Académique Documents
Professionnel Documents
Culture Documents
www.emeraldinsight.com/0954-7541.htm
2. Methodological framework
The RG is an interview technique developed by Kelly as a part of his Personal
Construct theory (1957). The personal construct theory was proposed by Kelly to
analyse the way people look at and evaluate the world surrounding them. According to
this framework people adopt a ‘‘scientific’’ attitude, formulating and testing hypothesis
about people, events and things and verifying the existence of possible relationships
between them. The general way these hypotheses are formulated is through bi-polar
constructs. These are implicit dimensions describing contrasting poles existing at
different levels of reality and are organized in a hierarchy (e.g. sympathetic vs.
antipathetic, traditional vs. modern and so on).
The RG technique of interview was developed as a tool to elicit the constructs that
people adopt to describe the ‘‘elements’’ (as people, events, things) the researcher is
interested in (Fransella et al., 2003). The adoption of a RG is a promising strategy for
exploratory surveys as it allows respondents to freely express their perceptions. This
is because no descriptive categories are suggested by the researcher. In fact, during a
RG interview people are asked to define by themselves the dimensions suitable to
describe differences between alternatives. The interviewer role is simply to submit the
alternatives (‘‘elements’’) to be described helping respondents to express the Consumers’
contrasting poles of the constructs they use.
Although the relevance of RG technique in marketing had been debated from a
perception of
methodological point of view (Marsden and Littler, 2000), there are several feature of wine packaging
this approach that make it attractive for consumer behaviour studies (Gains, 1994).
First, unlike other forms of interviews, the RGT eliminates the interviewer bias in the
definition of descriptive dimensions. Second, as the descriptive dimensions elicited
through RG in-depth interviews are expressed in a terminology defined by the
35
respondent himself, RG represents a valuable tool in the definition of marketing
strategies.
Finally the way a RG interview is developed reduces to a relevant extent the
difficulties that usually arise during non structured interviews when respondents are
asked to freely describe the characteristics of a specific product. Conversely, within the
RG interview the description of the products directly emerges from the elicitation of
the differences perceived by respondents when they face alternative expressions of the
same ‘‘element’’. For instance, following the so called ‘‘triadic’’ approach (Fransella et al.,
2003), the respondent is systematically asked to identify one of three alternative objects
(e.g. three different packages of the same product) that is perceived as different from
the other two. The choice is then synthesised with an expression (a word, a simple
proposition) representing the ‘‘construct’’ underlining the separation that has been
made.
The RG approach has been applied also in food marketing research, to study the
acceptability of foods (McEwan and Thomson, 1989) and the evaluation of beverages
with respect to different contexts of consumption (Gains and Thomson, 1990). In the
latter study the ‘‘grids’’ containing the constructs used to describe a set of alternative
canned lagers are used as input in a consensus analysis performed with the generalized
procrustean technique. This leads to a mapping of the different products in a
‘‘descriptive’’ space defined by the elicited constructs. The same approach was adopted
in our study in order to provide a first assessment of the relative positioning of a set of
competing wines that includes the two products of the winery that promoted the
research. In addition, due to the exploratory nature of the study, the multivariate
analysis of grids has been complemented by a content analysis of the transcript of the
interviews, with two aims: to clarify the meanings attributed by respondents to the
elicited constructs and to explore the attitudes and the motivations behind the elicited
system of constructs. In the next section the research design will be described in detail.
4. Consensus analysis
An example of RG for rosè wines is illustrated in Table I. To simplify the table
constructs are identified with only one of the two poles elicited by the assessor. The
higher the score assigned to a wine the more the pole indicated in the column header is
perceived as appropriate to describe the wine corresponding to the considered row.
Constructs
Really expressive Traditional Appropriate Label suggests
Wines Large bottle label label name for quality
Rosè 1 2 5 7 7 6
Table I. Rosè 2 3 7 4 5 5
An example of RG for Rosè 3 7 1 5 1 1
the rosè wines, assessor Rosè 4 4 4 1 2 1
no. 7 Rosè 5 1 2 2 2 2
Obviously, number of columns and column headers are different across grids as they Consumers’
depend on the set of constructs used by each respondent in describing the bottles
of wine.
perception of
Each grid can be seen as the representation of the object (the bottle) in an n-dimension wine packaging
space the dimensions of which are defined by the constructs. Due to the quantitative
nature of the collected data a multivariate technique could be used to compare these
representations. The procrustean approach offers a tool appropriate to this scope
(Krazanowski, 1998). In a procrustean analysis two or more alternative configurations of
37
the same set of objects in an n-dimension space are transformed through translation,
rotation and dilation of the space itself. All these operations, without influencing the
internal relationship existing between points, tend to create a new space in which the
individual configurations match as possible. The outcome of such transformation can
be analysed from two points of view: by describing the consensus configuration,
represented by the centroids of the individual points in the transformed space; and by
analysing the differences existing between individual configurations.
Following Gains and Thompson (1990), in order to perform a generalized
procrustean analysis and obtain a consensus configuration, the RG dimensions were
standardized. All the grids were enlarged to the maximum size defined by the assessor
which had elicited the maximum number of constructs, adding to the other grids the
appropriate number of zero columns.
The GP analysis was carried out separately for white and rosè wines using the
GENPROCRUSTES procedure of the Genstat package for statistical analysis. Table II
shows the percentage variance accounted for by projecting on the consensus axes the
final configuration for each assessor.
An acceptable homogeneity between individual profiles and with respect to the
consensus configuration seems to emerge. The columns show an increasing variation
of data across assessors moving from the first through the fifth factor.
Factors
Assessors Fac 1 Fac 2 Fac 3 Fac 4 Fac 5 Fac 1 + fac 2
0,4
light-coloured glass
dark-coloured glass
0,3 w3
w5 0,2
0,1 w4
0
w1
-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5
-0,1
-0,2
w2
-0,3
w6
-0,4
-0,5
Figure 1.
Consensus configuration small and gaudy labels with dark and original
for the white wines colours, innovative capsule
Factors Consumers’
Assessors Fac 1 Fac 2 Fac 3 Fac 4 Fac 1 + fac 2 perception of
1 55.9 20.9 15.0 8.2 76.8
wine packaging
2 24.4 30.5 35.2 9.9 54.9
3 32.6 51.1 10.0 6.3 83.8
4 37.9 20.2 22.9 19.0 58.1
5 47.7 27.3 18.8 6.3 74.9 39
6 31.5 39.9 10.3 18.2 71.5
7 46.2 25.1 20.2 8.5 71.3
8 57.5 32.3 6.4 3.8 89.8
9 36.6 30.8 10.9 21.7 67.4
10 46.5 18.7 23.0 11.9 65.2
11 48.4 21.2 11.1 19.3 69.5
12 41.6 31.8 17.7 8.8 73.4
13 46.5 20.0 19.0 14.6 66.5 Table III.
14 46.5 29.6 10.1 13.9 76.0 Percentage variance
15 33.8 42.7 13.7 9.8 76.5 accounted for by final
Consensus cv between 42.4 30.0 15.9 11.8 72.4 individual configurations
assessors 22.0 31.8 45.0 45.9 12.4 for the rosè wines
above all about more objective features of packaging (shape and dimension of bottle,
colours) the first seems to outline a more abstract dimension referring to the dyad
elegance/importance contrasted with anonymity (associated with cheapness).
To check for the existence of clusters of assessors with similar configurations a
principal coordinate analysis was carried out using the residuals of individual final
configurations with respect to the consensus configuration[2]. In Figures 3 and 4 the
assessors are plotted against the first two extracted factor, divided by sex. The first
two axes account respectively for 69 per cent of variance in the case of white wines, and
70 per cent for the rosè case. As a consequence, the plots can be considered a good
colours
careful made capsule, with company mark,
0,5
0,4
anonymous capsule, cheap bottle
elegant, important labels
0,3
r1 r5
0,2
0,1
r4
0
-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5
r2 -0,1
-0,2
-0,3
-0,4 r3
-0,5
Figure 2.
traditional colours, dark capsule, label Consensus configuration
suggests for quality for the rosè wines
IJWM
18,1 0,5
0,4
0,3
0,2
40 0,1
fac2
0
-0,6 -0,4 -0,2 -0,1 0 0,2 0,4 0,6
-0,2
Figure 3.
Principal coordinate -0,3
analysis on the residual of -0,4
individual configurations
for the white wines men women -0,5
fac1
picture of the relative position of each assessor with respect to the consensus
configuration.
No evident cluster seems to emerge. Only in the white wine case the first factor
(horizontal) seems to discriminate between men and women[3].
As a consequence a separate GPA by sex was carried out. A better interpretation of
the factors is achieved in the rosè case, with men reproducing the description of bottles
with the two fundamental dimensions of bottles and dress already found in the white
wine case; moreover the women’s consensus configuration gives rise to an interesting
opposition between the concepts of refined and youthful.
From the consensus analysis seems to emerge a quite clear picture of the main
descriptive dimension that the interviewed consumers use to describe and assess
different bottles of wine. However constructs are synthetic definitions of concepts that
can present a variable and, in some cases, high degree of complexity and abstractness.
In this context a more qualitative approach appears to be a necessary to complement
the multivariate analysis on RGs. In the next section some further findings from the
analysis of the transcripts of the interviews will be presented.
0,5
0,4
0,3
0,2
0,1
fac2
0,0
-0,6 -0,4 -0,2 -0,1 0,0 0,2 0,4 0,6
-0,2
Figure 4. -0,3
Principal coordinate -0,4
analysis on the residual of
individual configurations -0,5
men women fac1
for the rosè wines
5. Content analysis Consumers’
As recalled in the third section, both the structured and the unstructured phases of all perception of
interviews were recorded. After a complete transcription of the tapes, a qualitative
analysis of the answers was performed. Using NVivo, a dedicated software for content wine packaging
analysis, the texts were coded using the elicited constructs as a guide.
First of all a qualitative profile for each wine was drafted with reference to the main
descriptive dimensions emerged in the consensus analysis. Moreover, a comparative 41
analysis was carried out with reference to the different components of the packaging
(bottles, labels, capsules and so on).
Through the analysis of RGs at least two different levels of the ‘‘construing system’’
(Fransella et al., 2003) of the respondents, with a different degree of abstractedness, are
identified. Constructs referring to more concrete attributes of packaging contribute to
define constructs concerning more abstract ones. Bearing in mind this finding, coding
has been finalized also to the identification of quotations that allow a better
understanding of the exact meaning attributed by consumers to constructs and
hierarchical relationships existing between them. In the remaining part of this section
the interpretation of the most important descriptive dimensions used by the
respondent will be discussed. In the presentation of the content analysis we will move
from the more concrete attribute of the bottles towards the more abstract constructs.
At a very basic level of perception stand colours. All respondents talk about colours
both to stress differences and to express preferences. The following quotations well
represent the different role assigned to colours:
. . . it looks like a low valuable wine . . . glass and wine appear too light
I want to see the colour of wine, I do not like dark glass . . .
. . . the capsule’s green is warmer . . . I like the combination with the colour of label . . .
In the case of white wines the colour of glass is the material attribute of packaging
most frequently considered: it seems able to discriminate across wines with a different
perceived level of value. Colours are critical also for rosè wines but in a more complex
way: consumers look for harmony, coherence in the use of colours, also with regard to
the content of the bottle.
Among the material attributes of the packaging also the shape and the size of the
bottle are often cited by consumers as important features to be considered in the
comparison between alternative products. As is shown by the following quotation (as
well as by many others) the size is a clearly perceived visual signal but does not show a
clear correlation with particular messages: it is simply recorded at a very first level of
perception by consumers.
These are both shorter . . .
Conversely, in the white wines case the bordelaise was clearly considered the ‘‘traditional’’
type of bottle, contrasting to different shapes that are immediately observed:
These two bottles have the same long and narrow shape . . . it is an innovative feature
The flat bottom, in both white and rosè wines, has been associated to a lower value
assigned to the bottle, as stressed by the following quotation:
The bottom is flat, it looks like a bottle for water . . .
IJWM The level of abstractedness increases considering the constructs related to labels.
Labels are assessed with respect to their position, shape and size, usually jointly with
18,1 the bottle dimensions. Summarizing the message emerging from the transcripts, labels
seem to be assessed contrasting the information they bring to consumers with their
ability to evoke more abstract functions assigned to the consumption of wine. While
front label is usually considered for evocation, back label is expected to provide to an
informative function, containing the relevant technical information about the wine:
42
This one contains no information . . . that one has a complete description in the back label
. . . innovative labels, searching for a new message . . .
Front and back labels are often jointly assessed: the coherence between the two is
considered as a signal of care in packaging design. Moreover, there appears to exist a
problem of equilibrium and appropriateness in the use of the labels, well expressed by
the following sentence:
. . . the web site on the back label . . . it doesn’t work for an important wine.
Finally the (global) message about the wine mainly passes through the choice of
colours (particularly for rosè), materials and graphic elements of the labels:
. . . watermarked paper has a positive effect on me . . . gives me the sensation of a valuable
wine
The two following quotations can be used to introduce the analysis of the more
abstract (second level) constructs concerning the distinction of a wine.
. . . the bottle is made with care, is longer, with a thorough graphic, golden inscriptions, the
mark on the capsule . . .
. . . it looks cheaper, it seems made carelessly
The words important, polished, refined seem to refer to the same construct and
generally contrast with cheapness and anonymity. Often respondents link distinction
with the care spent to design the bottle, assessing the coherence in the use of
characters, clear printing, homogeneous graphical signs on different parts of the
bottles and so on. Some attributes are indicated as inconsistent with the distinction of a
wine: flat bottom of the bottle, vivid colours in labels and capsules and a light-coloured
glass in the case of white wines.
A recurring theme is also synthesised by the opposition between tradition and
innovation, often associated to the term classic opposed to modern. The following
sentence is a part of a longer quotation where the respondent also talks about bar codes
and ISO certification mark, elements that were present in one of the proposed bottles:
. . . the e-mail on the back is not right . . . it is better a simple, traditional feature . . .
The value attributed to traditional features, that can or cannot be connected with
distinction, is variable with the attitude of the respondent. In some cases tradition is
interpreted as a sign of reliability; in others as a sign of lack of innovation.
. . . it is too ‘‘alternative’’ . . . it doesn’t convey a sense of tradition . . .
. . . it is inspiring, the new shape of the bottle, so different from the canons of Tuscany, the
dark label . . .
Finally, innovativeness is generally appreciated if associate with care in details.
6. Concluding remarks Consumers’
In this work the results of an exploratory survey preliminary to the redesigning of the perception of
packaging of two wines have been presented. A sample of consumers was interviewed
adopting a RG approach to elicit the main dimensions through which they describe the wine packaging
differences between alternative bottles of wine. Collected data were analysed using
both multivariate techniques and the qualitative content analysis of transcripts.
The answer the study gives to the research questions can be summarized as follows. 43
The first glance description of packaging is built by respondents around two
fundamental dimensions. On the one hand consumers seem to be affected by shape,
size and colour of the bottle; on the other hand they consider the dress of the bottle,
represented by the set of the other packaging elements (labels, capsules). Although in
different ways, these two axes of description seem to be relevant for both white and
rosè wines.
The way consumers describe the perceived differences among alternative
packagings and their preferences between them points out the existence of a
conceptual hierarchy. At a very basic level we find concrete attributes like colour, shape
and size. These characteristics represent the very first signal perceived by respondents
in front of bottles. Consumers use them to define other constructs that refer to more
abstract characteristics they use while assessing alternative products and choosing
among them. In the case of the wines considered in this study, the most important
dimensions were the bi-polar dyads of tradition vs. innovativeness, distinction vs.
anonymity and care vs. carelessness. Obviously, consumers with different attitudes
can use the same concept to choose in opposite ways. In any case efforts are devoted
to assess the coherence between the single element/construct perceived and the
overall message the bottle seem to propose. There is, for instance, the case of light
colour of glass for white wines, considered incompatible with the distinction of
the bottle.
Some concluding remarks can be made from a methodological point of view as well.
The RG approach has been confirmed to be an effective tool for conducting exploratory
in-depth interviews. The feature of this methodology of main interest for exploratory
surveys is represented by giving respondents the opportunity to freely express their
perception. Moreover, the triadic technique of elicitation, with its repetitions, allows
respondents to progressively improve and specify their description of the elements.
In the context of this work the analysis of grids using multivariate techniques has
been interpreted as a useful tool in the interpretation of the answers collected during
the survey. As a consequence the consensus configuration obtained cannot be
considered in any way an interpretative model of how consumers rank the wines.
Notwithstanding, the description of constructs is inevitably connected with the
expression of preferences. For this reason the combination of the RG technique with
content analysis allows for a first glance interpretation of ‘‘motivation’’ existing behind
the consumers’ perceptions. This result is witnessed by the elicitation of constructs of
different degrees of abstraction.
Further developments are possible both using quantitative and qualitative
approaches. First the results seem to be a suitable basis for a survey on a
representative sample directed to model the preferences about packaging features. For
example the elicited constructs could be used to design a conjoint analysis that would
be an effective way to identify the preferred combination of attributes for the
packaging of the considered wines.
IJWM The analysis of preferences could be improved also within a qualitative perspective.
Starting from the elicited constructs and using laddering techniques of interview the
18,1 hierarchical structure of motivation linking characteristics, attitudes and values could
be further explored.
Notes
1. As the constructs are independently defined by each assessor, despite their being
44 described using the same words, they can express opposite concepts. For example the
‘‘innovativeness’’ of a bottle could be linked by a respondent to features that another one
links to ‘‘traditionality’’.
2. A symmetrical matrix of similarities calculated from the residuals has been used as the
input data set for the PCO procedure of Genstat.
3. The other adopted criteria of stratification do not show any clear connection with the
factors.
References
Britton, P. (1992), ‘‘Packaging: graphic examples of consumer seduction’’, Beverage Industry,
Vol. 83 No. 8, p. 21.
Fransella, F., Bell, R. and Bannister, D. (2003), A Manual for Repertory Grid Technique, John
Wiley and Sons, Chichester.
Gains, N. (1994), ‘‘The repertory grid approach’’, in McFie, H.J.H. and Thomson D.M.H. (Eds),
Measurement of Food Preferences, Blackie Academic and Professional, London, pp. 51-76.
Gains, N. and Thomson, D.M.H. (1990), ‘‘Contextual evaluation of canned lagers using repertory
grid method’’, International Journal of Food Science and Technology, Vol. 25, pp. 699-705.
Grunert, K.G. (1996), Market Orientation in Food and Agriculture, Kluwer Academic Publishers,
Norwell.
Hall, J. and Winchester, M. (2000), ‘‘What’s really driving wine consumers?’’, The Australian and
New Zealand Wine Industry Journal, Vol. 15 No. 4, pp. S68-72.
Krzanowski, W.J. (1998), Multivariate Analysis. A User’s Perspective, Clarendon, Oxford.
Marsden, D. and Littler, D. (2000), ‘‘Repertory grid technique. An interpretive research
framework’’, European Journal of Marketing, Vol. 34 No. 7, pp. 816-34.
McEwan, J.A. and Thomson, D.M.H. (1989), ‘‘The repertory grid method and preference mapping
in market research: a case study on chocolate confectionery’’, Food Quality and Preferences,
Vol. 1 No. 2, pp. 59-68.
Nomisma (2003), Wine marketing. Il Marketing del vino in Europa: consume, canali, distributori e
importatori, Agra Editrice, Roma.
Reynolds, T.G. and Gutman, J. (1988), ‘‘Laddering theory, method, analysis and interpretation’’,
Journal of Advertising Research, Vol. 19 No. 28, pp. 11-31.
Tootelian, D.H. and Ross, K. (2000), ‘‘Products labels: what information do consumer want and
will they believe it?’’ Journal of Food Products Marketing, Vol. 61 No. 1, pp. 25-38.
Corresponding author
Benedetto Rocchi can be contacted at: benedetto.rocchi@unifi.it