OF LITY

ETHODS I SENSO
SPECIAL EfERENCE
CTS

H. KRISHNA IYER
Central Institute of Fisheries Technology, Cochin- f L

INTRODUCTION eristics of the fishery products, new

Sensory evaluation is still the most techniques of preparation and processing
reliable method for evaluation of the like artificial dehydration, canning under
freshness of raw a::1cl processed fishery different media, freezing and rating of
products. Sophisticate J methods like Inte- the speciality products will make the problem
lectron Fish Tester, cell fragility tech- more complicated. Hence the existing
nique and chemical and bacterioligical meth0ds of sensory evluation of quality and
methods like estimation of trimethylamine, the use of str.l.istical techniques need consi-
hypoxanthine, carb'myl compounds, deration in evaluating the quality of raw
volatile acid and total bacterial count have and processed fisher; products.
no doubt been developed for accessing the
spoilage in fish products. However, they
Methods or tests t;scd in sensory evaluation:
are of. very limited use as; none of these Quite a good number of tests are used
methods is a complete test by itself. in sensory evaluation of qualities of foods
Further, the reproducibility and depend- and beverages. These tests are broadly
ability of these methods are known to be classified under four heads as difference
upset by icing and other types of preserva- tests, preference tests, despriptive tests and
tion employed. The quality of raw and sensitivity tesrs. Difference tests are used
processed fishery products depend on several in situations where we are interested in
factors like physiological conditions at the finding whether a particular . food item is
time of capture, morphological differences, different from a standard or existing one_
rigor mortis, species, rate of icing and Descriptive tests are more of flavour profile
subsequent storage conditions. In ice type. The flavour profile method provides
storage, some of the flavour bearing a way for the evaluation of flavour by
components which otherwise would have an analytical descriptive procedure. It
been measured obj~ctively might leach concerns wirh the different types and
out and protein denaturation could take intensity of aroma and flavour factors.
place. Also, fishery products are having This type of analysis are used in the
their own characteristic flavour and aroma development, improvement and control
which are mostly complex in nature, which of a product as well as to detect trouble
vary with species and type of treatment .shooting points. Preferenc tests are usally
applied and which none of these objective used for getting an idea about the prefe-
methods so far developed can singly bring rence of a particular food item. Consu-
out successfully. More than the complex mer's reaction about the acceptability of a
nature of the flavour and odour charact- food item also comes under this group.

104 fiSHERY TECHNOLOGY

Krisltna lyer: Metfwds of Senor!) Evaluation of Ouality and Application of Statistical JY':etfwds in
Sensory Evaluation 'Pro~lems !Vitti Special 'Reference to 'Yisltery. 'Products

Senstitivity tests i~clade stimulus, difference blem on hand a particular method can be
and terminal thereshould tests as \VCll as selected.
dilution tests. These types of tests are gen-
erally used for evaluating the odour and Statistical methods in s£nsory evaluation
flavour of specific characteristics of foods problem:
and beverages. Since sensory evaluation methods are
Different methods of sensory evalua- to a great extent psycho-physical in nature,
statistical methods are quite often used in
tion which receive frequent references are
the design, analysis and interpretation of
single sample, paired comparison, duotrio,
data pertaining to experimer..ts on sensory
dual standard, multiple comparison, tria-
evaluation of foods and beverages. Simpler
ngle, rank order and scalar scoring test.
as well as more sophisticated statistical
Single sample test is usually used where
food items having an after taste or fla- methods are used frequently depending
vour carry over. Here only one sample upon the type of problem. Binomial distri-
is presented at a time for evaluation. bution, ' t ' distribution, analysis of vari-
Though it is easy to conduct but statis- ance technique, sequential analysis, fracti-
tically it is cnsonsidered to be less efficient. ?nal replication, technique of confounding,
In paired comparison test the panel members mcomplete block designs and Chi-square
are given a reference sample first followed statistic are used in the design and analyisis
by unknown samples for comparison. The of experiments. The purpose of statistical
panel member's task is to state whether analysis is to bring out significant differences
the unknown sample is different from the when they exist after accounting for chance
reference sample or not. A slightly diff- cau_ses. The_ chance probability of detecting
erent forms of paired comparison test are a difference m paired. comparison, duo-trio
the duo-trio and dual standard tests. and dual standard are 1. each while in
In duo-trio method the panel members are tria?gular test it is f· 1-n multiple com-
first given a reference sample A followed panson the chance probability is still less
by two samples as unknown. The panel depending on the number of test samples.
members' task is to say which of the two Binomial distribution and Chi-square statis-
belonged to A. In multiple comparison tic are quite often used for analysing data
method, judges are given a number of where paired comparison, duo-trio, dual
samples at a time and asked to select the standard and triangular methods are used.
sample which showed marked difference in Wald's sequential method of analysis are
the group. In the triangular method each used to determine the average number of
panel member is given three samples of triangular tests to be conducted for selecting
which two are identicaL They are asked judges having a specified ability to detect
to select the odd sample out of the three. flavour differences. Analysis of variance
In rank order test the panel members are technique is used for analysing data on
given a number of coded samples at a scalar scoring tests. Analysis of variance
time and asked to rank them according to tables are prepared under the assumptions
the intensity of a particular characteristic. that scores are normaly and independently
Some times three or more samples are distributed, treatment effects and environ-
provided to judges in a random order at a mental effects are additive and the error
time and asked to score or rate the sam- variances are homogeneous. Some times
ples according to a pre-determined scale. the scores obtained from a taste panel
This method is called scalar scoring me- data may not satisfy these assumptions.
thod. Depending upon the type of pro- Taste fatigue and taste adaption often make

VoL IX No. 2 1972 105

'Kristina lyer: Metliods o/ 8ensory E:oaluati:Jn of Ouality and Application of 8tatistica! Metftods in
8ensory Cualuation ·'Prof5lems roitli 8pecial 'Reference to Fisiiery 'Products

deviations from these assumptions. Taste quality of fishery products. The concen-
fatigue arises whenever too many samples tration of the solutions used were
are presented at a time for evaluation. Sodium Chloride- 0.5%; Hydrochloric acid
In such cases balanced incomplete block
designs can be successfully used. Where -0.25% Sucrose - 1%, Sodium Carbonate
ranking method was adopted, rapid mehod - 0.25%, and Quinine Sulphate- 0.05%.
of rank analysis can be applied. Each of these solutions were further diluted
Primary taste test: to give four samples of varying concen-
trations as 0%, 25%, 50% and 100% of
As a preliminary to the constitution
the prepared solutions. These solutions
of a taste panel primary taste testing are
generally conducted. The primary taste were arranged in random order and supp-
test is intended to screen members who lied to the entire Institutional members.
are sensitive to basic tastes. Solutions of A score sheet was used for assessing the
Sodium chlo;_·idc, Hydrochloric acid, Suc- taste of the solutions used and arranging
rose, Sodium Carbonate and Quinine Sul- them in their increasing order of concen-
phate are some of the solutions generally
used to represent the basic tastes viz. trations. Those who scored 80% and
saltiness, acidity (sourness), sweetness, above were considered successfull. The
alkalinity anJ bitterness respectively. The following Table (Table I) gives the number
same basic principles have been applied of identifications in the different solutions
for selecting a panel for ~valuation of by the mem~~rs.

TABLE I DETAILS OF IDENTIFICATION BY INSTITUTIONAL MEMBERS IN DIFFERENT SOLUTIONS

Sugar Alkaline Acid Salt Quinine

All correct 16 27 34 24 34
One mistake 22 14 17 21 13

Constitution of a taste panel for frozen cause of the presence of a highly signi-
and canned shrimp:- ficant interaction meansquare resulting from
the characteristic complex flavour stimuli
Krishna Iyer et: al, (1969) described of the product, taste fatigue and taste
certain methods they have tried in th~ adaption and the latter because of the
laboratory of Central Institute of Fishe- high total deviation (Table II, III, IV & V).
ries Technology, Cochin, for constituting
a taste panel for frozen and canned shrimp. Descriptive numerical scoring system:
The authors have tried three methods of
panel selection viz. Scalar scoring method, Numerical scoring system were frequ-
Range and deviation method and Trian- ently used for evaluating or rating the
gular method. Out of these three meth- quality of fishery products. Hedonic scale
ods tried, they were of opmwn that, was prepared in the past for this. purp-
triangular method was more suitable for ose. The hed-::mic scale contains express-
panel formation of such fishery products. ions of the form "like very much",
According to them the fist two methods "like", "neither like nor dislike", "dislike",
were not quite successful, the former be- "dislike very much". These expressions

]06 FISHERY TECHNOLOGY

'Krislina lyer: Metliods of Sensory Evaluation of Oua lity and Application of Statistical Metfiods in
Senary C:v~luaticn 'ProBlems witfi Special 'Reference to Fisliery 'Products

were given corresponding numerical scorings Boggs, M. M. & Hanson, H. L. 1949.

+ 2, + 1, 0, - 1, - 2 respectively for Advances in Food Research, Vol. 2.
the purpose of statistical trea.tment, Torry Bradly, R. A. 1953. Biometrics, 9, 22.
Research Station, Aberdeen, had developed
Caul, J. F. 1957. Advances in Food Re-
a descriptive numerical scoring system for
search, Vol. 7.
the ice-stored, smoked and frozen cha-
racteristics of white fish such as cod for Cochran, W. G. & Cox G. M. 1957. Expe-
judging the colour, odour (raw and cooked), rimental designs, John Wiley & Sons,
texture and flavour of the product. Such Newyork.
a descriptive numerical scoring system Committee on sensory evaluation of the
would provide a more uniform rating of Instititute of Food Technologists, 1964.
these products by the panel members. Sensory testing guide for panel eva-
luation of foods and Beverages, Food
ACKNOWLEDGEMENT
Technol, 18, 8
The author expresses his deep sense FAO Fisheries Reports, No. 19, 1965.
of gratitude to Dr. V. K. Pillai, Director, FAO Technical Report No. TA 2308,
C. I. F. T., for critically going through 1967.
this article and offering valuable sugges-
tions and comments. Kempthorne, 0, 1967. The design and
analysis of experiments.
REFERENCES Kramer, A & Twigg, B. A. 1962. Funda-
Arthur D. Little Inc. 1958. Flavour rese- mentals of quality control for the
arch and food acJeptance, Rainhold food industry.
publishing Corporation, Newyork, Krishna Iyer, H., Choudhury, D. R. &
London. Pillai, V. K., 1969. Fishery Techno!,
Baines, C. R. & Shewan, J. M. "Sensory 6 (2)
methods for evaluating the quality of Peryam, D. R., 1958. Sensory difference
white fish. Torry Memoir No. 189. teste, Food Technol; 12, 5.

TABLE II RESULTS OF SCALAR SCORING TEST-ANALYSTS OF VARIANCE TABLE FOR FLAVOUR

SCORES

Source of Variation ss DF MS

Total 93.0893 125

Bet. samples 2.7143 2 1.3572*
Bet. panel members 38.5893 13 2.9684';'
Samples X panel members 21.2857 26 0.8187*~'

Error 30.5000 84 0.3631

* denotes significance at 5% level

** denote significance at 1% level

VoL IX No. 2 1972 107

Kristina lyer: Metfl.ods of 8enso:ry Eoaluation of Oulaity and .Application of Statistical Metfiods in
Sensory Evaluation 'Prol5lems witli 8pecial 'Reference to Fisfier!J 'Products

TABLE m RESULTS OF RANGE AND DEVIATION METHOD

(Flavour scores)

Panel Range of Sum of Ran~e of sum Sum of

members sum ranges Range of sum deviations

1 4 l 4.00 (Sig) 4
2 2 5 0.40 (N.S) 8
3 2 5 0.40 (,) 6
4 4 3 1.33 , 7
5. 4 3 1.33 8
"
6. 3 4 0.75 8
"
1 4 4 1.00 " 4
8. 5 2 2.40 (Sig) 2
9' 0 6 0.00 (N. S.) 8
1(}! 3 4 0.75 6
"
H 2 3 0.67 7
"
12 2 5 0.40 ,, 6
13 3 3 1.00 10
4 4 1.00 "
14 4
"
Sig. Significant at 5% level
N. S. not significant.

TABLE IV SELECTION OF PANEL MEMBERS USING TRIANGULAR METHOD FOR CANNED PRAWNS

Panel members No. of times odd sampl"s No. of times good

were detected correctly samples were detected

1 5 5
2 5 6
3 5 5
4 5 5
5 6 6
6 6 6

TABLF V SELECTION OF PANEL MEMBRS USING TRIANGULAR METHOD FOR FROZEN PRAWNS

Panel members No. of time odd samples No. of times good sumples
werE'! detected correctly were detected

1 5 5
2 5 5
3 5 5
4 6 6
5 5 5
6 6 6

108 FISHERY TECHNOLOGY