Overview and policies
This course addresses the ideal standards of judgment and
decision making, and the
ways in which people fall short
according to these standards. Understanding of the ideals and
our limitations can help improve judgments and decisions in such
fields as medicine, law, and public policy.
This syllabus will be revised often. Use the reload button on
your browser to make sure
you have the latest version.
Email, appointments, mailing list
I check my email several times per day. I try to answer all
questions, as do the TAs. To make an appointment with me, check my schedule in my web
page and then send me an email message saying when you want to
meet and what it is about.
Assignments may be submitted by email, or through the web (with
special forms provided for each one). I will convert "doc" files to
text (using "antiword"), unless you warn me not to. (I don't use a
word processor.) The assignments will be revised extensively to
accommodate the much-larger-than-usual class. This will include the
number of assignments, and their due dates. I have not yet begin
these revisions.
The course has a newsgroup, upenn.psych.psych153.
Students post to this with questions and comments, and I put
important announcements there. There is a
mailing list version, for which you can sign up, if you are
not already signed up. (You can unsubscribe if you prefer the
newsgroup version.) The list/newsgroup has
archives
so that you can see what you missed or lost.
The course meets the Quantitative Data Analysis and the
Formal Reasoning and Analysis components of the
College of Arts and Sciences General Requirement (for class of 2009 or earlier).
No kidding.
Exams, assignments, and grades
There are some assignments (which are graded). You may submit drafts
of assignments 4 and 5 for comments. (This may change.) I must
receive them at least a week before the due date. (No
exceptions. Don't ask.) You will get the draft back in time for
you to make revisions.
The midterm and final will consist of short essay
questions. They will include both topics in the reading and
topics covered in class. Exams are open-book, typed, time- and
page-limited, and submitted by electronic mail or web form. The
midterm and final will be designed for two hours of work
(including both writing and looking things up). The starting
time is somewhat flexible. I can instruct the computer to send
you the exam at a particular time. If you leave campus before
the final exam, you can do it remotely. I will send exams to
your Penn email address.
Grades will be based on a weighted sum of all work.
Approximate weights (in percent of grade) are:
A1 8%; A2 6%; A3 8%; A4 8%; A5 15%; midterm 25%; final (non-cumulative) 30%.
Each of these scores will be standardized and multiplied by its
weight, and then the scores will be added up.
Lateness will be penalized. I will follow the rules of the
University, including rules about incompletes, and the Code of Academic
Integrity.
Reading
The main text (T) is my book,
Thinking and deciding. It is available at the Penn Book Center (not
the Penn Bookstore).
I plan to revise this syllabus often. Do not just print it and
expect it to remain accurate. And make sure your browser is set to
reload pages that have been revised (or reload all pages).
Schedule
Here is the course schedule by week, with reading. The letter T
stands for
Thinking and deciding.
Links for the names of topic go to slides that I will present in
class. You should be able to print these if you want to, before
class. But please wait until the last minute to do that, as I am
constantly revising. The slides do not work properly in Internet
Explorer, although you might be able to print them. To print the
slides using Firefox without wasting paper, use the menus: "View / Use
Style / print," then "File / Print." The slides are not intended as a
complete outline of what I plan to cover.
1/16 to 1/23 (no class 1/21)
Thinking and logic: T1-4.
1/28
Probability: T5,
Clinical trials
1/30 to 2/4
Probability and
calibration: T6,
Posner on hindsight bias (NYT 8/29/04).
2/6
Hypothesis testing: T7,
Intelligent design
2/11
Correlation and contingency: T8.
2/13
Irrational belief persistence: T9.
2/15 Assignment 1 due at midnight
2/18 to 2/20
Utility: T10.
2/22 Assignment 2 due at midnight
(here).
2/25 to 2/27
Decisions under uncertainty: T11
3/3 Midterm exam (default time, 2-4)
3/5 to 3/17 (spring vacation 3/10-3/16
Decisions under certainty: T12,
Tritch, T. (2007).
Helping people help themselves.
3/19 to 3/24
Utility measurement and
decision analysis: T13, T14,
Keeney, R. L. (1992). Value-focused thinking: A path to
creative decisionmaking. Cambridge, MA: Harvard University
Press, pp. 3-87, 147-148 (in
Rosengarten), or
Keeney, R. L. (2002).
Common mistakes in making value trade-offs.
Operations Research, 50 (6), 935-945.
3/26
Judgment: T15.
3/31 to 4/7
Moral thinking, utilitarianism, and
fairness: T16, T17,
Million-dollar Murray (Gladwell)
4/4
Assignment 3 due at midnight (requires Firefox 1.5).
4/9
Negotiations (TW-R)
4/14 to 4/16
social dilemmas, and voting (TW-R):
T18,
G. Hardin (1968),
The tragedy of the commons. (Science, 162, 1243-1248).
Approval voting
(Wikipedia)
4/18
Assignment 4 due at midnight by email.
4/21
Intertemporal choice:
T19
4/23 to 4/28
Risk:
T20
Airport security follies
4/30
Assignment 5 due at 4 PM (by email or web form for JB, at Solomon for others)
5/9Final exam: 12-2.
Assignments [These will definitely change.]
Assignment 1: Exercise on probability.
This is an exercise in using probability theory to estimate
personal probabilities. The idea is to estimate each probability
(at least) two ways and then see (in assignment 2) which is more accurate. One way
is a direct, intuitive judgment. The other way is a calculation from other
judgments. There are two parts. The first is a
questionnaire.
The second is a set of
questions based on the email you get from completing the
first part. This will be done on the web.
Assignment 2: Further analysis of the class data. Answer
these questions.
Assignment 3: Utility judgments
After reading ch. 13 of T&D, do
this
assignment. You need
Firefox (version 1.5.0.1 or later)
for his assignment. After you are done, you may view your graphs again
by cutting and pasting the numbers with the plus signs (and not the commas)
into this
viewer.
Assignment 4: Decision analysis.
Carry out a multi-attribute analysis of some decision that you
might face. (Be imaginative if you wish. You might, for example,
be president of a country some day, or head of the IMF.) Make
sure that the decision has at least three options and at least
three attributes.
This is not a paper. You don't need to go on at great length
describing the decision in words. Just include the minimum to
make it clear what the decision is about, and what the attributes
are, if that is not clear from their names.
You will benefit most if you think about fundamental values in
Keeney's sense. So, if you make up an initial list of things
that you care about, ask yourself why you care about each one.
You may find that there is some deeper goal that motivates it.
Keeney gives several other helpful suggestions to discover
fundamental values. For example, one decision you might face is
choosing a career. Many of the things that people say when asked
for their values about this are superficial, e.g., that they want
their career to be ``interesting.'' What is interesting to you
may depend on your other fundamental values about what you want
your life to mean, what you want to use it for. The values in the
birth-control example below resulted from application of Keeney's
method by several students in a former class.
Repeated decisions, like what to eat, are good for this
assignment, because the emphasis is on the discovery of values.
Thus, you can put aside the complications that result from the
fact that one of your goals in many decisions is variety.
You should hand in something like one of the examples we discussed
in class. Do not worry about the precision of assigning
values to intermediate cases. Do worry about assigning sensible
weights. You should explain how you did that for one attribute
(relative to the most important one). Make sure you do the
reading before you do this.
The weights are determined first by picking the most important
attribute range and then comparing other ranges to that.
You can pick the most important range in two ways. First, you
might just ask yourself which is more important, the
difference between the top and bottom of one range (as you
have defined the top and bottom) or the difference between the
top and bottom of another. Second, you could make up a
hypothetical decision, e.g., between two option that are
identical except that one is best on attribute A and worst
on B and the other is the opposite.
The simplest way to determine weights is by direct judgment.
How big is the difference between the top and bottom of the
smaller range, compared to the difference between the top and
bottom of the larger range (or vice versa)? (See the text for
other methods.) Make sure to explain how you have done this for
one dimension.
In addition, you should carry out and explain at least one
test of consistency. A test involves estimating one of the
weights in two different ways and explaining how you resolved any
disagreement that you found. One way is to make judgments
relative to some dimension other than the largest and then check
ratio consistency: if the weight of B relative to A is .6 and the
weight of C relative to A is .3, then the weight of C relative to
B should be .5.
Here is an optional
form for submitting the assignment.
Assignment 5: Discussion paper.
Write a short (absolute maximum 200 lines, 70 characters per
line; minimum 25 lines) reflective discussion of some topic from
the course. (Here is a
form
if you want to use it.) This should require no additional
research, although you should of course acknowledge any sources
that you use because you already know them (including course
reading). Do not feel obliged to fill up the 200 lines; 125
should suffice if you write succinctly.
One possible format for this paper is a philosophical
discussion of some question from the course, such as whether
self-deception is rational (but probably not that one). Try to
avoid questions that have completely trivial answers and
questions so big that a book would be required to answer them.
Avoid mere recapitulation of arguments in the reading as well as
unsympathetic attacks on it. Avoid one-sided presentations that
ignore the arguments of the other side. Feel free to ask if a
topic is appropriate. A good model to follow is the discussion
of self-deception on pp. 71-73 of the text, especially the
discussion of whether self-deception can be rational. Feel free
to skip ahead in the reading if you wish to hand in the paper
early. If you choose this kind of topic, make sure to consider
(and rebut, or take into account) objections to arguments you
make.
Another possible format is something more like a research
proposal. Take a result from the course, propose alternative
explanations of it, and suggest ways of distinguishing them.
Last modified
