Академический Документы
Профессиональный Документы
Культура Документы
Experimental Psychology
VOL. 60, No. 5 NOVEMBER 1960
on expectation but also on the dispersion ideal point corresponding to his pre-
of the possible outcomes, i.e., for a given ferred amount of variance. The folded
expectation an individual may prefer J scale is called an I scale and corresponds
certain amounts of variance over others. to his preference ordering on the gambles.
That this is a significant variable has It is convenient to make the following
recently been urged by Allais (1953), distinction. If the J scale is the same
who criticizes the "American School" for all individuals at the level of an
for disregarding it; more recently, Roy- ordered-metric scale it is called a "quan-
den, Suppes, and Walsh (1959) studied titative J scale" whereas if it is the same
the interaction of variance preferences only at the level of an ordinal scale it
with expected value in the context of is called a "qualitative J scale."
measuring the utility of gambling. If the preference orderings of all 5s
Perhaps one of the principal reasons are generated from a common quantita-
for the long neglect of a concept of tive J scale, then the variety of orderings
variance preferences is that with the which can occur is, at most, one more
use of the expected utility model, any than the number of midpoints, (J) + 1.
particular variance preference may be If they have only the qualitative J
equally well interpreted as a utility scale in common, then the variety of
function on money. In fact, one of the admissible preference orderings is 2""1.
current methods for constructing such In each case, however, not only is the
utility functions is based upon prefer- number of admissible preference order-
ences among bets with the same expecta- ings limited but also the type; they must
tion and equiprobable outcomes but be only those obtained by folding a J
different variance. This method has scale between a pair of midpoints
been most thoroughly developed by this leads, for example, to the restriction
Davidson, Suppes, and Siegel (1957) that the least preferred alternative on
and has also been used by Coombs any I scale may only be one of the
and Komorita (1958). It becomes two end stimuli on the J scale.
difficult to test for the existence of This distinction3 between a quantita-
variance preferences after having used tive J scale and a qualitative J scale
them to construct a utility function. is relevant here because if the measures
It is proposed here, however, that a on the J scale were actual objective
test can be made which will permit a variance, the preference orderings would
choice between the variance preference be limited to those generated from a
and the expected utility interpretations. common quantitative J scale. If the
The argument is as follows. preference orderings were generated from
Consider a set of n alternatives, each a common qualitative J scale one might
alternative being a gamble with two speak of a scale of subjective variance.
outcomes, at probabilities of .5, and all For > 4, ! > 2"-1 > (?) + 1. For
gambles having the same objective example, for n = 6 there are 720
expected value. If the expected utility possible preference orderings, all leading
model is assumed, then for any prefer- to monotonically increasing utility for
ence ordering of these alternatives, a money, but only a specified subset of
utility function for money which is 32 of them may be interpreted as
monotonically increasing can be found. subjective variance preferences, and only
The number of possible orderings, then, 16 of these as preferences generated
which might be "explained" by a
utility function for money is n! 'Another way to formulate the distinc-
From the point of view of the unfold- tion between a quantitative and qualitative
J scale is in terms of a cardinal J scale on
ing theory of preferential choice (Coombs, which an individual's single-peaked prefer-
1952) the amount of variance of each ence function descends linearly in both
gamble may be considered a point on a directions from the ideal point (a quantita-
continuum of variance and each indi- tive J scale) or descends monotonically
vidual folds this Joint (J) scale at his (a qualitative J scale).
COMPONENTS OF RISK IN DECISION MAKING 267
TABLE 1
TABLE OF GAMBLES USED AS STIMULI
five sets. Set I is the column of 6 bets with To increase interest and make the situa-
a constant skewness of 5 = 1.15 (probability tion more realistic, two additional instruc-
of winning of p = 1/3). Set II is the column tions were given: (a) at the end of the last
of 6 bets with S = Q(p = 1/2). Set III is the session some of the items would be picked
column of 6 bets with S = -1.15(p = 2/3). and played with no exchange of money so
Set IV is the row of S bets with a constant that "you can see how you would have come
variance of $1.00. Set V is the row of 5 bets out if we had played for keeps", and (V)
with a constant variance of $25.00. Sets I, after that, 5s would be "given an oppor-
II, and III were used to study variance tunity" actually to play two of the bets for
preferences under constant probability treat- real money.
ment; Sets IV and V were used to study Procedure.The 5s appeared in a group
skewness (probability) preferences under on three successive evenings. On Day 1
constant variance treatment. In each of they went through the first two replications;
Sets I, II, and III the stimuli are labeled from on Day 2, the next three; and on Day 3,
A to F in order from low to high variance. the last three. There was considerable
In each of Sets IV and V the stimuli are difference between 5s in the amount of time
labeled from 1 to S in order from p = 1/6 to needed to complete the booklets, ranging
p = 5/6. from 28 to 74 min. on Day 1 and from 9 to
Within each set, the bets were exhaus- 60 min. on Day 2. Nevertheless, all 5s were
tively paired, producing 65 distinct pairs, paid the standard rate promised them, $2.00
as follows: Sets I, II, and III, 15 pairs each; per session.
Sets IV and V, 10 pairs each. Each of the At the end of Day 3, 10 pairs were chosen
65 pairs was presented 8 times; an additional at random and a die thrown for each to
56 pairs of bets, not used in this analysis, determine the outcome of each bet. The 5s
brought the total to 576 pairs. The stimuli were encouraged to keep track of what they
were presented in dittoed booklets which would have won playing the bets they had
were constructed so that one replication on picked. Then 5s were given $.50 credit and
all pairs was completed before the next was asked whether or not they would care actually
started. Within each replication, the stimuli to play a bet of the form: 1/2 to win $<:25,
were presented in a random order. 1/2 to lose $.25. This was done twice. (This
Instructions.The 5s were told to choose, bet had appeared eight times in the booklets.)
in each pair, the bet they "would prefer
to play" and to indicate this decision by
making a check in a box beside the bets. RESULTS
They were also told to treat each bet sepa-
rately, as if it were the only one offered. Analysis of Inconsistency of Preference
This was done in the hope of avoiding mul- Purification of the data.The first
tiple-bet strategies such as choosing a few
of each kind of bet to keep a "balanced step was to eliminate 5s whose judg-
portfolio." ments are not significantly different
COMPONENTS OF RISK IN DECISION MAKING 269
TABLE 7 Source df MS F