Agriculture Reference
In-Depth Information
7. Measuring preferences for food attributes
7.1. Data and methodology
The sample was designed following the protocol described by [29]. The sample was drawn
proportionate to population size by county in Georgia, North Carolina and South Carolina.
After specifying the sampling frame parameters, the required sample was purchased from
Survey Sampling Inc. Researchers designed and formatted a Fuzzy Pair-wise Comparison
(FPC) questionnaire to be compatible with the data collection protocol of Survey Monkey, and
trained enumerators to use the questionnaire to collect the data. Enumerators asked consumers
to make pair-wise comparisons of five food attributes: nutritional value, hygiene, taste,
affordable price and freshness, in order to determine their preference for one attribute over
the other. The selected attributes are consistent with the studies which have been done in the
U.S. [30]. Data were collected from a random sample of 412 respondents.
In this study, FPC was used to derive a measure of an individual consumer's preferences for
fresh fruit and vegetable attributes. The main reasons for using FPC are: 1) The FPC is similar
to traditional pair-wise comparisons. Consumers are asked to compare the attributes one pair
at a time. However, unlike the traditional pair-wise method, consumers are not forced to make
a binary choice between two attributes. Consumers are permitted to indicate the degree of
preference for one attribute over another, and response indicating indifference between
attributes is permitted. 2) Unlike the other methods, the scale values are based on the respon‐
dent's entire set of paired comparisons. 3) FPC more accurately represents the natural range
of response patterns that are possible. The consumer's fuzzy preference matrix R with elements
can be constructed as follows [40]:
ì
0
if i j i j
= "
, 1,...,
, 1,...,
=
n
=
í
R
(1)
ij
r if i j i j
¹ "
=
n
î
ij
In the FPC method, a measure of preference, μ can be calculated for each attribute by using
the consumer's preference matrix R. The intensity of each preference is measured separately
using the following equation:
1/2
æ
n
ö
(
)
å
2
(2)
m
= -
1
R n
/
-
1
ç
÷
j
ç
ij
÷
è
ø
i
=
1
where μj has a range in the closed interval [0,1]. A larger value for μj indicates greater intensity
of preference for attribute j. Consequently, fresh fruit and vegetable attributes are ranked from
most to least preferable by evaluating the μ values. Then, Friedman and Kendall's W tests were
