people risking the scorn of

429
00:31:28,770 --> 00:31:33,650
Speaker 1: the police and the enmity of Harold Shipman to stop him.

430
00:31:33,690 --> 00:31:38,690
Speaker 1: The statisticians, with their production line mathematics designed to inspect

431
00:31:38,730 --> 00:31:42,650
Speaker 1: condoms and chocolate chip cookies, might have stopped his murder

432
00:31:42,690 --> 00:32:12,050
Speaker 1: Spree more than a decade earlier. Essential sources on Shipman's

433
00:32:12,090 --> 00:32:16,330
Speaker 1: crimes are The Shipman Inquiry and Brian Whittle and Jeane

434
00:32:16,410 --> 00:32:22,170
Speaker 1: Rich's book Prescription for Murder, David Spiegelhalter's excellent book The

435
00:32:22,410 --> 00:32:26,290
Speaker 1: Art of Statistics covers the Shipman case, and my own

436
00:32:26,330 --> 00:32:29,850
Speaker 1: book The Data Detective makes a plea for taking the

437
00:32:29,930 --> 00:32:37,730
Speaker 1: numbers seriously. Other sources are at Tim Harford dot com.

438
00:32:37,890 --> 00:32:41,730
Speaker 1: Cautionary Tales is written by me Tim Harford with Andrew Wright.

439
00:32:42,050 --> 00:32:45,490
Speaker 1: It's produced by Ryan Dilley and Marilyn Rust. The sound

440
00:32:45,570 --> 00:32:48,930
Speaker 1: design and original music is the work of Pascal Wise.

441
00:32:49,650 --> 00:32:53,890
Speaker 1: Julia Barton edited the scripts. Starring in this series of

442
00:32:53,930 --> 00:32:59,890
Speaker 1: Cautionary Tales Helena Bonham Carter and Jeffrey Wright, alongside Nazzar Alderazzi,

443
00:33:00,530 --> 00:33:07,130
Speaker 1: Ed Gohan, Melanie Gutteridge, Rachel Hanshaw, copenaholbrook Smith, Greg Lockett,

444
00:33:07,330 --> 00:33:12,290
Speaker 1: Messiamunroe and rufless Right. This show wouldn't have been possible

445
00:33:12,330 --> 00:33:16,170
Speaker 1: without the work of Mia La Belle, Jacob Weisberg, Heather Fane,

446
00:33:16,570 --> 00:33:22,330
Speaker 1: John Schnarz, Carli mcgiori, Eric Sandler, Emily Rostick, Maggie Taylor,

447
00:33:22,730 --> 00:33:28,010
Speaker 1: An Yellow Lakhan and Maya Kanig. Cautionary Tales is a

448
00:33:28,050 --> 00:33:32,770
Speaker 1: production of Pushkin Industries. If you like the show, please

449
00:33:32,810 --> 00:33:35,890
Speaker 1: remember to rate, share and review.