13. Study 2C: Demonstration of underlying choice strategies
13.1. Study method and predictions
Four distinct types of brand attribute matrices were constructed in the orange juice
category. The brand attribute matrices were similar in structure to the matrix used in study 2A.
All brand-attribute matrices had three brands. There was one dominant brand (Tropicana) and
two non-dominant brands (Oasis and Simply-Orange). The pretests for brand dominance in the
orange juice category are detailed in study 3. The matrix had three equivalently important
attributes: nutrition rating, taste rating and freshness rating and one trivial attribute: extraction
process.
I label the four types of matrices as (a) mixed-trivial, (b) mixed-non-trivial, (c) brand-
only and (d) trivial-only. The target brand had the same ratings on the three important attributes
and the trivial attribute across all four conditions. For the first three conditions, Oasis was the
target brand. For the trivial-only-matrix an unknown brand: Sonera was the target brand. I
expand on the rationale for using a different target brand in the trivial-only-matrix below.
The mixed-trivial matrix had the identical structure as the ground coffee matrix of study
on one of the three important attributes. Specifically, Simply-Orange was rated high on nutrition,
Tropicana was rated high on taste and Oasis was rated high on freshness. Further, the target
brand Oasis had the trivial attribute: Patented Florida extraction while the other two brands had
regular extraction. The mixed-trivial matrix served as the control matrix in which all three types
of processing (brand based, lexicographic and trivial attribute based) were possible as there was
variability between brands, important attributes and trivial attributes.
The mixed-non-trivial matrix had the same structure as the mixed-trivial matrix with one
exception. All three brands had the trivial attribute: Patented Florida extraction. Hence, in the
mixed-non-trivial matrix people could engage only in brand based and lexicographic processing.
Trivial attribute processing was not possible as there was no variability on the trivial attribute
dimension across the three brands.
The third matrix i.e., the brand only matrix was so constructed that all three brands had
identical ratings on the three important attributes as well as the trivial attribute. This restricted
the processing type to only brand based processing. Participants could not engage in either
lexicographic or trivial attribute processing in this matrix as there was no variability in the
ratings of important attributes or trivial attributes across the three brands.
The final matrix was the trivial-only matrix. The matrix was so created that it favored
trivial-attribute processing. To achieve that I created a brand attribute matrix in which the
likelihood of brand based or lexicographic processing was minimal. To minimize brand based
processing, I used three brand names that were equally attractive (Solevita, Sunito and Sonera)
but that were not present in the Canadian market. Further, to avoid lexicographic processing, the
matrix was so designed that the three brands had identical ratings on the important attributes. The
Patented Florida extraction while the other two brands Solevita and Sunito had Regular
extraction. For comparability, the target brand Sonera had identical ratings (on important
attributes and the trivial attribute) as the target brand Oasis had in the previous three conditions.
See appendix 9 for the stimuli (brand-attribute matrices of all four choice sets) constructed for
study 2C.
I argued earlier that the overall choice share that a brand secures, is an aggregation of the
choice share it gains because of (i) prior brand cognitions, (ii) being attractive on an important
attribute and (iii) being attractive on a trivial attribute. Proportionally, greater the number of
contributing factors for choice, greater should be the choice of a target brand. Given that I
theoretically argue and empirically show that trivial attributes lead to a significant choice share
increase, I predict that the choice share of the target brand (Oasis) in the mixed-trivial matrix
[control condition] should be significantly higher than the choice share of Oasis in (i) the brand-
only matrix and (ii) the mixed-non-trivial matrix.
I further predict that the choice share of the target brand in the mixed-trivial matrix will
be less than the choice share of the target brand in the trivial-only matrix. This will happen as in
the trivial-only matrix, the only factor upon which people can make a choice is the trivial
attribute which favors the target brand. This should lead to extremely high choice-share for the
target brand in the trivial-only matrix. In contrast, in the mixed-trivial matrix, trivial attribute is
only one of the three factors affecting choice. While the trivial attribute favors the target brand,
prior brand cognitions and marginal attractiveness on certain important attributes will still
continue to drive choice towards other brands in the choice set. Hence, I expect a lower choice
share of the target brand in the mixed-trivial matrix relative to its choice share in the trivial-only
Participants and procedure. One hundred and eighty nine undergraduate students (66%
Female, 18.3 years) participated in the study. The study design was a 4 condition between
subjects design. The type of brand-attribute matrix was the manipulated variable with four levels
[mixed-trivial, mixed-non-trivial, brand-only, and trivial-only]. Participants were asked to
choose from three orange juice brands that had different product features. Participants evaluated
the brand-attribute matrix and selected one of the three brands. They next provided basic
demographic information and the study concluded.
13.2.
Results
The choice share of the target brand was the dependent variable. The findings of the
study were consistent with my predictions. First, the choice share of the target brand (Oasis) in
the mixed-trivial matrix (29.2%) was significantly higher than the target brand’s choice share in
the brand-only matrix (6.2%),
χ
2 (1, 96) = 8.64, p < .01, Φ = .30. The choice share of the target brand (Oasis) in the mixed-trivial matrix (29.2%) was also significantly higher than its choiceshare in the mixed-non-trivial matrix (10.8%),
χ
2 (1, 94) = 4.87, p < .05, Φ =.22.In my next analysis, I compared the target brand’s choice share in the mixed-trivial
matrix to its choice share in the trivial-only matrix. The target brand (Sonera) in the trivial-only
matrix secured the bulk of choice share (85.1%). Further, as predicted, the target brand’s choice
share in the trivial-only matrix (85.1%) was substantially higher than the target brand’s choice
share in the mixed-trivial matrix (29.2%),
χ
2 (1, 95) = 30.2, p < .001, Φ = .56. The choice share results of study 2C are presented in Table 8 below.Table 8: Study 2C Results – Brand Choice Share
Treatment
Condition Type of Processing Cell size Choice share
1 Mixed Trivial 48 Simply-Orange 33.3% Tropicana 37.5% Oasis 29.2% 2 Mixed Non-trivial 46 Simply-Orange 37.0% Tropicana 52.2% Oasis 10.8% 3 Brand only 48 Simply-Orange 33.4% Tropicana 60.4% Oasis 6.2% 4 Trivial only 47 Solevita 0 % Sunito 14.9% Sonera 85.1%
13.3.
Discussion
Study 2C confirmed that people adopt different choice strategies (i.e., brand based,
lexicographic and trivial attribute based processing) when choosing from a brand-attribute matrix
that has a trivial attribute. These findings provide experimental evidence that people engage in