The Sense and Nonsense of Taste Tests:

Are Consumers Blindly Loyal?

Abstract
This paper addresses the issues of validity, reliability and generalizability
of consumer product testing and its use in strategic decisions. Two common
methodologies, namely triangle taste tests and paired comparison tests, are
compared. We advocate the use triangle taste tests that incorporate a repeated
paired preference test. We provide correction factors to enhance the validity and
reliability of the test. The results of three triangle taste studies of cola beverages,
one in USA and two in Taiwan, are reported. The results of these studies show
that consumers have difficulty in discriminating between sodas and do not show
consistent preferences in blind taste tests. Moreover, consumers report a higher
confidence in their judgment than is warranted by their actual ability to
discriminate or report a consistent preference. These results hold in both
countries and across consumers regardless of their usage of the product
category or their stated brand loyalty.

Taste V. Freshness; Miller Rolls Out Latest Ads Focused on Direct Taste
Comparisons, As Anheuser-Busch Reacts With Massive Freshness Campaign.
New Ads Continue Highlighting Millers Taste Advantages, While A-B Exhorts
Distributors to Spend on Freshness Activities. Miller Also Expands Its Highly
Successful Blind Taste Challenge, Upping Its Target to Five Million Challenges.1
This press release from Miller Brewing Company in 2004 is just one
example of how taste tests have become a public relations, advertising and
competitive weapon for food and beverage manufacturers.

Introduction
Product tests on consumers are commonly performed by companies for a
variety of reasons. Most commonly, taste tests are used to test new product
formulations, and engage in comparative advertising versus an entrenched
competitor. The most infamous example of this use is the introduction of New
Coke in 19852. One of the best known examples of taste tests is the Coke-Pepsi
taste test. In the 1970s Pepsi launched blind taste tests to show their product
superiority. They found that the younger consumer liked the taste of Pepsi better
than the taste of Coke a finding that they spun into their famous Pepsi
Challenge commercials. To respond to this campaign, Coke conducted its own
blind taste tests and confirmed Pepsis results. They therefore decided to change
their formulation. After conducting over 191,000 blind taste tests in 13 cities
where they had consumers compare the taste of four new Coke formulations

1
2

http://www.millerbrewing.com/pressRoom/archiveDetails.asp?ideanumber=124
Fournier, Susan, Introducing New Coke, HBS Case Study 9-5000-067, October 31, 2001.

versus Pepsi and the original Coke formulation, they launched the New Coke a
marketing disaster, that was recalled in a couple of years.
Arguably, Coca-Cola used a flawed research methodology. The basic
tenets of good marketing research in product testing, as in every other domain,
are that it is:

Valid: that is, asks the right question,

Reliable: that is, gets the right answer; and

Generalizable: the answers are from the right people,

The Coke taste tests can be critiqued on the basis of their external and
predictive validity as they tested the effect of formulations without allowing for the
impact of brand on taste perceptions which means that their results were not
externally valid (as consumers do not buy product formulations, but rather
brands); and they measured taste preference rather than buying intentions. To
the extent purchases are made for reasons other than preference in taste, the
measure is an inadequate predictor of sales. They can be critiqued on their
reliability as they did not allow for the fact that people may not be able to
discriminate between the tastes of different sodas; they did not conduct a testretest reliability measure to ensure the stability of responses; and they did not
correct for chance effects of consumers guessing.
While these issues are specific to Coke and taste-testing, they speak more
generally to the issue of consumer research such as other types of sensory
testing (including tactile, olfactory, visual, and auditory tests), advertising copy
testing, concept testing, price discrimination studies, package design studies,

store layout studies and others. This is because one can use the taste-test as a
metaphor to understand how consumers make decisions on the basis of
information available to them in their environment: the basic issue that speaks to
the validity and reliability of research methods in general, and consumer testing
in particular.
By understanding the rigor involved with taste tests, other companies can
avoid the same errors of validity and reliability in measurement. It is clear, that
despite the infamy of the New Coke story, companies have yet to learn from
Cokes mistakes and may be on the same road to making the same errors that
Coke made almost two decades ago. Since the beginning of 2004, over 1 million
consumers have taken part in taste tests comparing Miller to Bud. While
releasing Millers new advertising campaign, Bob Mikulay, Executive Vicepresident of marketing at the Miller Brewing company confirmed that Miller would
conduct as many as five million taste challenges by the end of 2005.
The main purposes of this paper are to:
1.

Explore the strategic use of taste tests as a methodology that can help
decisions related to branding, pricing, product design, and advertising.

2.

Demonstrate that taste belongs to a group of sensory variables that is

subjectively experienced by consumers within the competitive context of
a companys brands as cultural icons, with objective product attributes
being merely one of the inputs into the consumer experience.

3.

Use taste tests as a domain to highlight the importance of conducting

consumer research that is valid and reliable. Specifically, we

recommend a research methodology, measures, and methods of

analysis for companies to use to conduct taste and other sensory tests.
4.

Demonstrate the generalizability of our recommendations by reporting

results from three studies using different tasks, and types of consumers.

Strategic Uses of Taste Tests beyond product introduction

The most common use of taste tests is to study how consumers evaluate
various product formulations and designs. However, taste testing can also be
used to advertise taste claims, and to understand the impact of branding on the
evaluation of the physical product. In the case of food and beverage product
categories, taste testing has received a considerable amount of attention in both
the trade and academic literature. A Google search in November 2004 on the key
words taste test yielded 8.7 Million entries. Not surprisingly, most market
research companies routinely offer taste-testing as part of their portfolio of
research products.
Use of taste tests for diagnosing problems and evaluating different
components of the marketing mix. Taste tests can be commonly used to
diagnose a problem and assess the value of different tactics. For example, a
Belgian chocolate bar manufacturer was unsure whether their poor sales were
due to a poor product, or a poor image versus competition. They conducted blind
and branded taste tests and found that their chocolate bar performed well when
the taste test was a blind one, but did not do well when it was branded. With this

information they were able to redirect money that they had been thinking of
spending on product development into image advertising instead.
Tests (using taste or other product benefits) can be conducted to examine
the benefits and costs of different packaging, different prices, different copy etc.
For example, if a manufacturer were interested in examining whether changing
the color of a bottle of beer made the beer in the bottle taste colder, they could
conduct taste tests where consumers were asked to rate the coldness of the beer
using different colored packaging. A comparison of ratings across the groups
would tell them to what extent consumers use the color of the package to draw
inferences about the sensory taste of coldness. The spectacular success of the
Mexican beer, Corona, could be, at least in part, to consumers attributing purity
of ingredients to the clarity of the bottle material.
Similarly, if a brand manager were interested in exploring how different
cues in advertising copy influence consumers olfactory sensations (e.g., the
scent of a product such as perfume, detergent etc.) they could ask people to
describe and rate the scent after being exposed to different executions of the
advertising. In fact, in the luxury consumer products industry, one of the last
product decisions made by fragrance manufacturers is the actual scent of the
perfume that they make. The scent merely has to be consistent with the package
shape and advertising theme that differentiates the perfume from others in the
market.
Similarly, taste testing could also be done to make labeling and pricing
decisions. For example, if a wine maker were interested in examining whether

they should price their bottle of wine in the $10-$15 range, or the $15-$20 range,
they could conduct blind taste tests describing the varietal and the price of the
wine, to assess whether consumers used price as a cue to make quality and
taste judgments.
Thus, at a general level, sensory testing, including taste testing, can be
used to assess whether a company should invest in changing its product
formulation (e.g., via research and development) or instead, develop their brand
name, package, and advertising.
Use of taste tests to assess the value of a brand. Taste tests can also be
used to assess the value of a brand: a difference in consumer preferences when
a taste test is blind versus branded shows the value of a brand: if people prefer
the taste more in a branded versus a blind taste test, this is evidence that the
brand has positive brand equity. Taste tests conducted with potato chips have
shown that consumers perceptions of the taste of the product (e.g., how crispy,
greasy, fresh, and flavorful a chip is) changed as a function of the brand
name associated with the potato chip. Taste tests conducted with coffee have
shown that the presence of a brand name improved the taste of coffee more for
users than it did for non-users. Thus, it is clear that consumers are not
particularly good at judging taste.
Use of taste tests to assess the value of a consumer. Taste tests could be
used to examine which segments display a stronger preference for a brand, are
more loyal to the brand, and have stronger attitudes that are more resilient to
competitive attacks. For example, if a company is interested in exploring whether

they should target the teenage market versus the young adult market, they could
examine differences in sensory ratings (such as taste) for branded and
unbranded versions of its product for each of these two groups both before and
after exposure to competitive claims derogating the products quality.
The ability to discriminate taste does not appear to increase with usage.
Taste tests conducted by Anheuser Busch (a large beer manufacturer)3
comparing its Bud, brand with that of two leading competitors: Coors and Miller,
showed that even among the regular 6-pack/ week buyers, less than 40% could
accurately discriminate between the taste of the three brands. As this segment is
their most profitable segment, the company uses this information to keep Bud
top-of-mind among its users. Its primary competitor, Miller, on the other hand,
has been advertising the taste superiority of its product in a recent series of ads
airing in Fall and Winter 2004. Rebuttals by Bud go back to peoples overall
preference for Bud rather than rebutting Millers taste claims. This is
presumably because Bud is aware that its loyal consumers cannot discriminate
(and do not necessarily care about the difference) between Bud and Miller.
Use of taste tests for legally defending claims. This knowledge has been
leveraged by companies. For example, Burger King used a taste test to defend
its competitive claim that its burgers were broiled, when McDonalds, its chief
competitor, took it to court saying that its burgers were frozen, micro-waved, and
then broiled. As taste tests showed that consumers could not differentiate
between the taste of a broiled burger, and a burger that was frozen, micro-waved,
and then broiled, the courts allowed Burger King to continue to use its claim.
3

Personal conversation, Brand manager.

Using taste tests to produce healthier product alternatives. In todays health

conscious society, it is incumbent on many manufacturers to offer healthier
alternatives to their existing line without compromising on the consumers taste
experience. For example, if manufacturers wish to reduce salt, lower fat or lower
cholesterol, they can use taste-tests to measure at what levels of product
modification are consumers able to discriminate between alternative formulations
(see Buchanan, Givon & Goldman, 1987 and Buchanan & Henderson, 1992 for
examples). Other than as an input to marketing decisions, such as those
discussed above, taste-testing can also be used for production decisions such as
how to reduce costs and maintain quality.
Using taste tests to reduce costs. A root beer manufacturer interested in
reducing the sugar content in its product (due to rising sugar costs), used a
series of taste tests to identify the product formulation where consumers would
not be able to differentiate between the original product and the product that
contained less sugar. They were able to reduce their costs by reducing the extent
of sugar they used by 10%, and increase their profits by 5%. Buchanan, Givon &
Goldman (1987) describe other examples of how manufacturers use taste-testing
to ensure that consumers are unable to differentiate between two product
formulations with one using a cheaper ingredient than the other.
Using taste tests for quality control. Taste-tests help manufacturers to
ensure batch-to batch consistency in their products (Moskowitz 1985). Tastetests are used to select or train expertssuch wine-testing experts or rice-

testing experts in Japan, who are more sensitive than the average consumer.
These experts then help to ensure product consistency.
To summarize, taste tests that have typically only been used to test new
product formulations, could also be used strategically by a company to assess
the value of different elements of its marketing mix by different customer
segments. We now turn to discussing the important of examining subjective
perceptions of experiences towards objective sensory attributes of a product
within its cultural and market context.

Managerial Takeaways for Strategic Decisions using Taste Tests

Managerial
Decision

Underlying
consumer
problem(s)/
processes

Managerial Solution

How to use taste

tests for product
formulation

Consumers have a
threshold within which
they cannot distinguish
differences.

Conduct taste tests using different

formulations to assess which is reliably
better (after controlling for statistical
guessing).
Reduce percentage of expensive
ingredients till the point that consumer are
able to discriminate between the taste of the
new and old formulations
Reduce percentage of less healthy
ingredients (e.g., saturated fat) till the point
that consumers still like the test and are
willing to bear the cost of lower taste for
better health.
Conduct taste tests in blind and (multiple)
non-blind formats: For example, to test the
effect of label color/ design/ logo etc.,
conduct taste tests using no label, label 1,
and label 2. Differences in preferences
across the three tests will demonstrate
whether consumers prefer one label to
another.

How to use taste

tests to reduce
costs
How to use taste
tests to produce
healthier foods.

Products with more

similar formulations are
more difficult to
distinguish.

How to use taste

tests for product
design

Consumers tastes are

contingent on contextual
cues (such as their
surroundings, and
aspects of the product
that do not involve taste:
e.g., packaging.)

How to use taste

tests to make price
decisions

Consumers infer quality

from price. Therefore,
higher priced items may
be perceived to taste
better.

Conduct taste tests after providing different

prices to different groups to assess whether
consumers use price to disambiguate taste.

How to use it to
identify problems
with sales; and How
to use the data to
determine budget
allocation

Consumers may be
unaware of the real
reason that they do not
like or buy a product.

How to use the data

to design
advertising

Consumers use
heuristics or rules of
thumb to process
information. A powerful
persuasion heuristic is
that If others like it, I will
too.
Consumers are
heterogenous in their
ability to discriminate,
show reliable preference,
and their awareness of
their own (in) ability.

How to use taste

test data in legal
situations

Conduct an experiment manipulating

various possible reasons that sales could be
poor. For example, if one possibility is poor
brand name conduct blind and branded
taste tests. If preference is higher in blind
taste tests, this suggests that the brand
image needs to be improved and money
should be directed to advertising. If the
reverse, this suggests that money should be
allocated to research and development to
design a better product.
Highlight results of taste tests that
demonstrate product, or brand superiority. If
taste is not the brands strong point, conduct
taste tests to show that in a branded test,
people prefer the product.
Be sure that the sample is not biased: it is
representative of the group to which you
wish to make generalizations. Beware of
large sample sized as they may mask a
small effect size (i.e., significant differences
may not be substantive differences);
UNLESS the goal is to show no difference.
In such a case, a large sample size with
effect sizes should be reported.
Closely critique methodology to see that a
triangle taste test has been conducted and
that the data has been corrected for
guessing.

Perception Impacts Sensation

Perceptions are by definition subjective; there is no such thing as objective
perceptions. The fundamental truth of this statement is the single most important
underpinning to collecting reliable data and making appropriate conclusions. By
explicitly recognizing, measuring and leveraging the subjectivity of the
consumers experience, a market researcher can extract maximum information
that is not only reliable, but also valid under a range of scenarios (i.e.,
generalizable).

10

The fact that brand information impacts consumer perception is neither

new nor surprising, though the modes and methods of documenting this are
constantly changing given new advances in scientific methods, hitherto
unexplored by traditional market researchers. A recent study4 of the impact of
brand on brain activity in the context of taste testing has given rise to the
intriguing possibility that brand influences not only the psychological responses to
a product but also the biological responses. The influence of brand on
discrimination and preference is particularly strong when the actual physical
differences in products being tested are weak. As the brain researchers observe:

Coca-Cola (Coke) and Pepsi are nearly identical in chemical composition,

yet humans routinely display strong subjective preferences for one or the
other. This simple observation raises the important question of how
cultural messages combine with content to shape our perceptions
Behavioral evidence suggests that cultural influences can insinuate
themselves into the decision-making processes that yield preferences for
one consumable or another.

There can be no doubt that Coke is a firmly established icon in American,

if not global, consumer culture. Indeed, it has been speculated that the
association of red as the color of Christmas and Santas clothing is a result of
4

Samuel M. McClure, Jian Li, Damon Tomlin, Kim S. Cypert, Latan M. Montague and P. Read
Montague (2004) Neural Correlates of Behavioral Preference for Culturally Familiar Drinks, Neuron,
44(2), 379-387.

11

Coca-Cola Companys decades of seasonal advertising in which Santa was

garbed in the corporate color.
The brain research study, conducted at Baylor University Medical Center,
found that when people were asked to taste two unlabeled cola samples (which
were actually Coke and Pepsi), an equal number chose one over the other.
However when one sample was identified as Coke and the other left unidentified
(both were actually Coke), there was a strong preference of when the labeled
product was identified as Coke, but no strong preference when the same was
identified as Pepsi. These results in themselves are well known in the marketing
literature. The brain research study for the first time demonstrates brand
marketing has a direct physiological effect on the brain, and consequently on the
brains capacity to make a choice. When a cola soft drink is tasted, whether
labeled with a brand or not, a part of the brain that responds to rewards, sugar in
the case of cola, lights up. However, when the cola product is labeled as Coke,
other regions of the brain not normally associated with gustatory sensation (viz.,
those controlling memory and cognition), also light up. No such effect was
recorded when the product was labeled as Pepsi. When you know what brand
you are consuming, your brain responds not only to the taste but also to what
else you know about the brand.
But what if consumers make judgments about unbranded products which
have similar or identical formulations? We suggest that in the absence of strong
sensory taste cues, the regions of the brain associated with memory and
cognitive control, along with any strong cues available to the non-gustatory

12

senses (sight, smell, hearing and touch) will offer guidance to the regions of the
brain associated with gustatory sensing. Could it be that branding, influences not
only the perception of taste but the taste itself? If you think Coke, do you taste
Coke? Investigating this conjecture on the basis of biometrics such as brain
activity is beyond our capability. Instead, we use consumers confidence in their
judgment as a surrogate for the taste that they are experiencing.
Figure 1:
The Role of Sensory & Brand Information on Consumption Experience

Product Formulation
Cultural cues

Brand

Taste
Expectation
Taste Sensation

Non-gustatory sensory
cues

Visual (sight)
Olfactory (smell)
Haptic (touch)
Auditory (sound)

Taste Judgments

Identification
Discrimination
Preference
Objective experience

Subjective experience

Accuracy
Consistency

Confidence

13

Recognizing that perceptions are subjective and contextual leads a

researcher to consider other factors affecting sensory experience. While the wine
industry appears to have studied these factors most closely, they are yet to
become standard contextual variables for other sensory experiences. These
include:

The location where consumption occurs (e.g., home, social, other).

The presentation of the product (e.g., shape, size, and material of glasses,
and temperature of servings can affect taste experiences).

Serving sizes (e.g., a sip versus a 1st or 2nd glass).

Context of complementary consumption items (e.g., items it is served with).

Social context (people surrounding the consumer at the time of consumption).

To summarize, we propose that sensory experiences are embedded

within a cultural context. Therefore, market research needs to explicitly control for
or measure aspects of the cultural context while measuring sensory experiences.

How to Design a Taste Test

There are several methods of taste-testing manufacturers and researchers
have used to measure discrimination ability.
The Triangle Test: The most common selection test is the Triangle test. In
triangle test, subjects taste three product samples (two As and B or vice-versa),
and then select the one that they think is different from the other two. The validity
and reliability of these tests have been questioned by many on the grounds of
adequacy, sample size requirements, and sensory fatigue. For example,

14

Morrison (1981) noted that to assess discrimination ability a minimum of two

tests per subject is required, with a minimum sample size of 600. Sample size
requirements can be cut in half if the number of tests per respondent is doubled,
but this causes sensory overload, which provides lower estimates of
discrimination ability than exist in the sample (Buchanan, et. al., 1987).
Repeated-Pair comparisons. Because of the problems caused by triangle
tests, many researchers have used repeated pair comparisons to measure
discrimination ability. In a paired comparison, subjects taste two different
products and then make a judgment about those products. After subjects have
made several choices, the consistency of their choices is used to assess their
discrimination ability. There are three different discrimination tasks that can be
performed in the paired comparison format:
1. Preference TaskSubjects are asked to choose the more preferred
one after they have tasted two different products, a procedure referred to as
consistent preference discrimination testing (Penny et al., 1972). Subjects who
have discriminating ability tend to choose the same product on every trial.
Buchanan, Givon and Goldman (1987) found that repeated paired comparisons
measured discrimination ability better than repeated triangle tests, and this is
because the task is simpler. The concerns with using preference as the task are
that it does not allow for consumers who are indifferent between the tastes of the
two products or for consumers whose tastes change across trials or whose
discrimination abilities improve across trials as a function of taking part in the
test.

15

2. Identification Task. Hopkins & Gridgeman (1955) suggested the use of

this method to test for product that varied in the intensity of their flavors. Subjects
tests two product samples (A and B) and judge which one has a particular
characteristic. The proportion of correct answers is used to measure
discrimination ability. The primary concern with this method is that it may tap
prior brand knowledge rather than discrimination ability (Buchanan and
Henderson 1992).
3. Discrimination Yes/No Task: In detection taste testing, subjects are
asked to test two products and answer whether they are the same or different.
Discrimination ability is measured by the proportion of correct answers (Coombs,
Dawes and Tversky, 1970). The primary critique against this method is that
judgments are based on prior beliefs as to the likelihood of a difference existing.
That is, subjects with high priors interpret any perceived difference as a true
difference in products, while those with low priors tend to discount perceived
differences (Coombs, Dawes and Tversky, 1970). Therefore, when participants
have unrealistic prior beliefs, they make more incorrect judgments, downwardly
biasing measures of discrimination ability (Buchanan and Henderson, 1992).
While each of the measures discussed above: preference, identification
and discrimination, all have their individual biases in assessing discrimination
ability, Buchanan & Henderson (1992) found that the discrimination task is the
least reliable, while the preference task introduces an upward bias in that it
shows the highest measured discrimination ability.
These methods are summarized in the table overleaf.

16

Pros and Cons of Alternative Taste Test Designs

Type of Test
Triangle taste test

Preference task
paired
comparison

Identification task
paired
comparison

Discrimination
Yes/ No task
paired
comparison

Key Features
Respondents taste three
product samples (two As and
B or vice-versa), and then
select the one that is different
from the other two.
Respondents choose which
product they prefer from two
options, on multiple trials. The
consistency of their choices is
used to assess their
discrimination ability.
A respondent tests two
product samples and identifies
which one has a specific
characteristic. Discrimination
ability = Proportion of correct
answers.
Respondents test two products
and judge whether they are
the same or different.

Estimated Biases
The triangle test is complex and causes
sensory fatigue. Therefore, it depresses
measured discrimination ability
It is simpler than a triangle test. However,
respondents with discriminating ability
may have no preference between the two
products; may change their preferences
across trials, or may learn across trials.
This would upwardly bias discrimination
estimates.
It requires sufficient product knowledge
to make a correct identification.

Has high measurement bias as

respondents are influenced by their prior
beliefs as to the likelihood of a difference
existing. Those who have unrealistic prior
beliefs make more incorrect judgments,
downwardly biasing measures of
discrimination ability.

Morrison (1981) suggested that triangle discrimination tests should be

used in conjunction with preference tests. This is the method we recommend.
The methodology corrects for the guessing element so as to estimate the true
proportion of people who can accurately discriminate between products and
consistently prefer the same product.5 The method is described below.

Morrison, Donald G. (1981), Triangle Taste Tests: Are the Subjects Who Respond Correctly
Lucky or Good? Journal of Marketing, 45 (Summer), 111-119; Buchanan, Bruce, and
Donald G. Morrison (1984), Taste Tests: Psychophysical Issues in Comparative Test
Design, Psychology and Marketing, 1(1), Spring, 69-91 ;Buchanan, Bruce, Moshe Givon,
and Arieh Goldman (1987), Measurement of Discrimination Ability in Taste Tests: An
Empirical Investigation, Journal of Marketing Research, 24(2), May, 154-163; Buchanan,
Bruce (1988), A model for repeat paired comparison preference tests, Psychometrika,
53(2), Jun, 209-221.

17

1.

The consumer must first be able to identify which product is

different (from a set of 3, where 2 are the same). None of the
three products, of course, have brand identifiers.

2.

Two preference tasks then assess whether the consumer

preference has test-retest reliability (i.e., they chose the same
product on two trials).

3.

Only consumers who are able to discriminate and show a

consistent preference pass the test.

This is the method we examine in greater detail in this paper.

How to Analyze a Taste Test: Objective Criterion

When consumers are asked to make discrimination and/ or preference
judgments in product comparisons, responses may be based on a combination of
the taste experience and guessing, with guessing increasing as objective
differences between formulations reduce. Therefore, all procedures should
correct for guessing when analyzing the data.
The accurate interpretation of taste test results rests on three main
presumptions: one, a person has to have the ability to discriminate prior to being
able to show a preference; two, they need to consistently prefer the same
product across repeated trials; and three, any overall numbers reflect both a true
ability to discriminate and show consistent preferences, and a guessing element,
for which the data needs to be corrected.

18

The arithmetic of correcting results for guessing is straight-forward. The

triangle taste test involves two distinct tasks, the discrimination task and the
repeated preference task.
1) Discrimination Task: Three glasses of a product, say soda, are provided.
Two contain the same soda and one contains a different soda. The
participant has to identify which one is the different soda. The underlying
logic behind this test is to show that there cannot be reliable preference
without there being some initial level of ability to discriminate. The actual
number of people who accurately differentiate the different soda needs to be
corrected for random guessing (as every person has a 1 in 3 chance of
choosing the correct glass if they guessed). The method to account for this
is to define the problem as follows:
P(d)

= Proportion who can truly differentiate

1-P(d) = Proportion who guess

Probability of guessing = 1/3 (because there are three choices).
C

= Proportion who mark the correct answer

= Proportion who discriminate + 1/3 Proportion who guess
= P(d) + 1/3 (1-P(d))
= 2/3 P(d) + 1/3

True percentage of people who can differentiate

corrected for guessing is P(d) = (C 1/3)/ (2/3)

19

2. Preference Task (repeated twice): Two unmarked glasses of soda, with

each containing a different soda are provided. The participant has to identify
which one s/he prefers. The preference task is repeated in the exact same
manner. The consistency in the responses between preference task 1 and
preference task 2 shows reliability. As there may be individuals who prefer
the same soda on two trials by pure chance, the actual number who shows
a consistent preference needs to be corrected for random guessing.
A = number of people who prefer Brand A in both trials
B = number of people who prefer Brand B in both trials
Total (A+B) = number who show an inconsistent preference
The indifferent can be in each of the four cells (AA; BB, AB, and BA) with
equal likelihood. Thus, the percentage who make inconsistent choices
(choose A once and B once) represent of those who are truly indifferent,
the other half being in the AA, or BB cells by pure chance.
 True percentage of persons showing consistent
preferences corrected for guessing
= (% reported consistent) - 0.5 (% inconsistent)

How to Complement Objective Taste Test Data with Subjective Measures

We have argued that perception is subjective. Explicitly recognizing this
consumer fact implies that market research measures supplement actual
objective taste data with subjectivity of taste assessment. We recommend that
taste-testers also collect measures of a respondents level of confidence in their

20

judgment. Consumer self-confidence has been frequently cited as an important

construct for understanding consumer behavior (Bearden, Hardesty and Rose
2001); and has been hypothesized to be an antecedent of subjective product
knowledge (Park, Mothersbaugh & Feick, 1994). Consumer self-confidence
reflects subjective evaluations of ones ability to generate a positive experience
as a consumer in the marketplace (Adelman, 1987). Therefore, like other
measurements of consumer behavior, self-confidence can be used to measure
consumers buying intention and purchase behaviors (Obermiller & Spangenberg,
1998).
Buchanan & Henderson (1992) investigated if subjects can provide any
information about their own performance on taste tests by examining their
confidence ratings. They found that subjects self-ratings of confidence were
significantly associated with their actual performance, and concluded that
subjects were capable of providing useful information about their own
performance. For example, those participants who were unable to discriminate
on a taste test indicated no confidence in their judgments, implying that
subjects can assess their abilities.
However, other studies in different domains have shown that consumers
may frequently be overconfident about their abilities. The overconfidence effect
suggests that people are more confident than is warranted by their actual level of
performance (Fischoff, Slovic and Lichtenstein, 1977). Classic tests of the overconfidence effect involve asking people general knowledge questions, and then
making them estimate the probability that they have answered these questions

21

correctly. In typical tests of the effect, confidence scores are higher than actual
accuracy scores, indicating overconfidence (for a review see Fischoff, 1982).
Alternately, researchers have asked people to estimate their ability to solve a
problem, make progress towards a solution, or draw an appropriate conclusion.
In these tasks as well, people are more confident of their problem-solving ability
than is warranted by their performance (see Metcalf, 1999 for a review).
Theoretically, our proposed model where we argue that peoples
sensations (e.g., taste) are contingent on their cognitive beliefs and can
overwhelm their actual experience suggests that actual performance may not
necessarily track subjective reality as measured by confidence in the accuracy of
a judgment. Examining whether peoples confidence is warranted by their actual
performance is an important managerial question as it speaks to the underlying
basis of brand loyalty. If consumers are overconfident in as much as they are
confident in their beliefs that one product is better than another, even when they
are unable to discriminate its test beyond chance likelihood, or able to
consistently prefer its taste, they belong to the category we call blindly loyal.
From a companys point of view this suggests that a brand manager would be
better advised to invest in their brands image for the consumers mind rather than
invest in product improvement. From a consumer welfare point of view,
companies can take advantage of this inability-overconfidence match by charging
them higher prices, reducing costs through diluting product quality, or wastefully
increasing advertising expenditures to maintain beliefs of one products
superiority over another.

22

Managerial Takeaways for the Design and Analysis of Taste Tests

Managerial
Decision

Underlying consumer
problem(s)/ processes

Managerial Solution

How to conduct
a taste test

Consumer ability to
discriminate and make
consistent preferences is low,
and lower than they believe it
is.
Consumers may arrive at
correct objective answers
through mere guesswork.
Taste is one, and perhaps not
the most important reason
that consumers purchase a
brand. However, they may be
unaware of their real
reasons for purchase and
may not be able to explicate
them.

Include a discrimination task prior to a

preference task.
Repeat the preference task to assess
consistency.

How to analyze
the data
What
conclusions
can be drawn
from taste tests

Correct for guessing.

P(d) = (C 1/3)/ (2/3)
P(c ) = (% consistent) - 0.5 (% inconsistent)
If you are interested in the impact on sales,
then include measures of choice, purchase
intention, recommendation intentions, and
repeat purchase behavior. If possible,
measure actual consumption amount in
follow up tests to see if more is consumed of
better tasting products. Do not stop at
gauging taste superiority. In all taste tests
also measure loyalty, consumption levels,
product category usage etc., so that you can
identify whether consumers who pass the
taste test belong to a different segment as
compared to those who do not pass the test.

Empirical tests6 have found that:

People do no better on the discrimination task than on the

preference-consistency task, and sometimes do worse.

Consumers are not well calibrated with their taste test judgment
abilities; while they perform worse at discrimination accuracy than
at preference consistency, they are more confident of their
discrimination ability.

There is a small, positive correlation between discrimination

accuracy and subjective confidence in discrimination, but not

Buchanan, Bruce, and Pamela W. Henderson (1992), Assessing the Bias of Preference, Detection, and
Identification Measures of Discrimination Ability in Product Design, Marketing Science, 11(1),
Winter, 64-75.

23

between consistency of preference and confidence in making a

consistent preference.
In the next section of this paper, we examine the robustness of these
effects across countries, and tasks.

The Cola Wars: A Triangle Taste Test Study

Taste tests of cola beverages (Coke vs. Pepsi, and Coke vs. Diet Coke)
were conducted in the United States and in Taiwan to address the following
questions:
1. Objective Ability: Do people have an ability to discriminate between cola
brands and show a consistent preference, based on blind tasting?
a. Does the accuracy of their ability to discriminate and the
consistency in their preferences depend on:
i. Their loyalty to a brand ?
ii. The extent to which they use the product category?
b. Are there cross-national differences in the answers to these
questions?
2. Subjective Perceptions: Is actual ability to discriminate or show a reliable
preference related to peoples confidence levels in their abilities?
a. Does brand loyalty and experience affect levels of confidence?
b. Are there cross-national differences in confidence levels?
c. Are the discrimination task vs. the preference task associated with
different levels of confidence?

24

We used the triangle taste discrimination test to estimate actual ability to

discriminate between two sodas and the repeated measures preference test to
measure the reliability of preference of one soda over another. After the taste
test, each respondent also reported the level of confidence in their taste test
judgments as well as their frequency of consuming cola drinks and their loyalty to
either Coke or Pepsi.
The overall schematic of the constructs and the manner in which they are
measured is reproduced below:
Objective
evaluation

Triangle Test
Discrimination
Task
And
Repeated Pair
Comparison

Discrimination
Accuracy
Vs.
Preference
consistency

Individual
differences
Brand Loyalty
Product Usage

Subjective
evaluation
Confidence

Discrimination Accuracy and Preference Consistency: Each participant

blind-tasted 7 glasses of soda. They first tasted three colas, of which two were
identical and the other different, and made a discrimination judgment as to which
two were the same. If they judged correctly, their discrimination judgment was
accurate, and if not, it was inaccurate. Next, they did the repeated pair
comparison. They tasted two brands of cola and judged which one they

25

preferred. They then repeated this preference judgment with the same two
colas, and if they had preferred the same cola as in the previous tasting, their
preference was consistent.
Confidence:

After the taste tests, each participant rated their confidence

in their discrimination and preference judgments respectively, on a 0 100 scale

by completing the following two statements.
I am _____ % confident that I identified the different soda in the trio.
I am _____% confident that I chose the same soda in my two stated
preferences.
Loyalty and Experience: Participants reported whether they were loyal to
either the Pepsi or Coke brand versus indifferent between the two. Finally, they
were asked to report the number of cans of cola that they drank in an average
week as an indicator of their familiarity with the product category.
Three taste test studies were conducted in two different countries among
consumers drawn from a population of business students. In the first study (US),
50 students enrolled in a marketing research class at a US business school
participated in the study; the brands used in the taste test were Pepsi and Coke.
In the second (TW 1), 62 students enrolled in marketing classes at a Taiwanese
university participated in a taste test also involving the Coke and Pepsi brands.
The third study (TW 2), conducted at the same Taiwanese university, had 66
participants but the brands used in the taste test were Coke and Diet Coke. In
reporting the results, the 3 studies are referred to as the US study (Coke vs.

26

Pepsi, n=50), the TW 1 study (Coke vs. Pepsi, n=62) and TW 2 study (Coke vs.
Diet Coke, n=66).7

1. Few Consumers Can Tell The Difference between Sodas

Overall, after correcting for guessing probabilities, the ability to
discriminate was low in all samples, although it was higher in the Taiwanese
samples than in the US sample (US 0, TW1 = 17.74%, and TW 2 = 27.27%, 2 =
4.47, p < .11). Note that even though as many as 32% of the US sample marked
the correct answers, the actual percentage who can discriminate is not
significantly higher than 0, as pure chance would have 33.33% guessing the right
answer. Overall, across all three samples, 15.73% of the sample was found to
have the ability to discriminate between the two sodas.
The ability to show a consistent preference was directionally higher than
the ability to discriminate for all samples, but did not differ across the three
samples (US = 35.71%, TW 1 = 21.31%, and TW 2 = 35.39%, 2 = 1.4, p= .50).
Again, note that even though 53.6% of the entire sample showed a consistent
preference, the percentage correcting for guessing is only 30.36%.
Overall, across samples, approximately 25% (45/ 178) passed the taste
test (a percent that did not differ by sample, overall 2 = 2.69, p > .25). However,
this number includes those who guess on either the discrimination task or the
consistency tasks. The lower bound is 15.73% and the upper bound is 30.36%.

Due to partial non-response to some questions, results in individual analyses may be based on
smaller sample sizes.

27

Overall Results of Taste Tests conducted in US and Taiwan

US: Coke

Taiwan: Coke

Taiwan: Coke

vs. Pepsi

vs. Pepsi

vs. Diet Coke

16/50

28/62

34/66

78/178

32%

45.2%

51.5%

43.8%

= (C 1/3)/ (2/3)

17.74%

27.27%

15.73%

# consistent in soda choice/ total

24/42

29/61

37/65

90/168

Percent consistent

57.1%

47.5%

56.9%

53.6%

PC corrected for guessing = %

35.71%

21.31%

35.39%

30.36%

# passing the taste test/ Total

12/50

12/62

21/66

45/ 178

Percent passing Taste test

24%

19.4%

31.8%

25.3%

Confidence in Discrimination

65.43

68.19

76.74

70.73

63.55

65.47

73.66

68.09

.82

.49

.56

.65

# who correctly discriminated/

Overall

total
Percent correct = C
PD corrected for guessing

consistent - .5 (% inconsistent)

ability (scale of 0-100)

Confidence in Consistency of
preference (scale of 0-100)
Correlation between confidence
levels

2. Discrimination ability improves with greater differences in formulation

Discrimination accuracy and preference consistency are higher in the
Coke vs. Diet Coke taste tests as compared to the Coke vs. Pepsi taste tests.
The sweeteners used in Coke and Pepsi are the same, namely sugar, whereas
Diet Coke has an artificial sweetener which most likely provides a more distinct
sensory signal. We conclude that when product formulations are different
enough to provide consumers identifiable and distinct sensory cues, the
consumer experience will be more likely to be driven by these objective signals.
On the other hand, when product formulations are very similar, the objective

28

cues are suppressed in guiding a consumers experience which is instead based

on psychological imputations that are made by the consumer.
3. Consumers are Blindly Loyal: loyalty does not improve discrimination
across brands, but does improve discrimination across formulations.
Those who professed a loyalty to either soda were no more able to
discriminate than those who were indifferent. A cross-tabulation of the accuracy
in the discrimination task with self-reports of loyalty (loyal to either Coke or to
Pepsi versus indifferent) did not reveal a significant relationship (Overall: 2 =
.06, US 2= 3.33; TW 1: 2 = .96; TW 2: 2 = 1.11, p > .05 for all; see table below):

Loyal consumers discriminate no better than indifferent consumers

Sample (Number who can
discriminate accurately/ Total)
US Coke-Pepsi challenge
Taiwan Coke-Pepsi challenge
Taiwan Coke-Diet Coke test
Total

Indifferent between

Loyal to Coke or

Coke and Pepsi

Pepsi

9/ 19 or 47.36%

7/ 31 or 22.85%

18/ 43 or 41.86%

10/ 18 or 55.55%

9/ 21 or 42.86%

25/ 44 or 56.82%

36/ 83 or 43.37%

42/ 93 or 45.16%

While loyalty did not improve participants ability to make a consistent

preference between brands, it did increase their ability to consistently choose the
same product variety within the brand. The same analysis for consistency in
preferences showed a significant relationship for the Coke-diet Coke test, but not
for the Coke-Pepsi tests (US: 2 = 1.42; TW 1 2 = .66, p > .50 for both, TW 2 2 =
10.96, Overall 2 = 4.89, ps < .05 for both; see table below):

29

Loyal consumers are more consistent only in the regular-diet taste test
Sample (Number who had

Indifferent between the

Loyal to Coke or

two sodas

Pepsi

US Coke-Pepsi challenge

11/16 or 68.75%

13/ 26 or 50.00%

Taiwan Coke-Pepsi challenge

19/43 or 44.19%

10/ 18 or 55.55%

Taiwan Coke-Diet Coke test

6/ 21 or 28.57%

31/ 43 or 72.09%*

36/ 80 or 45%

54/ 87 or 62.07%*

consistent preferences/ Total)

Total

4. Product Experience does not improve ability

Product category experience does not improve the ability to discriminate
or show a consistent preference. In the Coke-Pepsi challenge, there was no
difference in the number of average units of soda consumed per week between
those who were able to accurately discriminate (Means = 2.21 and 0.96 for US
and Taiwan respectively) versus those who were not (Means = 2.21 and 0.67, F
s = 0 and 1.36 respectively, p s > .20). However, counter-intuitively, in the CokeDiet Coke tests in Taiwan, those who were able to accurately discriminate drank
less soda than those who did not accurately discriminate between the sodas
(Means = 1.13 versus 0.59 for inaccurate versus accurate consumers, F (1, 62) =
5.47, p < .05).
Experience had no effect on the ability to make a consistent preference in
any of the three tests. Those who showed a consistent preference consumed no
more soda (Means = 2.14, 0.83 and 0.86 units per week for US, TW1 and TW 2

30

respectively) compared to those who showed an inconsistent preference (Means

= 2.28, 0.78 and 0.82 respectively for the three samples, all F s < 1, p s > .80)8.
To summarize, it appears that consumers ability to discriminate and make
consistent judgments is low overall (approximately 25%) and loyalty and
experience do not substantially improve this ability. We next examine whether
peoples confidence in their judgments calibrates with their actual accuracy and
consistency.

5. Consumers are Over-confident

Overall, there was a high level of self-confidence that people attribute to
their ability to discriminate between brands and to make consistent choices.
Overall, US participants showed the same level of confidence as Taiwanese
participants in the Coke-Pepsi tasting (average confidence in discrimination =
65.43 vs. 68.19; average confidence in consistency = 63.55 vs. 65.47; for US vs.
TW 1 respectively), but Taiwanese participants in the Coke-diet Coke test showed
a higher level of confidence overall (Discrimination = 76.74; Consistency = 73.66,
overall effect of test F(1, 168) = 4.44, p < .05). There was no difference in the
two measures of confidence in any sample (paired t = .682, .859 and 1.349 for
US, TW 1, and TW 2 respectively, ps > .10 for all). The confidence in the two tasks
was also highly correlated suggesting that subjectivity of taste may be a better
way to differentiate consumers than their actual performance on a taste test (r =
.824, .488, and .565 for US, TW 1, and TW 2 respectively, p < .001).
8

There was no substantive difference in these results when the usage levels were collapsed into usage
categories (non-drinkers, light drinkers, and heavier drinkers); or when performance in the taste test was
collapsed into two categories: those who passed the taste test and those who did not (ps > .05 for all tests).

31

Overall, results show that those who could accurately discriminate

between the two sodas were no more confident. The level of confidence in
discrimination ability for those who were accurate in the discrimination task was
no different from the confidence level of those who were not. A similar analysis
on the confidence levels in preference consistency showed that those with an
inconsistent preference were as confident as those with a consistent preference.
This suggests that peoples confidence is unwarranted their actual ability, and
reflects over-confidence in their ability to identify the taste of a soda.

6. Loyal and experienced consumers are more confident than others, but
only for a task perceived to be easy
We assessed whether the confidence in the two tasks was different for
those who professed loyalty to either of the test brands versus indifference
between the two. Overall, results show that those who are loyal to either Coke or
Pepsi or are indifferent have no higher levels of confidence in their ability to
either discriminate or reliably prefer Coke over Pepsi (in both the US and in
Taiwan).
However, consumers are more confident that they can discriminate
between diet Coke and regular Coke (Means = 69.24 vs. 80.32 for indifferent
versus loyal consumers, F(1, 63) = 5.22, p < .05), and make consistent
preferences between those two (Means = 64.67 vs. 77.95 for indifferent versus
loyal consumers, F(1, 63) = 6.51, p < .05).

32

Loyalty increases confidence levels only in the Coke-Diet Coke test

Study

Indifferent between the two

Loyal to Coke or Pepsi
sodas
Confidence in
Confidence Confidence in Confidence
Discrimination
in
Discrimination
in
Ability
Preference Ability
Preference
Consistency
Consistency
62.19
54.69
67.29
68.61
67.16
63.74
70.56
69.33

US Coke-Pepsi
Taiwan CokePepsi
Taiwan Coke69.24a
64.67 b
Diet Coke
a,b
differences in means significant at p<.05

80.32 a

77.95 b

A similar analysis examining whether confidence levels were affected by

product category consumption showed no difference in the level of confidence in
either task for those who drank no soda in an average week, light-drinkers, or
heavier drinkers in any of the three samples (all p s > .25). Overall, results show
that increased experience does not increase confidence levels.
To summarize, consumers are highly confident in their ability to
discriminate and make consistent preferences regardless of loyalty and amount
of product category consumed. The key findings with implications for brand
managers are reproduced in the table overleaf.

33

Summary of Main Findings with Implications for Managers

Main Findings

Implication

People cannot
discriminate

Discrimination ability is a precondition for preference. Thus,

taste tests should always include a discrimination test prior to
a preference test.

Discrimination
Unrelated to
consistency

In many situations, discrimination is a necessary, but not a

sufficient condition to show preference. This is because
people who passed a discrimination task may have guessed,
or because they may have no preference.

People are
inconsistent

Preference tests should be repeated at least twice to gauge

whether the data is reliable.

Confidence is high

People believe that they are better at judging taste than they
actually are, so they are unlikely to be influenced by
advertising saying that others failed a taste test, or chose the
less popular brand.

Confidence unrelated
to ability

People are unaware of their ability to judge taste. Therefore,

merely asking them to state a taste preference without
actually tasting a product may not be valid data regarding
taste. The preference judgment would include some overall
beliefs regarding the brand/ packaging etc.

No cross cultural
difference

These effects are likely to occur even among consumers who

are less prone to the unrealistic-optimism bias (those from
Pacific Rim countries for example) whereby people believe
that they are better than the average person at a task.

No differences as a
function of brand
loyalty

Consumers are loyal to brands for many reasons, with the

intrinsic features of the brand just one among them. A brand
manager should understand which are the key drivers to get
brand loyal consumers to repeat purchase, which ones are
likely to get those loyal to a competitors product to switch,
and which ones will get the indifferent to increase their loyalty
to one brand.

No differences as a
function of product
category experience

Those with regular and frequent experience with the product

category have no better ability than those with lesser
experience suggesting that ability does not improve with
experience. It also suggests that a brands major consumers
are as prone to the biases as the rest of the customer base.

Confidence across
the tasks is related.
Confidence is higher
for preference rather
than discrimination.

Confidence reflects an individual difference variable: people

are more or less confident.
Consumers perceive the preference task to be easier than the
discrimination task.

34

Implications for Brand Managers

Our overwhelming result is that consumers cannot accurately tell the
differences between brands nor make consistent preference choices when actual
differences in the product are small, as in the case of cola. Second, in spite of
this lack of discrimination ability and preference consistency, consumers have a
very high level of confidence in the evaluations and judgments that they make
about product experience. This result is robust over a wide variety of consumers
regardless of whether they profess to be brand loyal or not and the amount of the
product category they consume. The result is also robust over replication in two
different countries, U.S. and Taiwan.
It would appear that for some product categories, a consumers
experience of a product is determined not so much by the objective reality of that
experience but rather by the psychological reality. Many earlier studies have
shown that when branding is introduced into the total experience, the brand is a
powerful determinant of the psychological component of the experience. The
study reported here indicates that even when no brand information coupled with
the product experience, consumers will generate their own cues to guide their
experience.
The challenge for a brand manager is how to channel these selfgenerated cues. One may be to strongly link a brand with sensory cues other
than those associated with the primary benefit the product provides. So when
consumers cannot tell the difference in taste between food or beverage products,
brand advertising may channel their consumption experience towards visual,

35

olfactory, auditory or tactile clues. Pepsi Blue and the snap, crackle, pop of
Rice Krispies are examples of this approach. When brands in a product begin to
compete away the entire range of sensory cues, brand strategy must shift to
linking the brand to cues associated with a consumption situation rather than the
product itself. Martinellis sparkling juices association to the Thanksgiving or
Christmas feast is one example of this.
Second, this paper introduces the brand manager to the correction for
chance or guessing which must be made before interpreting the results of taste
tests. The triangle taste test methodology used in this the study reported in this
paper provides the necessary data to make this correction. The formulae for
making this correction are provided.
A third implication of the study is about the validity of blind taste tests as a
methodology for deciding the winning product formulation. Our results
demonstrate that consumer perceptions and preferences in some product
categories have little to do with the intrinsic aspects of the product itself.
Consequently, brand managers are cautioned against undue reliance on
attributing consumer preferences to product differences in alternate test products
and using such data to decide on the winning product design or formulation. At
the very least, it would be valuable to follow up the taste test with a diagnostic
interview to reveal on what consumers are actually basing their discrimination
and preference judgments. Finally, we provide a range of uses, other than
product formulation, that managers can use taste tests for.

36

Reference
Adelman, Pamela K. (1987). Occupational Complexity, Control, and Personal
Income: Their Relation to Psychological Well-Being n Men and Women,
Journal of Applied Psychology, 72(November), 529-537.
Bearden, William O., Hardesty, David M. and Randall L. Rose (2001). Consumer
Self-Confidence: Refinements in Conceptualization and Measurement.
Journal of Consumer Research, 28(June), 121-134.
Buchanan, Bruce and Pamela W. Henderson (1992). Assessing the Bias of
Preference, Detection and Identification Measures of Discrimination Ability
in Product Design. Marketing Science, 11(Winter), 64-75.
Buchanan, Bruce, Givon, Moshe M. and Arieh Goldman (1987). Measurement of
Discrimination Ability in Taste Tests; An Empirical Investigation, Journal of
Marketing Research, 24(2), 154-163.
Coombs, C. H., R. M. Dawes and A. Tversky (1970). Mathematical Psychology:
An Elementary Introduction. Englewood Cliffs, NJ: Prentice-Hall, Inc.
Fischhoff, B. (1982). Debiasing. In D. Kahneman, P. Slovic, and A. Tversky
(Eds.), Judgment under Uncertainty: Heuristics and Biases (pp. 422-444).
New York: Cambridge University Press.
Fischhoff, B., Slovic, P. & Lichtenstein, S. (1977). Knowing with Certainty: The
Appropriateness of Extreme Confidence. Journal of Experimental
Psychology: Human Perception and Performance, 3, 552-564.
Hopkins, J. W. and N. T. Gridgeman (1955). Comparative Sensitivity of Pair and
Triad Flavor Difference Tests, Biometrics, 11(March), 63-68.
McClure, Samuel M., Li, Jian, Tomlin, Damon, Cypert, Kim S., Montague, Latane
M. and P Read Montague (2004). Neural Correlates of Behavioral
Preference for Culturally Familiar Drinks. Neuron, 44(October), 379-387.

37

Metcalf, Linda(1999), Teaching Toward Solutions: Step-by-step strategies for

handling academic, behaviour and family issues in the classroom. Centre
for Applied Research. New York: Simon & Schuster.
Morrison, Donald G. (1981). Triangle Taste Tests: Are the Subjects who
Respond Correctly Lucky or Good? Journal of Marketing, 45(Summer), 111119.
Moskowitz, Howard R. (1985). New Product Testing and Sensory Evaluation of
Foods. Westport, CT: Food and Nutrition Press, Inc.
Obermiller, Carl and Eric R. Spangenberg (1998). Development of a Scale to
Measure Consumer Skepticism toward Advertising. Journal of Consumer
Psychology, 7(2), 159-186.
Park, C. Whan, Mothersbaugh, David L. and Lawrence Feick (1994). Consumer
Knowledge Assessment. Journal of Consumer Research, 21(June), 71-82.
Penny, J. C., I. M. Hunt and W. A. Twyman (1972). Product Testing Methodology
in Relation to Marketing Problems: A Review. Journal of the Market
Research Society, 14(January), 1-29.

38