34 THE JOURNAL OF CONSUMER RESEARCH
cision. However, the ability to select correctly the "best"
brand was demonstrably poorer at both low and high
levels of information load compared to intermediate
levels.
Because of the substantial implications that these
findings have for public policy decisions, replication at-
tempts using different subjects, products, and improved
methods are necessary. The fundamental ways in which
the current study differs from the earlier investigation are
as follows: (1) Housewives rather than students served
as subjects. (2) Rice and prepared dinners were utilized
as the products of interest rather than laundry detergent.
(3) The item of information presented to the subject
was made to conform to the information theoretic defini-
tion of an information "bit," by using dichotomous
information dimensions. (4) Maximum information
load was extended from 72 items to 256 bits. (5) Each
of the variables making up information load (that is,
number-of-brands and bits-of-information-per-brand)
was articulated in terms of four rather than three levels.
(6) Additional, qualitatively superior dependent
measures were introduced. (7) Finally, the current in-
vestigation proceeded beyond the. basic question of
information load to examine how package information
displays can best be organized.
METHOD
Subjects
The subjects were 192 paid housewife volunteers
residing in the greater Lafayette, Indiana community.
Selection procedures insured a socio-demographically
heterogeneous sample in terms of age, education, and
family composition.
Design
A 4 (number-of-brands) X 4 (number-of-bits-of-
information-per-brand) between-subjects analysis of
variance design was employed in which subjects were
randomly assigned to one of sixteen experimental cells
(n = 12 per cell). There were either 4, 8, 12, or 16
brands and either 4, 8, 12, or 16 bits of information per
brand presented to the consumers for each product.
All subjects responded to information regarding two
products: rice and prepared dinners. In order to avoid
providing subjects with ideas regarding the hypotheses
of the study and thereby generating possible demand
characteristics (cf. Orne, 1962), each subject remained
in the same "brand x information bits" treatment con-
dition for both products. The order of product presenta-
tion was counterbalanced for subjects within each cell.
Operationalizing the Independent Variables
Package Information Load. Examination of current
supermarket packages for various brands of rice and
prepared dinners, as well as discussions with knowledge-
able individuals,' revealed that packages for these pro-
ducts often display at least twenty different dimensions2
of information (price, container size, nutritional com-
ponents, storage instruction, etc.). Moreover, most of
these dimensions are relatively complex from an in-
formation theoretic perspective. Assume, for example,
that the actual number of "calories per serving" for a
particular brand of rice was. one of the 128 whole integer
values between 50 and 178 calories. To determine which
specific value is involved (e.g., 98 calories per serving)
can require as many as seven bits of information. Each
information dimension was therefore simplified so that
only a single bit would be required to determine the
specific value involved (e.g., high vs. low calories per
serving; high vs. low cholesterol). Sixteen such relatively
"simple" dichotomous information dimensions were
generated for each of the two products.
The specific information that subjects in each of the
cells received for rice was determined as follows. First,
a 16 (brands) X 16 (bits of information) matrix was
constructed so that the 16 bogus brands, labeled A
through P, appeared as an alphabetically ordered margi-
nal row across the top of the matrix, while the 16 in-
formation dimensions appeared as a marginal column
down the left-hand side. Next, the specific value of
each dimension for each cell was randomly selected
from the two values that each dimension was permitted
to have(such as high vs. low cholesterol content). The
information developed for the 16 X 16 matrix was that
presented to the twelve subjects in the most complex
cell. The specific information arrays for the housewives
in the remaining cells were generated by starting in the
top left-hand corner of the 16 X 16 matrix and then
working across and down until both the appropriate
number of brands and bits of information per brand
were obtained.
The information associated with each "brand" was
placed onto a separate 4" X 6" index card. A different
letter of the alphabet appeared at the top of each card
to identify that "brand." Thus, for example, subjects
in the "8 brands X 12 bits of information" cell each
received 8 index cards, labeled A through H, with twelve
bits of information taken from twelve different informa-
tion dimensions appearing on each card. It should be
noted that providing the product information in such
a manner presents a simpler crossbrand comparison
task for the consumer than would be the case if actual
product packages were utilized.
Mode of Information Organization. In order to study
the effects of "organized vs. disorganized" and "same
vs. different" informational arrays, both the specific
information and the order in which this information
1 Raymond C. Stokes, the Director of the Consumer Research
Institute, was particularly helpful in this regard.
2 To facilitate communication, the following terminology will
be employed throughout. The term "information dimension"
will be used to refer to a basic category of product information
such as price, calories-per-serving, or type of container, while
the term "information value" will refer to the specific points or
forms that information may assume along these dimensions
such as 32?, 64?, and 73? for price, 42, 50, and 70 for calories-
per-serving, and box, tube, plastic bag, or jar for type of con-
tainer. Thus, information dimensions are associated with pro-
ducts, and information values are associated with specfic brands.
This content downloaded from 194.254.129.28 on Sat, 21 Sep 2019 13:15:48 UTC
All use subject to https://about.jstor.org/terms