2. Experimental design
An economic experiment provides an appropriate tool to examine
which type of frequency (simple or relative) and which type of representation
(numerical or pictorial) give more accurate and fast responses
to managerial decision making. The experiment presents the following
managerial decision-making environment. Subjects receive the results
of a simple market research exercise andmust choose one out of two alternative
provision levels (number of units) of a good to be allocated in
a market. The market research provides information on the percentage
of the customers in the sample thatwill buy a unit of the good (only one
unit per customer) under the assumption that the survey is completely
reliable (a customerwill buy a unit of this product if and only if she or he
demonstrates her or his intention to buy during the survey). This information
is displayed using four statistical formats (Fig. 1).
Providing information with these four types of presentations avoids
possible word confusions among the four statistical formats. Moreover,
the cost of an unsold product unit (over-supply cost) and the cost of
leaving a subjectwithout the product (under-supply cost) are different.
Subjects choose between two possible answers in each round. These
two possible answers can be of three types: i) Correct and Focal; ii) Correct
and Opposite; and iii) Correct and Extreme. Correctmeans the right
answer to the decision problem, namely that the number of units of the
product to allocate to the market to maximize expected profit given the
sample information, the market size and the asymmetry of supply costs.
Focal refers to the option of allocating to the market a number of units
proportional to the number of clients that will buy the product in the
sample, with no other consideration on the asymmetry of over-supply
and under-supply costs (i.e., given a market size of 1000 potential
buyers, if 30% of clients in the sample buy the product, the number of
units to allocate to the total market will be 300, no matter if the oversupply
cost is larger or smaller than the under-supply cost). The answer
Opposite refers to an answer that is smaller than the focal one, in those
cases where the right answer is larger than the focal answer and vice
versa. The final option is Extreme, whereby the number of units allocated
to the market is larger than the right number in those cases where
the right answer is also larger than the focal answer. The converse is
also true, namely that a choice is extreme when the choice is lower
than the right answer and the right answer is also smaller than the
focal option.
The experiment considers two cost treatments (higher overprovisioning
cost and higher under-provisioning cost). Each treatment
includes two sample sizes (500 and 100) as well as two different proportions
of elicited buyers in the sample (30% and 70%).
As previously mentioned, subjects had to deal with 96 possible
decision-making problems (cards). The use of an orthogonal design reduced
the number of cards to 25 (a quarter of the total). Fromthemedical
literature the experiment borrows the methodology of orthogonal
designs, which is novel in the design of economic experiments. Such a
design keeps the experiment manageable without substantially reducing
its analytical power. Specifically, such a design enables the inclusion
of multiple factors in an experiment because the sample size remains
small, thus making the estimation of both the main effects and all the