Correlation Arguments
(Problems printing? Click here.)
Causal Claims
Causality is basically a relationship between one object or
event and another object or event. If we were to say that A causes B, we
would mean that there is a relationship between A and B such that whenever
A occurs, we should expect B to occur also. Causal claims are therefore
conditional predictions. They are predictions about what will happen next
IF something else has just happened. Causal arguments are therefore
attempts to get people to believe in these predictions. Causal arguments
cannot prove that anything other than a relationship exists. No
causal argument can prove that A exists or that B exists.
For this reason, causal arguments can only establish relationships between
known entities. If we don't already know that A exists and that B exists,
we cannot make a causal argument about A and B. (We will cover arguments
that prove things exist in a later chapter.)
Here are some causal claims.
a. Tobacco consumption, whether chewing or smoking,
causes cancer.
b. Abolition of the fairness rule caused and continues to cause the
American population to become more and more conservative.
c. The way America administers welfare tends to keep people on welfare
even when they would like to become independent.
d. Moderate drinking is actually better for you than not drinking at
all.
e. Rain causes accidents in Southern California
f. Rain causes more accidents when it comes after sunny days than it
does after rainy days.
Notice that none of the above statements are arguments.
Arguments come with supporting reasons, which are not present in
any of the above sentences. Thus, while some of the claims given above may
seem plausible to you, none of them as yet come with any reason
for you to believe them. This means that if you believe any of
these claims you are either taking my word for it (which you shouldn't do)
or you are believing on the basis of a feeling, (which you definitely
should not do!)
Similarly, neither of the following more elaborate claims is an argument.
g. Car accidents are mostly caused by accident
goblins, who are surrounded by a magic aura that causes drivers
to loose control of their vehicles. Accident goblins don't like hot
concrete or asphalt, but they love to dance in the rain, so when it
rains in Southern California, the accident goblins all come out to dance
in the rain. This means that on rainy days there are far more accident
goblins on the roads than usual, so of course the accident rate goes up.
h. The increase in accidents on rainy days is caused by Hydrogenic Ion
Depeletion. The speed and level of electrochemical activity in the brain
is influenced by the availability of positive ions in the bloodstream to
act as "electron sinks" facillitating axon recovery after nerve impulse
propagation. (This is the time an axon takes to get ready to carry the next
nerve impulse.) The main source of positive ions for this purpose is
atmospheric oxygen and nitrogen ions, which enter the bloodstream
through the lungs along with the far more plentiful non-ionic oxygen.
Free ions are very susceptible to surface effects, and are attracted to
the surfaces of raindrops as they fall. This depletes the atmosphere
of positive ions because it removes them from the air faster than new
ones can be made by sunlight. Although this condition is not dangerous
for animal life, it does result in a small but measurable slowing of
brain activity, with a concomittant degradation of judgement
and increase in reaction times.
The reason that I don't consider either of these statements to be arguments
is that neither of them comes with any reason to believe in any
part of the story. No evidence is given to support the existence of
accident goblins or hydrogenic ion depletion. Rather, the claim is just
elaborated with the addition of details, but details are not evidence.
It's true that one or other of these explanations may feel
plausible because it feels like the kind of explanation you're
used to, but a feeling is no reason to believe something.
Correlations
A "correlation" is a relationship between two kinds of event. If two types
of thing teend to occur together, they are correlated. For instance, in
most public places, clothes and people are correlated. At public parks,
for instance, we rarely see clothes without people, or vice versa. A
causal claim is basically a claim that, whether or not two types of event
were correlated in the past, they will be correlated in the future. The
following are all examples of claims about correlations.
i. Groups composed entirely of smokers have been found
to contain more cancer cases than groups of nonsmokers.
j. After the abolition of the fairness rule, the American population
began a slow but stready and continuous drift to the right.
k. People who try to get off welfare in America succeed less often than
people who try to get off welfare in European countries.
l. Moderate drinkers are healthier than nondrinkers.
m. The accident rate always goes up whenever it rains in Southern
California
n. The first day of rain after a dry spell always has more accidents
than later rainy days. This difference is more pronounced when rain
follows a _long_ dry spell.
The following are claims that do not involve correlations.
o. Tobacco is harvested in late June, early July, which
means that the astrological sign of all cigarettes is Cancer.
p. The Fairness Doctrine was abolished by Ronald Reagan, a far right
wing President.
q. Welfare officials like their jobs, and make more money when there's
more people on welfare.
r. Nondrinkers deprive themselves of the pleasures of intoxication,
noncoordination, somnambulation and regurgitation.
s. I aquaplaned once. Woooooooo! It was fun.
t. If I was an accident goblin I'd get the hell off the roads when the
accidents start happening.
Now there's two things I want you to think about here. The first thing is
that in every one of the first group of statements there's one type of
thing whose presence (or absence) seems to vary with the presence (or
absence) of another kind of thing. The second thing to think about is to
ask yourself whether any of the second group of statements gives any logical
reason to believe any of the causal claims
Exercise 1. If any of the claims ("o" through "t") made
immediately above were true, would any of them logically support
any of the causal claims ("a" through "f" made at the top of this chapter?
Causal claims can only be justified on the basis of established
correlations. It is often said that "correlation does not prove
causation." This is completely and utterly false.
Correlation is the only thing that ever proves
causation. It's true that not every correlation proves
causation, but every causal relationship that has ever been
proved, has been proved on the basis of a correlation . (Thus
the claim that "correlation does not prove causation" is logically
analogous to the claim that "evidence does not prove guilt" in a court of
law.) The trick here is to figure out which correlations prove causation,
and which don't.
For an illustration of this idea, see this
cartoon
But also read about some clearly spurious correlations www.tylervigen.com/
A correlation argument is an argument that attempts to establish a causal
relationship by pointing to a particular kind of correlation that is
claimed to exist between two types of event. For instance, the following
are all correlation arguments.
aa. Groups composed entirely of smokers have been found
to contain more cancer cases than groups of nonsmokers, so tobacco
consumption, whether chewing or smoking, causes cancer.
ab. Abolition of the fairness rule caused and continues to cause the
American population to become more and more conservative. We know this
because, after the abolition of the fairness rule, the American
population began a slow but stready and continuous drift to the right.
ac. People who try to get off welfare in America succeed less often than
people who try to get off welfare in European countries, from this it
follows that the way America administers welfare tends to keep people on
welfare even when they would like to become independent.
ad. Moderate drinkers are healthier than nondrinkers, so moderate
drinking is actually better for you than not drinking at all.
ae. Rain causes accidents in Southern California, because the accident
rate always goes up whenever it rains in Southern California.
af. The first day of rain after a dry spell always has more accidents
than later rainy days. This difference is more pronounced when rain
follows a _long_ dry spell, so rain causes more accidents when it comes
after sunny days than it does after rainy days.
But these arguments are definitely not correlation
arguments:
ag. Tobacco consumption, whether chewing or smoking,
causes cancer, because tobacco is harvested in late June, early July,
which means that the astrological sign of all cigarettes is Cancer.
ah. The Fairness Doctrine was abolished by Ronald Reagan, a far right
wing President, so abolition of the fairness rule caused and continues
to cause the American population to become more and more conservative.
ai. The way America administers welfare tends to keep people on welfare
even when they would like to become independent, because welfare
officials like their jobs, and make more money when there's more people
on welfare.
aj. Moderate drinking is actually better for you than not drinking at
all. This follows from the fact that nondrinkers deprive themselves of
the pleasures of intoxication, noncoordination, somnambulation and
regurgitation.
ak. I aquaplaned once. Woooooooo! It was fun, so rain causes accidents
in Southern California
al. Rain causes more accidents when it comes after sunny days than it
does after rainy days, I know this is true because if I was an
accident goblin I'd get the hell off the roads when the accidents start
happening.
Exercise 2. One of the following is a correlation
argument. The other isn't. Say which is which and explain why.
A. We compared 100 random moderate drinkers to 100
random nondrinkers, and found that the moderate drinkers were healthier
than nondrinkers, so moderate drinking is good for you.
B. Moderate drinking is better for you than nondrinking
because "teetotaling" sounds like "teetertottering," and that can't be
good.
Some Points About Causal Reasoning.
"Causal" reasoning is reasoning about cause-and-effect relationships.
Correlation arguments do not support claims about the existence or
non-existance of objects, nor about the occurance or non-occurance of
events. They just support claims about causal relationships. Thus
no correlation argument would support a claim that tile roofs exist, nor
would any causal argument support a claim that baldness happens. But if
someone wanted to support a claim that tile roofs cause baldness, well,
then, that claim would have to be backed up by a correlation argument.
(Notice that a causal claim is just a claim that A causes B. It
is not a claim that anyone knows how A causes B.)
Causal relationships are usually not one-to-one relationships. Causes can
have multiple effects, and effects can have multiple causes. Cigarette
smoking does not cause cancer, emphysema or heart attack in everyone
who smokes cigarettes. But we can bet that a group of a thousand people
who smoke heavily we will see more cases of cancer, emphysema and heart
attack than in a demographically identical group of a thousand nonsmokers.
And ciggies are not the only things that cause cancer, emphysema and heart
attack. So a cause is something that makes the effect more likely
to happen but usually does not guarantee that the effect will happen.
It is important to remember that causal arguments do not have to
come with explanations. When smoking was established as a cause of cancer,
nobody had any idea how smoking managed to cause cancer.
Nevertheless, the strong correlation between smoking cigarettes and
getting cancer convinced reasonable people that a causal link existed. It
didn't matter that nobody had any idea how smoking did it, all that
mattered was that the correlation between cigarettes and cancer was so
strong that no other explanation made sense. All that is
necessary to establish a causal link is that a strong correlation
be established. Explanations are immaterial to this correlation, so the
lack of an explanation is never a weakness in a causal argument.
The way to evaluate a causal argument is to ask whether the known
correlation is strong enough to justify the causal claim and, if it is, to
then ask if all other causal possibilities have been eliminated.
If it is, and they have, then your argument is good. If
it isn't, or they haven't, then the argument sucks.
While an argument that tile roofs cause baldness would have to start with
some correlation between tile roofs and baldness, it wouldn't get very far
unless it included reasons to eliminate all other possible causes
of baldness, like, for instance, salsa music. (We wouldn't have to
eliminate UFOs however. The fact that something has never been proved to
exist is generally enough to remove it from consideration when we're
talking causes.)
If somebody were to say that we should abolish tile roofs because they
cause baldness we would evaluate his "causal claim" by asking him to give
evidence of a correlation between baldness and tile roofs. If it turned
out that men started losing their hair the moment they went under a tile
roof, and that men who never lived under tile roofs never went bald, then
we would have a strong correlation, and good reason to think that tile
roofs cause baldness. If it turned out that men went bald or not
independently of whether or not they lived under a tile roof, then we
would have no correlation, and no reason to think that tile roofs cause
baldness. So a causal argument that comes without at least some evidence
of a correlation is always a bad argument.
Exercise 3. Answer the following questions, and explain
your answers.
i. Are causal arguments required to explain how the purported cause causes
the effect?
ii. Are causal arguments required to show that the purported cause is the
only thing that ever causes the effect?
iii. Are causal arguments required to show that the purported cause
absolutely always causes the effect?
iv. Can a causal argument succeed without showing that the purported cause
is at least sometimes correlated with the effect?
v. Can a causal argument be based on a "cause" that hasn't been previously
proved to exist?
An "anecdote" is a little story about some event that supposedly happened at some time to some person. The kind of anecdote we're concerned with here is an anecdote that says one thing happend at the same time as another time, or that a certain type of thing tends to happen after a certain other type of thing. Anecdotes generally concern unusual events and unexpected correlations (that's what makes them interesting), and they can be valuable in suggesting lines of research, but they can never prove causal claims all by themselves.
Before 1910, lung cancer was virtually unknown. A single case of lung cancer would be news in the medical community, and a doctor who saw a case would want to talk about it. Starting about 1918, doctors in industrialized countries began seeing, and talking about, the surprisingly new high rate of lung cancers they were starting to see. Some of these doctors also noted that, almost entirely without exception, the people who had these cases of lung cancer were smokers. For non-smokers, the rate of lung cancers remained almost vanishingly low. A non-smoker who got lung cancer was still a very, very rare thing. These facts, reported informally by doctors in letters to medical journals and by word of mouth, strongly suggested that smoking causes lung cancer. But they didn't prove it.
There are several reasons why we cannot take anecdotes, even a large number of anecdotes, as proving anything. For one thing, ill-intentioned people often make stuff up to discredit people of groups they don't like. Other times, people invent explanations for events and then pass these invented explanations off as fact. Anecdotes can also arise simply from mistakes. People are generally not good reporters of their observations, and will often report distorted versions of what they themselves saw. Also, anecdotal reporting is uncontrolled. If we hear ten reports of a smoker getting cancer, we have to determine whether these ten stories concern ten different smokers, or just one smoker whose story was heard by ten different people, each of whom then told the story to us. Even when the anecdotal evidence is of high quality, as when trained professionals, such as doctors, carefully report cases that are clearly distinct from each other, we still have to be careful, as there is the possiblity of error or unknown common cause.
Thus, anecdotal evidence can be invaluable in telling us what to think about and where to spend our research dollars, but it cannot prove anything directly.
Smoking
and Lung Cancer
Mill's Methods
John Stuart Mill figured out five different ways to support a causal
connection, depending on what kind of evidence you have. (These methods
may turn out to be "just common sense.") I've noticed that students often
have a hard time understanding explanations of Mill's methods. However,
when I give those students exercises designed to test their ability to
apply Mill's methods, they find those exercises ludicrously easy. This
suggests that what I grandly describe as "Mill's Methods" can be figured
out by community college students with no previous training in those
methods. So, if you don't understand these explanations, don't worry about
it. You will probably be able to do the exercises anyway.
First, some terminology. A positive correlation is when two things
tend to happen together. Tile roofs and baldness are positively correlated
if it is the case that tile roofs tend to show up every time baldness
occurs, and vice versa. It's when two things tend to be absent at the same
time. Tile roofs and baldness are positively correlated if it is the case
that tile roofs tend to be absent every time baldness is also absent, and
vice versa. (I'm not sure those "vice versas"
A negative correlation is where occurances of one thing are
correlated with non-occurrences of another thing.
Mill's method of agreement relies on positive correlations. It
basically says rule out everything that is not necessary to bring about
the effect, and what is left over is more likely to be the cause. Say we
have a population of one thousand bald men and the only thing all of them
have in common is that each of them spends long periods of time under a
tile roof. If this were true, then anything else that was offered as a
candidate for the cause could be eliminated by pointing out that somebody
got bald without being exposed to that thing. Since tile roof is the only
thing that cannot be so eliminated, we have good reason to think that tile
roof is the cause.
Mill's method of difference says rule out everything that is not
sufficient to bring about the effect. Say we have a population of one
thousand hairy men and one bald man who lives apart from them under a tile
roof. The only thing the hairy men all have in common is that none of them
lives under a tile roof. In this case any other causal candidate could be
eliminated by pointing out that somebody was exposed to that thing without
getting bald. Since tile roof is the only thing that cannot be ruled out
that way, we have good reason to think the tile roof is the cause of the
baldness.
Mill's "joint method" just means applying both methods together to
the same data set. If two things are both positively and negatively
correlated, then that could be strong evidence that one of them causes the
other.
Mill method of concomitant variation says look for something that
varies in proportion to the effect. Say that Al has lived under tile for 7
years and is 10% bald, Ben has lived under tile for 14 years and is 20%
bald, Cal lived under tile for 21 years and is 30% bald, Don has lived
under tile 28 years for 40% baldness, Ed has 35 years and 50%, Finn 42
years and 60%, Greg 49 years and 70%, Hal 56 years and 80%, Ian 63 years
and 90% while Jack has spent seventy years living under a tile roof and is
completely bald. Since the amount of baldness goes up and down with the
time of residence under tile, and if nothing else correlates in this way,
this would give us good reason to think that living under a tile roof
causes baldness. Remember that the corespondance doesn't have to be one to
one. It can be any ratio at all, just as long as that ratio is constant
across instances.
Mill's method of residuals is a way of figuring out how much of
some effect can be attributed to some particular cause. Say that we know
that polka music causes baldness in 4 percent of listeners, plaid shirts
cause baldness in 8 percent of wearers, and that jumping jacks cause
baldness in 16 percent of jumpers. And say that we know that these are the
only causes of baldness apart from tile roofs. Finally say that we know
that yesterday one million men wore plaid shirts to do jumping jacks to
polka music under a tile roof, and 40 percent of them became bald.
Ignoring the tile, we can account for no more than 32 percent of these
baldnesses, (4+8+16=28), so being under a tile roof must cause baldness in
at least 12 percent (40-28=12) of men.
Now apply Mill's methods to the following problems, and then evaluate the
arguments that follow the problems. See if you think the person making a
causal argument in each dialog is correctly applying Mill's methods.
Several patients at Chicago Soap Hospital are experiencing bizarre drug
side effects. The hospital administrator, Dr. Whatmeworry, is going crazy,
and has hired several people to figure out what is going on.
Unfortunately, none of them seem to know what they're doing. Your job is
to sort through the arguments and figure out who, if anyone, knows what
he's talking about. (To make this easier, assume that only one drug is
causing each effect, and that clinical trials were not done.)
Some patients in Ward 1 are turning into fairies and
have turned several doctors into animals. The patients who are
growing gossamer wings and buying pixie hats are on Unicorp, Verlag,
Xindeco and Zydigm. Those who are merely looking on in
horror are on Verlag, Zydigm and Unicorp.
The patients in Ward 2 have all acquired the ability to
shoot low energy laser beams from their eyeballs, which they use to
irritate their nurses, warm beverages and vaporize insects. Patient
Abel is on Terraplex, Omniken, Shegos and Permax. Patient
Baker is on Omniken, Shegos and Terraplex. Patient
Cooper is on Terraplex, Shegos and Permax. Patient
Dasia is on Shegos, Omniken and Permax.
Ward 3 Patients are channeling famous generals of the
past. They are now in great demand by TV networks as commentators on
various wars. The camera crews are disrupting hospital operations, and one
patient is faking symptoms (pretending to channel Buffy the Vampire
Slayer) in order to get on The McLaughlin Group. Patient Killian
is channeling Caesar and Napoleon. She is on Invap 40mg, Kolvox 100mg,
Minafax 300mg, and Nasdaq 1mg. Patient Lomax is
channeling Rommel, Wellington, Alexander and Sun-Tzu and she is on Invap
10mg, Kolvox 200mg, Minafax 600mg and Nasdaq 2mg. Patient Miles
is just channeling Vinegar Joe Stilwell and she is on Invap 30mg, Kolvox
50mg, Minafax 900mg and Nasdaq 1mg. Finally, Patient North
is channeling Grant, Zhukov and Sharon. She is on Invap 20mg, Kolvox
150mg, Minafax 600mg and Nasdaq 4mg.
Some patients in Ward 4 are growing extra limbs, which
are presenting interesting possibilities for getting around the ward and
possible future sporting events. Patient Eggar is
growing an extra left foot from his right elbow, and is on Quindor,
Dynastar, Comdex and Gennum. Patient Harris has no extra
limbs, and is on Dynastar, Fitel and Gennum. Patient Frank
amuses children with the extra arm growing out of his left kneecap, and is
on Comdex, Dynastar, Fitel, Quindor, and Gennum. Patient Irvin
is perfectly normal and on Fitel, Gennum and Comdex. Patient
Gregg makes interesting gestures with the extra hand that has
grown out of where his nose used to be. He is on Quindor, Comdex, Fitel
and Dynastar. Patient Jakes has two arms growing from
his tailbone, and can give himself a back massage. He is on Fitel, Gennum
and Comdex.
Now, there's two ways you can approach the following exercises. First, you
could start by ignoring the given arguments and apply Mill's Method's
yourself before comparing your answers to those given in the dialog
analysis exercises. (This means doing exercises 8 through 11 before
exercises 4 through 7.) Or, you could do the dialog analysis first and
Mill's methods after.
Dialog Analysis Exercises.
Find the weaker argument in each pair, and then explain in your own words
why it is weak.
4.size="12" face="Arial">Daryl.
Don't take Zydigm! All of the people who turned into fairies took
Zydigm, so it must be the cause!
Gwendolyn. Not everybody who took Zydigm turned into a fairy,
so it might not be the cause.
5.size="12" face="Arial">Aliya. The drug that gives
people the ability to shoot laser beams from their eyes must be Terraplex.
Able, Baker and Cooper all took Terraplex, and they all got laser vision.
Karson. Yeah, but Dasia got laser eyes too, and she didn't take
Terraplex.
6.size="12" face="Arial">Gianni. The drug that lets
you channel famous generals must be Nasdaq. Miles took 1 milligram of
Nasdaq and channeled just one general, that's a 1-to-1 correlation, so
Nasdaq must be the drug.
Susana. But Killian also took 1 milligram of Nasdaq, and she
channeled 2 generals. Lomax took two milligrams, but she channeled four
generals! Neither of those is a 1-to-1 correlation, so Nasdaq probably
isn't the drug. And anyway, it doesn't have to be one unit of the drug to
one general. We have to look for a situation where the number of generals
goes up and down in exactly the same way that the dosage of the drug goes
up and down.
7.size="12" face="Arial">Zoie. Dynastar must be the
drug that is causing the extra limbs. Eggar, Frank and Gregg all took
Dynastar and all grew extra limbs.
Tristian. But Harris also took Dynastar and didn't grow any extra
limbs, and Jakes grew extra limbs without taking Dynastar.
Mill's Method Exercises
8. What is most likely causing the fairification in ward
1, and which of Mill's methods is appropriate here?
9. What is most likely causing people to get laser eye
in ward 2, and which of Mill's methods is appropriate here?
10. What is most likely causing the general disorder in
ward 3, and which of Mill's methods is appropriate here?
11.What is most likely causing the limbular proliferation
in ward 4, and which of Mill's methods is appropriate here?
Direct and Counter Arguments in Causal Reasoning.
You will remember from the last chapter that a counter
argument says something explicitly about the logic of another
argument, whereas a direct argument gives independent
reasons for believing some particular conclusion. When a direct argument
is opposed by a counter argument, the counter argument gives a reason to
think that the direct argument hasn't proved it's conclusion, but does not
give us any reason to believe any other conclusion.
A. Tobacco consumption, whether chewing or smoking,
causes cancer, because tobacco is harvested in late June, early July,
which means that the astrological sign of all cigarettes is Cancer.
B. In astrology "cancer" is just the name of "the crab," which refers
to a certain portion of the sky. It has nothing to do with the disease
called "cancer."
Notice that argument B, while it kicks argument A right in the head, does
not give us any reason to think that tobacco doesn't cause
cancer. Now, when a direct argument is opposed by another direct argument,
the second argument gives us a reason to think that a different conclusion
is true, but it doesn't give us any reason to think there's anything wrong
with the first argument. So argument B is a direct
argument.
A. Moderate drinkers are healthier than nondrinkers, so
moderate drinking is actually better for you than not drinking at all.
B. People who drink ten or more drinks every day tend to have a wide
variety of serious health problems, so it follows that moderate drinking
will cause moderate health problems.
Notice that, again, the second argument in this pair doesn't give us any
reason to think that there's anything at all wrong with the first
argument. It says absolutely nothing about the logic used in argument A,
and so gives us no reason to think that that logic is bad. So it's a direct
argument.
A. Abolition of the fairness rule caused and continues
to cause the American population to become more and more conservative.
We know this because, after the abolition of the fairness rule, the
American population began a slow but stready and continuous drift to the
right.
B. Okay, abolishing the fairness rule might have caused the
rightward drift, but you haven't eliminated other possible causes, such
as wars, terrorism, scaremongering by politicians, or even just aging in
the population. Untill you've looked at and accounted for all other
possible causes, you're just not going to have a credible argument here.
Argument A refers to a strong correlation: As long as the fairness
rule has been absent, the country has been drifing to the right. Argument
B doesn't challenge that data, but points out that there may be other
causes for the rightward drift. (As in the Ferlinghetti example given
below.) This is getting into the nuts-and-bolts of argument A, and means
that argument B is a counter argument. Also notice that
argument B does not support the idea that abolishing the fairness rule didn't
cause the rightward drift, it just supports the idea that argument
A fails to prove that it did.
Exercises
For each of the following dialogs, figure out if the second argument is a
direct argument or a counter argument.
12. A. Moderate drinkers are
healthier than nondrinkers, so moderate drinking is actually better for
you than not drinking at all.
B. Now, that implies that a nondrinker who started
drinking moderately would experience an improvement in health, even if
he made no other changes. But have you thought about the possibility
that moderate drinking might be associated with a desire to enjoy life?
If moderate drinkers tended to be happy people with active lifestyles
who just had drinks when they socialized, and nondrinkers tended to be
depressed, sedentary people who didn't socialize, that might account for
the difference without any beenficial effects from the alcohol.
13. A. Rain causes accidents in
Southern California, because Caltrans figures show that the accident
rate always goes up whenever it rains in Southern California.
B. That can't be true. I have statistics here from the
National Association of Slip-And-Sliding Rain Dancers, and their records
show no increase in the accident rate when it rains.
14. A. Rain causes accidents in
Southern California, because I always see many more smashed cars on the
side of the road immediately after it has rained Southern California.
B. But Southern California is also very smoggy.
Smog obscures vision, and rain washes smog out of the air, making it
easier to see things. Which means it's possible that there's just as
many accidents when it's dry, but you just don't see them because
they're hidden by the smog.
Evaluating Causal Arguments
To evaluate a causal argument you must figure out two things. First, you
must figure out whether the causal claim offered is a good explanation for
the observed correlation. Second, you must figure out whether or not there
is any other reasonable explanation for that correlation. If it turns out
that the offered causal claim really isn't a good explanation for all of
the observed facts, then the argument is no good. If it turns out that
some other explanation for the observed facts is just as good an
explanation as the causal claim, then the argument is no good.
Since basically any bad causal argument can be called a "false cause"
fallacy, I'm going to continue my discussion of bad causal arguments under
that heading.
False Cause
The classic False Cause fallacy assumes a causal connection based on a
trivial correlation.
Interviewer. "Coach, can you tell me why the fans are tossing cans of
beans on to the field?
Coach. "Well, we haven't done well this season. We only beat
Milwaukee, Dallas and Fargo. Just before that Milwaukee game I was
embarrassed by a noisy attack of flatulence. Then, as I was psyching the
team up for the Dallas game..."
Interviewer. "That was you? I thought it was a low-flying jet!"
Coach. "No that was me..."
Interviewer. "And the cheerleaders who fainted before the Fargo
game?"
Coach. "Me again, I'm afraid."
Interviewer. "And so the beans..."
Coach. "If you'll excuse me, I think I can choke down another
can.
I love your sister, but every time she comes to California we have an
earthquake. So we can't invite her again until I've finished
earthquake-proofing my collection of decorative plates.
Just because something is the only cause you can think of doesn't mean it
has any causal relationship to the effect.
Pete: Since 1985, the
average SAT scores of the incoming freshmen at Miskatonic University
have consistently been about 10% better than they were before 1985.
Nothing important changed in or around 1985. In fact, we've checked all
the things that the schools could be doing that could possibly cause a
rise in SAT scores, and we found that none of them changed
significantly, so it isn't due to anything that local educators are
doing.
Ona: Have you forgotten that I‘ve been teaching at
Miskatonic since 1985. Obviously, that's the reason.
Pete: How do you figure that?
Ona: Well, duh, before I got here, the scores were bad.
After I got here, the scores got better.
Sometimes people commit false cause when they get confused about causal
relationships. People often assume that one thing causes (or must cause)
another thing merely because the two things are strongly associated in
their minds. But a mental association is a far cry from a causal
relationship.
Ignoring Possible Alternative Cause.
One way to look at causal arguments is to see them as claiming that the
only way to explain the frequency of a certain event is to assume that it
is caused by some particular other kind of event. This means that whenever
there's something else that's equally well correlated with the effect, it
follows that that something else could just as easily have been what
caused the effect, and so the argument fails to prove the causal
conclusion it was intended to prove.
Causal reasoning requires careful attention to detail, and a willingness
to follow the evidence wherever it actually leads. Consider the following
problem.
Exercise 15 The village of Ferlinghetti has been famous
for its locally produced mineral water for about 150 years. However, the
village also has a problem with adult asthma. Many of the town's adults
spend much of their time coughing and wheezing. Locals believe that the
mineral water reduces or prevents the asthma and have been drinking more
and more of it since the asthma problem was first identified in 1852. The
asthma problem has grown slowly but steadily worse, with no interruptions,
since the early 1850s. Asthma attacks do not get any worse or any better
at any particular time of day or time of year. The leading citizens in the
town have identified four possible causes of the asthma. A tin mine in the
mountains produces runoff, which runs into the local river, from which the
town draws all its drinking water. (Fortunately, the mineral water comes
from a deep mountain spring that is not at all affected by the runoff.) A
perfume factory in the next valley produces fumes, which drift over to
Ferlinghetti. Farmers in the nearby plain use a special kind of wheat,
which releases pollen that is carried into Ferlinghetti by the prevailing
winds. Bats from the local woods fly into the town at night to feed on
insects, and leave fur and skin flakes that hang in the air for about an
hour after they fly through. The tin mine has been in continuous operation
at the same level of production since Roman times. The perfume factory was
built in 1843, but was completely shut down for both world war one and
world war two. The special wheat was introduced in 1840. The local species
of bat was discovered and classified in 1845 when local kids found caves
containing both the bats and accumulations of bat droppings that proved to
be dozens of feet thick. You are a scientist called in to investigate the
situation and discover the cause of the asthma problem.
size="12" face="Arial">The town mayor argues, "it's
the tin mine in the mountains, because that runoff water does get into the
local drinking water. It must have started running off in 1852 and the
amount of water running off must have gotten larger and larger since
then."
The chief of police claims that "the perfume factory has
to be the cause. There must be something in the perfume fumes that causes
asthma, and they must have been using it continuously since 1852."
The town drunk confides in you that "the wheat must be
the cause. They've been growing it continuously since at least 1852, so
it's perfectly well correlated."
And the town librarian says "it's the bats!
Aaaaaaaaaaagh! They're after me! Keep them away, keep them away! Send for
Buffy now!"
Apply careful causal reasoning to the problem. You will find the answe
very clear once you write down and organize all the facts given
above. (You can print out just this problem if you download ferlinghetti.htm
or ferlinghetti.rtf.)
Ignoring a Possible Common Cause
Remember that a strong correlation, by itself, does not prove causality.
The right way to show causality is to demonstrate a strong correlation
that cannot be exaplained any other way. When there is another reasonable
explanation for the putative causal relationship, the argument commits the
fallacy of "ignoring a possible common cause." Remember,
it is sometimes the case that two things are correlated, not because one
causes the other, but because some third factor causes both.
Every time I put on a little weight, I lose a little hair. So if I
exercise and go on a diet, I'll stop losing my hair.
Ignoring a common cause might be easier to understand if it was called
"ignoring a possible common cause" because it is the fact that
some third thing might be causing the correlation that makes the
argument fail. Strictly speaking, all you have to to refute any causal
argument and it is a very popular fallacy. It is often commited by people
who want to believe that drug use, by itself, causes crime. If
there is any reasonable possibility that a person's criminality and drug
use have a common cause, such as poor upbringing or exposure to scofflaw
culture, such as could reasonably be thought to have caused criminalty
without exposure to drugs, then the correlation between drug use and
criminality cannot prove that the drug use caused the criminal tendencies.
Even when there is a strong correlation that cannot be reasonably
explained without assuming a causal realtionship between the two events,
people may be confused about which is the cause and which is the effect.
People can commit false cause by completely reversing a causal
relationship. That is to say, two things are correlated, but the wrong
thing is identified as the cause.
Reversing Cause and Effect
Have you noticed how Jane gets
depressed every time her cancer gets worse. If we could just cheer her
up, that would take care of that nasty cancer!
Here the arguer has reversed cause and effect. by
mistaking the effect for the cause and the cause for the effect. The two
things are undoubtedly correlated, and there is no third element that
could cause both things, but the wrong event is identified as the cause.
Reversing the Direction of the Effect
Finally, a more subtle, and perhaps more common fallacy is when an arguer
correctly identifies a cause, but reverses the direction of the effect.
The arguer mighy, say, have noticed that the masses of water shot out of
fire-hoses are associated with building fires, but erroneously assumes
that adding the water to the building causes the fire, or that
adding water makes the fire burn more fiercely. Because I can't
think of a better name for it, I call this reversing the
direction of the effect.
Why do we have drug laws? We have to
have them because of all those drug-related murders! Those dealers will
kill to keep their activities secret, you know!
Notice that the arguer here believes that laws against possession, use and
sale of drugs reduce the murder rate. Also notice that the
arguer does not offer any data whatsoever. Finally, notice that the arguer
ignores a well-known causal relationship between criminalization of
activities and murder. Students of history will recall that the
criminalization of alcohol trafficking in the United States was followed
by an enormous increase in the number of murders related to alcohol
trafficking. This murder rate was drastically reduced by the later
legalization of alcohol trafficking. Generally speaking, people do not
kill to conceal activities that are not illegal.
But the government of Slobovia had
to imprision those peaceful protestors! Don't you know that disagreement
with the government can escalate to violent opposition?
Here the arguer correctly connects political repression with changes in
the level of political violence. But he illegitimately assumes that
political repression reduces or prevents political
violence. The right way to prove such a thing would be to give evidence
that repressive countries tend to have lower levels of political violence
than non-repressive countries, which he has not done.
Begging the Question
I want to emphasize very strongly that a mental association is
no substitute for actual evidence. Just because two things are
associated in your mind, or you think that one thing must
cause another, you cannot even begin to have an argument to that effect
until you come up with some real facts. To argue that one thing must cause
another merely because you think that it must be so is simply to
beg the question. Consider the following.
size="12" face="Arial">Rylan. Of course welfare causes dependency! It just stands to
reason that people on welfare will want to stay on welfare for as long
as possible.
Aspen.
Are you crazy? I think it stands to reason that people on welfare will
want to get off welfare as soon as possible. We can't both be right, so
I don't think talking about "stands to reason" tells us anything about
welfare.
Notice that Rylan offers his own opinion as evidence.
Unsupported by evidence, his claim that his opinion "stands to reason" is
nonsense. Things only stand to reason when they are supported by
a logically compelling argument. Offering one's personal opinion as
evidence is not an argument. Aspen, however, has a logically compelling
argument. Notice that her conclusion is just that arguments like
Rylan's can't prove anything. True, her opinion that people on welfare
will want to get off welfare as soon as possible is just her opinion, but
it is a fact that she holds that opinion, which shows that Rylan's opinion
isn't the only one out there, which is enough to show that
Rylan's argument is no good. (Generally, people who tell you that
something "stands to reason" instead of giving you an argument are simply
trying to cover the fact that they have no argument.)
People can of course commit fallacies when attacking a causal
argument. The most common fallacy in this area is red herring.
Red Herring
The "inconvenient fact" method of attacking causal arguments doesn't
always work. Sometimes the fact in question isn't even remotely
inconvenient for the argument. When this happens, the person attacking the
argument has committed a red herring fallacy.
Suppose that the "Hacking Cough" study is a long-term study comparing a
group of heavy cigarette smokers to a group of nonsmokers has found that,
over time, cases of lung cancer crop up in the smoking group at a rate
five times that which they crop up in the non-smoking group. Some people
take the Hacking Cough Study to have proved that cigarette smoking causes
lung cancer. Other people claim that it doesn't prove that cigarette
smoking causes lung cancer. If we pretend that these people give the
following arguments, we would be pretending that they all commit red
herring fallacies, because all the following arguments are red herrings.
The Hacking Cough Study doesn't prove that smoking causes lung cancer
because it is a fact that some people who smoke heavily never
get lung cancer.
You cannot claim that the Hacking Cough Study proves that smoking causes
lung cancer because no one, including the authors of the study, can explain
just how it is that smoking manages to cause lung cancer.
Plenty of nonsmokers get lung cancer, so it can't be true that cigarette
smoking causes lung cancer.
The first argument is a fallacy because the statement "A causes B" only
means "A raises the probability that B will happen," not "A always
makes B happen." The second one is a fallacy because causal arguments don't
have to provide explanations. Strong evidence of a strong correlation always
establishes a cause and effect relationship, whether or not we can explain
that relationship. And finally, the third argument is also a red herring
because the study doesn't have to prove that smoking is the only
cause of lung cancer in order to prove that it is a cause of
lung cancer.
Consistency With Available Evidence
Sometimes people will insist that some causal claim is true even though
all the evidence they have implies that it is false. The following is a
paraphrase of an actual conversation. (Only the names have been changed.)
size="12" face="Arial">Beau.
It's obvious that welfare causes dependency, because someone who's
getting money without working is going to want to keep on taking it
easy.
Issac. You know, I don't think it's
fair of you to project your own preferences on other people. What makes
you think that people on welfare feel the same way you would in that
situation?
Beau. What are you talking about? I
wouldn't feel that way! I'd want to get off welfare as soon as possible.
Issac. Oh, sorry. I guess you were
basing this on people you know. Still, you shouldn't assume that the
people who end up on welfare are like the people in your family, and
your other friends and acquaintances.
Beau. Hey! None of my family are
like that! None of my friends would want to stay on welfare.
In fact, I don't know anyone who would.
Issac. So you're saying that
someone who's getting money without working is going to want to keep on
taking it easy, even though all of the actual human beings you
know about wouldn't want to keep on taking it easy, even if
they were getting money without working.
Beau. Now you've got it.
The first point is, of course, that Beau's conclusion is contradicted
by all the evidence he has. He's saying that all people are suceptible to
being made dependent by welfare even though all of the people he knows lack
that suceptibility. The second point to notice is that Beau is not even
using the right kind of strategy to establish a causal claim. His first
statement is actually a question begging fallacy, since
the statement "someone who's getting money without working is going to
want to keep on taking it easy" is simply a more general statement of the
claim "welfare causes dependency." The two claims are not different enough
to make the premise acceptable to anyone who rejected the conclusion.
Issac treats Beau's argument as a generalization as a way of getting Beau
to admit that he has evidence that contradicts his conclusion. Beau may
indeed be generalizing, since he might have an idea in his head
of what he thinks a typical welfare recipiant is like, and is generalizing
from this imaginary person to real welfare recipients. Of course,
imaginary evidence never counts. (Therefore arguments based on imaginary
evidence always commit the relevance fallacy of red herring.)
Causal Chains and Slippery Slopes
The fallacy of slippery slope combines a whole string of dubious
causal claims. The difference between a slippery slope fallacy and a
legitimate causal chain argument is that the legitimate
argument depends upon a series of well-established causal relationships
while the slippery slope includes at least one dubious or speculative
causal claim.
Censoring hardcore pornography will
soon lead to censorship of softcore pornography, which in turn will lead
to suppression of harmless erotica like swimsuit photos, and that itself
will finally cause censorship of nude pictures in medical textbooks.
If you make marijuana legal then it'll be magic mushrooms. If 'shrooms,
then speed, and after that, peyote! Then cocaine won't be far behind.
Which means crack! And what about heroin? And acid! Then there's angel
dust, and finally those designer drugs that make people's heads explode!
So if we make marijuana legal, decent people won't be able to sleep for
the sound of all the exploding heads.
Would legalizing marijuana cause 'shrooms to become legal? Would
legalizing 'shrooms cause speed to become legal? Would legalizing
speed cause peyote to become legal? Would legalizing peyote cause
crack to become legal? Would legalizing crack cause heroin to
become legal? Would legalizing heroin cause angel dust to become
legal? Would legalizing angel dust cause designer drugs to become
legal? Is there anything about legalizing one drug that makes it
impossible, or extremely difficult to avoid legalizing the next
drug in the sequence? If there's any point where we can legalize
one (relatively harmless) drug without having to legalize the
next (more harmful) drug, the causal chain snaps, and the slippery-slope
argument fails.
(Slippery-slope is what logicians call a presumption
fallacy, because the arguer illegitimately assumes that, because the first
element in a chain seems to him to be associated with the second element
in the chain, it will then follow that allowing the first element to exist
will cause the second element to come to pass. But if a causal
relationship has not been proved to exist between these two elements, then
he simply cannot assume that one will cause the other.
Reasonableness
Another important issue in causal reasoning is the reasonableness
of a causal prediction. A causal prediction is a claim that some action
that we contemplate taking will have some particular effect. For instance,
the claim that invading Iraq will make us more secure is a causal
prediction because it holds that if we invade Iraq (as we just did), the
foreseeable future will be more secure for us than if we had not invaded
Iraq. An opposing causal prediction might be that the invasion will make
us less secure. Such predictions can be partially tested by waiting until
after the fact and seeing if the claimed effect actually comes to pass. In
the case of the Iraqi invasion, for instance, the first prediction would
lead one to expect decreased security at airports, fewer terrorist alerts
and less money spent on homeland security, whereas the second prediction
would lead one to expect at least a continuing high level of security,
more terrorist alerts and more money spent on homeland security. But there
are two problems. The first is that there may be other factors involved,
so that the real effects are hidden. For instance, it may in fact be true
that the invasion increases our security, but some other independent
factor, (say, the final fruition of a terrorist plot that was begun well
before the invasion was announced, or the loss of competent security
administrators in a car accident) decreases our security, hiding the gain
from the invasion. The second problem is that we need to be able to
predict the effects of our actions before we act. We cannot undo an
invasion, so we have to be sure of its effects before we go
ahead and do it. For this reason, a great deal of causal reasoning is done
without the benefit of definitive evidence regarding the particular kind
of action contemplated. The best we can do sometimes is to think about
what we might reasonably expect to happen based on generally known causal
rules, and such thinking, like any thinking, always runs the risk of
committing non-causal as well as causal fallacies. So look out for any and
all of the fallacies we've learned previously as well as the causal ones.
If you're not already sick of hearing about the subject, here's a little
story about causal reasoning: causalstory.htm
Technically, a "fact" is a something that cannot reasonably be disputed,
an "opinion" is just something someone believes, and a "conclusion" is
something someone wants other people to believe. In terms of evidence,
we can define a "fact" as a claim that is supported by compelling
evidence, an "opinion" as a claim that someone thinks is
supported by compelling evidence, and a conclusion someone says is supported by
compelling evidence.
Unfortunately, the words "fact" and "opinion" are occasionally badly
misused, as in the following dialog:
Chuvaskaya.
So the teacher asked us to analyze this online debate between Doctor
Polyp and Professor Spleen. I think Dr. Polyp won the debate because
he pointed out that there are no documented cases of people dying as
a result of using marijuana, that increases in marijuana consumption
have not been followed by increases in disease the way the sharp
rise of cigarette smoking was followed by a sharp rise in lung
cancer, and that California has more-or-less legalized marijuana
without experiencing any noticeable increase in social problems.
Professor Spleen doesn't discuss these matters at all, and he gives
no evidence to support his claims that marijuana is more dangerous
than alcohol and tobacco, and should be illegal. So I think that
Doctor Polyp's argument for legalizing marijuana is better than
Professor Spleen's argument against it.
Flanders. I disagree. I think
your analysis is completely wrong. Look at the debate again. It's
true that Doctor Polyp says that no-one has died from using
marijuana, that marijuana hasn't increased disease, and that
California virtually legalizing marijuana hasn't caused any
problems, but Professor Spleen has pointed out that marijuana is
extremely dangerous and destructive, and it should be absolutely
illegal everywhere. So you can clearly see that Professor Spleen has
the facts while Doctor Polyp is just giving his own personal
opinion.
I want you to carefully examine the above dialog to determine who it is
out of Doctor Polyp and Professor Spleen who actually has the
facts and who is merely stating personal opinions. Notice that Spleen
claims that marijuana is bad and should be illegal, while Polyp disagrees
with this, so it follows that they have different conclusions.
but the mere fact that two people disagree says nothing about which one
of them has the facts. In fact, if all these two did was state opposing
opinions, then we would have to say that neither of them had
the facts. But also notice that while Spleen says nothing to support
his conclusion, Polyp says quite a deal. Polyp gives reasons
to believe his conclusion while Spleen does not. In general,
if one side of an argument gives reasons and the other side doesn't,
the side that doesn't give reasons cannot be
said to "have the facts." A side that does not give reasons does not
have the facts. Only sides that give reasons can ever be said
to have the facts, and, even then those reasons don't always turn out to
actually be facts. Furthermore, since Flanders doesn't mention
Spleen raising any objections to Polyp's claims, we should assume, based
on this dialog, that Spleen has not given us any reason to
doubt Polyp's reasons. Thus, if anyone "has the facts" here,
it is Polyp who has the facts, and Spleen who is
merely giving his personal opinion.
For the purposes of this text, I want you to define a "fact" as a claim that is not actually disputed by anyone. Thus, if one side says that marijuana has never killed anyone, that it does not cause disease, and virtually legalizing it in California hasn't caused problems, and the other side does not dispute these claims, then those claims are the facts in this particular case.
The bottom line is, in this course, you should look to see whether or
not a factual claim is disputed by the other side. If the
claim is ignored or accepted by the other side, then that claim is a fact,
at least as far as this class is concerned.
The following videos might be helpful, but I made them a long time ago, and I've reorganized the class since then, so some of what I say in these videos may fail to match this chapter.
More
Exercises
For each of the following argument pairs, I want you to do following
things:
For each argument, decide whether or not it makes a causal
claim.
For each argument that makes a causal claim, decide whether or not that
claim is based on a correlation argument.
For each correlation argument, identify the specific correlation.
For each pair of opposing arguments, decide whether or not the second
argument is a counter argument.
For each counter argument, identify the specific part of the other
argument that is being attacked.
For each pair of opposing arguments, figure out which one is the weaker
argument.
For each weak argument, name the fallacy committed by that arguments.
For each fallacy figure out the key fact that kicks that bad argument in
the head.
Remember that the right way to establish a causal relationship is to
show that:
1. The thing that's supposed to be the cause is very strongly correlated
with the effect.
2. Whenever the two things happen together, the thing that's supposed to
be the cause happens before the effect.
3. The correlation cannot reasonably be explained any other way.
Remember that the very best critiques will include both the
correct fallacy name and the crucial fact(s), explained in
such a way that a reasonable reader will be able to immediately see
why the weak argument is bad.
Once you have made your best effort to analyze all the arguments and
write your own critiques, check your answers at the end of the chapter
Exercise 16. Raegan.
How can you stand there and protest the war? Don't you know our boys are
over there?
Rohan. Well, if there wasn't a war, wouldn't they be
back here instead of over there?
Exercise 17. Glenn. Studies show that people who dance
are five times as likely to have knobby knees as people who don't dance,
so dancing causes knobby knees.
Savanah. Hah! That's ridiculous! Don't you know that
every one of those studies included a bunch of people who danced around
for years and never got knobby knees, so it must be blindingly obvious
that dancing doesn't cause knobby knees.
Exercise 18. Tobias. We must ban digital cameras
immediately! A major 17-country study has shown that people who use
digital cameras are 10 times more likely than non-digicam-users to
become lawyers, and that the increase in lawyerism in any area is
perfectly correlated with the local increase in digicam use. The study
has been checked over by 217 experts who all independently certified it
as perfectly conducted and demographically appropriate, with absolutely
no reason to think that the study might be flawed in any respect. So we
must ban digicams before we are buried in lawyers.
Leila. I've seen that study. I've read that study from
cover to cover. And I can tell you we don't need to worry. This study
fails to prove that digicam use causes lawyerism because it totally
fails to explain how digicam use causes lawyerism.
Exercise 19. Everett. I've just discovered some
disturbing news about Histafix, the medicine to relieve allergy
symptoms. Did you know that once allergy sufferers start taking it, they
almost never stop! Obviously, this means that taking Histafix causes
dependence on Histafix. I just heard that thousands of people are taking
Histafix even as we speak, so we should ban it now so that these people
can be cured of their dependence on Histafix.
Kiersten. Um, maybe all these people keep taking
Histafix because their allergies don't magically go away after they take
a few doses.
Exercise 20. Tate. I just found out that exposure to
Faux News causes blowhardery. It's been found that the more time people
spend watching Faux News Network, the more likely they are to become
irrational, fact-avoidant, gratuitously insulting pompous, brain-dead
blowhards!
Kelsie. That's rubbish. I know hundreds of blowhards
who never watched Faux News, so blowhardery has nothing to do with
watching that Fair And Balanced network.
Exercise 21. Mike. Gasolene is flammable, so I don't
think it's a good way to put out that fire.
Karley. The need to pour gasoline on the fire has never
been so clear. Ever since I started putting gasoline on the fire it's
been getting fiercer and fiercer! So obviously, we need to go out and
buy a lot more gasoline.
Exercise 22. Barry. I used to think that Reverend Jim
was wasting his time with his Pray for Peace campaign, but it's really
begun to foster a sense of community in our parish.
Julius. A sense of community is a good thing, but we
must set it against the needs of people in other countries. Since
Reverend Jim started his Pray-For-Peace campaign, four more wars have
started. We must shut down this Pray-For-Peace campaign before even more
countries are dragged into war!
Exercise 23. Wally. If we allow pornography in adult
book stores it won't stop there. Soon it will be in the regular book
stores, because they're book stores too. Then it will be in the
libraries, because they're just bookstores that lend books instead of
selling them. Schools have libraries, so it will soon be in the schools
too! Well did you know that churches have schools too, so if we allow
pornography in adult book stores it won't be long before it's in our
churches too! Obviously, pornography should be banned absolutely
Xena. The people who want to ban pornography are mostly
the same people who want to ban all discussion of sex. If we cave in on
pornography they'll have a precedent to point to, which will make it
easier for them to ban things like The Joy of Sex. If they succeed
there, they'll go on to other things, like sex education. If we want to
preserve free discussion of sexual matters, we should hold the line on
pornography.
Exercise 24. Cullen. I think it's pretty clear that
drinking bottled water causes mopery. A recent study has shown that
people who drink bottled water are four times as likely to mope around
as people who don't drink bottled water.
Sterling. Yes, but the study also showed that there's
plenty of people who mope around without ever touching bottled water, so
the study doesn't prove that drinking bottled water causes mopery.
Exercise 25. Quentin. I think we should go back to
pumping oxygen into that sunken submarine. It would help the sailors.
Elisha. Since we stopped pumping oxygen into that
sunken submarine the sailors have gotten sicker and sicker. This is a
very serious situation which we must address immediately by making sure
that there is absolutely no oxygen in that sunken submarine.
Exercise 26. Tanya. I've just heard some disturbing
news. A major recent study just compared two demographically identical
groups, one of which listened to classical music a lot, and the other
didn't listen to classical music at all. The group of classical music
listeners turned out to have a much higher percentage of white-collar
criminals than the non listeners, so it looks like listening to
classical music causes white-collar crime.
Alfred. I've seen the same study, and I can tell you
that it totally fails to prove that classical music causes white-collar
crime because nowhere in that study do they even begin to explain how
listening to classical music goes about causing that white-collar crime.
Exercise 27. Humberto. It's beginning to look like
drinking herbal tea makes people taller. They did a long-term study
comparing a million adult herbal tea drinkers with a million adults who
never, ever drink that disgusting rubbish. On average, the people in the
herbal tea group got three millimeters taller for every year they drank
herbal tea, while the demographically equivalent adults in the other
group didn't grow at all.
Camila. Herbal tea doesn't make people taller. Plenty
of people in that study drank herbal tea for years without getting any
taller.
Exercise 28. Answer the following questions, and explain your
answers.
vi. Can a correlation argument be based only on the fact that the "cause"
is always absent whenever the effect is absent?
vii. Can a correlation argument work if something else is just as strongly
correlated with the effect as the purported cause?
viii. Does the "cause" have to have a one-to-one correspondence with the
effect?
ix. Is "it's the only thing I can think of" ever a good causal argument?
Exercise 29. Answer the following questions, and explain your
answers.
i. Is it possible to make a successful correlation argument without any
evidence of a correlation between the effect and the thing that’s supposed
to be causing it?
ii. Is it possible to make a successful correlation argument without
providing an explanation of how the purported cause might cause the
effect?
iii. Is it possible to make a successful correlation argument without
providing any evidence that the purported cause even exists?
iv. Is it possible to make a successful correlation argument if it is
proven that the purported cause sometimes happens without being followed
by the effect?
v. Is it possible to make a successful correlation argument based on a
single incident?
vi. Is it possible to make a successful correlation argument if it is
proven that the effect sometimes happens even though the purported cause
is not present?
vii. Is it possible to make a successful correlation argument based only
on the fact that no-one else can think of anything that could cause the
effect?
For more practice, you can download
and do the practice/makeup exercises.
(Make sure the document margins are set to 0.5 inches or narrower.)
If you have time, I recommend the movie Fat
Head. It's well made, informative and very funny. It also blows the
lid off a bunch of pseudo-scientific myths that are very prevalent in our
society. (It also debunks a few myths I used to be very fond of!)
Exercise Answers.
1. No. Even if all the claims ("o" through "t") were true, none of
them would logically support any of the causal claims ("a"
through "f") made at the top of this chapter?
2. "A" is a correlation argument because it claims that moderate drinking
has been found to correlate with better health. "B" is not because it is
based on a similarity between two words, which has nothing to do with
correlations.
3.
i. No.
ii. Nope.
iii. Nuh-ah!
iv. Well, no.
v. Good god, no!
(Explanations for these answers can be found in the text preceding the
questions. If my explanation isn't clear, please ask about this in class
or by email.)
4. Daryl’s argument misses the fact that the patients who didn’t fairy-up
also took Zydigm. (Gwendolyn alludes to this in her criticism of Daryl’s
argument.)
5. Aliya’s argument is weak because she ignores the fact that Dasia got
laser eyes without taking Terraplex. It’s possible that both Terraplex and
another drug cause laser eyes, but it’s not likely.
6. Gianni fails because his “1-to-1 correlation” is not a correlation at
all, as Susana points out. The fact that a one-general patient took
exactly one milligram is meaningless because “milligram” is an arbitrary
measure. If the drug had been measured in “grains” (an old apothacary’s
standard) it’s mass would have been 64.79891 grains.
7. Zoie fails because she ignores the fact that Harris also took Dynastar
and didn't grow any extra limbs, and Jakes grew extra limbs without taking
Dynastar
8. There is one drug, Xindeco, that was only taken by the fairies, so
Mill’s method of difference supports Xindeco as the cause.
(It is also possible, but less likely, that this is a coincidence, and
that one of the other drugs causes the problem, but only in some people.)
9. Mill’s method of agreement much more supports the claim of Shegos to be
the cause because it is the only drug taken by all four patients.
10. Susana is right that we have to look for what Mill calls “concommitant
variation,” and we find it in Kolvox, which gives us one channeled general
per 50mg.
11. If you examine the evidence you will see that there is one and only
one drug that is always present when extra limbs are present and always
absent when extra limbs are absent, and that drug is Quindor. This way of
thinking is called "Mill’s combined method."
12. B is a counter argument. If we can reasonably explain the difference
without assuming a health benefit from moderate drinking, then the
differnce does not support the implication that there's a benefit.
13. B is a direct argument. The second argument just gives an alternative
source of data. It doesn't give us any reason to question the evidence or
logic given in argument A.
14. B is a counter argument. It offers an alternative explanation for the
observed data. If that explanation turns out to be reasonable,
argument A's data will fail to support argument A's conclusion. (The
alternative explanation doesn't have to be proved true, it just
has to be reasonable)
15. It’s the mineral water. It’s the only thing that correlates with the
asthma. As the villagers began drinking it more and more, the problem got
worse and worse. If it was the tin mine, the asthma problem would have
started in Roman times and continued at the same level since then. The
bats have been around longer than the tin mine, even though they were only
discovered in 1845, so it’s not them either. The perfume factory shut down
during both wars and the wheat only produces pollen in the summer, but the
asthma problem didn’t go away during either war, and it exists even when
there’s no pollen in the air, so neither of those is the cause.
Consumption of the mineral water is the only thing that perfectly
correlates with the asthma problem (key fact), so it’s the cause.
16. Both arguments make causal claims. Reagan claims that protesting here
endangers the boys over there. (Her argument can be paraphrased
as "Protesting endangers our boys. Endangering our boys is wrong. So
protesing is wrong.") Rohan claims that keeping the boys in a war zone
endangers the boys more. Neither offers any correlation. Rohan
offers a direct argument. He gives no reason to
question Reagan's argument. Raegan appears to be relying on the idea that
homefront protests encourage the enemy to fight harder which endangers the
combat troops. and it does seem reasonable that if the German people had
protested the 1939 invasion of Poland, the Poles might have been
encouraged to more strongly resist the blitzkrieg. The same applies for
other Saddam Hussein’s 1990 invasion of Kuwait, and so on. In fact, if
Mexico fell into the hands of an aggressive right-wing government that
somehow managed to mount a successful invasion of the US, protests against
this invasion by ordinary Mexicans would probably encourage Americans to
more strongly resist the invasion, thus endangering the Mexican soldiers
spearheading the invasion. Of course, there is the (key)
fact that, if the invading government changed its mind and stopped the
invasion, none of the soldiers would be in any danger at all, so if Raegan
really cares about the troops' lives, she should also be trying to stop
the invasion. The fact that it is the government (which Reagan
supports) that is actually putting the troops in danger means that Reagan
is technically comitting the fallacy of ignoring a possible
alternative cause .
17. Glenn makes a causal claim, Savanah does not. Glenn refers to
"studies," which presumably are based on correnations. The correlation
mentioned is between dancing and knobby knees. Savanah's argument is a
counter argument because she refers to information contained in the
studies Glenn is relying upon to support his claim. If there's something
wrong with those studies, Glenn's argument collapses. The key
fact here is that causal claims only have to show an
increased probability, so Savanah is wrong because Glenn is only
claiming that dancing makes knobby knees more likely, not that it always
causes knobby knees. Savanah is referring to something that isn't really
relevant, so she's committing the red herring fallacy.
18. Tobias makes a causal claim, Leila doesn't. Tobias indirectly refers
to correlations when he mentions the studies. Leila’s counter
argument criticizing those studies doesn’t work because a causal argument
only needs to prove the existence of a reliable correlation
between two kinds of event. The key fact is that lack of explanation
doesn’t matter so long as the correlation is proved. (Pretty
much all known causal relationships lacked explanation when they were
first proved.) Leila is making a fuss over something that doesn't matter,
so she's comitting a red herring fallacy.
19. Everett makes a causal claim backed by a correlation
argument. Kiersten attacks the logic of his argument, so she's making a counter
argument. Notice that while she doesn't question Everett's data, she does
question his interpretation of those facts. She's implying that
his data doesn't mean what he thinks it means, which makes her comment a counter
argument. What would it mean to say that someone is “dependant” on
something? Presumably it means that the person finds himself unable to do
without that thing even though he actually had no real need for that
thing. Thus someone who is dependant on Histafix would be unable to stop
taking Histafix, even though she actually did not need it to control
allergy symptoms. To make his argument work, Everett would have to show
that there is a significant group of people who provably do not have
allergies, and who still find themselves unable to do without Histafix.
Since he fails to even begin to prove the existance of even one person who
both lacks allergies and stays on Histafix anyway, Everett’s argument
fails completely. Since he correctly notes that people on the drug tend to
stay on the drug, but fails to notice that this could be because allergies
are permanant, he's committing the fallacy of ignoring a possible
alternative cause.
20. Tate makes a causal claim backed by a correlation
argument. He cites a correlation between watching Faux and blowhardery.
Kelsie attacks the logic of his argument, so she's making a counter
argument. Kelsie’s argument fails. The key fact is that
Tate is not claiming that exposure to Faux News is the only
cause of blowhardery, only that it is a cause. It is perfectly
possible for there to be more than one cause of blowhardery, so the
existance of non-Faux blowhards doesn’t invalidate Tate’s argument. Kelsie
is committing a red herring fallacy because the
existence of non-Faux blowhards is irrelevant to the issue. (Another way
to put this is that she think it's a key fact, but it isn't.)
21. Mike is not making a causal argument because he is not
supporting his causal claim that gasolene is flammable. Rather, he is
making that unsupported (but unquestioned) claim in support of his claim
that gasolene is not a good way to put out a fire. This is a direct
argument. Karley does not address his argument, but instead
makes contrary a claim based on independent evidence, thus also making a direct
argument. Karley has reversed the direction of effect.
She wants the fire to go down, and advocates adding gasolene to achieve
this, but (key fact) the addition of gasolene positively
correlates with the fire getting fiercer, not with it going
down. Notice that while Mike does not support his claim that gasolene is
flammable, this is okay because that claim is commonly accepted and not
challenged in this dialog, whereas Karley actuall cites evidence that
contradicts her conclusion.
22. Barry and Julius make unrelated causal arguments, and so both make direct
arguments. Barry claims that Rev. Jim’s Pray for Peace campaign
has begun to foster a sense of community, which implies that he's seen
this sense grow as the prayer campaign has progressed. This could
presumably be challenged in various ways, but Julius does not do so.
Instead, Julius points out that four more wars have started since the
campaign began. However, (key fact) he does not say how
many wars ended, or how many began before the campaign got going, and he
ignores all the other possible causes of those four new wars. This seems
like a weak correlation fallacy to me.
23. Both Wally and Xena attempt to make causal chain arguments.
These attempt to string together cascades of proven causal relationships,
so neither of them is a correlation argument. Neither criticises
the other's logic, so both make direct arguments.
Causal chain arguments only work if each link in the chain is a proven
causal relationship. Wally’s argument is based only on his own
definitions, so isn’t based on causal relationships at all. This makes it
a definite slippery slope fallacy . Xena’s argument, in
contrast, is based on claims about the real nature of the issues and the
people involved and so if her description of the situation is accurate,
her argument is at least plausible.
24. Cullen makes a causal claim based on a correlation
(in the study) between bottled water and mopery. He's not talking about
anyone else's argument, so it's a direct argument.
Sterling talks about the study and gives no independent evidence, so he
makes a counter argument. Sterling commits a red
herring fallacy because (key fact) Cullen is just
arguing that drinking bottled water causes mopery, not that it
is the only cause of mopery.
25. Quentin makes a causal claim (oxygen will cause the sailors
to get better), but it's not based on any cited correlation. Neither
argument refers to the other, so they're both direct
arguments. Elisha is reversing the direction of effect
here. The key fact is that cutting the oxygen made the sailors sicker,
so his evidence supports the claim that it is a lack of oxygen
that is making the sailors sick.
26. Tanya makes a causal claim based on a correlation between classical
music and white-collar crime. This is a direct argument.
Alfred makes a counter argument by pointing out what he
takes to be a flaw in the study. However, his argument is a red
herring because of the key fact that you
don’t have to be able to explain something in order to prove that it
exists. If you can show a solid correlation, you’ve proved the causal
relationship.
27. Humberto supports a causal claim with a correlation
implied by the word "study," which is a direct argument.
Camilla makes a counter argument by arguing with
Humbert's interpretation of the data in Humberto's argument. She commits a
red herring fallacy because the key fact
is that Humberto does not have to prove that herbal tea always
make people taller, only that it sometimes does.
28.
vi. No way.
vii. No. Are you kidding?
viii. Noooooooo.
ix. Ha! No.
(All answers are "no.")
29.
i. No.
ii. Yes.
iii. No.
iv. Yes.
v. No.
vi. Yes.
vii. No.
If you need to make up the quiz for this chapter, use the Make
Up Exercise
For more practice, you can download
and do the practice/makeup exercises.
(Make sure the document margins are set to 0.5 inches or narrower.)
Copyright © 2013 by Martin C. Young