1
00:00:00,000 --> 00:00:02,000
Tare Tuesday.

2
00:00:02,000 --> 00:00:10,320
So you know, sometimes people learn something that makes them feel super smart.

3
00:00:10,320 --> 00:00:14,840
And then so they decide to go ahead and apply it every chance they get, even when they don't

4
00:00:14,840 --> 00:00:19,360
fully understand what the hell it is that they're really talking about.

5
00:00:19,360 --> 00:00:20,360
Okay?

6
00:00:20,360 --> 00:00:26,080
For me, from teaching, one of the things that comes in, especially in my critical thinking

7
00:00:26,080 --> 00:00:30,880
classes, is this distinction between correlation and causation.

8
00:00:30,880 --> 00:00:31,880
Okay?

9
00:00:31,880 --> 00:00:34,600
So let's first get the basics out of the way.

10
00:00:34,600 --> 00:00:35,600
Okay?

11
00:00:35,600 --> 00:00:37,560
Correlation does not equal causation.

12
00:00:37,560 --> 00:00:38,560
Okay?

13
00:00:38,560 --> 00:00:39,840
You've probably heard this before.

14
00:00:39,840 --> 00:00:42,560
I hear it from students all the time.

15
00:00:42,560 --> 00:00:44,440
You're absolutely correct.

16
00:00:44,440 --> 00:00:46,120
Correlation does not equal causation.

17
00:00:46,120 --> 00:00:47,120
What does that mean?

18
00:00:47,120 --> 00:00:51,280
Does that mean that just simply means that just because two events are correlated, right?

19
00:00:51,280 --> 00:00:55,440
So if one happens, the other doesn't happen or every time one thing happens, something

20
00:00:55,440 --> 00:00:56,440
else happens.

21
00:00:56,440 --> 00:00:57,440
Okay?

22
00:00:57,440 --> 00:01:02,520
That correlation does not guarantee that there's some sort of causal relationship, like one

23
00:01:02,520 --> 00:01:04,120
is causing the other.

24
00:01:04,120 --> 00:01:08,360
It could just be that there's coincidence for one, but if it happens repeatedly, more

25
00:01:08,360 --> 00:01:11,160
likely than not, there might be some third variable.

26
00:01:11,160 --> 00:01:15,920
There's something else that is maybe either causing both things to happen or things that

27
00:01:15,920 --> 00:01:19,920
are distantly related somehow, but not causally related.

28
00:01:19,920 --> 00:01:20,920
Right?

29
00:01:20,920 --> 00:01:25,000
It just happens that every time one thing happens, it's also other factors that suggest

30
00:01:25,000 --> 00:01:28,320
that something else might also happen, but they're not causally connected at all.

31
00:01:28,320 --> 00:01:29,320
Okay?

32
00:01:29,320 --> 00:01:30,320
All right.

33
00:01:30,320 --> 00:01:31,320
So I'm not going to disagree with any of that.

34
00:01:31,320 --> 00:01:32,320
That is all correct.

35
00:01:32,320 --> 00:01:33,320
That is all correct.

36
00:01:33,320 --> 00:01:34,320
Okay?

37
00:01:34,320 --> 00:01:41,360
What pisses me off though is when people say correlation is not causation as a way to dismiss

38
00:01:41,360 --> 00:01:48,440
the weight that correlation has in drawing conclusions about causation.

39
00:01:48,440 --> 00:01:49,440
What do I mean by this?

40
00:01:49,440 --> 00:01:50,440
Okay?

41
00:01:50,440 --> 00:01:57,560
Because correlation does not equal causation does not mean that correlation is not some

42
00:01:57,560 --> 00:02:03,040
sort of indicator that it is plausible that there is some causation going on.

43
00:02:03,040 --> 00:02:04,040
Okay?

44
00:02:04,040 --> 00:02:05,600
Does that make sense?

45
00:02:05,600 --> 00:02:06,600
Let me break it down for you.

46
00:02:06,600 --> 00:02:07,600
Okay?

47
00:02:07,600 --> 00:02:08,600
All right.

48
00:02:08,600 --> 00:02:09,600
So here's a correlation.

49
00:02:09,600 --> 00:02:10,600
Okay?

50
00:02:10,600 --> 00:02:11,600
Here's a correlation for you.

51
00:02:11,600 --> 00:02:12,600
Okay?

52
00:02:12,600 --> 00:02:17,840
Every time there's somebody sets a fire to a house, okay, suppose that the news, the

53
00:02:17,840 --> 00:02:20,880
public news is out there with their cameras and everything and their video taping, it's

54
00:02:20,880 --> 00:02:23,880
like, ah, another like some fire struck this neighborhood again.

55
00:02:23,880 --> 00:02:24,880
Okay?

56
00:02:24,880 --> 00:02:29,000
And then every time that fire, there's a new fire in the neighborhood, in the city.

57
00:02:29,000 --> 00:02:30,000
Okay?

58
00:02:30,000 --> 00:02:37,280
In that video shot from the news, you see me in the audience and it's only me that seems

59
00:02:37,280 --> 00:02:41,600
to show up in the audience at every single one of these fires.

60
00:02:41,600 --> 00:02:42,600
Okay?

61
00:02:42,600 --> 00:02:44,300
One fire, two fires.

62
00:02:44,300 --> 00:02:46,040
Maybe you think it's a coincidence.

63
00:02:46,040 --> 00:02:48,280
Three fires, four fires, five fires.

64
00:02:48,280 --> 00:02:49,280
Right?

65
00:02:49,280 --> 00:02:50,280
Okay.

66
00:02:50,280 --> 00:02:51,280
Now we're getting a correlation here.

67
00:02:51,280 --> 00:02:52,280
Okay?

68
00:02:52,280 --> 00:02:53,280
I'm at all of these events.

69
00:02:53,280 --> 00:02:54,280
All right?

70
00:02:54,280 --> 00:02:58,880
Now, okay, does that mean I caused the fire?

71
00:02:58,880 --> 00:02:59,880
Not necessarily.

72
00:02:59,880 --> 00:03:00,880
Right?

73
00:03:00,880 --> 00:03:06,200
It could be that I am just some dude that has a sick obsession with houses on fires.

74
00:03:06,200 --> 00:03:10,200
And so anytime there's a fire, all right, maybe I just have my ears to the ground and

75
00:03:10,200 --> 00:03:12,160
I'm just like, I just know where the fires are happening.

76
00:03:12,160 --> 00:03:13,960
And I just zoom right there as quickly as I can.

77
00:03:13,960 --> 00:03:18,000
I just happen to be in the shot when the newscasters are panning the area.

78
00:03:18,000 --> 00:03:19,000
Okay?

79
00:03:19,000 --> 00:03:20,920
So it could be that it's a correlation.

80
00:03:20,920 --> 00:03:23,240
Again, it's not causation necessarily.

81
00:03:23,240 --> 00:03:24,720
Right?

82
00:03:24,720 --> 00:03:27,720
Because I just happen to really like fires and then whenever there's fires, I just happen

83
00:03:27,720 --> 00:03:28,720
to be there.

84
00:03:28,720 --> 00:03:29,720
Okay?

85
00:03:29,720 --> 00:03:31,960
I guess you could kind of call it like a causation, right?

86
00:03:31,960 --> 00:03:34,600
The fires caused me to be there, but that's a weird way of thinking about it.

87
00:03:34,600 --> 00:03:35,600
Right?

88
00:03:35,600 --> 00:03:36,600
I think of it more like a correlation.

89
00:03:36,600 --> 00:03:37,600
Okay?

90
00:03:37,600 --> 00:03:38,600
But here's the thing.

91
00:03:38,600 --> 00:03:39,600
Here's the thing.

92
00:03:39,600 --> 00:03:47,120
If you're a detective and you noticed a pattern and you see that I am the only person that

93
00:03:47,120 --> 00:03:50,880
is at every single one of these events, right?

94
00:03:50,880 --> 00:03:55,920
Every single one of these events where somebody deliberately set a fire to a house.

95
00:03:55,920 --> 00:03:56,920
Okay?

96
00:03:56,920 --> 00:03:57,920
Somebody.

97
00:03:57,920 --> 00:03:58,920
Okay?

98
00:03:58,920 --> 00:04:00,240
Wouldn't you want to bring me in for questioning?

99
00:04:00,240 --> 00:04:08,200
I'm not saying there's a guarantee that I'm the one who caused those fires, but it seems

100
00:04:08,200 --> 00:04:12,960
like good detective duty to at least bring me in for questioning.

101
00:04:12,960 --> 00:04:16,160
I'm not saying that you should arrest me and then you should put me in jail.

102
00:04:16,160 --> 00:04:17,160
Okay?

103
00:04:17,160 --> 00:04:18,160
I would argue in court.

104
00:04:18,160 --> 00:04:19,160
I'd be like, correlation is not causation.

105
00:04:19,160 --> 00:04:20,520
Correlation is not causation.

106
00:04:20,520 --> 00:04:21,520
Okay?

107
00:04:21,520 --> 00:04:23,200
But here's the thing.

108
00:04:23,200 --> 00:04:24,320
Okay?

109
00:04:24,320 --> 00:04:28,960
Correlation is one of the ways that we try to detect causation.

110
00:04:28,960 --> 00:04:31,480
It is not the only thing that we use.

111
00:04:31,480 --> 00:04:32,480
Okay?

112
00:04:32,480 --> 00:04:38,200
When more and more correlations start to stack up, then you start becoming more and more

113
00:04:38,200 --> 00:04:41,920
suspicious that there's some sort of causal link between them.

114
00:04:41,920 --> 00:04:43,640
That is not a mere accident.

115
00:04:43,640 --> 00:04:45,160
That is not a third variable.

116
00:04:45,160 --> 00:04:47,480
You could try to eliminate all the other variables.

117
00:04:47,480 --> 00:04:51,600
And if that correlation is still there, that's one of the ways we figure out whether or not

118
00:04:51,600 --> 00:04:53,200
there could be a causal factor here.

119
00:04:53,200 --> 00:04:54,200
Okay?

120
00:04:54,200 --> 00:04:55,200
There's a causal relationship.

121
00:04:55,200 --> 00:04:56,200
Okay?

122
00:04:56,200 --> 00:04:58,200
So all I'm saying is this.

123
00:04:58,200 --> 00:04:59,200
Okay?

124
00:04:59,200 --> 00:05:03,720
You can acknowledge if somebody is making the error, and it is an error, to just simply

125
00:05:03,720 --> 00:05:07,240
go from correlation and then shift directly to causation.

126
00:05:07,240 --> 00:05:08,240
Okay?

127
00:05:08,240 --> 00:05:09,720
That is a mistake.

128
00:05:09,720 --> 00:05:15,200
But you should not shit on people for thinking correlation is not causation.

129
00:05:15,200 --> 00:05:19,880
But correlation does often raise a lot of flags.

130
00:05:19,880 --> 00:05:22,200
It at least gives us some stuff to look into.

131
00:05:22,200 --> 00:05:23,200
Okay?

132
00:05:23,200 --> 00:05:26,720
It's going to give that detective goddamn good reason to go ahead and question me for

133
00:05:26,720 --> 00:05:29,320
all the arsons happening in the neighborhood.

134
00:05:29,320 --> 00:05:30,320
Okay?

135
00:05:30,320 --> 00:05:32,960
The police better question me because if they don't, they're not doing their job.

136
00:05:32,960 --> 00:05:33,960
Okay?

137
00:05:33,960 --> 00:05:36,280
If they see that I'm there at all those sites of the fires, right?

138
00:05:36,280 --> 00:05:40,480
They should bring me in and be like, hey, we noticed that you are the only one that

139
00:05:40,480 --> 00:05:43,000
happens to be at every single one of these fires.

140
00:05:43,000 --> 00:05:44,000
Right?

141
00:05:44,000 --> 00:05:45,640
Why is that?

142
00:05:45,640 --> 00:05:46,640
Oh.

143
00:05:46,640 --> 00:05:47,640
Strange.

144
00:05:47,640 --> 00:05:48,640
Okay?

145
00:05:48,640 --> 00:05:50,280
You should at least try to get to the bottom of it.

146
00:05:50,280 --> 00:05:51,280
Okay?

147
00:05:51,280 --> 00:05:53,400
Now, at the end of the day, again, it might just be something else.

148
00:05:53,400 --> 00:05:54,720
It might be just a really bad coincidence.

149
00:05:54,720 --> 00:05:59,080
I might have terrible luck, but if you're not being suspicious, something wrong with

150
00:05:59,080 --> 00:06:00,080
you.

151
00:06:00,080 --> 00:06:01,080
Okay?

152
00:06:01,080 --> 00:06:02,080
You're not reasoning well.

153
00:06:02,080 --> 00:06:03,080
Okay?

154
00:06:03,080 --> 00:06:04,080
So again, all right?

155
00:06:04,080 --> 00:06:10,080
Correlation does not equal causation, but a lot of correlation does raise a lot of suspicions

156
00:06:10,080 --> 00:06:13,040
about the potential for causation.

157
00:06:13,040 --> 00:06:14,480
That is not irrational.

158
00:06:14,480 --> 00:06:15,480
That's good reasoning.

159
00:06:15,480 --> 00:06:16,600
All right?

160
00:06:16,600 --> 00:06:18,080
That's not the end of the story.

161
00:06:18,080 --> 00:06:24,360
There's still more investigating to do, but don't blame somebody for suspecting the possibility

162
00:06:24,360 --> 00:06:28,400
of causation when they see a whole lot of correlation.

163
00:06:28,400 --> 00:06:29,400
Okay?

164
00:06:29,400 --> 00:06:34,400
So, don't think you're a little bit, little smarty pants by learning this distinction.

165
00:06:34,400 --> 00:06:36,060
Correlation does not equal causation.

166
00:06:36,060 --> 00:06:40,280
And then just leave it at that and then think, oh yeah, and therefore correlation has nothing

167
00:06:40,280 --> 00:06:41,800
to do with causation.

168
00:06:41,800 --> 00:06:42,800
That's ridiculous.

169
00:06:42,800 --> 00:06:43,800
Okay?

170
00:06:43,800 --> 00:06:47,520
Correlation is one of the main ways we try to identify causation.

171
00:06:47,520 --> 00:06:51,520
This is a technique used in a majority of the social sciences.

172
00:06:51,520 --> 00:06:52,520
Okay?

173
00:06:52,520 --> 00:06:56,280
It's at least a starting point in a lot of cases in the medical sciences, and sometimes

174
00:06:56,280 --> 00:06:58,040
that's all they have to go on.

175
00:06:58,040 --> 00:06:59,040
Okay?

176
00:06:59,040 --> 00:07:03,680
So, it is a technique used across the board and has helped us make a lot of advances.

177
00:07:03,680 --> 00:07:05,560
So don't shit on correlation.

178
00:07:05,560 --> 00:07:06,560
Okay?

179
00:07:06,560 --> 00:07:08,200
Correlation helps.

180
00:07:08,200 --> 00:07:09,200
Got it?

181
00:07:09,200 --> 00:07:11,720
People being a smart ass.

182
00:07:11,720 --> 00:07:25,360
You can be a smart ass.