FACULTY OF APPLIED SCIENCE

FST568
AS246

PRACTICAL 4 : Difference Test

NAME MATRIX NUMBER

Nurulain Binti Mohd Asri 2022912321

GROUP:
AS2465B

LECTURER:
PROFESOR DR ZAIBUNNISA ABDUL HAIYEE

DATE OF SUBMISSION:
28 MAY 2023
INTRODUCTION

Food sensory evaluation is an essential component of the beverage industry (Marques et al.,
2022), enabling researchers and manufacturers to assess the sensory attributes of food
products and understand consumer preferences. One common approach in sensory
evaluation is the difference test, which aims to discern perceivable differences between food
samples. By utilizing difference tests, researchers can specify sensory variations, evaluate
product quality, and make informed decisions regarding product development and
improvement. For instance, beverages such as coffee (Chapko & Seo, 2019; Kalschne et al.,
2019), apple juice (Da Silva et al., 2019), iced tea (Jaeger et al., 2020), alcoholic cocktails
(Pierguidi et al., 2021), and soy-free protein drinks (Aschemann-Witzel et al., 2019) have
benefited from the evolution of sensory evaluation methods, enhancing quality, and producing
beverages closer to consumers’ preferences (Marques et al., 2022).

Difference testing is practicable in sensory evaluation to ascertain the presence of

sensory discrepancies among food samples. According to (Difference Testing, n.d.) there are
three distinct types of difference tests are typically employed: the triangle test, the duo-trio
test, and the paired comparison test. The Triangle Test is based on the principle of presenting
three samples to participants, two of which are identical and asking them to identify the sample
that differs in sensory attributes (Choi, n.d.). This test assesses the ability to detect differences
between similar products. The Constant Reference Duo-Trio Test modifies the Triangle Test
by introducing a reference sample and two alternative samples, with participants tasked to
identify the sample that is different from the reference (Johnson, 2021). This test allows for
maintaining a constant reference point while detecting differences. The Directional Paired
Comparison Test evaluates both the presence and direction of sensory differences by
presenting sample pairs to participants, who must determine both the existence and direction
of differences (Choi, n.d.). This test provides insights into the magnitude and preference of
sensory differences between products.

By examining the principles and applications of these difference tests, this report aims
to offer a comprehensive understanding of their roles and effectiveness in food sensory
evaluation. The Drinho soy drink is likely easy to differentiate as the higher thresholds (30%
less sugar) than the soy drink from Yeos for the triangle test. Besides, the duo-trio test for
orange juice containing citric acid is slightly difficult to detect the contrast in sourness. On the
other hand, grape juice with extremely low threshold difference of sugar is also intricate to
differentiate in paired comparison tests.
OBJECTIVES

1. To determine whether detectable sensory differences exist between two different brands
of soy drinks and two formulation of orange juice.
2. To indicate which of two formulation of grape juice is sweeter.

INGREDIENTS

INGREDIENT AMOUNT
Drinho soy drink (less sugar) 1L
Yeos soy drink 1L
Sunquick orange cordial 60 ml
Nitric acid 0.3 g
Sunquick grape cordial 40 ml
Sugar 43 g
Water 4L

PROCEDURES

A. Triangle Test
A. Three samples coded with a three-digit random code number were assigned and
placed in random order of presentation on the tray.
B. Two of these samples were the same one is different.
C. The coded samples were evaluated in the sequence presented on the tray, from left to
the right.
D. Mouth was rinsed with water before begin tasting and between each sample tasting.
E. In google form, the number of the sample that is different was ticked. If no difference
is apparent, best guess was ticked.
F. The answer was compared with the model answer and data from all panellists were
gathered to conclude the findings.
G. Triangle test statistical chart (Appendix 3) was used for a quick test to determine if
there is statistically significant difference between the two samples.
 Figure 4.1: Triangle test set-up

B. Constant Reference Duo-trio Test

1. One sample given was marked R as reference sample and another two samples were
coded with random three-digits number.
2. One of the coded samples was identical to R and the other is different.
3. Mouth was rinsed with water before begin tasting and between each sample tasting.
4. The samples were tasted in the order presented on the tray (sample marked with R
first then followed by coded samples on the left).
5. The code number was ticked corresponding to the sample that is similar to R.
6. The answer was compared with the model answer and data from all panellists were
gathered to conclude the findings.
7. Two-sample test statistical chart (two-tail test) (Appendix 4) was used for a quick test
to determine if there is statistically significant difference between the two samples.

Figure 4.2: Duo-trio test set-up

C. Directional Paired Comparison Test

 1. Two samples coded with a three-digits random numbers were given.
2. The samples were tasted in the order presented from left to the right.
3. Mouth was rinsed with water before begin tasting and between each sample tasting.
4. The sweetness of these two samples were evaluated.
5. The answer was compared with the model answer and data from all panellists were
gathered to conclude the findings.
6. Two-sample test statistical chart (one-tail test) (Appendix 5) was used for a quick test
to determine if there is statistically significant difference between the two samples.

Figure 4.3: Simple paired comparison test set-up

Table 4.1: Sample preparation for difference test

SAMPLE PREPARATION
Soy drink (original)
Commercial beverages
Soy drink (less sweet)
30 ml orange cordial + 1 L water
Orange juice
30 ml orange cordial + 0.3 g nitric acid + 1 L water
20 g sugar + 20 ml grape cordial + 1 L water
Grape juice
23 sugar + 20 ml grape cordial + 1 L water

• Three-digits random numbers for each sample referred to the Table of Random Digits
(Appendix 1)
RESULTS AND CALCULATIONS

A. Triangle Test

MASTER SHEET

Sample Code and Order of Presentation Total

Panellist Permutation Number of
No. No. Sample A Sample A Sample B Sample B Correct
Answer
1 1 990 1 132 3 605 2
2 2 990 1 132 2 605 3
3 5 990 2 132 3 605 1
4 5 990 2 132 3 605 1
5 2 990 1 132 2 605 3
6 5 990 2 132 3 605 1
7 2 990 1 132 2 605 3
8 5 990 2 132 3 605 1
9 1 990 1 132 3 605 2
10 1 990 1 132 3 605 2
11 5 990 2 132 3 605 1
12 5 990 2 132 3 605 1
13 1 990 1 132 3 605 2
14 2 990 1 132 2 605 3
15 1 990 1 132 3 605 2

• Possible serving order:

Permutation 1: ABA

Permutation 2: AAB

Permutation 3: BBA

Permutation 4: BAB

Permutation 5: BAA

Permutation 6: ABB
Table 4.2: Data collected from triangle test

Sample Code and Order of Presentation Total

Panellist Permutation Number of
No. No. Sample A Sample A Sample B Sample B Correct
Answer
1 1 990 1 132 3 605 2 ✓

2 2 990 1 132 2 605 3 ✓

3 5 990 2 132 3 605 1 ✓

4 5 990 2 132 3 605 1 X
5 2 990 1 132 2 605 3 ✓

6 5 990 2 132 3 605 1 ✓

7 2 990 1 132 2 605 3 ✓

8 5 990 2 132 3 605 1 ✓

9 1 990 1 132 3 605 2 ✓

10 1 990 1 132 3 605 2 ✓

11 5 990 2 132 3 605 1 X
12 5 990 2 132 3 605 1 ✓

13 1 990 1 132 3 605 2 ✓

14 2 990 1 132 2 605 3 ✓

15 1 990 1 132 3 605 2 ✓

13 / 15
Figure 4.4: Statistical table for triangle test
From the table above, 13 panellists were correctly identified that sample 605 was different
from another 2 samples.

B. Constant Reference Duo-trio Test

MASTER SHEET

Sample Code and Order of Presentation Total No. of

Panellist
Reference A Correct
No. Permutation
Sample A Sample B Answer
No.
1 / 1 729 1 387 2
2 / 1 729 1 387 2
3 / 2 729 2 387 1
4 / 1 729 1 387 2
5 / 2 729 2 387 1
6 / 1 729 1 387 2
7 / 2 729 2 387 1
8 / 2 729 2 387 1
9 / 1 729 1 387 2
10 / 2 729 2 387 1
11 / 2 729 2 387 1
12 / 2 729 2 387 1
13 / 1 729 1 387 2
14 / 1 729 1 387 2
15 / 1 729 1 387 2

• Possible serving order

Reference choose = Sample A

Permutation 1 RA = AB
Permutation 2 RA = BA
Table 4.3: Data collected from constant reference duo-trio test

Sample Code and Order of Presentation Total No. of

Panellist
Reference A Correct
No. Permutation
Sample A Sample B Answer
No.
1 / 1 729 1 387 2 X
2 / 1 729 1 387 2 ✓

3 / 2 729 2 387 1 ✓

4 / 1 729 1 387 2 ✓
5 / 2 729 2 387 1 X
1 2
6 / 1 729 387 X
7 / 2 729 2 387 1 X
8 / 2 729 2 387 1 ✓

9 / 1 729 1 387 2 ✓

10 / 2 729 2 387 1 ✓

11 / 2 729 2 387 1 ✓

12 / 2 729 2 387 1 ✓

13 / 1 729 1 387 2 ✓

14 / 1 729 1 387 2 ✓

15 / 1 729 1 387 2 ✓
11 / 15
Figure 4.5: Statistical table for two sample test
From the table above, 11 panellists were correctly identified that sample 387 was similar to
the reference.

C. Directional Paired Comparison Test

MASTER SHEET

Sample Code and Order of Presentation Total Number

Panellist No. of Correct
Permutation
Sample A Sample B Answer
No.
1 2 655 2 427 1
2 1 655 1 427 2
3 2 655 2 427 1
4 2 655 2 427 1
5 1 655 1 427 2
6 1 655 1 427 2
7 1 655 1 427 2
8 2 655 2 427 1
9 1 655 1 427 2
10 2 655 2 427 1
11 2 655 2 427 1
12 2 655 2 427 1
13 1 655 1 427 2
14 1 655 1 427 2
15 2 655 2 427 1

• Combination of permutation:
Permutation 1: AB
Permutation 2: BA
Table 4.4: Data collected from directional paired comparison test

Sample Code and Order of Presentation Total Number

Panellist No. of Correct
Permutation
Sample A Sample B Answer
No.
1 2 655 2 427 1 ✓

2 1 655 1 427 2 ✓
3 2 655 2 427 1 X
4 2 655 2 427 1 ✓
5 1 655 1 427 2 X
6 1 655 1 427 2 ✓

7 1 655 1 427 2 ✓

8 2 655 2 427 1 ✓

9 1 655 1 427 2 ✓

10 2 655 2 427 1 ✓

11 2 655 2 427 1 ✓
12 2 655 2 427 1 X
13 1 655 1 427 2 X
14 1 655 1 427 2 X
15 2 655 2 427 1 X
9 / 15

From the table above, only 9 panellists were identified that sample 655 had sweeter flavour
compared to 427.
Figure 4.6: Statistical table for paired difference test
DISCUSSION

For triangle test, from the table 4.2, 13 panellists (86.67%) were correctly identified
that sample 605 was different from another 2 samples. According to Statistical Chart (Fig. 4.4),
9 number of panellists out of 15 in a triangle test were required to give correct judgements at
5% significance levels (95% confidence level). In conclusion, there is a detectable difference
between both soy drinks from different brands which were Drinho and Yeos, at 5% significance
levels. This may be due to huge difference in sugar content in Drinho soy drinks which 30%
less sugar content (12.3 g) in Yeos soy drinks is 38% sugar (19 g).

Besides, for duo-trio test, the table 4.3 shows 11 panellists (73.33%) were correctly
identified that sample 387 was similar to the reference. According to Statistical Chart (Fig.
4.5), 12 number of panellists out of 15 in a duo-trio test required to give correct judgements at
5% significance levels (95% confidence level). In conclusion, there is no detectable difference
between the two formulations in terms of sour flavour at 5% significance levels. This due to
the addition of little citric acid (0.3 g), hence, make it difficult for the panellist to detect the
difference.

Furthermore, for paired comparison test, the table 4.4 indicate that only 9 panellists
(60%) can identify that sample 655 had sweeter flavour compared to 427. According to
Statistical Chart (Fig. 4.6), 12 number of panellists out of 15 in a simple paired comparison
test required to give correct judgements at 5% significance levels (95% confidence level). In
conclusion, there is no detectable difference between the two formulations in terms of sweet
flavour at 5% significance levels. This because there were only 3 g difference of sugar content
in the two formulations.

On the other hand, from the above discussion, can be concluded that the three-product
(triangle) test justifies its assertion of the most sensitivity as the likelihood probability of only
1/3. Nevertheless, this test is more complicated because the panellist must recall two products'
sensory attributes before assessing a third and making a decision.

However, there might be some errors while conducting this sensory evaluation. For
instance, a panellist might have no strong likes or dislikes for the beverages to be tasted. Next,
the number of samples to be tasted should be only 6 samples per session to avoid taster
fatigue. Besides, poor health conditions such as fever and flu due to the weather during the
evaluation. Apart from that, panellists were given too much information about the sample, and
nature of the experiment, and the type of product before the test is conducted by reading the
lab manual before the evaluation. Respond of one panellist will influence other panellists which
leads to mutual suggestion. Next, the lack of motivation happens when the panellist has to
wait a very long time for their turn to enter the sensory evaluation room. Therefore, to avoid
stimulus error, all samples must be as uniform as possible and mask unwanted differences.
Furthermore, use a random presentation to avoid errors of central tendency.

CONCLUSION

In summary, there is a detectable difference between both soy drinks from different brands
which were Drinho and Yeos, at 5% significance levels as there were 13 panellists able to
detect sample that is different. However, there is no detectable difference between the two
formulations in terms of sour flavour at 5% significance levels as only 11 panellists able to
detect sample that identical to the reference. Besides, there is also no detectable difference
between the two formulations in terms of sweet flavour at 5% significance levels as there were
only 9 panellists that able to identify the sweeter sample. The objectives are achieved.

REFERENCES

Aschemann-Witzel, J., Varela, P., & Peschel, A. O. (2019). Consumers’ categorization of food
ingredients: Do consumers perceive them as ‘clean label’ producers expect? An
exploration with projective mapping. Food Quality and Preference, 71, 117–128.
https://doi.org/10.1016/J.FOODQUAL.2018.06.003

Chapko, M. J., & Seo, H. S. (2019). Characterizing product temperature-dependent sensory

perception of brewed coffee beverages: Descriptive sensory analysis. Food Research
International, 121, 612–621. https://doi.org/10.1016/J.FOODRES.2018.12.026

Choi, S. E. (n.d.). Sensory Evaluation (Vol. 110, p. 84). Difference testing is practicable in
sensory evaluation to ascertain the presence of sensory discrepancies among food
samples. In the sensory laboratory, three distinct types of difference tests are typically
employed: the triangle test, the duo-trio tes….

Da Silva, T. M., Marinoni, D. T., Peano, C., & Giuggioli, N. R. (2019). A New Sensory Approach
Combined with a Text-Mining Tool to Create a Sensory Lexicon and Profile of
Monovarietal Apple Juices. Foods 2019, Vol. 8, Page 608, 8(12), 608.
https://doi.org/10.3390/FOODS8120608
Difference Testing. (n.d.). Byu. Retrieved May 28, 2023, from https://ndfs.byu.edu/sensory-
lab/difference-testing

Jaeger, S. R., Beresford, M. K., Lo, K. R., Hunter, D. C., Chheang, S. L., & Ares, G. (2020).
What does it mean to check-all-that-apply? Four case studies with beverages. Food
Quality and Preference, 80, 103794. https://doi.org/10.1016/J.FOODQUAL.2019.103794

Johnson, M. (2021, February 2). Duo-Trio Test. Society of Sensory Professionals.

https://www.sensorysociety.org/knowledge/sspwiki/Pages/Duo-Trio%20Test.aspx

Kalschne, D. L., Biasuz, T., De Conti, A. J., Viegas, M. C., Corso, M. P., & Benassi, M. de T.
(2019). Sensory characterization and acceptance of coffee brews of C. arabica and C.
canephora blended with steamed defective coffee. Food Research International, 124,
234–238. https://doi.org/10.1016/J.FOODRES.2018.03.038

Marques, C., Correia, E., Dinis, L. T., & Vilela, A. (2022). An Overview of Sensory
Characterization Techniques: From Classical Descriptive Analysis to the Emergence of
Novel Profiling Methods. Foods 2022, Vol. 11, Page 255, 11(3), 255.
https://doi.org/10.3390/FOODS11030255

Pierguidi, L., Spinelli, S., Monteleone, E., & Dinnella, C. (2021). The combined use of temporal
dominance of sensations (TDS) and discrete time-intensity (DTI) to describe the dynamic
sensory profile of alcoholic cocktails. Food Quality and Preference, 93, 104281.
https://doi.org/10.1016/J.FOODQUAL.2021.104281
 APPENDIX

Appendix 1

Table of random digits