Solutions to Problem Set 1.

Problem 1 (20p)

Figure 1 Figure 2

Figure 3 Figure 4

Figure 5 Figure 6

Figure 7 Figure 8

Figure 9 Figure 10

Figures 1-10 above depict preference relations of ten different people on a set of three or four
alternatives. (For each decision maker her set of alternatives constitutes the entire domain of choice.)
Alternatives are marked as dots and preferences are marked with arrows: the direction of an arrow is
the direction of the preference relation, no arrow between two dots means that an individual is
indifferent between these two alternatives.

This study source was downloaded by 100000832570404 from CourseHero.com on 07-11-2022 08:13:28 GMT -05:00

https://www.coursehero.com/file/16896616/homework-1-solutions/
 Page 2 of 4

In each of the ten figures preference relation either is or is not transitive and also, independently of
the transitivity of the preference relation, the indifference relation is or is not transitive.

In the following table put “true” or “false” in each of the 20 empty cells.

Preference relation is Indifference relation is

transitive transitive
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10

SOLUTION:

Preference relation is Indifference relation is

transitive transitive
Figure 1 False False
Figure 2 False True
Figure 3 True True
Figure 4 False True
Figure 5 False True
Figure 6 True False
Figure 7 True False
Figure 8 True True
Figure 9 True True
Figure 10 True False

Problem 2 (10p)
Assume a voter has a strict preference over any two candidates in the set of Clinton, Obama and
Edwards; in other words, it is not the case that he is indifferent between some two candidates in this
set. In addition, assume that his preference relation is not rational.

(1) Prove that a voter who prefers Obama over Clinton has to prefer Edwards over Obama and
Clinton over Edwards. Present your reasoning.

(2) Prove that if all voters are not rational then either (i) all of them have identical preferences or (ii)
the set of all voters is split into two subsets such that in each subset all voters have identical
preferences.

This study source was downloaded by 100000832570404 from CourseHero.com on 07-11-2022 08:13:28 GMT -05:00

https://www.coursehero.com/file/16896616/homework-1-solutions/
 Page 3 of 4

SOLUTION: With three elements in the domain (O,C,E) and strict preferences between any two
elements there are 8 possible preference systems. (If you cannot see why, drawing all possibilities is
a good idea.) Only two of them are not rational: clockwise cycle and counter-clockwise cycle. From
this observation (1) and (2) follow in an obvious way.

Given Problem 3 below, you may want to note that if a decision maker were setting his preferences
at random, his probability of ending up with rational preferences will be 0.75 (6 out of 8.)

Problem 3 (8p)
Let’s see now how likely would we observe a rational outcome if a decision maker acted randomly.
Consider the case of three alternatives X, Y and Z. Assume that deciding between any two of these
alternatives, e.g., X and Y, the decision maker rolls a die and depending on the outcome decides Y 
X or X  Y or YX, each with equal probability. (More specifically, assume that a decision maker
takes any two alternatives, say X and Y, and rolls a die; if one or two dots come up he decides that he
is indifferent between X and Y, if three or four dots come up he decides that he prefers X over Y and
if five or six dots come up he decides that he prefers Y over X. Then he does the same for X and Z
and for Y and Z.)

ADVICE/HIT: Write out all possible outcomes. Each outcome should be depicted as a graph with
three dots corresponding to x, y and z and arrows, or lack thereof, between any two dots. Now, each
graph depicts a rational or a non-rational preference system. Which graphs correspond to rational
systems? What is the relative frequency of the rational ones in the set of all systems? Note that this
frequency is the probability that a preference relation that is determined at random, as described
above, will end up being rational?

SOLUTION: How many different preference systems can we get? We have 1 with three
indifferences, 6 with two indifferences (all 6 are not rational), 12 with one indifference (6 are not
rational), and 8 with no indifferences (2 are not rational), thus a total of 27 with 14 being not rational.
Thus the probability that his preference is rational is 13/27 ≈ 48%.

Problem 4 (3 extra credit points)

Let’s go back to explaining the survey data in which 60% chose Edwards over Obama, 60% chose
Edwards over Clinton, 60% chose Clinton over Obama and then when asked to cast a single vote for
one of the three candidates, 45% chose Clinton, 30% Obama, and 25% Edwards.

(1) Prove that it is not possible that all voters are rational (have transitive preferences.)

(2) Suppose now that the distribution of votes was not 45% chose Clinton, 30% Obama, and 25%
Edwards but 38% Clinton, 32% Obama and 30% Edwards; pair-wise data remains unchanged.
Prove that it is possible now that all voters are rational (have transitive preferences) and calculate the
percentage of votes for each of the six different orderings of the three candidates.

SOLUTION:
Part (1): First, note that the way the survey was done forced voters to have strict preferences i.e.,
they could not be indifferent between any two candidates. Second, if the voters were rational, then
all preferences would have to be transitive. For example, if someone prefers Edwards over Clinton
and Clinton over Obama then he has to prefer Edwards over Obama. From this we conclude that the
set of all voters can be partitioned into the following six sets:

This study source was downloaded by 100000832570404 from CourseHero.com on 07-11-2022 08:13:28 GMT -05:00

https://www.coursehero.com/file/16896616/homework-1-solutions/
 Page 4 of 4

CEO (c)
COE (c')
EOC (e)
ECO (e')
OEC (o)
OCE (o')
where E, O and C stand for Edwards, Obama and Clinton and order in which the candidates are
listed, e.g., CEO, means that C is the most preferred while O is the least preferred of the three and
letters in parentheses stand for the percentage of the group with the corresponding preferences. Note
that c + c' + e + e' + o + o' = 100. All numbers here are percentages.

Suppose now that the results, as assumed in (1), are possible when all voters are rational.

This means that:

since Edwards gets 60% against Clinton we must have: e + e' + o = 60; (1)
since Clinton gets 60% against Obama we must have: c + c' + e' = 60; (2)
since Edwards gets 60% against Obama we must have: e + e' + c = 60; (3)
since Clinton got 45% when voters were choosing from the set of three: c + c' = 45; (4)
since Obama got 30% when voters were choosing from the set of three: o + o' = 30; (5)
since Edwards got 25% when voters were choosing from the set of three: e + e' = 25.(6)

Now, from (1) and (6) we infer that o = 35. But this stands in contradiction with (5). Hence it is not
possible to get (1) thru (6) with all voters being rational.

Part (2): In this part conditions (1) thru (3) are the same and conditions (4) thru (6) change to the
following:
since Clinton got 45% when voters were choosing from the set of three: c + c' = 38; (4')
since Obama got 30% when voters were choosing from the set of three: o + o' = 32; (5')
since Edwards got 25% when voters were choosing from the set of three: e + e' = 30.(6')

From (6') and (1) we get o = 30. From this and (5') we infer that o' = 2.
From (6') and (3) we get c = 30. From this and (4') we infer that c' = 8.
From (4') and (2) we get e' = 22. From this and (6') we infer that e = 8.

And so we conclude that it is possible that all voters are rational and that the respective percentages
have to be as derived above.

This study source was downloaded by 100000832570404 from CourseHero.com on 07-11-2022 08:13:28 GMT -05:00

https://www.coursehero.com/file/16896616/homework-1-solutions/
Powered by TCPDF (www.tcpdf.org)