Critical Thinking: Inductive Reasoning
Critical Thinking: Inductive Reasoning
Critical Thinking: Inductive Reasoning
Chapter 11
Inductive Reasoning
11-2
Inductive Generalizations
Generalization: statement made about all or
most members of a group.
Inductive generalization: inductive argument
that relies on characteristics of a sample
population (i.e., a portion of the population)
to make a claim about the population as a
whole.
i.e., an inductive argument with a generalization
as a conclusion.
Example: All the bass Hank caught in the
Susquehanna have been less than 1lb. So,
most of the bass in the Susquehanna are
less than 1lb.
11-3
Making Inductive Generalizations
stronger by making conclusions
weaker.
Notice…
All the bass Hank caught in the Susquehanna
have been less than 1lb. So, all of the bass in the
Susquehanna are less than 1lb.
..is a pretty weak argument. Even if Hank
fishes often, the Susquehanna is a big river
and his catches are not enough to justify
such a “sweeping conclusion.”
However, if we changed the conclusion to
“most of the bass are…” or, better yet,
“many of the bass are…” the argument
would be much stronger.
11-4
Evaluating Inductive Generalizations
11-7
Reference Class
In a statistical argument, if you find out
more information about the person in
question, you “narrow” the group (class)
the person is in.
Example:
1. You are a college student who likes essays.
2. 85% of college students who like essays want
cumulative fails.
3. Thus you probably want a cumulative final.
This additional information weakened our
justification for believing that you don’t
want a final.
11-8
Argument by Analogy
Analogy: comparison of things based on
similarities.
Argument from analogy: an argument that
suggests that the presence of certain
similarities is evidence for further
similarities.
Common Form:
1. A and B have characteristic X
2. A has characteristic Y
3. So B probably has characteristic Y too.
Example:
1. Tiffany and Heather are both tall and play
basketball.
2. Tiffany also plays volleyball.
3. So, Heather probably plays volleyball too. 11-9
Evaluating Arguments from Analogy
Are the premises true?
Are the similarities relevant?
Since being tall is helpful in volleyball, the fact that both Tiffany and
Heather are tall is relevant to the previous conclusion.
The more relevant similarities there are, the better.
If we also learn that they both get scholarships if they play more than one
sport, our conclusion is more supported.
Are there relevant dis-similarities?
Irrelevant dis-similarities: hair color
Relevant dis-similarity: job status
The more examples which are also similar, the better.
If Amber and Krissy are also tall and play both basketball and volleyball our
conclusion is even further supported.
The more diversity in the examples, the better.
If Tiffany, Amber and Krissy are different in many ways, except for the fact
that they are all tall and play basketball and volleyball, it seems more likely
that their being tall and playing basketball is relevant to their playing
volleyball. Thus, Heather’s being tall and playing basketball is better
evidence that she also plays volleyball.
Is the conclusion too specific?
“Heather probably plays volleyball” is better supported than “Heather must
play volleyball.”
11-10
Arguing by Analogy
11-11
Induction and Causal Arguments
Causes precede and are constantly conjoined with
their effects.
On a billiard table, we constantly see the cue ball’s
striking of the other balls just before they move. We
assume a causal connection.
And if we were to argue that they are causally
connected, we would cite the fact that the cue ball’s
striking of the other balls always precedes and is
constantly conjoined with the movement of the other
balls.
But as we learned in chapter 6, two things being
constantly conjoined isn’t enough to conclude a
causal connection. Additionally, one thing preceding
another is not enough to conclude a causal
connection.
So how can we argue and conclude that two things
are causally connected? 11-12
Two kinds of causal arguments
Arguments about a single instance:
Example: My car wouldn’t start but I haven’t replaced
the battery in six years. It must have been the battery.
This is just an argument about the cause of one event.
Arguments about a general relationship.
Example: There is a high correlation between smoking
and lung cancer. Smoking must cause lung cancer.
This is suggesting a causal relationship between
smoking and lung cancer (not about a specific person’s
lung cancer).
Notice that it isn’t saying that everyone who smokes
gets lung cancer. The generous interpretation reads
it as: smoking makes one more likely to get lung
cancer.
11-13
The dangers of correlation
Truth be told, we can never PROVE (beyond doubt)
that there is a causal relation between two things;
Hume taught us that.
The best we can do is observe correlation.
But correlation is the best we can do when it comes to
arguments for causation.
But when arguing from correlation, one must be
careful.
We have a tendency to see things as correlated when
we already think they are causally related.
Superstition, belief about bad luck, etc.
But assuming they are connected when trying to prove
they are, begs the question.
To really establish a “link,” we must be careful to
eliminate bias and expectation.
This is why, to eliminate the “placebo effect” medical
scientists do “double blind” studies.
11-14
Correlation and Cause
Even large amounts of correlation are not enough to establish
a causal connection.
Example: big-feet and math (p.327)
News reporters have this problem all the time.
When arguing from correlation, you need to make sure that
there aren’t any other factors that might account for the
correlation.
Example: Vitamin C study (p.329)
Positive Correlation: if A is found together with B 50% or more
of the time B is found, then A is positively correlated with B.
The higher the percentage, the more likely A has a causal
connection to B.
Negative Correlation: if A is found together with B 50% or less
of the time, then A is negatively correlated with B.
The lower the percentage, the more likely that A prevents B.
But all in all, correlation is most often due to coincidence…
Even if x is correlated with y, it could be due to the fact that they
are both the causal result of some other thing z
…so it is wise to always be suspicious.
11-15
Probability
Epistemic Probability expresses how likely we think
something is, given what we believe.
“There is a pretty high probability that I’ll go to the
beach sometime this summer.”
Relative Frequency Probability takes information
about a group as a whole and applies it to individual
cases.
“There is a 90% chance that the operation will be
successful.”
This is derived from the fact that, for 90% of the people
on which the operation was performed, the operation
was successful.
A priori probability are statements that can be
calculated prior to sensory observations.
“There is a 50% chance of getting tails on this coin
toss.”
The nature of the coin determines the “objective”
probability of getting tails.
11-16
Some Clues:
11-17
More on a priori probability
How to figure a priori probability
The probably of either A or B is “Pr(A) + Pr(B)”
Probability of drawing either a King or a 7 is
a standard deck: 1/13+1/13 [i.e, 2/13
(15.4%)]
The probably of getting both A and B is “Pr(A)
x Pr(B)”
Probably of drawing a K and then a 7 (or a
7 and then a K) is 1/13x1/13 [i.e., 1/169
(.59%)]
11-18
Gambler’s fallacy:
Thinking that previous chance occurrences affect
future ones.
The probably of a roulette wheel coming up black
is always 47.37%, even if it just came up black
28 times in a row.
Granted, if you haven’t starting spinning the
wheel yet, the probably of it hitting black 29
times in a row is low. But, if you have already hit
black 28 times, the probably of getting 29 in a
row now is the same as the probably of hitting it
once (because one more is all you need for 29):
47.37%.
11-19
Bet values
Expected Value: The payoff or loss you can expect
from a bet.
How to figure expected value: take the payoff and
multiple it by your odds.
The expected value of a 1/100 chance at $100 is $1.
1/100x$100=$1
If there are multiple payoff, you average them:
1/3rd chance at 0, 1/3 chance at $50, 1/3 chance at
$100. Expected value $50.
(0+50+100)/3=$50
Deal or no deal?:
The “banker” always offers less than expected value
(the average of the amounts left), until the end when
he wants them to take the deal.
11-20
Relative Value
Of course, there are other reasons to take bets other than
payoffs.
Your own needs, preferences and resources can affect the
“value” of a bet as well.
The value a bet has, given such considerations, is the
“relative value.”
Example: The relative value of betting $100 for a long shot at
a billion is high for a millionaire (he can afford it), but low for
a homeless person (who wouldn’t want to risk money he can
use to eat on a long shot at a billion).
Diminishing marginal value: as quantity of bets increase,
the relative value of the bets tends to decrease.
If you are really hungry, you are willing to buy a piece of
pizza for $10 (if that is all that is available). It’s relative value
is really high. But after you eat it, and your hunger subsides
somewhat, its relative value drops and you are less willing to
pay so much. Buy enough pizza, and you won’t be willing to
pay much at all. Buy too much (and eat it) and you won’t take
it for free.
11-21