Sensory Evaluation Lab Report: Taylor Zwimpfer Nutrition 205, Section 3 28 October 2013
Sensory Evaluation Lab Report: Taylor Zwimpfer Nutrition 205, Section 3 28 October 2013
Taylor Zwimpfer
Nutrition 205, Section 3
28 October 2013
Zwimpfer 2
Abstract
The purpose of this study was to utilize and evaluate the importance of sensory analysis tests for
determining food acceptability, preference, and characteristics. In a food laboratory at San Diego State
University, a total of 89 untrained panelists between the ages of 19 and 35 years (79 females, 10 males) were
subjects in seven different sensory evaluation tests including: descriptive analysis, beverage association, paired
comparison, triangle testing, duo trio, scoring/rating testing, and a ranking test. Tests were conducted to
illustrate the importance of the five senses (sight, smell, taste, touch, and hearing) in determining food
acceptability and preference. Data and results were compiled and analyzed within this study to explain the
different methods of sensory analysis and their relevance to food producers and consumers. Sensory evaluations
have been proven useful in many food and beverage studies and aid in predicting the successfulness of products
within the market. This study demonstrates the significance of chemical compositions of a food in relation to
consumer liking, wanting, and behavioral attitude/perceptions toward it. Results of the scoring/rating test,
paired comparison test, triangle test and ranking test demonstrated that panelist were able to differentiate sour
intensity between apple juice samples containing different amounts of citric acid concentration. The beverage
association test results explained and inverse relationship between perceived beverage artificiality and
naturalness as color changes from light yellow to emerald green. The analysis of goldfish, raisins,
marshmallows, and almonds using descriptive terms yielded results depicting flavor and aroma as being two of
the most agreed upon categories to label within the panel. Some of the tests obtained results that correlated well
together, while others proved to be limiting. Certain errors could have come into play for each of the tests which
are explained within the study.
Introduction
Appearance, flavor, texture, aroma, consistency and mouthfeel are just a few of the factors influencing
food selection and preference. In order to get a better understanding of why consumers select the foods that they
do, sensory analysis is employed. Sensory analysis is an experimental approach to measuring consumer
feedback of foods through the senses of sight, smell, taste, touch, and hearing. This type of evaluation is known
as subjective because it relies largely upon the opinions of selected individuals, not on measurable, quantifiable
data. Sensory Analysis can further be broken down into analytical (effective) tests and affective tests. For the
purpose of this study analytical tests were used through a panel of untrained students in a food science class at
San Diego State University. These analytical tests were used to detect differences among various samples of
beverages and foods. Certain analytical tests differentiate between samples, these tests fall under the category of
difference tests. Among these tests include triangle, duo-trio, paired comparisons, ranking, and scoring/rating.
Others are descriptive tests that quantify differences and are “used to detail the specific flavors (garlic, vanilla,
caramel, boiled milk) or textures (smoothness, springiness, moistness) of a food or beverage” (Brown 2011).
Zwimpfer 3
The Beverage Color and Association Test was the first to be conducted and was utilized to look into the
influence of color on our perception of the flavor of food and drink. Five different beverages were placed on the
front table of the laboratory and given color identities (light yellow, dark yellow, chartreuse, dark chartreuse,
and emerald). Panelists were not told what was inside of the samples but were asked to rate them on a one to
five scale in the categories of sweetness, sourness, artificiality, naturalness, prefer, and dislike. Panelists were
also asked to record at what temperature they would drink the beverage sample (hot, warm, tepid, or cold) and if
they would drink it (yes or no). According to a study by Shankar et al. “color conveys critical information about
the flavor of food and drink by providing clues as to edibility, flavor identity, and flavor intensity”. Color seems
to constitute a key role in our perception of both food identity and flavor. In addition, color maintains symbolic
and associative information as a result of lifetime experience. The study by Shankar et al. assessed the
spontaneous color-flavor associations held by particular groups of individuals in order to examine how
associations of color vary in different cultures. It was hypothesized that human flavor perception may not only
be influenced by multisensory integration but also by higher level cognitive information such as expectations.
These expectations are created through past experience and repeated co-occurrence of specific colors with
specific tastes, flavors, and aromas. Twenty British Participants and 15 Taiwanese participants took part in the
experiment. Seven drinks were presented to the participants, of which contained 125mL of water with varying
amounts of food coloring to create brown, blue, orange, yellow, red, green, and clear samples. Participants
were instructed to look at each drink that was presented and “based solely on its color, write down the first
flavor or drink that came to mind upon viewing this color in the data table” provided (Shankar et al. 2010). The
goal of the experiment was to see if color has a symbolic effect on the way individuals or groups perceive
something to taste which would subsequently lead to an approach or avoidance behavior toward that food or
beverage.
Descriptive analysis in sensory evaluations is used to describe sensory attributes or qualities of a certain
food. For the study done in the nutrition lab, descriptive analysis was performed on marshmallows, almonds,
goldfish, and raisins. Panelists analyzed their samples of each using descriptive terms for appearance, flavor,
Zwimpfer 4
texture, aroma, consistency, and mouthfeel. Typically this kind of test is done to help producers determine what
consumers like or dislike about a particular product and the discernible attributes that it possesses. In a study by
Varela et al., descriptive analysis was used to uncover the drive behind coffee liking. A total of ninety-six
consumers of coffee took part in the study (31 men and 65 women, from 20 to 67 years of age). They received
six samples of coffee including varying amounts of added sugar and milk. Participants were asked to rank the
samples in order of preference and then leave “open comments” as to what they liked or disliked about the
individual samples. Similar descriptive terms were stated by the participants in both the like and dislike
categories, thus alluding there are strong traits of coffee that are seen amongst numerous people. Having this
information and feedback demonstrates how descriptive analysis is useful in determining the prevailing sensory
The triangle test was utilized to differentiate between three samples, two of which were the same and
one was different. The triangle test simply asks of the evaluator, which sample is different from the other two?
This type of test is beneficial to producers because it provides consumer feedback on whether or not two
products contain recognizable differences. For this study, three coded beverage samples were given to the
panelists, two containing apple juice with 0% citric acid added to them and one containing apple juice with 1%
citric acid added to it. Panelists were unaware of the contents of the beverage and were asked to state which
sample was different from the other two. At the University of California Davis, a triangle test was used in a
study by Rousseau, et al. for commercially available Dijon style mustards. Each of the twenty participants
participated in 8 triangle test sessions where they were presented with sample mustards given in different
sequences, BAA, ABB, AAB, BAB, ABA, and BBA where A and B are indicative of different Dijon mustard
brands. The results demonstrated that samples provided in certain sequences were more easily detected as being
“same” or different”. This would allude that the sample Dijon mustards provided possibly did not have a great
The duo-trio test is similar to the triangle test in that it involves three samples, two of which are the
Zwimpfer 5
same and one is different. Participants in the duo-trio test are asked to determine which sample is different and
what attribute the odd sample has that makes it different. For this study, three cookies of similar appearance
were presented to the panelists. Two cookies were the same and one was different. Panelists were asked to
determine which cookie was odd and what made it stand out from the other two in relation to dryness,
crunchiness, or having less vanilla. The duo-trio test in this regard might be useful for producers creating an
imitation product to a name brand or creating a product fortified or enriched with additives for health purposes.
One example of using the duo-trio test for fortified foods occurred in a study by Beinner et al. The study
conducted was used to examine the perceived differences between conventional rice and rice fortified with iron
among Brazilian people. The duo-trio test was applied under controlled conditions for thirty-seven untrained
panelists. Each panelist received a standard rice sample and two unknown samples of rice. One was the same as
the standard and one was different. Upon determining which rice was different, panelists described that
difference using sensory analysis. These descriptions for the variance in product taste, texture and color help to
determine consumer acceptability. This is one benefit of using duo-trio testing (Beinner et al.).
One difference test employed by this study was the ranking test, in which various samples are ranked by
the intensity of a certain characteristic that they possess. The study used five different samples of apple juice,
each with assorted amounts of citric acid added to them. Panelists were given the samples and allowed to try
them in any order so long as they ranked them from least sour to most sour. This method is used by producers to
understand and determine consumer acceptability for a certain sensory characteristic. Numerous other studies
have been conducted using ranking tests, including one by Kildegaard et al. on the preference, liking, and
wanting for beverages in children and the role of sourness perception in beverage selection. It was hypothesized
that sourness perception is a strong factor in the rejection of a food. Research shows that “generally preferences
increase with decreasing levels of citric acid”. One hundred ninety-five children completed the study by tasting
two types of beverages, apple juice and a Danish fruit drink made of 11 different kinds of fruits and berries.
Acid content was held constant for both drinks but the amount of dry matter content was altered, there were no
Zwimpfer 6
visible color changes within samples. Children were told to rank 8 different samples (4 of each juice type) from
most sour to least sour. Results demonstrated that dry matter content added to juices did in fact affect perceived
sourness of the juice. The ranking test helped to determine a correlation between preference and perceived
The paired comparison test was used in this study to determine which of two samples has more of a
certain characteristic. Two coded sample beverage were given to panelists for tasting to evaluate which sample
had a greater sour intensity. One of the beverages was apple juice with 0% citric acid and the other was apple
juice with the addition of 1% citric acid. This difference test is a beneficial source of food analysis because it is
easily performed due to the requirement of a small sample size and is basic enough to be performed on an array
of food products. Food and beverage producers would find the paired comparison test to be advantageous
because it allows them to measure the intensity of a certain characteristic that their product may contain through
Scoring tests allow different samples to be arranged on a numeric scale relative to each other when
being tested for a certain characteristic. For instance, in this study a reference sample was given and told that it
was rated with the number 4 for its sour intensity. Two more samples were given after and panelists were asked
to rate or score these two samples relative to the reference sample. The reference sample was apple juice with
2.5% citric acid and the other two samples were apple juice with 5% and 1% citric acid. Producers can use the
results of a scoring test to measure the intensity of products and determine which will be accepted by consumers
Methods
Panelists
Experiments were conducted in a nutrition lab at San Diego State University amongst students enrolled
in Introduction to the Science of Food (Nutrition 205). The experiments consisted of various sensory evaluation
tests performed using an array of foods and beverages. Panelists were each given a paper labeled “Demographic
Zwimpfer 7
Questionnaire” and asked to answer all questions independently and without any discussion with other panelists.
The questionnaire asked panelists to provide information regarding age, gender, major, student status, marital
status, living arrangements, smoking, and food allergies. Once all questionnaires were completed, panelists’
information was input into a single computer and compiled by the instructor of the course. In order to do this,
panelists closed their eyes and the instructor asked each question separately. Answers to each question were
given by students raising their hands when their answer was called out. A total of eighty-nine panelists from
four different lab sections participated in the demographic questionnaire and subsequent sensory evaluation
tests. Ages ranged from 19-35 years old with the average age being 23.4 years old. Of the eighty-nine panelists,
78.65% were between the ages of 19 and 23 and the remaining 21.35% were above the age of 23. Gender was
split unevenly with females accounting for 88.76% of panelists and males accounting for 11.24%. When asked
their marital status, 93.26% of panelists reported themselves as single, 5.618% married and 1.124% divorced.
The panel was comprised of 95.51% undergraduate students and 4.494% graduate students. All panelists were
foods and nutrition majors. Living status ranged from 12.36% of panelists living alone, 26.97% living with one
roommate and 60.67%sharing a home with two or more roommates. The percentage of non-smokers among
panelists was 98.88%. When asked if any food allergies were present only 10.11% of panelists said that they did
have allergies from foods including: shellfish, salmon, cinnamon, eggs, mayonnaise, dairy, gluten, soy, and
preservatives.
Environment
All sensory evaluation tests were conducted in a nutrition laboratory at San Diego State University.
Panelists sat at individual desks that were arranged in five lines facing the front of the room. They were
instructed to answer all questions and conduct all evaluations individually. The instructor sat at the front of the
class and input data into the computer. All samples were placed on an island in the front of the class from which
the first individual in each row collected them. The laboratory was well lit and had normal room temperature.
Zwimpfer 8
five glass beakers from the refrigerator and lined them up evenly spaced on the island in the front of the room.
Each beaker contained different colored beverages ranging from light to dark (light yellow, dark yellow,
chartreuse, dark chartreuse, and emerald green). Panelists received a questionnaire form to record their
individual perception. The first question asked was “Do you drink Apple Juice?” to which the panelists
answered either yes or no. After answering this, each panelist was asked to rank the beverages at the front table
based on a scale of 1 (lowest) to 5 (highest) for the following parameters: sweetness, sourness, artificiality,
naturalness, prefer, and dislike. Participants then were asked at what temperature they would drink each
beverage. They chose from the temperatures of hot, warm, tepid, or cold. The final question panelists answered
was “would you drink it?”, for this the answers were either yes or no.
Panelists filled out their questionnaires individually. Once each person was done, panelists were
prompted to close their eyes for the instructor to perform a blind tally of raised hand votes for each question and
category. When all results were collected and input into the computer, the beverage names were revealed. The
light yellow was Mountain Dairy Lemonade, dark yellow was Xtremo Citrico Vibrante Gatorade, chartreuse
was 350 mls. Lemon Lime Gatorade plus 150 mls., dark chartreuse was Green Squall Powerade, and emerald
This descriptive test was used to detail specific properties of almonds, marshmallows, gold fish and
raisins. The first individual of each row of panelists approached the front of the classroom and obtained all
sample cups for the rest of the panelists in their row. Each panelist received one sample cup for each food
product being evaluated containing two of each item (ie. two raisins per person). Participants were prompted to
evaluate each food product based on the categories: appearance, flavor, texture, aroma, consistency, and mouth
feel. They performed the evaluation by assigning a term from a predetermined list of descriptive words to each
Zwimpfer 9
product. In the test for goldfish, two panelists did not participate and in the test for marshmallows one
individual did not participate in some of the categories due to dietary restrictions. Results were first recorded in
panelist’s individual notebooks and then compiled by the lab technician into the computer. This required all
panelists to close their eyes and raise their hands for the different categories and descriptive terms that they
chose.
Difference Tests
Paired Comparison
This difference test was conducted to determine which of two samples given to the sensory panelist was
identified as having a greater intensity than the other. For this comparison, the trait being tested for was
sourness. The instructor allowed an individual from the beginning of each row of participants to collect samples
and pass one of each to the rest of the individuals in their row. Panelists were presented with two samples in
plain white paper cups labeled with arbitrary codes of 635T1 and 573T2. They were instructed to taste each one
and record their perception of having a “greater” or “lesser” intensity based on the sourness characteristic.
Panelists recorded their individual perceptions and then closed their eyes. The lab technician prompted for all
participants to keep their eyes closed as they raised their hands to report which sample had a greater and lesser
intensity. The lab technician counted hands and input the results into the computer. At the end panelists were
informed that sample 635T1 was 0% citric acid solution and sample 573T2 was 1% citric acid solution.
Duo-Trio Test
This difference test was given to panelists in the form of three sample cookies. The lab technician passed
out one cookie (Sample 8175) and told panelists that it was the standard. Two more cookies (Samples 6104 and
1108) were presented to the panelists and they were informed that one of the two cookies was the same as the
standard and one was different. Participants were not told what this difference was but were asked to first taste
the standard cookie and consider the characteristics they notice about its taste and mouth feel. Panelists were
then able to eat the next two samples and drink water in between each sampling to cleanse their pallet. When
Zwimpfer 10
finished tasting the samples, panelists were told to record which sample they thought was the same as the
standard. They then answered in their opinion what the major factor was that differed between samples: dryness,
crunchiness, or vanilla content. All answers were given to the lab instructor by students raising their hands and
recorded in the computer like previous tests. At the end, panelists were informed that the standard 8175 cookie
was a Nabisco Nilla Wafer as well as sample 1108. Sample 6104 was First Street (Smart and Final) Vanilla
Scoring Test
Three beverage samples were introduced to the panelists: Sample 420M, Sample S723, and Sample
0110 (reference sample). An individual from the front of each row approached the front of the class and poured
one of each sample for themselves and the others in their row. Samples were passed out by the individual and
the seated panelists were informed which sample was which. Samples were ranked on a scale of 1 (more sour)
to 7 (less sour). The reference sample was assigned an arbitrary score of 4. Panelists were instructed to rate the
sour intensity of the other two samples relative to the reference. Panelists raised their hands as the instructor
called out the various scores and samples so their answers could be recorded in the computer. When all
responses were recorded, panelists were told what the samples consisted of. Sample S723 contained 5% citric
acid solution, 420M contained 1% citric acid solution and the reference sample 0110 contained 2.5% citric acid
solution.
Ranking Test
The ranking test was conducted using 5 similar beverage samples of varying intensity of sourness and
assigned sample codes (695F8, 495P2, 192L3, 543K8, and 555D7). One person from the front of each row of
panelists approached the front of the room and poured enough of each sample for all others in their row and
distributed the samples. Panelists were instructed to taste each sample and rank them in descending order of
sourness from #1 (most sour) to #5 (least sour). They then were to rank the samples in order of preference. No
ties were allowed; however, panelists were allowed to retaste any sample. Panelists were encouraged to drink
water in between tasting each sample. Results were recorded by the lab technician by having panelists close
Zwimpfer 11
their eyes and raise their hands when each ranking number for intensity and preference for each sample was
called out. It was revealed that each sample contained varying amount of citric acid: 695F8 (2.5%), 495P2 (0%),
Triangle Test
The triangle test was utilized to have panelists differentiate between three beverage samples and
determine which sample was not identical to the other two. All samples appeared similar and were given to
panelists simultaneously by the first panelist in each row. Each sample had a code on it (777C1, 542E2, and
112H9) that the panelists were told to record and state which two samples were the same and which one was
different. Results were given to the lab technician through a show of hands as the sample numbers were called
out as being the same or different. Samples 777C1 and 542E2 were revealed to be the same with 0% citric acid
in them and Sample 112H9 was the odd sample with 1% citric acid.
Statistical Analysis
A statistical analysis of results from four different laboratory groups of panelists was conducted by the
head laboratory technician. Data and results from all participants were compiled into an excel sheet and
reported as percentages. Percentages were calculated for the following sensory evaluation tests: duo-trio,
triangle, paired comparison, ranking, scoring, descriptive evaluation, and color association/perception of
Results
when comparing associations of sourness vs. sweetness (Figure 1), most natural vs. most artificial (Figure 2),
and prefer the most vs. dislike the most (Figure 3).
In Figure 1, it can be seen that there was an inverse relationship formed between color intensity and
perceived sourness of the sample beverages as color changed from light to dark. Light yellow was perceived as
Zwimpfer 12
being the most sour (44.94% of votes), followed by dark yellow (17.98%), chartreuse (14.61%), dark chartreuse
(12.36%) and emerald (10.11%). Color did not greatly affect perceived sweetness; however, as emerald
received the most votes for perceived sweetness (43.82%), followed by light yellow (25.84%), dark yellow
(19.1%), chartreuse (4.49%), and dark chartreuse (4.49%). There was no difference in perceived sweetness
Figure 2 demonstrates a relationship between natural and artificial perception of the colored beverage
samples. As color intensity increase from light yellow to emerald, voting for perceived naturalness increased.
Light yellow received an overwhelming majority of panelist votes at 96.63% for seeming to be the most natural.
The degree of perceived naturalness decreased as color ranged from dark yellow (2.25%), chartreuse (1.12%),
dark chartreuse (0%), to emerald (0%). The most artificial colored beverages received votes in almost reverse
order: emerald (79.78%), dark yellow (8.99%), dark chartreuse (5.62%), chartreuse (4.49%), light yellow
(1.12%). Dark yellow was the only sample that caused questioning of an inverse relationship between perceived
artificiality and perceived naturalness, as it was ranked second to emerald for being most artificial.
100
Percent of Votes Recieved
50 90
45 80
40
35 70
30 60
25 50
20 40
15 30
10
5 20
0 10
0
Sweetest
Most Natural
Most Sour
Most Artificial
Colors of Sample Beverages Observed Color of Sample Beverages Observed
Figure 3 demonstrates the percentage of panelist votes in regards to preference and dislikes of the
various beverage samples. Preference increased from light yellow (68.54%) to chartreuse (13.48%), to dark
chartreuse (7.87%). The two least preferred samples were emerald (5.62%) and dark yellow (4.50%), which
Zwimpfer 13
coincidentally were also viewed as being the two most artificial samples. Dislike of beverages was greatest
among these two samples with emerald receiving the most votes (59.55%) and dark yellow receiving the second
most votes (26.97%). dark yellow (5.62%), light yellow (4.50%) and chartreuse (3.37%) followed next,
When asked if they would drink the five beverages, panelists responded with either a yes or no. Yes
votes are demonstrated in Figure 4. The greatest amount of “yes” votes was allotted to light yellow (35.10%),
followed by chartreuse (22.86%), dark yellow (18.37%), dark chartreuse (14.29%) and emerald (9.39%).
80
70 9.39%
60
50 Light Yellow
40 14.29% 35.10% Dark Yellow
30
20 Chartreuse
22.86%
10 Dark Chartreuse
18.37%
0
Emerald
Prefer the Most
Dislike the Most
Table 1 depicts the temperatures at which panelists preferred to drink the varying beverages. All
beverages were seen as being best served cold obtaining 94.38% of votes for light yellow, dark yellow,
chartreuse, emerald and 95.51% of votes for dark chartreuse. The second most accepted temperature was
tepid, receiving 5.62% of votes for emerald, 4.49% of votes for light yellow and dark yellow, and 3.37% of
votes for chartreuse and dark chartreuse. Hot and warm temperatures acquired the least amount of votes for
Panelists defined the qualities of goldfish that stood out most to them in the categories of appearance,
flavor, texture, aroma, consistency and mouthfeel. Table 2 illustrates the top three descriptive terms for each
sensory attribute evaluated. According to the panelists, the appearance of goldfish is best described as being
golden brown (38.2%), dry (26.97%) and symmetrical (6.74%). The some of the other appearance descriptions
included rough, asymmetrical, puffy, and light brown (26.97%). The flavor goldfish was found to be salty
(77.52%), stale (6.74%), and sharp (5.62%). Texture evaluation determined that goldfish are perceived as being
crunchy (50.56%), crisp (37.08%), gritty (3.37%) and other (5.62%). The aroma of goldfish is was not distinct,
causing 53.9% of panelists to vote it as being “nothing”. The other top two descriptive terms for goldfish aroma
were burnt (29.2%) and spicy (8.99%). The consistency of goldfish is brittle (48.32%), cheesy (23.60%), and
thin (13.48%). Crisp (44.94%), crunchy (37.08%), and gritty (11.24%) were the three most used terms to
Table 2: Top Three Descriptive Terms Associated with Goldfish for Each Sensory Category
Appearance Flavor Texture Aroma Consistency Mouthfeel
Golden Brown- 38.2% Salty-77.52% Crunchy- 50.56% Nothing- 53.9% Brittle- 48.32% Crisp- 44.94%
Dry- 26.97% Stale- 6.74% Crisp- 37.08% Burnt- 29.2% Cheezy- 23.60% Crunchy- 37.08%
Symmetrical- 6.74% Sharp- 5.62% Gritty-3.37% Spicy- 8.99% Thin- 13.48% Gritty-11.24%
Other- 26.97% Other- 5.62% Other-5.62% Other- 6.74% Other- 11.24% Other- 2.25%
Table 3 depicts the top three terms used to express the sensory qualities identified by panelists as being
most prevalent for raisins. Raisins were believed to have an appearance that was best defined as sunken
(23.60%), glossy (17.98%), or dark (16.85%). The flavor of raisins was best described as sweet (51.69%), fruity
Zwimpfer 15
(34.83%), and flowery (6.74%). The texture of raisins was characterized as chewy (42.70%), gummy (23.60%),
and rubbery (13.48%). Fruity (42.70%), sweet (39.30%), and flowery (7.87%) were the most prevalent terms
reported as the aroma. Consistency was represented by the terms chewy (47.19%), gummy (24.72%), and
rubbery (22.47%). Mouthfeel was designated as sticky (61.80%), slimy (12.36%), and smooth (11.24%).
Table 3: Top Three Descriptive Terms Associated with Raisins for Each Sensory Category
Appearance Flavor Texture Aroma Consistency Mouthfeel
Sunken-23.60% Sweet- 51.69% Chewy- 42.70% Fruity- 42.70% Chewy-47.19% Sticky- 61.80%
Glossy-17.98% Fruity-34.83% Gummy- 23.60% Sweet- 39.30% Gummy-24.72% Slimy- 12.36%
Dark- 16.85% Flowery-6.74% Rubbery-13.48% Flowery- 7.87% Rubbery-22.47% Smooth- 11.24%
Other- 41.6% Other-6.74% Other- 20.22% Other- 10.1% Other- 6.07% Other-14.6%
Table 4 demonstrates the top three terms used to describe almonds through sensory evaluation. The
appearance of almonds was designated as dry (26.97%), golden brown (26.97%), and light brown (25.84%).
Flavor attributes of almonds were majorly nutty (80.90%), flat (12.36%) and stale (3.37%). Texture was
indicated as hard (25.84%), crunchy (24.72%), and firm (22.47%). 91% of votes for the aroma of almonds were
allotted to the term “nothing”. The following top two terms were burnt (5.62%) and flowery (2.25%). The
consistency of almonds from the panels perspective was mostly thick (56.18%), chewy (34.83%), and rubbery
(3.37%). Crunchy (61.80%), gritty (21.25%), and crisp (7.87%) best express the mouthfeel of almonds.
Table 4: Top Three Descriptive Terms Associated with Almonds for Each Sensory Category
Appearance Flavor Texture Aroma Consistency Mouthfeel
Nutty-
Dry- 26.97% 80.90% Hard- 25.84% Nothing-91% Thick- 56.18% Crunchy- 61.80%
Golden-Brown- 26.97% Flat- 12.36% Crunchy- 24.72% Burnt- 5.62% Chewy- 34.83% Gritty- 21.35%
Light-Brown- 25.84% Stale- 3.37% Firm- 22.47% Flowery- 2.25% Rubbery- 3.37% Crisp- 7.87%
Other- 20.20% Other- 3.37% Other- 26.97% Other- 1.12% Other-5.62% Other- 8.99%
Marshmallows were judged using descriptive analysis for all the sensory qualities listed in Table 5. The
appearance of marshmallows was identified most as puffy (83.15%), rounded (6.74%), and smooth (2.25%).
Flavor was denounced by three similar terms: sweet (68.54%), floury (14.61%), and sweet (68.54%).
Marshmallow texture was perceived as springy (22.47%), gummy (22.47%), and velvety (16.85%). A sweet
Zwimpfer 16
aroma was found most strongly in marshmallows with 86.50% of votes for the category. The following terms
used to describe aroma of marshmallows were nothing (6.74%) and flowery (3.37%). Consistency was found to
be gummy (32.58%), chewy (26.97%), and thin (12.36%). Mouthfeel resulted in descriptive terms including
Table 5: Top Three Descriptive Terms Associated with Marshmallows for Each Sensory Category
Appearance Flavor Texture Aroma Consistency Mouthfeel
Puffy- 83.15% Sweet- 68.54% Springy- 22.47% Sweet- 86.50% Gummy- 32.58% Smooth- 43.82%
Rounded- 6.74% Floury- 14.61% Gummy-22.47% Nothing- 6.74% Chewy- 26.97% Sticky-29.21%
Smooth- 2.25% Pasty-12.36% Velvety- 16.85% Flowery- 3.37% Thin- 12.36% Slimy- 19.10%
Other- 4.49% Other- 0% Other- 34.83% Other- 0% Other-24.72% Other-3.37%
Difference Tests
Paired Comparison
The paired-comparison test resulted in 98.88% of panelists voting that sample 573T2 containing apple
juice with 1% citric acid was the sample of greater sourness. Only 1.12% of panelist votes lead to the perception
that sample 635T1 containing apple juice with 0% citric acid was of greater intensity for sourness. Those who
voted for the sample apple juice with 1% added citric acid as being of greater sour intensity were correct in this
paired comparison test as depicted in Figure 5. The chance of selecting the correct sample was one out of two.
Sample Codes
573T2
98.88% 635T1
Zwimpfer 17
Duo-Trio Test
In the duo-trio test, both samples 8175 (standard) and 1108 were Nabisco Nilla Wafers brand cookies.
96% of panelists were able to determine that these two samples were the same while 4% of panelists thought
sample 6104 was equal to the standard. This cookie was actually an off brand, First Street (Smart and Final)
Vanilla Wafers. Figure 6 illustrates the reasoning of panelists as to why the off sample differed from the
standard. 57% of panelists attributed the difference as the odd sample having less vanilla, while 27% thought it
had a greater crunchiness. Dryness was explained by 15% of votes as being the source for differentiation among
sample cookies.
15%
Dryness
57% 27%
Crunchiness
Less Vanilla
Scoring/Rating Test
The reference sample (2.5% citric acid) for this scoring test was given an arbitrary number of 4 on the
1(more sour) to 7 (less sour) scale. As depicted in Figure 7, the majority of panelists voted 420M (1% citric acid
sample as having a score above the reference sample, and therefore perceived it as being less sour. 60.6% of
panelists rated the 1% citric acid sample as a 6 and 19.1% of panelists rated the sample as a 7. Approximately
5.61% of panelists thought the 1% citric acid sample was more sour than the reference and 1.12% thought it
was of equivalent sourness. In regards to sample S723 (5% citric acid) only 1.12% of panelists votes perceived
Zwimpfer 18
it as being less sour than the reference. A majority of voters (58.43%) gave the 5% citric acid sample a rating of
1 for being more sour, followed by 34.83% of votes rating the sample as 2.
80
Percentage of Panelist Votes
60
1% Citric Acid Sample
40
5 % Citric Acid Sample
20
Reference Sample (2.5%
0 Citric Acid)
1 2 3 4 5 6 7
More Sour Less Sour
Sour Intensity
Ranking Test
The ranking test results for sourness and preference of apple juice beverages with varying amounts of
citric acid are portrayed in Figure 8 and Figure 9. Figure 8 demonstrates a clear relationship between rank level
of sourness and levels of citric acid in apple juice. As the amount of citric acid in the apple juice increases, so
does the level of sourness. Panelists ranked apple juice with 0% citric acid as a 5 on average (86.52%). Apple
juice with 1% citric acid was mostly associated with a rank of 4 (86.52%) and apple juice with 2.5% citric acid
was associated with a rank of 3 (87.64%). Apple juice with 10% citric acid was ranked number one for being
the most sour by the majority of voters (96.6%), followed by apple juice with 5% citric acid ranked as number 2
by 88.76% of votes.
Figure 9 demonstrates that as the percent concentration of citric acid in apple juice increases, the rank of
preference decreases. 94.38% of votes ranked 10% citric acid apple juice as a 5 on the preference scale, and
78.65% of votes for apple juice with 5% citric acid were allotted to a ranking of 4. Apple juice with 2.5% citric
acid was found to be the second most preferential, receiving a rank of 2 from 62.92% of its votes. Apple Juice
Figure 8: Panelists Rank of Sourness for Figure 9: Panelists Rank of Preference for
Apple Juice Beverages With Varying Apple Juice Beverages Containing
Concentrations of Citric Acid Varying Amounts of Citric Acid
Triangle Test
In the triangle test, 97.75% of panelists voted sample 542E2 as being the same as sample 777C1. Both
of these samples contained apple juice with 0% citric acid. Only 2.25% of panelists found apple juice sample
112H9 as being the same as sample 777C1. In actuality, these two samples varied as 112H9 was apple juice
with a 1% citric acid concentration and 777C1 had 0% citric acid concentration. The results are depicted in
Figure 10 to visually demonstrate the percentage of correct matches and incorrect matched in the triangle test.
The probability of determining the same sample was one out of two.
Discussion
In reference to the beverage color and association tests, the results demonstrated that color has an effect
on perceived sweetness, sourness, naturalness, and artificiality of a beverage, which in turn influences a
person’s willingness to drink that beverage. The temperature of such beverages is affected as well by the color
and individuals wanting to drink it. The results of this study demonstrated that, for the most part, as the color of
the beverage became darker, perceived sourness of a beverage decreased. This perception is likely due to
already existing beverage drink associations that panelists maintained. For example, light yellow is indicative of
a beverage such as lemonade (very tart/sour) whereas dark chartreuse could be related to green Nyquil, which is
not perceived as being sour to the average individual. Results for perceived sweetness of a beverage did not
strongly correlate with color changes. Emerald and light yellow were one either extreme ends of the color
spectrum and were both thought to be the sweetest among the five beverage samples. This may be due to
external stimuli within the panelists as well. In the study by Shankar et al, it was found that the most common
food associations made for green beverages were mint, lime, apple, and kiwi. Lemon beverages were associated
most with lemon, pineapple, yellow soda, and white wine. Any of these plus more food affiliations could
provide reasoning as to why panelists voted as they did in each of the categories. This reasoning is likely
accountable for perceived artificiality and naturalness of colored beverages as well. This study found that as
beverage color transitions from light yellow to emerald, perceived artificiality increases (with the exception of
dark yellow) and perceived naturalness decreases. Overall, beverages were preferred most on the lighter side of
the spectrum (light yellow to chartreuse) and disliked the most on the darker side of the spectrum (dark
chartreuse to emerald). Shankar et al. determined in their experiment that flavor perception is “best
conceptualized as the interplay between multisensory integration and expectancy effects”. Panelists prefer the
light yellow beverage because they possibly expect it to taste like lemonade whereas they dislike emerald the
most because they expect it to taste artificial do to its unpopularity among natural drink sources (Shankar et al.
2010).
Zwimpfer 21
The results indicated that none of the colored drinks would on average be enjoyed as a temperature that
was above cold. For each colored samples, at least 94% of panelists said they would prefer the beverage to be
cold. The greatest percentage of panelists said that they would drink the light yellow beverage, likely due to the
expected flavors that it possessed. Emerald was the least likely to be drunk by individuals. It could be inferred
that this is due to its seemingly artificial nature based on color (ie. there are not many beverages associated with
The results of the descriptive analysis test performed in the lab demonstrated that different qualities of
foods stand out to different individuals. Flavor and aroma proved to be the most easily examined because for
these two categories, the greatest percentage of panelists used the same term to describe the attributes of the
various samples. For example, when choosing a term to describe the flavor of almonds, 80.90% of panelists
deemed the food to be nutty and 91% agreed that they had no aroma. The most inconsistent results were found
when evaluating appearance of the different foods. This is exemplified by only 23.60% of panelists naming the
appearance of raisins to be sunken. Though this was the term that the greatest number of panelists voted for, it is
still a small percentage in relation to the total population. In the study by Varela et al. on what factors stimulate
the liking of coffee it was found that most individual’s preference was influenced by the flavor and aroma of
coffee. Some examples of descriptive terms used to explain a liking for certain samples of coffee were right
sweetness (26.9%), coffee flavor (9.6%), milky flavor (8.8%), soft flavor (8.5%), and good flavor (8.0%).
Coffee aroma and good aroma were other comments used to describe a liking for the beverages. These findings
allowed the authors of the study to conclude the qualities of coffee that were consistently found among
consumers. Both the study on coffee and the study in the nutrition laboratory proved that descriptive analysis is
relevant to finding the strong qualities that a product may possess, especially relating to flavor and aroma.
There were also certain drawbacks to using the descriptive analysis test in the laboratory. Having a
predetermined list of terms to use for the analysis could have created bias or swayed panel members opinions
Zwimpfer 22
within the various sensory categories. Also, all evaluation was performed in an open room where panelists
could see and possibly interpret others reactions to the various samples. This would alter results for all
categories of sensory evaluation in descriptive analysis. In addition, these sample evaluations did not ask of
panelists to give feedback as to whether they liked or disliked the qualities evaluated within the samples.
Therefore, it is not known whether or not the qualities mentioned are good or bad for product acceptability
ratings.
The triangle test data revealed that 97.75% of panelists could accurately detect that sample 777C1 and
112H9 were different from each other and that sample 777C1 and 542E2 were the same. The likelihood of
selecting the odd sample by change was one in two. For this reason, it should not have been difficult to detect
which sample was different from the other two. In this test however, bias could have been created as panelists
could see the reactions of each other tasting the apple juices with 1% and 0% citric acid concentration. If a sour
face was made when one panelist was trying a sample, which could have led the other panelists to believe that
was the odd sample when it may not have been. In the triangle test used by Rousseau et al in determining a
difference in taste between commercial Dijon mustards, it was found that sequencing samples in different orders
led to different outcomes of correct and incorrect assumptions. Participants in the Dijon mustard study were
given six tests within numerous sessions of mustard triangle test tastings. One sequence of order proved to be
accurate far more times than the other five. This example explains how triangle tests might not be the most
beneficial for sensory analysis. Though they are easily performed, results can be altered based on sequencing of
samples as well as bias if other individuals are present during the testing.
The duo-trio test results suggested that 96% of panelists could tell that the Smart and Final Vanilla
Wafers sample was different from the standard Nabisco Nilla Wafers and 4% could not accurately determine
which of the two samples provided was the same as the standard. Panelists evaluated the difference they found
between the two wafer brands and determined that the greatest difference was in vanilla content, followed by
crunchiness and dryness. In the case of the study performed by Beinner et al. it was conceived that the
Zwimpfer 23
differences observed between the iron fortified rice and conventional rice included alteration in texture, color
difference, and a noticeable taste in one of the samples. Overall however, the iron fortified rice was well
accepted. The duo trio test in this case was used to help a producer find out if improving a products nutritional
content for health related reasons would affect consumer acceptance of that product. Duo trio tests can also be
beneficial for store brands to compare their product to name brand foods and beverages for competitive reasons.
One impediment to the duo trio test is that it merely compares products through differentiation; it does not
measure to what extent the products differ as a ranking or rating test would (Beinner et al.)
The ranking test results provided insight into the inverse relationship between sourness of a beverage
and preference for that beverage. The results indicated that with the addition of citric acid concentration to apple
juice, ranking levels increase (from 5 to 1) for sourness and levels decrease (from 1 to 5) for preference. Not all
panelists ranked their beverages in this way, but this assumption is based off of the average trends seen within
the data. The study by Kildegaard et al. had similar findings when testing children’s sourness perceptions and
preferences for different apples juices and fruit drinks. The findings of the study showed that “in a citrus
flavored model beverage system, the addition of sugars resulted in a suppression of sourness perception”. They
also found that preference increased as sourness decreased in both fruit drinks and apple juices. (Kildegaard et
al 2010).
Multiple factors could have affected panelist’s ranking choices, including sensory and memory fatigue.
Sensory fatigue is likely to occur within ranking tests because panelists are asked to try numerous foods or
beverages back to back and they are no longer sensitive to the stimuli that they are testing for. Drinking water in
between samples is used to combat this problem but is not always effective. Memory fatigue also occurs when
panelists are attempting to compare various items. If those items are not allowed to be retested, than a panelist
may forget all together what something tasted like or how strong it tasted. For this reason samples should be
allowed to be retested and should also be kept separately from each other to avoid confusion or swapping of
samples.
Zwimpfer 24
The paired comparison test resulted in 98.88% of panelists detecting that sample 573T2 (apple juice 1%
citric acid) contained more citric acid, and was therefore had a greater sour intensity than sample 635T1 (apple
juice with 0% citric acid) . This test was incredibly simple too produce due to its use of only two samples and
ease of data analysis. It is advantageous to food producers for that reason. Things that might interfere with the
results of a paired comparison test might include mixing up of samples, taste fatigue, or using two samples that
only have slight variations between them. Paired comparison also only allows for two samples to be used so it
can be limiting. This test merely asks which sample contains more of a distinct characteristic, if more
information is needed as to why the two samples are different, other tests would need to be conducted.
Results of the scoring/rating test were somewhat consistent among panelists. The reference sample (2.5%
citric acid) was placed at a rating of 4 in the middle of the rating scale. In the case of the 5% citric acid sample,
panelists could generally tell that I was had a greater sour intensity than the reference sample as all but 1.12%
rated it on the less sour side of the scale. For the 1% citric acid solution however, ratings ranged all over the
scale, indicating that panelists could not tell a prevalent difference between the reference and the sample at hand.
For this reason, the scoring/rating test does not appear to be the most effective difference test. Misjudgment
may have been caused by taste fatigue or confusion of instructions as to how to rate each sample. In the future,
it might be better to allow panelists to rate the reference samples for themselves, so they are not basing their
Sensory analysis is affected by many external factors, which can be beneficial or harmful to a study.
Because there are many different tests that can be manipulated in sensory analysis, it is important to choose the
right one. Certain tests such as paired comparison, duo trio and triangle tests are good to used when comparing
two items because the small sample sizes are easy to obtain feedback and results for in a short amount of time.
Other tests such as ranking, scoring, and descriptive analysis are good to use when taking into consideration
larger samples, but can also have their limitations. Many factors influence each of these tests and can alter the
results easily so analysis should be conducted in a controlled environment to yield the most accurate results.
Zwimpfer 25
References
Shankar M, Levitan C, Spence C. 2010. Grape expectations: The role of cognitive influences in color–flavor
interactions. Consciousness and Cognition 19 (1): 380-390. Available from Science Direct.com.
Posted 13 October 2009.
Beinner M, Soares A, Barros A, Monteiro M. (2010). Sensory evaluation of rice fortified with iron. Ciencia E
Tecnologioa De Alimientos, 30(2): 516-519.
Brown A. 2011. Understanding Food Principles and Preparation. 4th ed. Belmont: Wadsworth Cengage
Learning. 625 p.
Kildegaard H. et al. 2010. Preference, liking and wanting for beverages in children aged 9–14 years: Role of
sourness perception, chemical composition and background variables. Food Quality and Preference 22
(7): 620-627. Available from ScienceDirect.com. Posted 21 March 2011.
Rousseau B, Rogeaux M, O'Mahony M. 1999. Mustard discrimination by same–different and triangle tests:
aspects of irritation, memory and τ criteria. Food Quality and Preference 10 (3): 173-184. Available
from ScienceDirect.com. Posted 14 January 1999.
Shankar M, Levitan C, Spence C. 2010. Grape expectations: The role of cognitive influences in color–flavor
interactions. Consciousness and Cognition 19 (1): 380-390. Available from Science Direct.com.
Posted 13 October 2009.
Varela P, Beltrán J, Fiszman S. 2013. An alternative way to uncover drivers of coffee liking: Preference
mapping based on consumers’ preference ranking and open comments. Food Quality and Preference,
Available from ScienceDirect.com. Posted 13 March 2013.
Zwimpfer 26