Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Hi, Troy Sology listeners. It's Katie.
0:02
We have an opportunity for you to
0:04
help the show. Stick around until
0:06
the end of this episode, and I'll give you the details
0:08
on how to do that. Now
0:10
on to our show.
0:19
When it comes to the effect of coffee
0:21
on our health, some studies have linked consumption
0:23
to heart risk factors such as raised
0:25
cholesterol or blood pressure sensitive. those
0:27
who drank
0:27
six or more cups of coffee a day had
0:30
a twenty two
0:30
percent higher risk of developing
0:32
a cardiovascular disease.
0:33
After twenty t five cups of coffee
0:36
a day had no ill effects on your
0:38
arteries. The study out of Australia found
0:40
that two to three cups of coffee a day
0:43
is not only associated with a lower risk
0:45
of heart disease and dangerous heart rhythms,
0:47
but also with living longer.
0:55
It
0:56
seems as if every year there's a new study
0:58
on whether coffee is good or bad for
1:00
you. You've
1:01
probably seen other contradictory reports
1:04
on how vitamins or water consumption
1:06
or salt or sugar affect some other health
1:08
outcome you care about.
1:10
So why is it that we see so many conflicting
1:12
reports in the media? Why
1:14
is it so difficult to determine
1:16
whether something like coffee is good or bad
1:19
for, say, your heart?
1:21
It has a lot to do with the challenge of separating
1:23
out all the factors that influence our health.
1:26
Things like age, weight, height,
1:29
genetics, mindset, sleep
1:31
patterns, stress, exercise,
1:34
and, well,
1:35
you get the idea.
1:36
Health is the sum of many things. and
1:39
coffee is just one small piece
1:41
of a much larger puzzle. Teasing
1:44
out its influence is no easy feat.
1:46
In
1:46
this episode, we'll dig into a tool
1:48
for tackling a common mistake that affects
1:51
how we think about everything from
1:53
coffee to medicine, to
1:55
education. You'll hear
1:57
a colorful story from history that illustrates
1:59
how this tool and I'll speak with
2:02
renowned economist John list about
2:04
how the very same tool can
2:06
generally help us all think more
2:07
clearly and make better decisions.
2:10
and
2:13
I'm
2:15
doctor Katie Milkman, and this is Choiceology,
2:18
an original podcast from Charles Schwab.
2:21
It's a show about the psychology and economics
2:23
behind our decisions.
2:24
We bring you true stories involving
2:27
dramatic choices and then we explore
2:29
how they relate to the latest research and behavioral
2:31
science. We do it all to
2:33
help you make better judgments and avoid costly
2:36
mistakes.
2:52
Think of somebody with an
2:54
enormous powdered wig that goes several
2:56
feet above their head and maybe has a ship in
2:58
it. That's that's the time that we're talking about.
3:00
This is Mara. Hi, my name is
3:02
Mara Ratcliffe. I write books for
3:04
kids. usually about strange
3:06
and fascinating things in history that
3:08
have been forgotten.
3:09
This particular strange and fascinating
3:12
story takes place in France.
3:13
So this is not very
3:16
long before the French revolution when
3:18
Louis the sixteenth was in Paris and his
3:20
queen Marie Antoinette they
3:22
wore these very fancy
3:25
outfits, and the ladies of the court
3:27
had these giant hairstyles
3:29
The year was seventeen seventy
3:31
eight, and the French aristocracy was
3:33
about to discover an extraordinary German
3:36
physician.
3:37
This guy came to Paris from
3:39
Vienna, and everybody goes a little bit crazy
3:41
over him. His name was Franz
3:43
Mezmer, He's a very dramatic
3:46
kind of character. He's elegant and
3:48
mysterious. He thought he was pretty
3:50
important. He wears a powdered wig
3:52
and a fine coat of purple silk. and
3:54
he carries an iron wand. And
3:57
he says he's discovered this astonishing
3:59
new
3:59
force, a force called
4:01
animal magnetism. This
4:04
force worked on people the way magnets work
4:06
on metal. And he said it
4:08
was this invisible force
4:10
that you couldn't see or smell
4:12
or taste, but it was all over the universe
4:14
and it just flowed from the universe
4:17
into his body and then out through his
4:19
magic wand. and
4:21
he got this reputation for being
4:23
able to perform miracle
4:24
cures.
4:25
Messmer said I dare to flatter myself
4:27
that the discoveries I have made will
4:29
push back the boundaries of our knowledge
4:31
of physics as did the invention
4:33
of microscopes and telescopes for the
4:35
age preceding our own. So
4:37
he thought that he was probably
4:39
the most
4:40
important scientist in the world. It
4:42
might sound a little strange in today's
4:44
world, but in eighteenth century,
4:46
France, A phenomenon like animal
4:48
magnetism didn't strike people
4:50
as terribly far fetched. There
4:52
was a good reason why people
4:54
might have believed that
4:56
something like animal magnetism could
4:59
exist. So many unbelievable
5:01
things were actually happening. that
5:03
it was really hard to know what
5:06
to believe. It was hard to know what could
5:08
be true and what couldn't be true.
5:10
For instance, Antoine Lavoisier
5:13
was a famous French scientist, he's
5:15
known as the father of modern chemistry, and
5:17
he had just done experiments with
5:20
hydrogen and oxygen, to secure these
5:23
things that nobody can see or
5:25
smell or taste and yet
5:27
suddenly he's setting
5:29
fire to them and, you know,
5:31
what appears to just be air is actually
5:33
this invisible force.
5:35
If somebody said, hey, there's this
5:37
force out there. You can't
5:39
see it. You can't tell
5:41
that it's there, but it's there and it has
5:43
these powerful effects. It
5:46
was plausible
5:47
mesmer claimed he could use the invisible
5:49
force of animal magnetism
5:50
to cure any sickness.
5:53
And there was some evidence that it worked.
5:56
Wealthy patients flocked to mezmir and
5:58
eventually named this treatment after him.
6:00
They called it
6:02
mezmirism. So
6:04
pretty soon, everybody has anybody in
6:06
Paris wants to be mesmerized. So
6:09
dukes and countesses pull up it
6:11
Dr. Messmer's door in their fancy
6:13
carriages and they disappear
6:15
into this room. He would be doing
6:17
this music
6:19
to create a spooky atmosphere. Others
6:22
use velvet curtains and the lights
6:24
would be low and it would be kind of
6:27
airless and everybody
6:29
sits around this sort of big
6:31
wooden tub with iron rods.
6:33
It's this very odd looking thing and looks
6:35
very scientific. and so people
6:37
would come into this very
6:39
dramatic scene with
6:41
this very dramatic guy and then
6:43
he's staring into their eyes and he's waving
6:46
hands and they start having
6:48
all these reactions. Shrieks, tears,
6:50
hiccups, and excessive laughter. People
6:53
were fainting and screaming and
6:55
falling all over the
6:55
place. And then they would
6:58
say that they felt better. And
7:00
maybe they did but
7:01
some people were not fans of
7:03
this new trend. Not everybody
7:06
was absolutely delighted by what was going on
7:08
with doctor Messmer and people whose noses
7:10
were really out of joint were the doctors
7:12
because nobody wanted their treatments
7:15
anymore. So they went and complained to
7:17
the king.
7:18
King Louis the sixteenth decided
7:20
to establish a commission to
7:22
investigate this new
7:23
medical phenomenon, and
7:24
he appointed a famous outsider to lead
7:27
it. His name was
7:29
Benjamin Franklin. Benjamin
7:30
Franklin who was
7:32
a celebrity in France.
7:35
He was very respected by
7:37
all the best scientists in France and at the
7:39
same time who's super popular with the
7:41
people. Ben Franklin had been in
7:43
France for
7:43
a couple of years. where he had
7:45
helped achieve official diplomatic recognition
7:47
for the United States in the revolutionary
7:50
war. So Franklin
7:52
was a pretty old man at this point
7:54
he had gout, he had kidney stones,
7:56
and he was not able
7:58
to get into a carriage and go
8:00
jostling over the cobblestones in Paris.
8:03
to go see mezmir. He was living
8:06
outside of Paris in the country. So
8:08
he asked for mezmir to come to him and
8:10
mezmir refused because
8:11
Messmer was in
8:14
his own eyes an extremely important person,
8:16
and
8:16
he wasn't gonna go to him. But
8:19
Messmer's
8:20
second in command, Charles Des escalation
8:23
went out there to demonstrate
8:26
for Franklin and the other members of the commission
8:28
how mesmerism worked. Franklin,
8:30
of course, the first thing he did
8:33
was had it tried on himself. It
8:35
must have been quite the scene with
8:37
Benjamin Franklin submitting to this strange
8:39
mesmerism from Charles Deslon.
8:41
It
8:41
sort of makes some woo woo gestures at
8:44
you, either with his hands or his
8:46
wand, Ben
8:46
Franklin and some of the members of the commission
8:49
just stand there and say, I don't
8:51
feel anything. And so the word
8:53
got back to doctor Messmer and doctor Messmer
8:55
said, well, there must be something strange about this
8:57
American because it's not
8:59
working on him for some reason.
9:01
At
9:01
this point, Ben
9:02
Franklin observed Charles Dezelan mesmerizing
9:04
some regular patients of doctor
9:06
Messmer's. And the people
9:09
would scream that they felt like their body
9:11
was burning all over and, you
9:13
know, they would fall down in a fade and all
9:15
this kind of thing. So Franklin and
9:17
the Commission were at a crossroads. On
9:19
one hand, the procedure seemed to
9:21
work on some patients. On the
9:23
other hand, when Charles Des laude
9:26
attempted to mesmerize Franklin and the
9:28
other commission members, they felt
9:30
nothing. Benjamin
9:31
Franklin needed a way to figure out what
9:33
was going on. Franklin
9:35
was skeptical in the
9:37
first place, but he kept an open
9:39
mind even though he was the world's most
9:41
famous scientist, he was open
9:43
to things. He just wanted to find out, well,
9:45
is it real or is it not real?
9:48
he observed what was happening to
9:50
himself and other people. And
9:52
he asked himself, well, could this
9:54
be in their minds? So
9:57
he's created this hypothesis
9:59
and he needs to figure out, well, how can I
10:01
test that? That was when
10:03
Franklin and the other members of the commission came
10:05
up with the idea of blindfolding
10:07
the patients so that
10:09
they wouldn't know what was being
10:11
done. This decision to
10:13
blindfold the patients was important to
10:15
the tests. into the future of
10:17
scientific research. So one
10:19
of the tests that Franklin
10:21
ran was they took this young boy
10:23
who was one of doctor of mesmer patients
10:25
and they blindfolded him and they took him outdoors
10:28
into a grove of apricot
10:30
trees. This boy was
10:32
supposed to be especially sensitive.
10:34
to animal magnetism. They
10:36
told him that one of the
10:38
trees had been mesmerized, meaning
10:41
this tree had had a wand
10:43
waved over it and had had
10:44
this invisible force funneled
10:46
into it. And they said, find the
10:48
tree. It's been mesmerized. he
10:50
started moving from tree to tree.
10:52
And first he's coughing,
10:55
and then he complains of a
10:57
headache. And then he says, 0II
10:59
feel really dizzy. It must be getting
11:01
closer. And finally, he
11:03
gets to the last tree and he just
11:05
faints dead away. It
11:07
was a dramatic moment, but
11:09
In fact, he hadn't gone
11:12
anywhere near the particular avocado
11:14
tree that had been quote
11:16
unquote mesmerized. It
11:19
seemed that the boy couldn't detect where the
11:21
mesmerized tree was at all if he was
11:23
blindfolded. Ben Franklin
11:25
repeated the experiment on several other
11:27
patients. There was this one patient
11:29
who reacted very strongly
11:31
when he wasn't blindfolded and
11:33
was being mesmerized. So they blindfolded
11:35
him and Franklin
11:37
said to him, hey, you're being
11:39
mesmerized right now. Can you feel it?
11:41
And he said oh, yes. Yes.
11:43
And in fact, Charles Des escalation
11:45
wasn't even in the room at that time.
11:47
And so then they had Charles come
11:49
back into the room very quietly
11:52
without this patient knowing he
11:54
was there. You can imagine this
11:56
patient just standing there blindfolded while
11:59
Charles Des escalation is just pulling out
12:01
all the stops, going around him and waving
12:03
his hands, and pointing his
12:05
wand, and staring at him in a
12:07
mezamer kind of way. And the guy
12:09
has no idea that he's there and he's just not responding
12:11
at all, which was
12:14
really a shock because Desilne
12:16
was a true believer in mesmer
12:18
and he expected him to
12:20
respond as people normally did. It
12:22
had never occurred to him that they were
12:24
reacting that way because they expected to.
12:26
Ben Franklin had his answer.
12:29
Through multiple blindfolded
12:32
tests like that, they were able to
12:34
show that it wasn't actually
12:36
what Charles was doing that mattered. It
12:37
was what the patient believed. All
12:40
of these patients who had responded
12:42
so dramatically to mesmerized
12:44
treatments once they're blindfolded that
12:47
just went away or their response was
12:49
clearly not a direct
12:51
result of the
12:51
treatment. They proved that
12:53
animal magnetism as such didn't really
12:56
exist, but that there was
12:58
something going on that came out of
13:00
the
13:00
patient's minds rather than in an
13:03
invisible force that was flowing out of a
13:05
wand. This may
13:06
be the first time a blind
13:09
trial was used to test a scientific
13:10
hypothesis. This
13:12
kind of test
13:13
would become the basis for proving that
13:16
one thing causes another.
13:18
Instead of simply examining
13:20
whether mesmerizing people seemed to
13:22
work.
13:23
Franklin realized it was critical to
13:25
assign some people to be mesmerized and
13:28
others not to be. and it
13:29
was crucial that people not know which
13:31
group they were in. This
13:33
was in
13:34
essence a controlled scientific experiment.
13:38
Today
13:38
we run these kinds of tests with far more
13:40
people, and we use random number
13:42
generators to decide who will get a treatment,
13:45
like being mesmerized, and who will
13:47
be in a control
13:47
group. In
13:48
this case, merely being told they're
13:51
mesmerized. The procedure
13:53
allows us to tease out cause and effect.
13:56
if there different outcomes for the people in the
13:58
treatment and control groups. Well,
14:00
then we know it's due to the treatment.
14:03
If
14:03
not, well, It
14:05
was all in our head. The
14:07
scientific method was well established
14:09
already. The idea that
14:11
you observe situation and then you ask a
14:13
question and then you make a hypothesis
14:15
and then you you test that
14:17
hypothesis. But what
14:19
had not been done before that Franklin
14:22
invented here was the blind protocol,
14:24
the blind test, and that
14:26
was really an important
14:28
development Ben Franklin
14:30
and the Commission published their findings in
14:32
a report. It was an
14:34
immediate bestseller. Twenty
14:36
thousand copies were snatched up right
14:38
away. And so
14:40
mezmer who had been a celebrity in a good
14:42
way was now sort of infamous
14:44
and mocked. He was the subject of parodies.
14:47
was this one stage play where
14:49
they show mesmerizing the patient and
14:51
the patient says, please doctor
14:53
tell me does animal magnetism really
14:55
do any good and the guy playing mezmer
14:57
jingles some coins and says, well, I can
14:59
assure you it does me a lot of good.
15:01
So this report of completely
15:04
ruined his reputation. He fled
15:06
Paris. He went back to
15:08
Germany where he spent his
15:10
last days with at Canary which would
15:12
wake him up every morning by landing on his
15:14
head. Mezmarism fell out of
15:16
favor. Mezmarism got
15:19
such a bad reputation that
15:21
eventually when scientists wanted
15:23
to go back and work with it somewhere, they
15:25
had to rename it hypnotism. so
15:28
that it wouldn't be associated with this
15:30
big scandal. But the
15:32
real lasting impact of this story
15:34
comes from Franklin's use of
15:36
blind trials. The blind
15:38
protocol is basically the gold standard
15:40
now for any kind of new medication.
15:43
You have to test every medication
15:45
against a placebo. It needs
15:47
to be blind, meaning the
15:49
patient needs to not
15:51
know whether they're getting the placebo
15:53
or the vacation. And that way,
15:56
if the medication has better effects,
15:58
then you know it's not
15:58
just because
15:59
they believed that it was going to help.
16:02
And today, we have the double blind protocol,
16:04
which is even better in which
16:05
the doctors who are giving
16:08
out
16:08
the medication don't know whether they're get giving
16:10
the medication or placebo so that they can't
16:13
have an impact on the results
16:15
either. It's what the FDA requires
16:17
before any new medication would come on
16:19
the market. It's super important.
16:23
Mara Rockliff is the author of several
16:25
historical books for children and teens,
16:27
including the award winning mesmerized, how
16:30
Ben Franklin solved a mystery that
16:32
baffled all of France. You can
16:34
find links in
16:34
the show notes and at schwab dot com
16:36
slash podcast.
16:44
The
16:44
story of Benjamin Franklin debunking
16:47
Franz Messmer's discovery of
16:49
animal magnetism may be the
16:51
earliest recorded example of a blinded
16:53
experiment. And
16:54
today, I want to focus on the amazing
16:56
power
16:56
of experiments like Franklin's to
16:58
cut
16:58
through the challenges we usually face
17:00
when we try to understand cause and effect
17:02
in the world. It's easy to
17:05
be tricked just like Messmer's
17:07
patients into
17:07
thinking one thing causes another
17:09
when in fact it doesn't.
17:11
For example, maybe you
17:13
heard for years that coffee was great
17:15
for some health outcome, only to read
17:17
a headline later that, oops,
17:20
that wasn't true.
17:21
What happened? Well, the
17:23
kinds of people who drink coffee aren't
17:26
exactly the same
17:26
as other people. If
17:28
they're just a little say
17:30
wealthier than average, It's easy to
17:32
form the false impression that
17:35
coffee leads to great health outcomes
17:37
when it's actually wealth that's so good
17:40
for you.
17:40
And when researchers around to doing an experiment,
17:42
cause and effect can be
17:45
untangled. In
17:47
an experiment, some people are randomly assigned
17:50
to drink
17:50
coffee and others aren't, and then
17:52
health outcomes are measured.
17:55
Differences in things like wealth are all
17:57
washed away by a random assignment experiment
17:59
because wealthy people are just as
18:02
likely to be randomly assigned to
18:04
drink coffee as to abstain from
18:06
it. And
18:06
so you're left
18:07
with the truth, which
18:08
is often that there's no causal
18:10
relationship between a food or
18:12
drug that was believed to have superpowers.
18:14
and the outcomes we seek. I remember
18:16
one headline pronouncing that abstaining
18:18
from alcohol entirely leads
18:20
to a shortened lifespan.
18:23
maybe. But
18:24
a much easier explanation is that
18:27
people who abstain from alcohol completely
18:29
do so because they're a little different.
18:32
Maybe a decent
18:32
chunk of abstainers have a health issue
18:34
that prevents
18:35
them from drinking, and that's the reason
18:37
they, on average, die younger.
18:40
A
18:40
full proof way to get around all this
18:42
mess is with the experimental method.
18:45
My
18:45
next guest is a renowned economist
18:48
and his area of expertise is
18:49
acting truly experimental economics.
18:52
That means he uses the experimental
18:54
method in essentially all of his
18:56
work. John list
18:58
is the chief economist at Walmart
19:00
and the Kenneth C Griffin distinguished
19:02
service professor of Economics at
19:04
the University of Chicago.
19:06
Hi, John. Thank you so much for joining
19:08
me today. Hey, Katie.
19:10
Thanks so much for having me. How's
19:12
everything going? Everything is great.
19:15
Okay. I'm gonna dive right in because I have
19:17
so many questions for you today.
19:21
So
19:21
my first question is if you could just describe what
19:23
it
19:23
means for two variables to be
19:26
correlated with one another. Howard Bauchner: Yeah, that's
19:27
a good question. I think if you
19:30
ask people To
19:31
define correlation, if you ask thirty people,
19:33
you'd probably get thirty different answers.
19:36
My preferred definition is
19:39
two variables that move
19:41
together, either directly,
19:43
like when one goes up, the
19:45
other goes up, like, crime and ice
19:47
cream sales, say, exactly
19:50
or like ice cream sales
19:52
and drownings. That would
19:54
be something that one one goes
19:56
up, the other goes up, And correlation
19:59
can
19:59
also be when one goes up, the
20:02
other goes down. Something
20:04
like when the price of a
20:06
good goes up, the
20:07
quantity demanded in economics goes
20:10
down. Now, I would say that's causal.
20:13
But, of course, correlations
20:15
simply means that two variables are moving together. Great.
20:17
I love that you brought up causality
20:20
because that was my next question. Could
20:22
you describe what it means for two
20:24
variables to be causally related, meaning one
20:26
causes the other. Causality is
20:28
a special form of correlation where
20:32
when one variable moves,
20:35
that
20:35
causes another variable
20:38
also to move. So
20:40
again, it could be one variable
20:42
goes up that could cause
20:44
another variable to go
20:46
up. or when one variable
20:48
goes up, it causes another
20:51
variable to go down. A
20:53
causal relationship now is
20:56
fundamentally different than
20:58
a relationship that is merely
21:01
correlational. And course,
21:03
a huge amount of your work and
21:05
mine too is about trying to
21:07
figure out what's
21:08
causal and what's not. and
21:10
I'm gonna get there in just a second.
21:13
But I wanted to ask
21:15
why you think it is so hard
21:17
to disentangle causation and correlation
21:19
and why people get them mixed
21:21
up?
21:21
Yes. So I think it's hard because
21:23
in many cases,
21:25
you don't know
21:27
assignment. So what
21:28
I mean by that is if
21:31
you ask yourself does
21:33
head start work? Head head
21:35
start being the early childhood education
21:37
and health program for low income
21:39
children and families. Exactly. So
21:41
what head start did early
21:43
on was they looked at
21:45
outcomes third
21:46
grade test scores, kindergarten readiness,
21:49
etcetera, of kids who went
21:51
to head start versus
21:53
kids who did not go to head
21:55
start. And
21:56
what they reported was that
21:58
kids who went to head start
22:00
had
22:00
better outcomes. So
22:02
they argued that head start
22:05
is good
22:05
for kids.
22:07
now Now what
22:08
you have mingle in
22:10
here is that parents
22:12
who really care about their child's
22:15
education are more
22:17
likely to put their child in head
22:19
start. So that's the
22:21
assignment mechanism that I'm talking
22:23
about. It's actually parents
22:26
choosing or kids selecting
22:28
into that particular
22:30
program And in many cases, it's
22:33
that that happens to be the most
22:35
important not bad start. People
22:37
think just because there's
22:39
a relationship there, that
22:41
it's causal. So it kinda
22:42
makes sense. Yeah. That's such a
22:45
great example of a situation where
22:46
a random assignment experiment could
22:48
help clarify things. So the
22:51
beautiful aspect of randomization is
22:53
you can go into
22:55
a really dirty environment.
22:57
And
22:57
as long as you control
23:00
the assignment mechanism, and what I
23:02
mean by that is I
23:05
control who
23:05
goes into control, And
23:08
who goes into treatment? If you wanna think
23:10
about a medical trial, you can. If
23:12
you wanna think about an early childhood
23:14
program, you can. So who
23:16
gets had start, who doesn't, as long as you control
23:18
that assignment. Exactly.
23:20
You have a bunch of parents who say,
23:22
I want my kid in head start,
23:25
and let's say I'm oversubscribed so
23:28
that I wanna be
23:30
fair, so
23:30
I use a lottery system.
23:32
and I randomly put some of them
23:34
in control and some in treatment. I
23:37
I realized the world is a
23:39
messy environment But what's nice
23:41
about randomization is it
23:43
balances that dirt
23:45
across
23:45
the treatment and control groups
23:47
So then when you difference off the
23:50
outcomes, you also
23:52
difference off the dirt because the
23:54
dirt is equally represented in
23:56
each of the two groups. Meaning like the
23:58
the kids that, you know, the lottery
23:59
assigned higher numbers and lower numbers
24:02
whose parents
24:02
all wanted them to be in head
24:04
start. The same number kids have older parents
24:06
and younger parents and kids who have high IQs and
24:08
low IQs because it's just a flip of a
24:10
coin. There's nothing different about the two
24:13
groups. a hundred percent. So you have ambitious
24:15
parents represented in both groups.
24:18
You have siblings of
24:20
two or two boys and two girl siblings
24:22
represented in both groups.
24:24
So these are the background features
24:26
that might matter And in
24:28
many cases, we want to measure those as
24:31
well because those will give us an
24:33
indication of You
24:36
know, who who does the program work the best
24:38
for? And are there
24:40
certain moderators of
24:42
the relationship that we need to be
24:44
aware of, that then when we
24:46
roll it out to the big time, we scale
24:48
it up, we sort of know which
24:50
types of families our program
24:52
work the best for? Who should we scale to?
24:55
And which kinds of families that really doesn't
24:57
work that
24:57
well for? And maybe then we need to scale
24:59
a different program
25:00
to those types of families. So
25:03
the nice thing
25:04
here that we're talking about is, first
25:07
of all,
25:08
I have an approach here that you
25:10
and I and many others are
25:13
doing called experimentation, and
25:15
it allows us to control
25:17
the assignment mechanism, which allows
25:19
us to establish an
25:22
estimate that has internal validity.
25:24
Do
25:24
have a favorite research study you've run
25:27
that doesn't
25:27
tangles correlation from causation? Oh,
25:29
gosh. I
25:30
would say everyone. So
25:33
let's think about charity. A lot of
25:35
your
25:35
listeners might be interested in charitable
25:38
giving. And
25:39
and what
25:41
happened when I started my
25:44
own research in the late nineties
25:46
on charitable giving is that there
25:49
was sort of this bible
25:51
that was written by Kent Dove.
25:53
And the bible basically
25:56
said that when
25:56
you're trying
25:57
to raise money, you
26:00
should
26:00
use a matching
26:03
grant And
26:03
what that basically means is we all hear this on NPR
26:06
when they fundraise. If you give a
26:08
one hundred dollars today, we
26:10
will match it with a hundred dollars from
26:13
an anonymous donor. Okay.
26:15
So the
26:18
fundraising Bible argued that
26:21
if you use a two:one
26:24
match, that will be better
26:25
than a one:one match And
26:28
if you use a three to one
26:30
match, that
26:30
will be better than a two to one match.
26:32
So three to one's really good. Right? You
26:34
give a hundred bucks. It's matched for
26:36
three hundred bucks. Now,
26:39
I talked to fundraisers back then
26:41
and I said, what is
26:42
the empirical evidence? And
26:45
they would show me evidence that
26:47
was, you know, in some cases, like
26:49
around Christmas time, they do
26:51
three to one, worked really
26:54
well, whereas in the summer, they
26:56
did one to one, and it didn't
26:58
work so well. And I said, well, do you ever
27:00
have data in the same time period?
27:03
They said, don't any studies like that.
27:06
So I tried
27:06
it. I worked with Dean Carlin
27:10
from Northwestern. And
27:11
Dean and I decided to help a
27:14
charitable organization raise money.
27:16
And we wanted to
27:17
test the theory about one
27:19
to one versus two to one versus three to one.
27:21
So what we did is we took
27:23
thousands of households and
27:26
put some in a control group, which
27:28
just received a letter with no
27:30
match. And then another group was one
27:32
to one, another group was two to one, another group
27:34
is three to one. And people were randomly
27:36
assigned to those groups. Yes. And we find
27:38
kind of two things that jump out.
27:40
First, having a match
27:42
matters a lot. So
27:44
if you just have match money
27:47
available, you raise more money.
27:49
And we can say that in a
27:51
causal sense, because we found that
27:53
the one to one, two to one and three to
27:55
one groups raised a lot more
27:57
money than the control. Okay.
27:59
Now
27:59
what about the one to one versus two to one
28:02
versus three to one? What we find
28:04
is that makes no
28:06
difference. So
28:07
the one to one raises
28:09
the exact same amount of money as
28:11
the two to one, which raises the exact
28:13
same amount of money as the three to
28:15
one. So what we can say
28:17
now is what they were telling
28:19
us before
28:21
was a correlation and they
28:23
were finding a result because of Christmas
28:26
cheergiving. The three to one
28:28
was working better in
28:30
December because a
28:31
lot of people give more anyway,
28:33
and they never really had a control
28:36
group to compare their three to one
28:38
width. So now we can say in a
28:40
causal way that
28:41
a higher match rate does not influence
28:43
giving patterns, but
28:46
having a match does. And
28:48
because we used a field
28:50
experiment can make a strong
28:52
causal statement. Yeah. I love that. That's a great
28:54
example, and it's one that really matters. Right?
28:56
Because we don't want to
28:58
be matching three to one and
29:00
trying to raise that kind of extra
29:03
capital when it's actually not necessary
29:05
to motivate our donor base if we're
29:08
an organization. No,
29:09
a hundred percent. So really, you're just throwing
29:12
away,
29:14
let's say rewards. that
29:16
you don't realize it, but
29:18
they're not helping you. You could take the three
29:20
to one dollars, give everyone one to one,
29:22
and then you can use those dollars for
29:25
for a new drive. And those
29:28
dollars, the parts go a long
29:30
way. So what
29:30
that means is I can make
29:32
a strong causal statement If
29:35
I understand the assignment
29:37
mechanism, I need a few other
29:39
assumptions too, like compliance
29:41
and attrition. Those are our exclusion
29:43
restrictions in the experimental world.
29:45
After those are in place, I
29:47
can be
29:48
confident that I'm estimating a causal
29:50
relationship. I also find it.
29:51
I'm curious if this is true for you
29:54
too that once you learn to think this
29:56
way, it helps you be more skeptical
29:59
of
29:59
the information that
29:59
others are feeding to you. And and it's
30:02
easier to poke holes and when someone's
30:04
giving you use
30:06
full data or useful information or
30:09
information that they've merely constructed to
30:11
align
30:11
with their goals.
30:14
So I'm sort of curious to what
30:16
extent doing this kind of work has changed the way you think about
30:18
the information the world is feeding you?
30:21
No. I think you're a hundred percent
30:23
correct. Another side
30:26
benefit of understanding
30:28
correlation versus causality
30:30
is that it's really
30:33
much easier for
30:35
you to understand what's happening
30:38
in the world. You could ask yourself,
30:40
what are the incentives that the
30:42
person has who's giving
30:44
me the information or who
30:46
has generated the information to they
30:48
have certain incentives to
30:51
give me a particular result.
30:54
If so, you should
30:56
think twice whether
30:57
the decision making is to vote
30:59
for a particular candidate or to
31:02
believe the information that
31:04
you're receiving to
31:06
think about is that truly a
31:08
causal result or is there a lurking
31:11
variable? I think all of this really
31:13
helps make you a better
31:15
decision maker as well.
31:16
That is a wonderful place I
31:18
feel to wrap. John, thank you so much
31:20
for taking the time to talk to me today. I
31:22
really appreciate it. Katie, it was so
31:24
great to be here I can't wait to come back. Job
31:27
list is the Kenneth
31:29
C Griffin distinguished service professor
31:31
of Economics at the University
31:33
of cargo. He's also the chief
31:36
economist for Walmart. You can
31:38
find a
31:38
link to his terrific new book, The Voltage
31:40
Effect how
31:42
to make good ideas great and great ideas
31:45
scale in the show
31:45
notes and at schwab dot com
31:48
slash podcast. Whether
31:51
you're interested in better understanding correlations between
31:54
market and economic data or just
31:56
want to make smarter financial decisions,
31:59
say around your
31:59
own charitable giving. Check
32:02
out
32:02
the financial decoder podcast.
32:04
It's a
32:04
great resource for digging into the
32:06
financial implications of the phenomena we
32:08
or here on Choiceology. You can
32:10
find it at schwab dot com slash
32:13
financial decoder or wherever you get
32:15
your podcasts. As
32:17
John List mentioned, becoming
32:20
more discerning about the distinction between
32:22
correlation and causation can help
32:24
you better understand the world and make
32:26
stronger decisions. When I teach my
32:28
wharton MBA students about the
32:30
experimental method and its power to help
32:32
disentangle correlation from causation,
32:34
The
32:34
first thing I suggest is that they
32:37
start greeting causal claims in news
32:38
headlines and from friends and colleagues
32:41
with a healthy dose of skepticism.
32:43
When
32:43
someone tells you that eating garlic can prevent headaches
32:45
or that owning more
32:46
books will improve your kid's lives.
32:49
No matter how plausible their
32:51
story, it's important to ask yourself,
32:53
could there
32:54
be some other explanation for this
32:56
besides a causal one?
32:58
Next,
32:58
if the claim is important enough,
33:01
I ask, what
33:02
kind of evidence would convincingly prove
33:04
this is true? The ideal
33:06
evidence, of course, would be experimental.
33:09
and sometimes you'll find it. After
33:12
all, experiments are how
33:14
doctors test new vaccines and medications.
33:17
Increasingly, experiments are being used
33:20
by economists and companies
33:22
to test everything from the value of
33:24
charitable matching campaigns
33:25
to micro finance.
33:27
In
33:27
fact, the economic Nobel
33:30
Prize in twenty nineteen was
33:32
awarded to a group of development economists
33:34
for their new experiment based approach to
33:36
fighting global poverty. And
33:39
pioneers
33:39
like John List are bringing experiments
33:41
inside big companies like Walmart.
33:44
to improve decision making. But experiments
33:46
are still rarer than they should be in business
33:48
and policy making given the importance
33:50
of understanding cause and effect.
33:53
Experiments can be
33:54
costly and complex, but it's
33:56
often worth the effort to determine what's
33:59
real and what's
33:59
not when you're facing high stakes
34:02
decisions. A key takeaway is that you
34:04
should constantly be on guard for correlations
34:07
misrepresented as causal relationships. Just
34:10
remind yourself how easy it is to be mesmerized and
34:13
don't fall for it.
34:18
You've been
34:24
listening to
34:26
Choiceology. an original
34:28
podcast from Charles
34:30
Schwab. If you've enjoyed the
34:31
show, we'd be really grateful if you'd leave
34:33
us a review on Apple
34:36
podcasts. You can also follow
34:36
us for free in your favorite podcasting app.
34:39
And if you want more of the kinds
34:41
of insights we bring you on
34:43
choice ology about how to improve your decisions. You
34:45
can order my book, how to change,
34:47
or sign up for my monthly newsletter,
34:49
milkman delivers at katie
34:52
milkman dot com slash
34:53
newsletter. That's it for this season. We'll have new episodes
34:55
for you in early twenty twenty
34:57
three. I'm doctor
35:00
Katie Milkmann,
35:01
Talk to you soon.
35:08
For disclosures, See the
35:10
show notes, or visit schwab dot com slash podcast.
35:13
As I mentioned at the
35:15
beginning of this episode, we're looking for
35:17
a little help with trisology. We're
35:19
launching a listener survey, so you can share your thoughts
35:21
on the show and help us better understand
35:23
our audience. Check it out at schwab
35:26
dot com slash
35:28
podcast survey.
35:28
We'd really appreciate you taking a few minutes to provide feedback. That's
35:32
schwab dot com
35:32
slash podcast survey.
35:34
Thanks, and we hope to hear
35:36
i'm healthier for me from you.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More