1
00:00:00,320 --> 00:00:05,020
A town puts up new streetlights, and that
same year burglaries drop.

2
00:00:05,780 --> 00:00:10,540
The argument concludes the lights scared
off the burglars, and it feels airtight.

3
00:00:11,420 --> 00:00:16,130
It is also one of the most common ways the
LSAT gets you to nod along at something

4
00:00:16,140 --> 00:00:17,370
that proves nothing.

5
00:00:18,319 --> 00:00:19,888
But I want to nod along.

6
00:00:20,528 --> 00:00:23,168
The lights went up, crime went down.

7
00:00:24,108 --> 00:00:26,628
What is actually wrong with connecting
those two?

8
00:00:27,718 --> 00:00:32,326
Hold that instinct, because by the end of
this you will see at least three different

9
00:00:32,447 --> 00:00:36,307
things that could be hiding behind that
drop, and not one of them is the

10
00:00:36,367 --> 00:00:36,867
streetlights.

11
00:00:37,997 --> 00:00:40,047
Let me reach back a couple of episodes
first.

12
00:00:41,222 --> 00:00:45,742
Quick retrieval, all the way back to
Episode 1 and then Episode 4.

13
00:00:46,622 --> 00:00:51,822
When the LSAT asks you to weaken an
argument, what are you actually attacking?

14
00:00:51,882 --> 00:00:54,822
Not the premises, not the conclusion.

15
00:00:54,942 --> 00:00:57,402
Take a second before I say it.

16
00:00:59,232 --> 00:01:00,181
The assumption.

17
00:01:00,621 --> 00:01:03,481
The bridge between the support and the
conclusion.

18
00:01:04,080 --> 00:01:09,211
We were poking the gap, the thing the
author needs to be true but never said out

19
00:01:09,241 --> 00:01:10,461
loud.

20
00:01:10,829 --> 00:01:16,448
That is the whole game today, because
causal arguments smuggle in one specific,

21
00:01:16,608 --> 00:01:22,038
predictable assumption, and once you can
name it, weaken and strengthen stop being

22
00:01:22,078 --> 00:01:22,538
guesswork.

23
00:01:23,378 --> 00:01:25,118
So here is the objective.

24
00:01:25,138 --> 00:01:30,018
By the end of this episode, the second a
stimulus concludes that X caused Y,

25
00:01:30,638 --> 00:01:34,878
you will name the hidden assumption and
run a three-question attack on it in under

26
00:01:34,978 --> 00:01:35,938
fifteen seconds.

27
00:01:37,498 --> 00:01:39,868
So this is not a brand-new question type.

28
00:01:40,258 --> 00:01:45,088
It is a pattern that keeps showing up
inside the types we already learned.

29
00:01:46,683 --> 00:01:48,493
And that is exactly why it pays.

30
00:01:49,273 --> 00:01:54,333
Logical Reasoning is now two of the scored
sections, roughly two-thirds of the test,

31
00:01:54,833 --> 00:01:57,453
and causal flaws are scattered all through
it.

32
00:01:58,313 --> 00:02:03,293
Learn this one shape and you are not
fixing one question, you are fixing dozens.

33
00:02:04,473 --> 00:02:06,113
Three moves today.

34
00:02:06,193 --> 00:02:09,653
One, define a causal claim and find its
gap.

35
00:02:10,673 --> 00:02:13,913
Two, the three ways to attack it.

36
00:02:14,053 --> 00:02:19,073
Three, how to defend it, which turns out
to be the same three moves run backwards.

37
00:02:20,092 --> 00:02:23,971
Picture two lines on a chart that rise and
fall together.

38
00:02:24,681 --> 00:02:27,581
Ice cream sales and, let's say, sunburns.

39
00:02:28,041 --> 00:02:30,281
They move in lockstep all summer.

40
00:02:31,041 --> 00:02:34,981
That togetherness has a name, and the name
is correlation.

41
00:02:35,981 --> 00:02:40,151
Now picture an arrow, one thing reaching
out and producing the other.

42
00:02:40,831 --> 00:02:45,121
That arrow is causation, and it points one
direction only.

43
00:02:46,817 --> 00:02:49,787
So the claim is not that the two things
happened together.

44
00:02:49,827 --> 00:02:53,067
The claim is that one of them made the
other happen.

45
00:02:54,325 --> 00:02:55,935
That is the whole distinction.

46
00:02:56,465 --> 00:02:59,645
Causation is directional and asymmetric.

47
00:03:00,325 --> 00:03:04,305
X to Y, never just X and Y side by side.

48
00:03:05,105 --> 00:03:07,965
And here is the move the test runs on you.

49
00:03:08,105 --> 00:03:13,665
In almost every LSAT causal argument, the
correlation is the premise and the

50
00:03:13,725 --> 00:03:15,965
cause-and-effect statement is the
conclusion.

51
00:03:16,665 --> 00:03:21,775
The author hands you two things moving
together, then quietly leaps to one of them

52
00:03:21,825 --> 00:03:22,655
causing the other.

53
00:03:24,237 --> 00:03:26,887
And that leap is the gap from Episode 1.

54
00:03:27,891 --> 00:03:30,431
It is the same gap wearing a costume.

55
00:03:31,331 --> 00:03:36,871
Correlation does not equal causation is
just a fancier way of saying there is a gap

56
00:03:36,881 --> 00:03:38,731
between the premise and the conclusion.

57
00:03:39,671 --> 00:03:44,371
So name the hidden assumption with me,
because it is the spine of everything else.

58
00:03:45,271 --> 00:03:50,461
When an author says X caused Y based on a
correlation, the author is secretly

59
00:03:50,491 --> 00:03:52,871
assuming three things at once.

60
00:03:52,931 --> 00:03:56,011
That X is the only explanation for Y.

61
00:03:56,071 --> 00:03:59,191
That the direction is right, X to Y and
not the reverse.

62
00:03:59,951 --> 00:04:02,331
And that it is not a fluke.

63
00:04:02,371 --> 00:04:05,111
Defend any one of those and you
strengthen.

64
00:04:05,171 --> 00:04:07,011
Break any one of those and you weaken.

65
00:04:08,128 --> 00:04:09,257
One pushback.

66
00:04:09,357 --> 00:04:11,487
You said the only explanation.

67
00:04:12,197 --> 00:04:13,377
That feels too strong.

68
00:04:13,957 --> 00:04:16,557
Real authors do not always claim it that
hard.

69
00:04:17,962 --> 00:04:21,993
That is the right place to push, and you
have caught a real limit.

70
00:04:22,703 --> 00:04:27,053
The only-cause assumption holds when the
author is fully certain,

71
00:04:27,533 --> 00:04:30,733
plain correlation in, confident cause out.

72
00:04:31,533 --> 00:04:37,323
But if the author hedges, says a
contributing factor or partly responsible,

73
00:04:37,323 --> 00:04:40,993
the assumption softens and your attack has
to soften with it.

74
00:04:41,613 --> 00:04:44,353
So check the author's confidence before
you swing.

75
00:04:45,153 --> 00:04:49,213
For the big, certain causal conclusions
the test loves, though,

76
00:04:49,253 --> 00:04:51,273
the three-part assumption is dead on.

77
00:04:52,367 --> 00:04:57,677
Here is your checklist, three questions
you run on every causal stimulus.

78
00:04:58,547 --> 00:04:59,147
Question one.

79
00:04:59,827 --> 00:05:01,987
Could something else have caused Y?

80
00:05:02,667 --> 00:05:04,447
That is the alternative cause.

81
00:05:05,387 --> 00:05:06,327
Back to the streetlights.

82
00:05:06,947 --> 00:05:10,387
Suppose the town also added police patrols
that same year.

83
00:05:11,067 --> 00:05:14,487
If the patrols cut the burglaries, the
lights get no credit.

84
00:05:15,097 --> 00:05:18,547
The correlation survives, but the causal
claim wobbles.

85
00:05:19,818 --> 00:05:23,468
And there is a sneakier version of
question one, isn't there.

86
00:05:23,928 --> 00:05:25,458
The hidden third thing.

87
00:05:27,236 --> 00:05:28,506
The cleanest example there is.

88
00:05:29,126 --> 00:05:35,116
Picture a city where ice cream sales and
drowning deaths rise and fall together all

89
00:05:35,146 --> 00:05:36,246
summer.

90
00:05:36,306 --> 00:05:40,726
It is tempting to say sugar makes people
reckless near the water.

91
00:05:40,746 --> 00:05:44,846
But there is a third variable sitting
underneath both of them.

92
00:05:44,966 --> 00:05:45,966
Heat.

93
00:05:46,066 --> 00:05:51,246
Hot days drive ice cream sales up, and hot
days drive people into the water,

94
00:05:51,606 --> 00:05:52,586
where a few drown.

95
00:05:53,486 --> 00:05:58,466
Ice cream and drowning correlate, but
neither one causes the other.

96
00:05:58,526 --> 00:06:00,426
A common cause drives both.

97
00:06:01,506 --> 00:06:05,516
That third-variable move is just a flavor
of alternative cause,

98
00:06:05,906 --> 00:06:09,316
so keep it in the same bucket rather than
counting it as a fourth attack.

99
00:06:10,530 --> 00:06:11,579
That keeps the list short.

100
00:06:12,219 --> 00:06:13,139
What is question two?

101
00:06:14,141 --> 00:06:15,021
Question two.

102
00:06:15,701 --> 00:06:16,881
Could it be backwards?

103
00:06:17,441 --> 00:06:18,681
Reverse the causation.

104
00:06:19,421 --> 00:06:24,941
People who drink coffee report sharper
focus, so the argument says coffee improves

105
00:06:25,001 --> 00:06:25,341
focus.

106
00:06:26,141 --> 00:06:27,181
But flip the arrow.

107
00:06:28,001 --> 00:06:31,881
Maybe the people who already need to
focus, the students cramming,

108
00:06:32,181 --> 00:06:36,151
the workers buried in deadlines, are the
ones reaching for coffee in the first

109
00:06:36,221 --> 00:06:36,461
place.

110
00:06:37,321 --> 00:06:41,181
The need to focus caused the coffee, not
the other way around.

111
00:06:42,021 --> 00:06:44,281
Same correlation, opposite arrow.

112
00:06:45,424 --> 00:06:48,213
This is the one I used to walk right past.

113
00:06:48,733 --> 00:06:54,153
I would hunt for an alternative cause and
never even ask whether the whole thing ran

114
00:06:54,193 --> 00:06:55,313
the other direction.

115
00:06:56,293 --> 00:07:00,872
Most students walk past it, which is
precisely why the test plants it.

116
00:07:01,562 --> 00:07:04,282
And there is a tiny tell that helps.

117
00:07:04,462 --> 00:07:07,782
A cause has to come before its effect in
time.

118
00:07:08,462 --> 00:07:13,822
So if the so-called effect actually
happened first, the arrow cannot point the way

119
00:07:13,862 --> 00:07:18,542
the author wants, and you file that under
reverse causation.

120
00:07:18,602 --> 00:07:19,182
Question three.

121
00:07:19,982 --> 00:07:21,122
Could it just be a fluke?

122
00:07:21,862 --> 00:07:23,742
Coincidence, or bad data.

123
00:07:24,822 --> 00:07:26,402
Back to the streetlights one last time.

124
00:07:27,082 --> 00:07:31,342
Suppose burglaries fell across the entire
state that year, every town,

125
00:07:31,752 --> 00:07:32,772
lights or no lights.

126
00:07:33,512 --> 00:07:37,102
Then this town's drop is just one ripple
in a statewide wave,

127
00:07:37,602 --> 00:07:39,292
and the streetlights were along for the
ride.

128
00:07:40,265 --> 00:07:44,164
So the three are: something else did it,
it runs backwards,

129
00:07:44,634 --> 00:07:46,194
or it is a coincidence.

130
00:07:47,634 --> 00:07:49,123
That is the entire arsenal.

131
00:07:49,843 --> 00:07:50,883
Alternative cause.

132
00:07:51,623 --> 00:07:52,263
Reverse it.

133
00:07:53,183 --> 00:07:53,803
Coincidence.

134
00:07:54,783 --> 00:08:00,143
Three questions on every causal stimulus,
and once you run them the credited answer

135
00:08:00,283 --> 00:08:02,883
usually walks up and introduces itself.

136
00:08:03,830 --> 00:08:07,460
Let me work one fully, then I hand you the
last step.

137
00:08:08,570 --> 00:08:10,860
A company rolls out a new wellness app.

138
00:08:11,390 --> 00:08:15,590
It finds that employees who use the app
take fewer sick days.

139
00:08:15,650 --> 00:08:18,410
The conclusion: the app made them
healthier.

140
00:08:19,150 --> 00:08:20,000
Pause it right there.

141
00:08:20,830 --> 00:08:24,710
Before I weaken it, which of the three
attacks feels strongest here?

142
00:08:25,310 --> 00:08:26,310
Try it yourself first.

143
00:08:27,577 --> 00:08:29,757
I will commit, and I will defend it.

144
00:08:30,557 --> 00:08:32,937
Reverse causation, easily.

145
00:08:33,577 --> 00:08:38,517
The argument says the app made them
healthier, but I think it is backwards.

146
00:08:39,096 --> 00:08:41,837
Being healthier is what made them use the
app.

147
00:08:42,437 --> 00:08:45,107
Healthy people are the ones who bother
with a wellness app,

148
00:08:45,557 --> 00:08:48,877
so the health caused the app, not the app
the health.

149
00:08:49,637 --> 00:08:50,987
I am confident on that one.

150
00:08:52,197 --> 00:08:56,257
You committed hard, so let me show you
exactly where it cracks.

151
00:08:56,837 --> 00:08:57,617
Watch the arrow.

152
00:08:58,317 --> 00:09:04,257
Reverse causation needs the effect, fewer
sick days, to literally produce the cause,

153
00:09:04,717 --> 00:09:05,477
opening the app.

154
00:09:06,137 --> 00:09:11,337
But taking fewer sick days does not reach
out and make someone download an app.

155
00:09:11,377 --> 00:09:15,707
The two are not the same event, just both
downstream of something else.

156
00:09:16,417 --> 00:09:21,337
That awkwardness is your signal to drop
reverse causation here.

157
00:09:21,377 --> 00:09:23,837
The strongest attack is the third
variable.

158
00:09:24,577 --> 00:09:28,257
Picture the kind of person who downloads a
wellness app the day it launches.

159
00:09:28,777 --> 00:09:29,417
Health-conscious.

160
00:09:29,937 --> 00:09:32,377
Already exercising, already sleeping well.

161
00:09:33,237 --> 00:09:37,697
That same conscientiousness is what makes
them take fewer sick days in the first

162
00:09:37,757 --> 00:09:38,057
place.

163
00:09:38,777 --> 00:09:44,237
So a common cause, the employee's existing
health habits, drives both the app use

164
00:09:44,617 --> 00:09:45,667
and the low sick days.

165
00:09:46,327 --> 00:09:49,617
The app pockets credit it never earned.

166
00:09:50,007 --> 00:09:51,197
And now I see it.

167
00:09:51,737 --> 00:09:56,057
So the credited weaken answer might not
even say the word cause.

168
00:09:56,517 --> 00:10:01,737
It might just quietly mention that the
early adopters were already the healthiest

169
00:10:01,817 --> 00:10:02,937
employees on staff.

170
00:10:04,190 --> 00:10:06,280
That is the trap inside the trap.

171
00:10:07,060 --> 00:10:11,820
Weaken answers almost never announce
themselves with causal vocabulary.

172
00:10:11,830 --> 00:10:16,880
They slip in a new factor, or note a
timeline, or describe a case where the cause

173
00:10:16,920 --> 00:10:18,840
showed up and the effect never did.

174
00:10:19,620 --> 00:10:23,580
You have to recognize the move, not wait
for the word cause.

175
00:10:24,500 --> 00:10:26,160
Now your turn, fresh argument.

176
00:10:27,100 --> 00:10:30,630
Neighborhoods with more bookstores have
higher average incomes,

177
00:10:31,220 --> 00:10:35,640
so the argument concludes that opening
bookstores raises a neighborhood's income.

178
00:10:36,620 --> 00:10:41,400
Take a beat, run the checklist, and tell
me which attack you reach for before you

179
00:10:41,440 --> 00:10:42,840
solve it.

180
00:10:43,649 --> 00:10:45,678
Alternative cause, the third-variable
flavor.

181
00:10:46,458 --> 00:10:49,157
Wealthier neighborhoods attract
bookstores.

182
00:10:49,698 --> 00:10:52,618
The wealth was already there and pulled
the bookstores in.

183
00:10:53,118 --> 00:10:56,218
So the bookstores did not create the
income.

184
00:10:56,228 --> 00:11:01,868
A common cause, the existing wealth,
explains both the stores and the high incomes.

185
00:11:03,447 --> 00:11:04,986
Now finish the self-explanation.

186
00:11:05,676 --> 00:11:09,316
Say why that wrecks the argument instead
of just nitpicking it.

187
00:11:10,307 --> 00:11:15,307
Because it offers a complete rival story
for the correlation that does not need the

188
00:11:15,367 --> 00:11:16,637
author's arrow at all.

189
00:11:17,327 --> 00:11:22,487
If the wealth is what draws the
bookstores, then the bookstores-raise-income claim

190
00:11:22,627 --> 00:11:24,207
loses its only support.

191
00:11:24,677 --> 00:11:29,207
The two things still move together, but
the author's explanation is no longer the

192
00:11:29,287 --> 00:11:30,087
one doing the work.

193
00:11:31,074 --> 00:11:33,054
And notice what you did not have to do.

194
00:11:33,774 --> 00:11:36,874
You never proved bookstores can't affect
income.

195
00:11:37,834 --> 00:11:40,334
One plausible alternative is enough.

196
00:11:41,114 --> 00:11:45,374
A weakener does not have to destroy the
argument, it only has to make the causal

197
00:11:45,454 --> 00:11:47,114
conclusion less likely.

198
00:11:48,283 --> 00:11:49,803
Now flip the job.

199
00:11:49,863 --> 00:11:53,463
If weakening pries doors open,
strengthening shuts them.

200
00:11:54,023 --> 00:11:56,223
Same three assumptions, opposite task.

201
00:11:56,763 --> 00:11:57,663
Two moves to know.

202
00:11:58,443 --> 00:12:01,183
Move one, rule out an alternative cause.

203
00:12:01,843 --> 00:12:06,363
Move two, show the effect tracks the
cause, present when the cause is present,

204
00:12:06,783 --> 00:12:08,223
gone when the cause is gone.

205
00:12:08,923 --> 00:12:13,083
Think of it as a control group, the cause
switched on and switched off.

206
00:12:14,398 --> 00:12:18,688
So strengthening is just closing the same
doors that weakening opens.

207
00:12:20,587 --> 00:12:21,607
That is the sentence to keep.

208
00:12:22,617 --> 00:12:24,897
Back to the streetlights for the on-off
test.

209
00:12:25,857 --> 00:12:30,357
Imagine we learn that on the one block
where a streetlight broke and stayed dark for

210
00:12:30,377 --> 00:12:34,967
months, burglaries climbed right back up,
and only on that block.

211
00:12:35,817 --> 00:12:37,577
Cause present, effect there.

212
00:12:38,357 --> 00:12:40,147
Cause absent, effect gone.

213
00:12:41,257 --> 00:12:46,057
That is the presence-absence test doing
real work, not just repeating that lights

214
00:12:46,077 --> 00:12:47,107
and safety go together.

215
00:12:48,051 --> 00:12:52,111
And to rule out the alternative cause, we
would want something like,

216
00:12:52,591 --> 00:12:56,071
the police patrols were exactly the same
as the year before.

217
00:12:56,411 --> 00:12:59,231
So patrols cannot be what explains the
drop.

218
00:13:00,030 --> 00:13:01,699
Now you are driving.

219
00:13:02,159 --> 00:13:06,839
That eliminates the rival explanation and
forces the credit back onto the lights.

220
00:13:07,699 --> 00:13:12,279
Here is the classic strengthen trap,
though, so pause and spot it before I do.

221
00:13:12,879 --> 00:13:15,039
Which of these does nothing for the
argument?

222
00:13:15,619 --> 00:13:20,319
Option A, burglaries also dropped on three
more streets that got new lights.

223
00:13:20,939 --> 00:13:25,779
Option B, the streets with new lights and
the streets without them had the same drop

224
00:13:25,799 --> 00:13:26,559
in burglaries.

225
00:13:27,465 --> 00:13:30,865
Option B does nothing, and worse than
nothing.

226
00:13:31,265 --> 00:13:35,755
If lit and unlit streets dropped the same
amount, the lights are not tracking

227
00:13:35,805 --> 00:13:38,145
anything, so that actually weakens it.

228
00:13:38,825 --> 00:13:43,395
And A, more streets where lights and the
drop appear together,

229
00:13:43,445 --> 00:13:47,125
that is just repeating the correlation,
which never proves the cause.

230
00:13:48,345 --> 00:13:52,085
Both halves are exactly the trap the test
sets.

231
00:13:52,785 --> 00:13:56,145
Piling on more correlation is not
strengthening.

232
00:13:56,885 --> 00:14:01,995
A real strengthener has to rule out a
rival or show the on-off pattern.

233
00:14:02,745 --> 00:14:08,055
Repeating that X and Y go together just
hands you the premise a second time and

234
00:14:08,125 --> 00:14:09,085
calls it support.

235
00:14:10,750 --> 00:14:13,920
Here is the whole episode crushed into one
heuristic.

236
00:14:14,680 --> 00:14:19,590
When you see X caused Y, attack the arrow
three ways.

237
00:14:20,460 --> 00:14:22,600
Is something else causing Y?

238
00:14:22,640 --> 00:14:23,920
Is the arrow backwards?

239
00:14:24,640 --> 00:14:25,560
Is it just a fluke?

240
00:14:26,320 --> 00:14:28,360
Attack the arrow three ways.

241
00:14:29,100 --> 00:14:33,480
That is the move, and it is the only
sentence you have to carry out of here.

242
00:14:34,471 --> 00:14:36,971
And to strengthen, I defend the arrow.

243
00:14:37,511 --> 00:14:42,151
Rule out the something else, and show the
effect switches on and off with the cause.

244
00:14:43,069 --> 00:14:44,637
So back to the cold open.

245
00:14:45,198 --> 00:14:48,208
The town swore the streetlights stopped
the burglars.

246
00:14:48,858 --> 00:14:51,958
But you can now see the patrols that went
up the same year,

247
00:14:52,517 --> 00:14:57,828
the statewide drop that lifted every town,
the broken-light block that tells the

248
00:14:57,958 --> 00:14:58,738
real story.

249
00:14:59,598 --> 00:15:02,578
The author saw a correlation and drew an
arrow.

250
00:15:03,418 --> 00:15:08,238
Your job is never to draw it for them, it
is to ask whether they earned it.

251
00:15:09,299 --> 00:15:10,779
One warning I want to keep.

252
00:15:11,279 --> 00:15:14,859
Don't diagram a causal claim like a
conditional.

253
00:15:14,899 --> 00:15:20,559
If I eat salty food I retain water sounds
like an if-then, but a cause-and-effect

254
00:15:20,599 --> 00:15:22,099
claim is a different system.

255
00:15:22,589 --> 00:15:26,519
I attack it with the three arrows, not
with sufficient and necessary.

256
00:15:27,556 --> 00:15:31,126
Hold onto that, it saves people real
points on test day.

257
00:15:31,856 --> 00:15:34,426
Between now and next time, one tiny drill.

258
00:15:35,236 --> 00:15:39,076
In any conversation today, catch one
because someone says out loud,

259
00:15:39,626 --> 00:15:44,136
then silently ask yourself, is that the
cause, or just the thing that came before.

260
00:15:44,996 --> 00:15:48,786
Train the reflex off the clock so it is
automatic on the clock.

261
00:15:49,856 --> 00:15:54,516
Next episode we stop attacking arguments
and start building from them.

262
00:15:54,596 --> 00:15:55,496
Inference questions.

263
00:15:56,036 --> 00:16:01,136
What must be true versus what is most
strongly supported, and why those two phrases

264
00:16:01,416 --> 00:16:03,636
send you to two completely different
answers.