Extension of the consistency of the data obtained with the Ideal Profile Method:
Would the ideal products be more liked than the tested products?
Thierry Worch
a,b,
⇑
, Sébastien Lê
b
, Pieter Punter
a
, Jérôme Pagès
b
a
OP&P Product Research, Utrecht, The Netherlands
b
Laboratoire de Mathématiques Appliquées, Agrocampus Ouest, Rennes, France
article info
Article history:
Received 26 October 2011
Received in revised form 20 February 2012
Accepted 29 March 2012
Available online 7 April 2012
Keywords:
Consumer
Ideal Profile Method
Liking, Consistency
Permutation test
PLS
PCR
abstract
The Ideal Profile Method is a sensory method in which, for each product tested, consumers are asked to
rate both the perceived and ideal intensities of a list of attributes. In addition, they are also required to
indicate how much they like each product. At the end of the task, three blocks of data are collected from
each consumer: the product profiles, their ideal profile and the liking ratings.
The ideal profiles can be used to help improving the existing products. However, this information
should be carefully managed since (1) it is obtained from consumers, and (2) it describes a virtual prod-
uct. In order to use the full potential of the ideal profiles, and to avoid a possible misinterpretation of the
data, one has to ensure that the information collected is consistent.
The process checking for the consistency of the ideal profiles proposed here is based on the liking rat-
ings: an ideal product should achieve higher hedonic ratings than the tested products, if it would be
tested. But since the liking scores of the ideal products are unknown, they are estimated first. However,
the comparison between liking scores (estimated for the ideals, measured for the tested products) would
only make sense if the ideal descriptions have not been randomly given. For that matter, a hypothesis test
checking for the significance of the ideal profiles is defined.
In the perfume example provided, it appears that most of the consumers did not describe their ideals
randomly. In addition, the estimations of the ideals liking scores are high compared to those given to the
tested products. Hence, for most of the consumers, the ideal profiles are considered as consistent accord-
ing to the potential liking of their ideal profiles.
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
The Ideal Profile Method (IPM) is a method which aims at
acquiring sensory data from consumers. During this test, products
are presented in a sequential monadic order taking care of order
and carry over effects (MacFie, Bratchell, Greenhoff, & Vallis,
1989) to consumers, who are asked to describe them on a set of gi-
ven attributes for both perceived and ideal intensities. During the
task, the same consumers are also asked to provide hedonic ratings
of the products.
In this sense, the IPM can be seen as a combination of QDA
Ò
(profiling products; Stone, Sidel, Oliver, Woosley, & Singleton,
1974) and JAR scaling (providing ideal profiles).
The application potential of the data provided from the IPM is
large as three types of information are obtained from each con-
sumer: the sensory profiles of the products (i.e. how consumers
perceive the products), the vector of hedonic scores (i.e. how much
consumers like the products) as well as the ideal profiles (i.e. what
are the consumers’ expectations) (Van Trijp, Punter, Mickartz, &
Kruithof, 2007).
Gathering this diverse information, and more specifically the
ideal profiles, is crucial as it can help manufacturers to improve
existing products (Worch, Dooley, Meullenet & Punter, 2010). In
this sense, the IPM can also be seen as an alternative to statistical
methods such as the external preference mapping (PrefMap;
Carroll, 1972) or the Landscape Segmentation Analysis (LSA; Ennis,
2005) which aims at estimating global or individual ideal profiles
from sensory profiles (usually provided from experts or trained
panels) and consumers’ liking scores.
However, this information (the ideal descriptions) is delicate as
(1) it comes directly from consumers and (2) it describes virtual
products.
In order to use the full potential of the ideal descriptions, and to
avoid any possible misinterpretation of this data which could lead
to an incorrect reformulation of the tested products, one has to be
sure that the information collected is relevant. Hence, it is impor-
tant to check the consistency the ideal data before use. The proce-
dure proposed here is performed in two steps.
0950-3293/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.foodqual.2012.03.010
⇑
Corresponding author. Address: OP&P Product Research, Burgemeester Reiger-
straat 89, NL-3581KP Utrecht, The Netherlands.
E-mail address: thierry@opp.nl (T. Worch).
Food Quality and Preference 26 (2012) 74–80
Contents lists available at SciVerse ScienceDirect
Food Quality and Preference
journal homepage: www.elsevier.com/locate/foodqual