Chapter 12: Judgment: Drawing Conclusions from Evidence
- Induction: Process to go beyond available information – drawing inferences about a pattern, based on few
examples
- Inductive conclusions are never guaranteed to be true but if done properly, conclusion are likely to be
true
- Descriptive account of human induction – telling us how the process ordinarily proceed
- Normative account of human induction – telling us how things ought to go
Judgment Heuristics
i.e. Process of education – connection and back-and-forth b/w teacher and student, capacity to learn depends on
the person who has the experience and the person’s memory
- We draw conclusions based on experiences gathered over long time
Attribute Substitution
- Attribute substitution: Form of strategy to try and evaluate some point but don’t have easy access to
target information, rely on other aspect of experience that’s more accessible
i.e. doing well Organic Chemistry “How well have my friends done, who got good grades or did poorly?” – seeking
information about frequencies – many not be easily accessible for you so you have trouble estimating relevant
frequencies
- Availability Heuristic: Relying on availability – ease w/ which things come to mind – as a substitute for
frequency
i.e. Judging f David is lying or not. Try to remember things about David and liars and see if David fits in category
“liar” BUT might be easier to rely on attribute substitution
- Representativeness Heuristics: Substitution using resemblance in place of information about category
membership (i.e. does David resemble your notion of typical liar?)
The Availability Heuristics
- Heuristics: Reasonably efficient strategies that usually lead us to the right answer – but can give us errors
- Availability and representativeness heuristics – attribute being used is easy to assess, being relied on is correlated
w/ target dimension to serve as proxy for target
- i.e. “Are there more words in the dictionary beginning w/ R (i.e. rose, rabbit, rock) or more words w/ R in 3rd
positions (i.e. tarp, bare, throw) More people say there are more beginning w/ R = WRONG
People get this wrong b/c of availability – trying to search for word starting w/ R = MANY vs. words w/ R in 3rd
position = LES
The Wide Range of Availability Effects
- People use strategies like availability in a wide
range of other cases
i.e. people regularly overestimate frequency of
evens that are quite rare – plays role in people
buying lottery tickets, physicians overestimate
likelihood of rare disease and fail to pursue more
appropriate diagnoses
- Why? Unusual events more likely to catch your
attention so they will be well-recorded in memory
– make these events more easily available to you
i.e. participants asked to recall number of examples
in their lives where they acted in asserted fashion.
½ recalled 6 while ½ recalled 12. Then all were
asked some more general questions, including how
assertive they thought they were.
6 examples – easier time “Those examples came
easily to mind; therefore there must be a large
number of examples, I must be can assertive
person”
12 examples – harder time “If these examples are so difficult to recall, I guess the examples can’t be typical for how
I act”
i.e. People would choose to spend resources on more frequent causes of death, rather than investigating rare
problems: Should we spend more on preventing death form car accidents or death form stomach cancer? People
think car accidents and homicide are more frequent, although the opposite is more frequent
Why produce this error? People are heavily influenced by pattern of media coverage – it is the availability
data that people use, making the media bias as important as memory bias
The Representativeness Heuristic
- Representativeness Heuristic – heuristic amounts to an assumption that the categories we encounter are
relatively homogeneous – leads us to act as if each member of category is “representative” of the category
i.e. if someone is a lawyer, we give in the stereotype and expect her to have the traits we associate w/ lawyers – if
someone looks like a typical lawyer – “lawyer traits” – we conclude she is a lawyer
- Assumption of homogeneity leaves us willing to draw conclusions form a relatively small sample – if each
member of a group is representative of group, it’s fair to draw conclusions about the whole after 1 or 3
observations
Reasoning from the Population to an Instance
i.e. tossing a coin and say that the coin has landed head up 6 times in a row – people think it is more likely to come
up tails than heads – “gambler’s fallacy”
There is no mechanism which the history of previous tosses will influence the next one – likelihood of tail on toss
number 7 is 50-50
i.e. Sample size: if we examine a large number of cases, we will find patterns close to overall population – “law of
large numbers” (law stating that the larger the sample, the greater resemblance will be b/w the properties of the
sample and the properties of the population at large). However, there is no “law of small numbers”
Reasoning form a Single Case to the Entire Population
- Hamill, Wilson and Nisbett (1980) showed participants videotaped interview where person identified as prison
guard described his job
1st condition: Guard was kind and compassionate w/ concern for rehabilitation
2nd condition: Guard expressed contempt for prison inmates, no concern for rehabilitation
Some participants were told the 2nd guard was atypical, chosen for the interview b/c of his extreme views
Participants asked question about view on criminal justice system, data showed they were influenced by interview:
those that say humane guard thought prison guards were decent people and v.v
- “Man who/Woman who” arguments: “What do you mean cigarettes cause cancer? I have an aunt who smoked for
50 years and she runs in marathons!” These arguments are persuasive but have force only by virtue of
representative heuristic
Anchoring
- We rely on shortcuts that involve one form or another of attribute substitution - we rely on availability to judge
frequency, indicator of category membership, also use other substitutes (i.e. use emotional response to stimulus as
basis for judging danger)
- Anchoring: Sometimes we don’t’ know the answer to a particular question but have a “ballpark”
idea/answer, we use the initial idea as in “anchor” and then reach our answer by making some suitable
adjustment to that anchor
- We might adjust to little and are more influenced by the initial anchor than we should be
I.e. participants had to estimate how old Gandhi was when he did – i.e age 140 – NO! asked how long he lived –
average 67.
Others asked whether Gandhi lived past 9; then had to estimate how he lived. Given this lower anchor, they gave
lower estimates – average 50
All estimates were off target – Gandhi was 78 when he died
- When judgment errors arise for other reasons, anchoring serves to cement them in place (i.e. trying to
judge which is more frequent in US – death by homicide or death by diabetes. You know homicides tend to be
front-page news, but diabetes is rarely reported unless victim is someone famous and anchoring plays its role. The
initial response, even though misguided, will influence subsequent judgments, making it difficult to undo the error)