Sensory Evaluation Notes 2010

SENSORY EVALUATION

Reference Notes

2010

Prepared by

Stephen Nottingham

Agri-Science Queensland Department of Employment, Economic

Development and Innovation

Sensory Evaluation: reference notes 2010 DEEDI and The University of Queensland

2

ACKNOWLEDGEMENTS I would like to thank Michael O’Mahony for his permission to include copies of the statistical tables from his book “Sensory Evaluation of Food: Statistical Methods and Procedures” and material supplied by Innovative Food Technologies, DEEDI, Brisbane and The University of Queensland. Copyright The Department of Employment, Economic Development and Innovation (DEEDI) seeks to maximise the economic potential of Queensland’s primary industries on a sustainable basis. © The State of Queensland, Employment, Economic Development and Innovation, 2009. Except as permitted by the Copyright Act 1968, no part of the work may in any form or by any electronic, mechanical, photocopying, recording, or any other means be reproduced, stored in a retrieval system or be broadcast or transmitted without the prior written permission of DEEDI. The information contained herein is subject to change without notice. The copyright owner shall not be liable for technical or other errors or omissions contained herein. The reader/user accepts all risks and responsibility for losses, damages, costs and other consequences resulting directly or indirectly from using this information. Enquiries about reproduction should be directed to [email protected] or telephone +61 7 3225 1398.


3

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ....................................................................................................2

INTRODUCTION....................................................................................................................5

THE HUMAN SENSES IN SENSORY EVALUATION .....................................................7

THE SENSES - AN INTRODUCTION .................................................................................7

SENSE OF SIGHT- .................................................................................................................8

THE SENSE OF SMELL......................................................................................................12

CHEMICAL / TRIGEMINAL FACTORS .........................................................................14

THE SENSE OF TASTE.......................................................................................................15

THE SENSE OF HEARING.................................................................................................20

THE SENSE OF TOUCH .....................................................................................................20

SENSORY INTERACTION.................................................................................................21

OPERATIONAL PRINCIPLES OF SENSORY TESTING .............................................24

DESIGN OF A SENSORY TESTING AREA.....................................................................25

THE FOOD.............................................................................................................................29

THE PEOPLE ........................................................................................................................31

TESTING METHODS ..........................................................................................................36

SENSORY EVALUATION METHODS .............................................................................37

ANALYTICAL SENSORY TESTS: ....................................................................................37

DIFFERENCE TESTING.....................................................................................................37

SIMPLE DIFFERENCE TEST............................................................................................38 TRIANGLE TEST.............................................................................................................38 DUO-TRIO TEST..............................................................................................................41 TWO-OUT-OF-FIVE TEST.............................................................................................44 “A” – “NOT A” TEST.......................................................................................................44 DIFFERENCE-FROM-CONTROL TEST (DFC) .........................................................45

DIRECTIONAL DIFFERENCE TESTS ............................................................................48 PAIRED COMPARISON TEST ......................................................................................48 RANKING TEST...............................................................................................................50 RATING TEST ..................................................................................................................53


4

SIMILARITY TESTING ......................................................................................................56

AFFECTIVE TESTS.............................................................................................................57

SPECIFIC TEST METHODS ..............................................................................................57 PAIRED PREFERENCE TEST.......................................................................................57 RANKING FOR PREFERENCE.....................................................................................60 RATING FOR PREFERENCE........................................................................................63

SENSORY EVALUATION IN CONSUMER TESTING..................................................66

DESCRIPTIVE TESTING ...................................................................................................73

SELECTION, TRAINING AND MOTIVATION OF A PANEL.....................................78

REPORTING .........................................................................................................................83

SELECTED BIBLIOGRAPHY............................................................................................84

JOURNALS............................................................................................................................86

STATISTICAL TABLES......................................................................................................87


5

INTRODUCTION Sensory evaluation - A scientific discipline used to evoke, measure, analyse and interpret reactions to those characteristics of foods and materials as they are perceived by the senses of sight, smell, taste, touch and hearing. Sensory evaluation was one of the earliest methods of quality control and it is still widely used in industry. However, the level of application depends on the situation (e.g. beer and wine tasting to operators sampling of products from production line). Four variables affect sensory evaluation:

The Food The People The Testing Environment Methods

Sensory evaluation terminology

Sensory evaluation Sensory Analysis Organoleptic Analysis Taste Testing Taste Panels Psychophysics Subjective Evaluation

Advantages

Gives real answer regarding consumer quality Relatively cheap process (depending on how it is done) Rapid Many applications Simple to perform Sensitive Objective methods are more reliable, accurate and reproducible. However, they must

be correlated to sensory evaluation to indicate a consumer response. Disadvantages

Time consuming and slow Complex Expensive to run Method selection Analysis Interpretation

Industry applications of sensory evaluation

Product development Product matching


6

Product improvement

Process change Cost reduction New raw materials selection Quality control Storage stability Product grading / rating Consumer acceptance Consumer preference Panel selection / training Correlation subjective / objective Research projects

o Effect of treatments eg macadamia and pecan maturity studies o Varietal assessments o Genomics relating the genetic marker to the flavour trait

Sensory Standards

Australian Standard

Year Title

AS 2542.1.1 2005 Sensory analysis of foods - General guide to methodology –Types of tests

AS 2542.1.3 1995 Sensory analysis of foods - General guide to methodology -Selection of assessors

AS 2542.2.1 2007 Sensory analysis of foods - Specific methods - Paired comparison test

AS 2542.2.2 2005 Sensory analysis of foods - Specific methods - Triangle test

AS 2542.2.3 2007 Sensory analysis of foods - Specific methods - Rating

AS 2542.2.4 2005 Sensory analysis of foods - Specific methods - Duo-trio test

AS 2542.2.5 1991 Sensory analysis of foods - Specific methods - 'A not A' test

AS 2542.2.6 1995 Sensory analysis of foods - Specific methods - Ranking

AS 2542.3 1989 Sensory analysis of foods - Glossary of terms

AS 2609.1 2005 Methods for assessing modifications to the flavour of foodstuffs due to packaging – sensory analysis

There are other standards (eg British Standards Institute, Standardization Administration of China etc) that may also be applicable, particularly those from ISO (International Organization for Standardization) and ASTM. (American Society for Testing and Materials) e.g. ISO 8589 – 2007 Sensory analysis -- General guidance for the design of test rooms

THE HUMAN SENSES IN SENSORY EVALUATION THE SENSES - AN INTRODUCTION The sensory properties of foods are related to three major attributes:

Appearance - colour, size, shape; Flavour - odour, taste; and Texture - mouth feel, viscosity and hearing.

These attributes are expressed as a continuum and not as finite properties. It is impossible to rate each one individually unless special precautions are taken, e.g. blindfolds, nose clips, coloured lights, purees. Humans possess about 30 different senses. However, the sensory properties of foods are perceived through the senses of:

Sight; Smell; Taste; Touch; Hearing and Trigeminal response

To study how the senses work it is useful to have a model (see below) of the process to show the steps in the chain of sensory perception. These include the stimulus which hits the receptor within the sense organ and causes a sensation that is converted into a nerve signal and transmitted to the brain. In the brain this is interpreted, based on previous experiences in the memory and integrated into a perception. Finally a response is formulated based on the subject’s perceptions.

Stimulus

Receptor

Sensation

Perception

Response Stimuli A stimulus is any chemical or physical activator that causes a response in a receptor, e.g. retina of the eye is a receptor for light, ear drum is a receptor for sound. An effective stimulus produces a sensation, the dimensions of which are:

Intensity/strength; Extent/separation; Duration/retention; and Hedonics/like-dislike.


7


8

Receptors Receptors are the stimuli detecting cells of the sense organ, e.g. taste buds on tongue, light receptors in retina of eye. Perception Perception is the psychological interpretation of sensations determined by comparison with past experiences, e.g. the sour taste of lemons is the perception of the sensation received by the receptors (taste buds) from a chemical stimulus (citric acid). Response The response is based on the subject’s perceptions and often humans can give varying responses to the same stimulus. This can be due to a difference in the sensation because the sense organs differ in sensitivity or a difference in the perception through lack of knowledge of the taste or odour or lack of training in expressing the perception. Thresholds For some of our senses (particularly true for our sense of smell and taste) there are thresholds based on each person’s sensitivity. Individuals can have an absolute or detection threshold and a recognition threshold. These are defined more in later sections of the notes. SENSE OF SIGHT- The appearance of food Stimuli = visible light Receptor= retina of the eye Perception=sight, vision, appearance The appearance of foods is a major factor governing its acceptability and can be subdivided into three main categories:

Optical properties- colour, gloss and translucency Physical form-shape and size Mode of presentation-lighting packaging etc

Optical properties Vision Vision is a complex phenomena consisting of several basic components. A stimulus, light, from an external source interacts with the object and is brought to focus on the retina of the eye. The retina is the receptor of vision and contains two types of cells. The rods are responsible for vision in dim light and the cones are responsible for colour vision. Light incident on these cells causes a photochemical reaction that generates an electrical impulse which is transmitted to the brain via the optic nerve. Colour blindness is caused by loss or lack of colour receptor cells in the cones. Approximately 8% of the population have some defect with relation to colour; mostly males. For females this is between 0.4% and 1% and in general Asians and Africans have a lower percentage, around 5%. Most common is red-green and less common is blue-yellow.

Light Visible light is that part of the electromagnetic spectrum which radiates between wavelengths of 380 - 770 nm. Different wavelengths produce different colours

380 - 450 nm = violet 450 - 475 nm = blue 500 - 575 nm = green 575 - 590 nm = yellow 590 - 770 nm = red

[NOTE: All electromagnetic radiations are physically the same. However, the optical system of the eye is such that only the visible range of wavelengths is absorbed by the lens.] Light sources Incandescent lights consist of a tungsten filament which is heated in an inert gas. The higher the temperature, the more light produced. Light from this source tends to be harsh and tends


9


10

to highlight the red end of the spectrum. Fluorescent lights operate by electrical excitation of atoms that produces spectral lines at specific wavelengths which then impinge onto fluorescent materials which convert the incident light into light at a longer wavelength. Light produced is softer but can produce colour distortion at particular wavelengths. Natural light is too variable for use in evaluating appearance of foods. The appearance of food is a major factor governing its acceptance by consumers. It can be subdivided into three main categories.

Optical properties – light - object interactions Physical form Mode of presentation

Optical properties - Light - Object interactions Light incident on an object may be:

Absorbed; Reflected; Transmitted; and Refracted.

The relationship between and within each of these components is responsible for the colour and gloss characteristics of the food. The main light/object interactions produced are:

Lightness/brightness value; Colour/hue; Chroma/purity Gloss Translucency

Physical form The second class of product appearance is physical form that can be subdivided into three parts:

Shape; Surface texture; and Visual consistency.

Shape and size are important from a food technologist's point of view because these can be altered during the manufacture of processed products. Some examples include:

Sliced, diced, pieces whole Length of frozen French fries Cut of beans Extrusions

Surface texture can indicate product texture. Some examples include:


11

Open dry structure of meat Wrinkling of peas Wilting of lettuce

Visual consistency can indicate product viscosity as in:

Setting of a jelly Syrups of different concentrations Pastes and purees

Mode of presentation This aspect should be considered from a marketing point of view and is important because it influences sales. Mode of presentation is applicable on the supermarket shelf (at retail level) and also in terms of presentation at the table (home and restaurant). Factors to be considered are:

Product description - name, price, ingredient, etc; Packaging - shape, design, colour; Contrast - phenomena of adjacent colours; and Illumination - affects apparent product colour.

Summary Appearance is an important aspect of food quality as it is the first subjective evaluation made of food quality. The product has to pass the visual assessment before the consumer can or will consider the other parameters such as taste and texture. Factors that should be considered in evaluating product appearance include:

use of standard conditions: light source (type, intensity, colour); background; and style of presentation (unless tested). selection of appearance attribute(s) for inclusion on scoresheet; using appearance to reduce tasting load; should be masked to eliminate unwanted interactions when assessing parameters

involving other senses; and colour charts/standards help rating.

THE SENSE OF SMELL (Odour/olfaction) Stimuli = volatile chemicals Receptors = olfactory cells in the nose Perception = smell, odour, aroma, flavour Smell is one of our most primate senses. Supposedly prehistoric people were more influenced by smell than other senses. The human nose is capable of detecting thousands of different odour substances. However, our sensitivity is much less than other animals. (Animals use smell - food, mating, territory etc). Smell is detected both before and during eating. Smell is an important aspect of flavour. There are 20x106 olfactory receptors, but only about 1000 taste receptors. Odour description requires the development of an odour/flavour memory, e.g. fishy, flowery, woody. This is the basis of flavour/odour memory development by wine judges and milk/cheese graders. Individuals vary a great deal in their sensitivity to different odours/aromas. Those with no sense of smell (odour blindness) suffer from anosmia but this is quite rare. Anatomy of olfactory system

Image from http://news.softpedia.com/news/How-Fine-is-Human-Olfaction-2.jpg


12


13

From the diagram it can be seen that most of air misses the olfactory area. Only 5-10% of inspired air passes over olfactory receptors. However, this amount can be increased by sniffing harder; obviously the more air which passes over the receptors the better the response. The large number of olfactory receptors (20x106) enable detection of :

More odours than tastes; A greater variety of odours; and Odours at much lower concentration (10 molecules/mL).

The human nose is 10, 000 times more sensitive than our sense of taste and it is rarely neutral as it is linked to our memories and emotions so people either like or dislike an odour. In order for odour to register:

Substance needs to be volatile enough to get into air in the sensory region. Substance needs to be partially soluble in mucus covering of receptors. Minimum number of odorous molecules needs to be present. Need to be in contact with receptors for minimum time.

Olfactory intensity Human nose is about 10-100 times more sensitive to odours than any physico-chemical analysis (e.g. gas chromatography). It has been demonstrated that human nose is capable of detecting ethyl mercaptan at a concentration of 0.01 mg/230m3 of air, which is equivalent to about 8 molecules/receptor. Olfactory threshold Detection threshold is the concentration where smell is detected. Recognition threshold is the concentration where the smell is recognised. Olfactory interactions Nature of the response may change with concentration (e.g. perfumes at low concentration are pleasant but at strong concentration may be unpleasant). Interaction of odours:

Additive - increase intensity; Suppressive - decrease intensity; and Blending - when new odour unrelated to originals. Usually can only discriminate 4 odours in one mix even for experts although they are

generally more reliable than other assessors Olfactory adaptation Initial sensation maybe strong - but weakens and makes identification difficult; this is due to adaptation of olfactory receptors. In testing we therefore need to allow for this by:

Taking first impression of odour and/or Waiting between tests to allow receptors to recover. You can smell your wrist in

between samples

Summary

Smell is a major component of food flavour. Human nose is much more sensitive than analytical instruments. Foods contain numerous compounds of varying volatility that can make analytical

interpretation difficult (e.g. strong peaks may produce weak odour whereas weak peaks may produce a strong odour).

Smell measures perception of a mixture; analytical testing does not. CHEMICAL / TRIGEMINAL FACTORS In the nose and mouth, there is a general chemical response that is mediated by the trigeminal nerves (also known as fifth cranial nerve). A number of everyday experiences are due to trigeminal stimulation. Ginger, chilli peppers, horseradish, salt and menthol, among others, stimulate trigeminal nerve ends resulting in a number of sensations such as heating, pungency and cooling. The trigeminal tracts have an impressive size compared to other chemical sense nerves. The papillae at the front of the tongue contain three times as many trigeminal nerves than taste nerves. It is an important chemical sense that is often overlooked. The response to mild trigeminal irritants, such as a high concentration of salt in snack foods resulting in mouthburn, may contribute to product acceptance.

Image from www.medical-look.com/human_anatomy/organs/trigeminal_nerve.gif


14


15

THE SENSE OF TASTE (Gustation) Stimuli = soluble chemicals or chemicals which are solubilised during chewing Receptors= taste buds in mouth Perception=taste, flavour What is commonly referred to as taste/flavour is actually a combination of:

Taste; Smell; Touch sensation and Temperature.

Strictly speaking taste involves only those sensations mediated by the Gustatory Nerve Fibres and these sensations have five (5) basic qualities:

Salt; Sweet; Sour; Bitter and Umami

Taste stimuli Taste response requires an aqueous solution of the substance (stimulus) to contact the taste buds. Therefore, saliva secretions are important in terms of ensuring contact between the product and the taste buds. Saliva production is generally stimulated by chewing, as well as the appearance and odour of the food. The tongue is important as it brings the food into contact with the taste buds and also provides a mixing action which enables an even distribution of food about the taste buds as well as preventing the development of concentration gradients.

Taste receptors The receptors for taste are the taste buds and these are mounted on papillae (folds in the skin of the tongue). The area of greatest response is the top of the tongue. Other areas in the mouth and throat where taste buds are situated include: palate, pharynx, larynx, tonsils, epiglottis, lips, cheeks, underside of tongue and floor of mouth.

Source from Meilgaard, M., Civille, G.V. and Carr, B.T. (1987) "Sensory Evaluation Techniques."

Taste buds are mainly located at the tip, sides and rear of tongue. There is very little response in the centre of the tongue. Any one of the five classical taste qualities (sweet, salt, sour, bitter and umami) can be perceived on any area of the tongue, so the old fashioned map of the tongue with different tastes in different areas is not necessarily accurate. Taste cells constantly degenerate and regenerate. Their life cycle is 10 days and they are easily destroyed by heat. The five basic tastes A basic taste is one for which specific taste buds have been identified as being physiologically responsible for the particular taste sensation. There is talk of a sixth basic taste – fat, but as yet there is not enough evidence for it to be recognised as a basic taste.


16


17

Sourness This is the simplest taste as only acids (H+) produce sourness and as the (H+) increases the sourness increases However there are some anomalies to this:

organic acids are more acidic than expected. sourness of aliphatic organic acids relates to chain length. some amino acids are sweet (aspartame) picric acid is bitter sugar may enhance/depress sourness sourness is also affected by pH and acid presence of buffers affects sourness

Sweetness The common substances that produce the sweet taste are the sugars and other hydroxy compounds such as alcohols and glycols. Other substances such as lead salts, amino acids, proteins, non-nutritive sweeteners (cyclamates, saccharin and aspartame) also taste sweet. Most of the artificial sweeteners were discovered by accident. Saccharin was discovered in 1879 when a John Hopkins worker inadvertently licked his fingers and is only sweet to humans. A graduate student at the University of Illinois in 1937 was smoking a cigarette that had come in contact with Cyclamate and noticed a sweet taste. Saltiness Many crystalline water-soluble salts yield a salty taste, but only sodium chloride gives a pure salty taste. Other substances taste salty but also bitter, alkaline, sweet and salt in various combinations. Bitterness Many chemically different compounds have a bitter taste. However, bitterness is mainly associated with alkaloids such as caffeine, quinine, strychnine and nicotine. Originally it was thought that bitterness was an indication of danger (poison). However, many alkaloids are used as drugs (e.g. codeine) and many other bitter substances are harmless (glycosides, esters and aldehydes and tannins in wines and tea). Bitterness is generally perceived at very low concentration and a relationship appears to exist between sweet and bitter as many sweet substances produce a bitter aftertaste (saccharin). Bitterness is the taste which most people have difficulty in detecting and response level varies greatly from individual to individual. Umami Umami is the taste that has been shown to be associated with substances that contain glutamate. The most notable example is mono-sodium glutamate (MSG). MSG is well known as a flavour enhancer and can cause adverse reactions in some sensitive individuals. However, there are many other compounds which contain glutamate and which are capable of producing the savoury, spicy, brothy taste associated with MSG. Many foods contain naturally high levels of glutamate.


18

Taste thresholds and sensitivity Absolute/Detection threshold - Concentration of stimulus at which a subject can detect a difference between two samples in a paired test. Recognition threshold - Concentration at which the specific taste can be identified. Recognition threshold is generally higher than detection threshold. Both absolute threshold and recognition threshold will vary between individuals. Most people can detect taste within 0.2 - 0.6 seconds and therefore if there is no response within this time the level is sub-threshold. However, recognition times vary between the basic tastes

Salt = 0.3s Sweet = 0.4s Sour = 0.5s Bitter = 1.0s Vision = 0.02s Hearing = 0.01s Touch = 0.005s

Reaction times also relate to retention times for example; bitterness has the longest reaction time (1.0s) and the sensation lingers considerably after tasting. There is great variability between individuals in their levels of sensitivity. Sensitivity is affected by:

Temperature; Sleep; Hunger; Age Gender Flavour enhancers Flavour suppressors Taste interactions Age

Adaptation and fatigue During exposure to a stimulus, sensitivity decreases due to adaptation and fatigue. This loss in sensitivity varies considerably with the taste (sweet, sour, salty or bitter) and also with the compound. For example, tasting a series of acids causes the sensitivity to be reduced by the preceding acids. However, recovery is usually rapid because most common organic acids are very soluble. Taste interactions Having described the 5 basic tastes it is obvious that foods are a very complex system which contains many different taste compounds and therefore many different tastes. The fact that there are only 5 basic tastes and yet we are able to detect hundreds of different taste sensations is due to a series of complex taste interactions that can range from simple 2 way interactions to complex 5 way interactions


19

Interactions between the 4 basic tastes were previously described simplistically by the taste tetrahedron proposed by Hans Henning in 1916. Supertasters Some people have more that the normal number of taste papillae (and taste buds) and have an increased density of fungiform papillae and are extremely sensitive to hot spicy food. About 25% of the population are supertasters, while 50% are classed as normal and 25% as non-tasters. More females than men are supertasters. More thin people are supertasters. Food also tends to taste sweeter to supertasters. Tests are available to determine which class of taster you belong to. Summary

Five types of taste receptors - salt, sweet, sour, bitter and umami. All areas of the tongue where there are taste bud containing papillae respond to

different sensations. Substances must be dissolved for taste buds to detect them. Flavour of the food is a complex interaction of different tastes and odours. Sensitivity to taste varies between individuals and is affected by their physiological

state.


20

THE SENSE OF HEARING (Audition) Stimuli = physical movement of sound waves in a medium (air) Receptor= ear drum Perception=sound, hearing Hearing Sound is the perception by humans of vibrations in a physical medium (air). The sound of food when it is being eaten is an important aspect in determining quality. Positive aspects:

Snap, crackle and pop; Fizz of champagne or beer; Crispiness of lettuce or celery; and Tapping a melon for quality.

Negative aspects:

Noisy environment may distract tasters or mask product sounds. Noisy panellists can also distract other tasters

THE SENSE OF TOUCH (Texture, Kinesthetics) Stimuli = physical contact between the food and body tissue Receptors = muscles and nerves in mouth and fingers Perception = touch, feel, texture, viscosity Texture usually relates to solid food while viscosity relates to homogeneous liquid foods and consistency relates to non-homogeneous liquid foods. Instrumental methods only measure one aspect of "texture" and again cannot relate the complex interactions which produce the perception of food texture. Food texture can be divided into finger feel and mouthfeel. Finger feel Firmness/Softness indicates the eating quality of some food products:

Ripeness level of fruit such as avocado and mango; Crumb texture of bread; Firmness of cheese; and Spreadability of butter or spread.

Juiciness can be used as a subjective quality index (eg the “thumbnail” test for corn). If a corn ear is pierced and a clear liquid comes out then it is immature, thick and milky then over mature and if it is thin and milky then it is ready to harvest.


21

Mouth feel Liquids

Viscosity - thin to viscous, e.g. milk, cream. Consistency - thin to thick, e.g. fruit yoghurts.

Solids Classification of textural characteristics - assessed mainly by chewing. Textural Terminology Mechanical Characteristics Primary and secondary parameters (Common terms) Hardness Soft, firm, hard, e.g. fruit ripeness, cheese maturity. Brittleness Crumbly, crunchy, brittle, e.g. muesli bars and biscuits Chewiness Tender, chewy, tough, e.g. meat.

Geometrical Characteristics Grittiness Gritty, grainy, coarse, e.g. stone cells in fruit, "sand" in

ice-cream. Fibrousness Fibrous, cellular, e.g. string/fibre in vegetables. Other Characteristics

Moistness Dry, moist, wet, e.g. cracker biscuit, cheeses, water melon.

Oiliness/Greasiness Oily, greasy, fatty, e.g. french fries, chips. SENSORY INTERACTION As has been indicated previously when eating or tasting food there is a continuous relationship between the senses and unless steps are taken to separate the individual senses or stimuli, interactions may occur. It is not known whether interactions occur at the receptor site or the brain. However, the second option would appear to be more likely. Interaction between senses This is the ability of a response from one modality to influence or affect the response from another. There are two aspects of this:

1. Positive - interactions giving clues to possible identity, e.g. pink milkshake being strawberry flavoured.

2. Negative - If clues are not correct this may lead to confusion and a wrong judgement,

e.g. pink milkshake with pineapple flavour. Types of sensory interactions


22

Taste - odour Receptors for these two senses are very close so that interactions between these senses are highly likely and these may be important in classifying a particular taste. Taste - tactile The taste threshold for sugar, salt, caffeine has all been shown to be lower in water than in tomato sauce. This may be due to the fact that in more viscous solutions the chemicals do not react with the receptors as easily as in pure solutions. Taste - sight This is a very important aspect because vision is the first sense affected and appearance of a product will have a major influence on absolute quality. Bright colours indicate strong flavours whereas dull colours indicate mild flavours. Other interactions include:

Odour - Sight Odour - Tactile Taste - Hearing Odour - Hearing

Multiple interactions Multiple interactions between more than two modalities are also possible. Example: Tasting food pureed, blindfolded and with nose clips gives a different response than when interactions are allowed. Interactions between stimuli These interactions are more difficult to define and measure but are just as important as interactions between the senses. Some examples include:

suppression of one flavour by another, e.g. sweetness is suppressed by acidity. This is the basis of ensuring brix/acid ratio for fruit juices are constant;

neutralisation of one flavour by another;

blending to produce a totally different flavour, e.g. garlic flavoured cheeses;

partial blending producing a new flavour and the original flavours;

no effect; original flavours are distinct and separate, e.g. fruit in cheese;

intensification resulting in enhancement of flavours, e.g. salt and MSG on food

improves the natural flavours. Similar situations may exist for all other stimuli. Summary


23

Interaction must be considered when designing sensory panels. If only one sense or stimulus is to be evaluated then all others must be masked. However, if interactions are required then ensure this can be achieved by means of sample preparation.


24

OPERATIONAL PRINCIPLES OF SENSORY TESTING When evaluating properties of foods using people as measuring instruments it is important to control the methods and conditions of testing as rigidly as possibly. This helps to eliminate the numerous errors or biases that can be caused by psychological and physiological factors. The mental attitude and physical condition of a taster and the atmosphere of the testing environment all influence their judgements. There are therefore a number of basic rules which should always be applied, as stringently as circumstances allow, when running taste panels (see also Australian Standard 2542.1.1-2005 Sensory Analysis – General guide to methodology – types of test). These relate to:

Preparing the testing environment; Preparing and serving samples; Selection of panellists; Designing the experiment and testing methods

General principles that should always be followed are: Never ask anyone to taste food they do not like; Make sure that the "correct" panellists are selected (see section on panel selection and training) and that they know in advance when they will be required. Keep a strict control over all variables except those being tested (e.g. sample size and temperature). Make sure the environment gives optimum opportunity for concentration. Tasting properly is a difficult job. Train panellists to be silent while tasting. This prevents panellists from influencing one another. Make tasting interesting and desirable. Use rewards to motivate taters, vary these and choose foods that contrast with those being tasted. Motivated tasters are more efficient. Give feedback on results whenever possible. Avoid giving any unnecessary information to panellists that may influence their scores. Tasters usually find what they expect to find; e.g. in a storage test they expect to find samples deteriorating. Plan your experiment in advance. Which will be the best test to use? Consider all aspects including how you will get the information required from your results (statistics). Run preliminary tests, i.e. practise and choose the best method for:

Sample size - adequate but not excessive; Serving temperature - standardise for all samples. It must be maintainable, and be an

acceptable temperature for the food; Serving vessels; Eating utensils.


25

DESIGN OF A SENSORY TESTING AREA The main considerations to keep in mind when preparing an area for sensory testing concern the requirements for an atmosphere conducive to concentration, where conditions can be controlled. Sensory panellists need somewhere comfortable and free from distractions if they are to be able to "tune in" to the sensations triggered by the stimuli in the food products they taste. Product characteristics can be markedly affected by temperature and humidity, and appearance is affected by lighting intensity. The conditions should be controlled in order to:

Reduce bias Improve accuracy Improve sensitivity (compare to the conditions used in an analytical laboratory)

International standard (ISO 8589-2007 Sensory analysis -- General guidance for the design of test rooms) The standard looks at the design of the testing area for both new and existing buildings. It also specifies which recommendations are considered essential and which are only desirable. Important points summarised from the standard are listed below. If designing an area that is to be dedicated solely to taste panel work, these should be seriously considered. Total area should include:

Testing area with individual booths and a group area; Preparation area/kitchen; Office; Cloakroom; Rest room; and Toilets.

General testing area

Easily accessible but in quiet position.

Location - close proximity to preparation area, but separate entrance, and with complete "close-off" capability.

Temperature and relative humidity - constant, controllable, and comfortable.

Noise - keep to a minimum, soundproof area as much as possible.

Odours - keep area free from odours (air conditioner with carbon filters, slight

positive pressure).

Use odourless materials in construction and decoration.

Use odourless cleaning agents.

Decoration - use neutral, light colours for walls and furniture (e.g. off-white, light grey).

Lighting - ambient lighting must be uniform, shadow-free and controllable. For consumer testing - as close to home conditions as possible.

Booths Number - minimum three, normally five to ten - six is a useful number since it fits in well with balanced ordering of 3 samples. Space - allow sufficient space for movement of tasters and for serving samples. Set-up - permanent booths recommended. Temporary is acceptable. If adjacent to preparation area include openings in the wall to pass samples through. Size and style specified. Consider space for samples, utensils, spittoons, rinsing agents, scoresheets and pens, computerised equipment. Include comfortable seats. Lighting - uniform, shadow-free, controllable, adequate intensity for assessing appearance. Devices to mask appearance (e.g. dimmers, coloured lights or filters).

Image from DEEDI booths at Hamilton Qld Group work area General Necessary for discussion and training purposes. Include large table and several chairs. "Lazy Susan" useful. Include board for discussion notes, etc. Lighting As for general area, with coloured lighting options like booths.


26

Image from DEEDI meeting room at Hamilton Qld Preparation area General Located close to assessment areas but no access to tasters. Design for efficient work-flow. Well ventilated. Flexible services (i.e. plumbing, gas, electricity). Equipment Depending on testing required. Include working surfaces, sinks, cooking equipment, refrigerator, freezer, dishwasher, etc. Storage space for crockery etc. Crockery, glassware etc for serving samples.

Image from DEEDI sensory kitchen at Hamilton Qld Office area General Separate but close to testing area, reasonable size, desk, filing cabinet,


27


28

computer, bookcase. Photocopying service needed. Additional areas Useful to include rest room, cloak room and toilets. Practical alternatives The requirements specified in the International Standard (ISO 8589) will obviously provide a suitable area, but they are not always feasible, either from the point of view of financial resources or physical space available. Very few industries are able to start from scratch, designing new premises solely dedicated to sensory analysis work. I therefore would like to abbreviate the list proposed in the standard to one which I consider includes the bare essentials. Minimum of 2 areas: Preparation area and office area. If possible position these at opposite ends of the room to avoid messy paperwork! Testing area with entrance separate from preparation area. Preparation area requires

Adequate storage for utensils and equipment; Adequate working surfaces to set out samples; Washing up facilities - minimum double sink with hot and cold running water; Refrigerator - minimum 2 door with separate freezer, preferably at least auto-defrost; Cooking equipment - depending on sample requirements; Rubbish bin - large with liner bags; Source of boiling water; Hand washing facilities.

Testing area requires

Comfortable chairs for panellists; Minimum space - 4 panellists; Table which can be easily divided into booths if required; All equipment likely to be needed while a panellist is tasting, e.g. pencils, spittoons,

toothpicks, tissues/serviettes; Well placed, efficient lighting; Waiting area with noticeboard - for tasters to wait for booths to become free and to

collect rewards after tasting. A system using collapsible booths can work quite well if it is not possible to keep an area solely for sensory work. These may be made of painted wood, heavy duty cardboard, or "corflute". They can be made specifically to fit any available benches or tables and folded and stored when not in use. The type of facility will depend on:


29

Finance Available space Frequency of use Tests conducted

THE FOOD Sample preparation and serving Serve tasters promptly and make sure they have everything they need. Run a taste panel as you would expect a good restaurant to be run, i.e. Give courteous friendly service, be efficient, and serve good food. Keep accurate records of any cooking or preparation methods used. Record temperatures and size of samples served and any special conditions (e.g. coloured lighting). It is important that panellists do not see the samples being prepared as this may indicate quality difference. Sample preparation should be uniform:

Temperature Cooking Thawing Size and shape (provided this is not a variable)

Sample should be randomly allocated to:

Avoid bias Overcome any non—uniformity

Sample size should be adequate:

30g solids 30mL liquids

Samples should be served immediately after preparation to reduce:

Flavour loss Discoloration Textural changes

Sufficient samples should be prepared to allow for seconds.


30

Containers for presentation Containers for presentation and tasting should be:

Odourless and tasteless Clean Identical for all samples and sessions Disposable containers or re-usable Coloured to mask product appearance (if required) Relevant to product

Serving temperature

Serve at room temperature where possible Preference tests use normal temperature Difference tests may alter temperature to accentuate flavours/odours Do not overheat:

too hot to taste drying out off flavours browning

Dilutions and Carriers Most foods should be served in the way they are normally eaten. However, some products such as spices, chillies, alcohol, onions, etc. may require dilution before testing. If dilutions are used they must be uniform in terms of diluent and concentration. Carriers are substances that are added to assist tasting of certain products. Carriers are a problem because they can be:

Expensive Time consuming Variable quality Difficult to control product/carriers ratio uniformity.

For example: developing a cake icing individually may not allow for interaction with flavour or it may be incompatible with the cake (affects texture or falls off). Number of samples/sessions The number of samples presented at any testing session will depend on:

Type of product - strong flavour —> less samples Type of test Rating scale may require fewer samples Test may dictate sample number

o eg: triangle test = 3 samples, paired comparison = 2 samples Type of panel — trained / experienced -> more samples Experimental design


31

As a general rule usually not more than 6 samples/sessions. Sessions / Trials Before starting your scheduled tasting sessions run two preliminary sessions. These will familiarise your panel with the scoresheet, the products to be tested and the procedures you wish them to follow. It also gives you practice at preparing and serving the quantity of samples needed, and a last chance to iron out any unforeseen problems. In calculating the number of sessions consider the following:

Total number of samples for tasting Statistical design Taster fatigue Motivation Type of panel (trained/untrained)

THE PEOPLE

Tasters should like the food so never force someone to consume something they don’t like.

‘Correct’ panellists should be selected. Trained for descriptive panels and consumers for consumer studies.

Keep panellists motivated by using a number of techniques. Panels should be run professionally Use “healthy” panellists. No colds, flus etc

Psychological factors Because sensory evaluation is a subjective system, it is necessary to allow for any psychological factors that may influence results and possibly lead to errors. Motivation Good results can only be obtained from a co-operative, responsive panel. Tasting becomes a chore when there are large numbers of samples/sessions involved. Motivating panellists can reduce this problem by:

Stressing importance of work Stimulating company expansion Greater profits More pay Ensuring panellists know what is involved with the trial i.e.: sessions, products, when

and where tasting will be conducted Having adequate facilities Using effective methods and designs Publicising results obtained from work Rewarding panellists


32

Sample Coding Remove possible bias or influence from samples codes. Do not use.

Single digit numbers Consecutive letters Same codes at consecutive sessions

Randomly or statistically generated three digit number codes are best. Order of Presentation Always use either a random order of presentation or a statistically balanced design to avoid:

Donkey vote (first is best; last is worst) Position bias - in triangle tests middle one is different Contrast effect — good after bad appears better, or bad after good appears worse.

Devise your own system for remembering orders, e.g. 3 digit numbers - put in sequence of one of digits. Keep it a secret! Always work systematically in coding, labelling, setting up, e.g. as in reading a page (1) Left to Right (2) Top to Bottom This provides an automatic check if something goes wrong. Balance presentation of samples whenever possible. This avoids contrast effect. i.e.. 2 samples A, B. - Half panel taste A first, other half taste B first.

- Half panel receive A on the left, other half receive B on the left.

3 samples 6 different orders in which they are tasted. Use

every order the same number of times. Number of tasters is a multiple of six.

- Position of samples on plate must also be balanced.

4 samples 24 different orders: use them all if possible (see

table on next page).

4 samples Generate random order. Write out set of cards and shuffle them.

When you cannot use balance to eliminate bias, use randomisation.


33

Four sample balanced orders

1 A B C D 2 A B D C 3 A C B D 4 A C D B 5 A D B C 6 A D C B 7 B A C D 8 B A D C 9 B C A D 10 B C D A 11 B D A C 12 B D C A 13 C A B D 14 C A D B 15 C B A D 16 C B D A 17 C D A B 18 C D B A 19 D A B C 20 D A C B 21 D B A C 22 D B C A 23 D C A B 24 D C B A

Expectation Error Any information a panellist receives before a test will influence the results. This is called expectation error. To overcome this:

Do not give detailed information about treatments Do not use people on panel who know what the treatments are Sample coding and design can prevent expectation error

Logical / Stimulus Error Tasters look for clues to get the “right” answer eg: a difference in sweetness may be associated with sample differences such as size, shape and colour. This error can be overcome by ensuring sample preparation is uniform or use masking. Halo Effect When more than one factor in a sample is evaluated at one time the result obtained may be different than if factors evaluated separately. This can be overcome by tasting each aspect separately. However, this is time consuming and would only be done if extremely accurate results were required. Testing one aspect at a time in preference does not simulate the “real situation” i.e.: consumers do not taste every aspect separately.


34

Mutual Suggestion Influence of other panellist may bias or influence results. This can be prevented by:

Using booths Not allowing talking in tasting area Reducing outside distractions

Contrast Effect

Good before bad – worse Bad before good – better

Tasting design can help overcome this. Phsiological factors in taste testing Time of Tests

Monday and Friday are recognised as being bad days for tasting Normally taste 1 hour before meals and 1 - 2 hours after Sometimes this becomes difficult in practice due to:

o Unavailability of tasters o Number of sessions

Smoking / Taste Affecting Substances As indicated earlier, smoking affects sensitivity to flavours —therefore should either:

Not use smokers Ensure they do not smoke for at least 1 - 2 hours before tasting Chewing gum, mints and spices etc may also influence taste

Illness Sensitivity of people suffering from illness is reduced -particularly those with colds or flu (physical and psychological) Likes / Dislikes In preference testing a series of treatments within a specific product type, it is legitimate to eliminate people who dislike the product (or those who are not discriminatory). Palate Clearing It is a good idea to get panellists to cleanse their palate:

Before tasting to remove any lingering tastes Between samples to reduce adaptation of taste buds. Warm water, biscuits, bread, apples may be used as a palate cleaning agent.

Palate clearing can be optional but whatever is done must be constant. The time between samples should also be kept constant if possible


35

Perfumes / Spices Ask panellists to refrain from wearing strong perfumes or breathing spicy odours wherever possible.


36

TESTING METHODS Wherever possible use standard methods i.e. either Australian Stds or ISO stds or those published in books or papers etc Make sure you have the appropriate method to answer your aims and objectives. Once you know your aims and objectives it is much easier to choose appropriate test. For example if you want to know if two products are different or not then the triangle test may be appropriate but if you want to know by how much and in which direction the difference lies then the triangle test is not appropriate. If using a triangle test for product matching you may find a difference but don’t know what to alter. How many samples are to be tasted all together? This may alter your testing method as for large number of samples you may require an incomplete block design or you may just want to perform a ranking test. What sort of detail is required from the Sensory Evaluation? If you require a lot of detailed information about the product then a sensory flavour or texture profile may be required. If it is only a simple difference then only a duo-trio or paired comparison is sufficient. Consider the amount of time to do the testing in relation to the number of sessions. Time is an important commodity and it is best to optimize the time but not compromise the results by fatiguing the panel with too many samples without enough time for palate refreshing and cleansing. Choose the appropriate panel type i.e. trained or consumers. Do not ask trained panelists how much they like or dislike a product or what their preferences are. If you use consumers then make sure they satisfy the selection criteria Questionnaire design Questionnaire design should be simple and easy to follow in terms of design and language and make sure tasters know how to use it. You may need to include some instructions on the scoresheet itself, but it is usually better to give instructions verbally to your panel first. The questionnaire should generally not be more than one page and include:

Name Date Time Product Sample codes Instructions Comments section

SENSORY EVALUATION METHODS There are two main types of sensory methods: Affective: tests which involve consumer preference or acceptance Analytical: tests which are involved with analysing specific product attributes in terms of:

discrimination/difference description


37

Triangle

Duo trio

Same Different

Difference from Control

Overall difference

Paired comparison

n - AFC

Attribute difference tests

Difference tests

Simple descriptive

Profiling

Quantitative tests

Analytical tests

Preference

Acceptability

Hedonic tests

Sensory testing methods

ANALYTICAL SENSORY TESTS: In general, analytical panels are used as “measuring instruments and therefore need to be:

Valid (able to measure appropriate parameters) Reproducible

Panellists can be trained or untrained depending on the degree of difference expected. Consumers will not detect the small differences that a trained panel would. An untrained panel would require 20-100 panellists while a trained panel would require 5-20. DIFFERENCE TESTING Difference tests may be sub-divided into 2 classes:

Simple difference tests are those that have no direction or characteristic associated with the difference between the products. Examples of simple difference tests are:

Triangle test Duo Trio test Two-out-of-five test A not A Difference from control

Directional difference tests are those that have a direction or characteristic associated

with the difference between the products. Examples of directional difference tests


38

are:

Paired comparison test Ranking Rating Similarity testing

In all difference tests ensure that samples are:

Representative of the product Prepared in exactly the same way Identified with a 3-digit random code Presently randomly to panellists

Panellists should complete tests in the specified order unless it is a ranking test. SIMPLE DIFFERENCE TEST TRIANGLE TEST (Australian Standard 2542.2.2 - 2005) Scope and Application Used to determine whether a perceptible difference exists between two samples. The difference can involve one or several sensory attributes, but no direction or magnitude of the difference is measured. The triangle test is an effective method to determine whether a change in ingredient, processing, packaging or storage has resulted in product differences. These situations may arise in product and process development, product matching, in quality control or as a preliminary test prior to quantitative descriptive testing. A triangle test can also be used for the selection and monitoring of panellists. With products that produce sensory fatigue, carryover effect or adaptation effects, the triangle test has limited application. Principle Three samples, two of which are identical, are presented simultaneously to each panellist for testing in a predetermined order. The panellist is told that two samples are identical and one is different (odd). The panellist is required to identify the different sample. The triangle test is a forced choice test. Preparation and Procedure The samples should be representative of the product and all prepared in exactly the same way. Select four 3-digit random number codes, two for each product. Prepare scoresheets to provide equal numbers of the following orders: AAB BBA ABA BAB BAA ABB

Make up sets of 3 samples to match the score sheets so that half contain 2 samples of product A and half contain 2 samples of product B (Total number of sets should be a multiple of 6.). Make up sets in multiples of the six arrangements as required for the number of panellists. If total number of panellists or quantity of products available is insufficient to provide equal numbers of the 6 orders, you still need to make sure there is a balance between sets with 2 ‘A’s and 2 ‘B’s. The triangle tests should be presented at random to the panellists. Instruct each panellist to examine in the specified order (e.g. left to right) and remind them that they must make a decision. Count the number of correct responses (those that select the odd sample) and compare the result with those presented in Table 2. Questionnaire Specimen answer form for the triangle test

Product Date Time Assessor One of the three samples presented is different from the other two. Please examine in the order requested and place a circle around the code of the sample which is different. 293 594 862

You must make a choice

Analysis of results The total number of correct responses is counted as well as the total number of responses and compared to the statistical tables (Table 2). This is based on the probability that if there is no real difference the odd sample will be chosen a third of the time. Example A company wishes to put a new dessert topping on the market. The product development section has two different thickening agents available to them, one which is considerably cheaper. They wish to know if there is any difference in the products made using the 2 different thickeners. Two batches (A, B) are prepared using the two different thickening agents and samples are presented to 17 assessors. As each assessor will only make one assessment, it will be necessary to prepare 27 samples of A and 27 samples of B, and arrange them to provide three of each of the six possible arrangements as indicated above. One set is discarded and the remaining 17 sets are randomly distributed between the assessors. The number of correct responses is 10, i.e. the number of panelists who correctly selected the odd sample from the 3 samples presented. The test organizer will accept a risk of error of 5% (P<0.05), that the test will reveal a difference when there is none. Table 2 indicates that for 17 assessors at P<0.05, 10 correct responses are required for significance.


39


40

It can be concluded that the product from the two thickening agents are significantly different (P<0.05). What should the test organizer do next???


41

DUO-TRIO TEST (Australian Standard 2542.2.4 - 2005) Scope and Application Used to determine whether a difference exists between two samples. The difference can involve one or several sensory attributes, but no direction or magnitude of the difference is measured. A duo-trio test can be used when one of the products is an existing standard or reference. A duo-trio test can be applied to determine whether changes in ingredients, processing, packaging or storage have resulted in differences between products. The duo-trio test finds application in the selection of panellists, product and process development, product matching, quality control and as a preliminary test prior to analytical descriptive testing. Statistically the duo-trio test is less powerful than the triangle test because the chance of guessing a correct result is one in two. The Duo-Trio test is therefore only used when it is required to form a judgement. This is the case when tasting a product with a lingering after-taste such as bitterness, spicy or chilli. Principle Three samples, two of which are identical, are presented to each panellist. One sample is identified as the reference sample and panellists are instructed to assess the reference sample first and then identify which of the two samples is the same as the reference. It is a forced choice test. Preparation and Procedure The samples prepared should be representative of the product and prepared in exactly the same way. If possible, samples are presented simultaneously or if required, sequentially. There are two forms of this test: balanced reference mode and constant reference mode. Balanced reference mode This is used when both the samples are unfamiliar and so both the samples are used as the reference sample. Select two 3-digit random codes, one for each product. Prepare scoresheets to provide equal numbers of the following orders: RA A B RB B A RA B A RB A B Make up sets of 3 samples (reference plus two samples) to match the scoresheets so that half contain 2 samples of product ‘A’ and half contain 2 samples of product `B'. (Total number of sets should be a multiple of 4). If total number of panellists or quantity of products available is insufficient to provide equal numbers of the 4 orders, then you will still need to check that there is a balance between sets with 2 ‘A’s and 2 ‘B’s.

The sample sets are allocated at random to the panellists. Instruct each panellist to assess the reference sample first, followed by the two other samples in order (e.g. left to right). Remind them that they must make a decision. Constant reference mode The constant reference duo-trio test is useful when you have trained panellists. In this test, one of the samples is a familiar product or designated standard. It is therefore the only one used as a reference sample. The number of possible presentation orders is thus restricted to: RA A B RA B A Select two 3-digit random codes, one for each product and prepare the scoresheets so that equal numbers of the two orders are presented. (Total number of sets should be a multiple of 2). Randomly allocate sample sets to panellists and instruct each panellist to assess the reference sample first, followed by the two other samples in order (e.g. left to right). Remind them that they must make a decision. Analysis of results Count the number of correct responses as well as the total number of responses and use the statistical Tables 3. Questionaire

DUO-TRIO TEST Name: Date: Time: Product: . You are provided with three samples. The left-hand one is a reference; one of the other two is the same as the reference. Taste the samples in the order left to right and circle the number of the sample which is the same as the reference. Reference Sample code:. . . . . . . . . Sample code: . . . . . . . . . YOU MUST MAKE A CHOICE Comments:

Example A duo-trio test was used to determine if methional could be detected when added to cheddar cheese in amounts of 0.125 ppm and 0.250 ppm. Each tray had a control sample marked R and two coded samples, one with methional added and one with no methional. The duo-trio test was used in preference to the triangle test because less tasting is required to form a judgment. This fact is important when tasting a substance with a lingering aftertaste, such as


42


43

methional. The test was performed on two successive days using eight judges. Each day the judges were presented with two trays. One tray contained a sample with 0.125 ppm methional and two control samples and the other contained a sample with 0.250 ppm methional and two control samples. A total of 16 judgments were made at each level. The results are shown in the following table. Duo trio test on cheddar cheese containing methional. 0.125 and 0.250 ppm. Day 1 Day 2

JUDGES 0.125 0.250 0.125 0.250

1 X R R R 2 R R R R 3 X R X R 4 R X X R 5 R R R R 6 X R X X 7 R R R R 8 R R R R TOTAL 5 7 5 7

X = wrong R = right 0.125 ppm = 10 out of 16 correct judgments 0.250 ppm = 14 out of 16 correct judgments Consult Table 3 for 16 judges in a two sample test. This chart shows that 12 correct judgments are significant at the 5% level. The conclusion is that methional added to cheddar cheese can be detected at the 0.250 ppm level but not at the 0.125 ppm level. What would you do next?? Advantages used where a reference standard is available less tasting required than triangle test can be used with trained or untrained assessors

Disadvantages No indication of character or degree of any difference Statistically less powerful than triangle test

Applications Quality control — use normal product as control Product matching Product or process improvement


44

Panel selection or training TWO-OUT-OF-FIVE TEST Used to determine whether there is a sensory difference between two samples and to select and monitor panellists. It is statistically very efficient as the probability of guessing correctly the different two samples from the five samples presented is low. It can be useful when only a small number of panellists are available. However, sensory fatigue and memory effects may affect the test. As with the triangle and duo-trio tests, assign 3-digit random codes to the samples and then make up the scoresheets, taking care to prepare the samples in an identical fashion. There will be 20 possible combinations. Panellists are instructed to assess each product from left to right and select the two samples that are different from the other three. Statistical tables exist to determine the significance of the result. “A” – “NOT A” TEST (Australian Standard 2542.2.5 – 1991 under review) Used to determine whether test samples in a series are the same as or different from the reference sample. It is an especially useful test where triangle and duo-trio tests cannot be used. This may be the case where comparisons are required between products that have a strong or lingering flavour/aftertaste when you will need to control the time between sample presentations or if there are differences in appearance. It is also useful to determine assessor sensitivity to a stimulus. Initially, panellists require familiarisation with the reference or “A” sample. Panellists are then presented with a series of samples, some of which are the reference sample “A” and some “not-A”. Generally, the panellist does not have access to the reference “A” while evaluating the test samples. The panellist must determine whether the sample is the same (“A”) or different (“not-A”) so it is a forced-choice test. Only one type of “not-A” sample exists per test series. Panellists may test one, two or up to 10 samples in series (depending on fatigue factors). The samples are presented randomly with 3-digit codes and one at a time (an assessment is made and recorded before proceeding to the next sample). All samples are prepared in an identical way and are representative of the product. The analysis of the data is quite complex.


45

DIFFERENCE-FROM-CONTROL TEST (DFC) Also called the degree of difference (DOD) test. There is no Australian Standard this test however further information can be obtained in Meilgaard, Civille and Carr and Aust et al. 1985. A new approach incorporating control lot variability was presented by Pecore et al 2006. Scope and Application This test is used to determine whether or not a perceptible overall difference exists between one or more samples and a control sample and also to give an indication as to the size of any difference perceived. In quality control situations, trained panellists may also be able to rate the degree of difference for individual attributes. A difference from control test is a useful test to use when other difference tests, such as triangle or duo-trio are not suitable because of the normal heterogeneity of the products to be tested. For example with products such as meats, baked goods and horticultural products it can be difficult to obtain a homogeneous sample which is necessary for a triangle or duo trio test. When used in conjunction with consumer acceptability testing and descriptive testing using a trained panel, the DFC test is useful for quality control and shelf life testing. In these cases the relative size of the difference is important for deciding whether the product is an accept or reject. It can be used to check production samples for the degree of difference from a recognised control or standard product. In this situation the panellists must be familiar with the range of differences expected and will require some training with reference samples and the use of the scale. The test can also be useful in product development situations to determine which sample is closest to a target product. Principle Each panellist is presented with an identified control sample plus one or more test samples. The panellists are asked to rate the size of difference between each test sample and the identified control sample. Panellists are informed that some of the test samples may be the same as the control sample. The mean difference from control for each test sample is compared with the mean difference from control obtained from the blind presentation of the control sample. The blind control sample is included as a measure of the placebo effect as it is very rare that the blind control will actually be rated as absolutely identical to the identified control. Panellists Generally 20-50 people are required. Panellists may be trained or untrained but not a mixture of the two. For some applications such as in a quality control, the panellists would require some training. All panellists should be familiar with the test format, how to use the scale and also be aware that some of the samples will be blind controls. Preparation and Procedure All samples should be representative of the product and all prepared in exactly the same way. Label an identified control sample for each panellist. Label additional blind control samples as well as the test sample(s) with 3 digit blinding codes. Where possible the control sample

and samples for assessment should be presented simultaneously. Each panellist evaluates the identified control sample first. The panellists then rate the degree of difference for each test sample of which some samples will be the blind control. The order of presentation of the test and blind control samples should be balanced. For example, half the panellists assess the samples in the order: 1. Identified control vs blind control 2. Identified control vs test sample While the other half assess the samples as: 1. Identified control vs test sample 2. Identified control vs blind control Examples of scales that may be used for the difference from control test: Line scale no difference very large difference Verbal Category Numerical category Scale

No difference 0 = No difference Very slight difference 1 Slight/moderate difference 2 Moderate difference 3 Moderate/large difference 4 Large difference 5 Very large difference 6

7 8 9 = Very large difference


46

Analysis of Results Calculate the mean difference from the identified control for each of the test samples and the blind control samples. If several samples have been evaluated, use a randomised block analysis of variance using the panellists as blocks. If only one test sample has been evaluated use a paired t-test to analyse the results. Example A company suspects a flavouring ingredient may have been left out of a batch of chunky vegetable soup. They want to know if this batch of soup is perceived to be different or not from a control batch of soup. Due to the natural degree of batch to batch variability with the product, a triangle test or other forced choice difference would be unsuitable due to the risk of yielding false positives or false negatives.

DIFFERENCE FROM CONTROL TEST Name: Date: Time: Product: ....................................................................................... . Assess the sample marked “control” first. Assess sample 386 and score the overall sensory difference between the two samples using the scale below. not different very different REMEMBER THAT A DUPLICATE CONTROL IS THE SAMPLE SOME OF THE TIME.


47


48

DIRECTIONAL DIFFERENCE TESTS PAIRED COMPARISON TEST (Australian Standard 2542.2.1 - 2007) Scope and Application Used to determine how a specific sensory property differs between two samples. It can be applied to determine a directional difference (e.g. which sample is sweeter). A paired comparison test has numerous applications in product or process development, eg substitution of a new low-calorie sweetener, in quality assurance as well as in storage tests and in product matching. A paired comparison test can also be used to determine if a more advanced sensory test should be applied. The paired comparison test can be used for multiple comparisons, but this results in a large number of pairs to assess which uses a lot of sample and can cause sensory fatigue. In this situation it is better to use a rating test. Before the sensory testing commences, it is necessary to decide whether the results will be treated as a unilateral or bilateral test. The most common paired comparison tests are two-sided (bilateral) where there is no prior expectation of the result. Conclusions that can be drawn are that there is no evidence of a difference or that one sample has a greater intensity of the chosen attribute or is preferred. One sided tests (unilateral) also exist when there is prior expectation of the direction of difference. Conclusions to be drawn include that there is no evidence of a difference or that the previously declared sample is greater in the attribute intensity or is preferred. The wording used on bilateral and unilateral score sheets is different. Test principle Two coded samples are presented. The panellists complete the scoresheet questions that have been previously determined by the test objective. Panellists The test is fairly simple requiring minimal training but the panellists must understand the attribute that is being tested. However, trained panellists may be selected if appropriate. Twenty is a reasonable number when the panellists have been screened. Statistically, numbers can be reduced to 7 for a trained panel, but when using completely untrained tasters such as consumers, then much larger numbers (100+) are needed. Preparation and Procedure Two 3-digit randomly coded samples, one of each product, are presented. The sample presented is representative of the product and all samples are prepared identically. Equal numbers of AB and BA are randomly allocated to the panellists. Panellists are instructed to assess the samples in a specific order (left to right) and identify which has the higher level of


49

a particular attribute or is preferred. The test is a forced choice test and ‘no difference’ responses are not allowed. Analysis of results Use standard statistical tables for unilateral tests (Table 3) and bilateral tests (Table 4). Count the number of replies identifying a particular sample most frequently. Compare this value with the number shown in the statistical table for the number of panellists used. Questionaires

BILATERAL PAIRED COMPARISON TEST Name……………………….Date………………………….Time………… In front of you are two coded samples of orange juice.

Please assess them in the order shown below from left to right and indicate which sample is sweetest by circling the appropriate code. Please cleanse your palate between samples.

Sample code 016 Sample code 983

YOU MUST MAKE A CHOICE Comments……………………………………………………………………….

UNILATERAL PAIRED COMPARISON TEST Name……………………….Date………………………….Time………… In front of you are two coded samples of orange juice.

Please assess them in the order shown below from left to right and indicate if sample 016 is sweeter than sample 983. Circle the response below. Please cleanse your palate between samples.

YES NO

YOU MUST MAKE A CHOICE Comments……………………………………………………………………


50

Examples Bilateral test Two drinks ‘A’ and ‘B’, are offered to a panel of 30 assessors. The two samples are presented under a random number, eg: ‘789’ and ‘379’. The test supervisor accepts a 5% level of significance (i.e.: P< 0.05%). He does not know which of the two samples contains more sugar. Question: Which sample is sweeter? Replies 18 opt for sample ‘A’ 12 opt for sample ‘B’ From Table 4 it can be concluded that there is no significant difference in the sweetness of the two drinks. Unilateral test Two drinks, ‘A’ and ‘B’, are offered to a panel of 30 assessors. The two samples are presented under a random number eg: ‘789’ and ‘379’. The test supervisor accepts a 1% level of significance (i.e.: P<0.01%). He knows that drink ‘A’ contains more sugar than drink ‘B’. Question: Is sample ‘A’ sweeter than sample ‘B’? Replies 22 yes and 8 No. From Table 3, it can be concluded that drink ‘A’ is significantly sweeter than drink’B’. Advantages/Disadvantages See paired preference. Applications Product Development Quality Control Shelf Life Measurement

RANKING TEST (Australian Standard 2542.2.6 – 1995 under review for adoption of ISO std) Scope and Application The ranking test can be considered an extension of the paired comparison test. It is used to place a series of three or more samples in a rank order to determine whether differences exist between samples. Samples are ranked for a specified criterion, e.g. an attribute (bitterness, crunchiness, hardness) or a preference. The criterion needs to be understood by the panellists. The data obtained is ordinal and therefore provides directional differences between samples but does not provide information about the degree of difference. The ranking test is a simple way to compare samples and is useful for reducing the number of


51

test samples prior to performing another test and to evaluate panellist ability. In product development, a ranking test can be used as a quick method of indicating the effects of different raw materials, processing, or packaging and storage treatments. Test principle Samples are presented to the panellists simultaneously and are placed by the panellists into a rank order relative to one another according to the specified criterion. Panellists Minimum of 8 but a larger number of panellists is better. Preparation and Presentation Three or more 3-digit random coded samples are presented to panellists simultaneously for assessment in a balanced or random (if more than 4 samples) order. All the samples are prepared and presented identically. The maximum number of samples will depend on the type of product. They are instructed to arrange the samples in rank order according to the level of the specified criterion, and are instructed whether to assign rank 1 for the lowest or highest level. It is a forced choice test and tied rankings are not permitted. A separate scoresheet is used and completed separately if the rank order is required for more than one criterion. As a panellist, it is often easier to perform this test by arranging the samples in a provisional order first and then to re-evaluate them before assigning final ranks. Analysis of results Rank totals are calculated for each sample and used to generate test statistics which are compared to statistical tables. The Friedman test is the preferred method of analysis. As samples are evaluated only in relation to each other, results from one test cannot be compared to those from another unless they both tested the same samples. Example A cordial manufacturer has been provided with two new samples of lemon flavour that are cheaper than the existing flavour. The manufacturer wants to know if cordials are made at the same flavour intensity, would it be cheaper to use either of the two new flavours. Samples are prepared at the same concentration but in order to test this from a sensory perspective the 3 samples are presented to 30 assessors who are asked to rank them in order of flavour intensity. The results are presented below:


52

Cordial Samples Assessor

A B C 1 1 2 3 2 1 3 2 3 2 1 3 4 1 3 2 5 3 2 1 6 1 2 3 7 1 3 2 8 3 2 1 9 3 2 1 10 3 2 1 11 2 1 3 12 3 2 1 13 1 2 3 14 3 1 2 15 3 2 1 16 1 2 3 17 2 3 1 18 1 3 2 19 1 2 3 20 2 1 3 21 3 1 2 22 2 1 3 23 1 3 2 24 3 2 1 25 1 3 2 26 1 3 2 27 3 2 1 28 3 2 1 29 1 2 3 30 2 1 3

Rank Sums 58 61 61

The Friedman value F needs to be calculated as follows.

F

JP PT J Pp

12

13 12

( )( )

where J - number of judges P - is the number of samples (products) T1,T2,T3 - are the rank totals for each sample

So )13(303)616158()13(330

12 222

F

= 360.2-360


53

= 0.2

For our example we get a F of 0.2. Now when the number of panellists is large (>15) or the number of samples exceeds 5 then F follows the chi-squared distribution with P-1 degrees of freedom. So looking up the chi-squared table (table 7) gives a critical value of 5.99. Since our calculated F is less than this we can not reject our question and conclude that there are no significant differences between the samples. Pairwise comparisons can be made using the formula below.

6

)1( PJPz where z corresponds to a two-tailed normal probability of α

α = 0.05 at the 5% level so z = 1.96 (table 8). Two samples are different if the difference between their rank sums is greater than or equal to 15.18. The technician must retain the null hypothesis that there is no difference between the flavour strength of the three products. RATING TEST (Australian Standard 2542.2.3 2007) The rating test can be used to measure the perceived intensity of sensory characteristics eg degree of strawberry flavour in a strawberry milkshake. For this type of test the basic principles of sensory evaluation should be followed eg coded samples, controlled test environment, number of samples tested. Panellists should be selected based on their ability to give consistent ratings to the same sample and to discriminate between samples checked by statistical analysis. The number of panellists used depends on the degree of training but generally a minimum of eight highly trained, more if less trained. Selection and training of panellists will be discussed later in a separate section. The response scale used for rating may be in the form of a category scale or a line scale. A category scale is a series of 7 – 15 boxes labelled to identify levels of intensity. With a line scale, panellists respond by marking a position on a horizontal line labelled with “anchors” at each end. An advantage of this type of scale is that panellists responses are not limited to a number of categories on the scale and therefore it may be possible to identify more differences between samples.


54

Example of a category scale. Strawberry flavour

Sample number 495 128 Extremely strong Very strong Moderate Slight Absent

An example of a line scale. Strawberry Flavour

None Very strong Analysis of results Ratings must be converted to numerical scores for analysis and interpretation. For category scales, successive integers are assigned to successive categories and these are used in analysis, e.g. with a 9-point scale, the integers 1-9 would be used. For graphic scales, the distance, e.g. in mm, between the response mark and one end of the scale serves as the response score. The arithmetic mean and standard deviation, when obtained for each sample, serve as measures of central tendency and variability, respectively. For statistical analysis, the analysis of variance technique is appropriate (or a t-test in the case of one or two samples). Correlation or regression analysis may be used for subjective/objective correlations. Advantages More than one sensory attribute can be examined. Size and direction of differences can be identified.

Disadvantages Selecting realistic terminology Agreement and understanding between assessors in descriptive terms Scales are not linear i.e.: 13 = extremely sweet is not twice as sweet as 7 = moderately

sweet Applications Product Development


55

Product matching Quality Control Research Storage Trials Panellist assessment

Example The scoring method was used to determine if there was a difference in bitterness in cheddar cheese made using three different milk-coagulating enzymes. Samples of cheese from each treatment were coded and presented to 12 judges for evaluation. The order in which the three samples were tasted was balanced so that each possible order was used twice: ABC, ACB, BAC, BCA, CAB and CBA. The ratings assigned by the judges were given numerical values, ranging from 0 points for ‘not bitter’ to 5 points for ‘extremely bitter’. The results are shown in the following table.

Samples Judges A B C Total

1 3 0 1 4 2 2 2 2 6 3 3 1 2 6 4 1 1 0 2 5 3 1 3 7 6 2 1 1 4 7 3 2 2 7 8 2 0 1 3 9 3 1 2 6 10 4 2 3 9 11 1 1 0 2 12 2 2 2 6

Total 29 14 19 62 Mean 2.42a 1.17b 1.58b

The results were submitted to analysis of variance. Any two values not followed by the same letter are significantly different (P<0.05). Sample A is significantly (P<0.05) more bitter than sample C and B. There is no significant difference (P>0.05) in bitterness between samples C and B.


56

SIMILARITY TESTING Sometimes we are looking for equivalence in samples and not a difference. Eg interchange ingredients and this leads to a category of testing known as similarity testing. Similarity testing considers the probability of rejecting the null hypothesis (Ho) when it is true i.e. alpha Type I error and the probability of accepting Ho when it is false i.e. beta Type II error. There is also the concept of the proportion of discriminators. When someone does a difference test like a triangle or duo trio then some people are true discriminators who see the true difference and select the correct “odd” sample and there are non-discriminators who see no difference and guess. Non-discriminators include people who guess correctly and those who guess incorrectly. Therefore the total number of correct choices by the panel reflects the sum of the discriminators plus the fraction of the non discriminators who guess correctly. This leads to this formula C = D + 1/3(N – D) where for example N (total number of panellists) = 45 C (number who were correct) = 21 D is the number of discriminators so 21 = D + 1/3(45-D) and solving for D 21 = D + 45/3 – D/3 21 – 45/3 = 2/3D D = (21- 15)3/2 D = 9 So 9 out of 45 people or 20% were likely to have really seen the difference. Other tests you may come across and may be useful for certain circumstances are the R- index signal detection method and the 3 AFC (alternate forced choice) and is sometimes used for threshold testing. AFC is when you ask which of the three samples is the sweetest for example.


57

AFFECTIVE TESTS Preference infers a preference for one product over another; either overall or in relation to a particular parameter. Acceptance infers actual utilisation/purchase of the product. Panel selection Select panel on basis of end use:

Age Race Religion Sex or Random selection for overall

Panel training No need for training, in terms of technique or ability. However, panellists should be instructed/briefed in terms of:

Method Questionnaire Length of trial Number of samples

Panel size

1. 20 to 100 people 2. 20 = pilot consumer panel 3. 100 = consumer panel

SPECIFIC TEST METHODS PAIRED PREFERENCE TEST (Reference: Australian Standard 2542.2.1; 2007) Application: to establish whether there is a preference between two samples. Principle: a pair of samples (one may be a control) is presented to each assessor. The assessors are asked to choose the sample they prefer. This test is a ‘forced choice’ i.e.: the assessors must select one sample as being more preferable. Responses indicating no preference are not permitted. Statistically based on null hypothesis that there is no preference between the samples. i.e.:PA = PB = 50%= 0 5


58

Bilateral Test - no expectation of preferences Specimen Answer form for bilateral paired preference test

PRODUCT………………….DATE…………..TIME………ASSESSOR………………….. Which sample do you prefer? Please examine code 349 first. Please tick the appropriate box.

Code 922 349 Place tick

YOU MUST MAKE A CHOICE

Conclusions

no preference A preferred to B B preferred to A

Question — which of the two samples do you prefer? Count the number of replies citing one of the two samples the more frequently. Conclude that this sample is significantly preferred to the other if the number obtained is greater than or equal to that shown in Table 4. Example: Two drinks ‘A’ and ‘B’ are offered to a panel of 30 assessors. The two samples are presented under random number eg: ‘789’ and ‘379’. The test supervisor accepts a 5% level of significance (i.e.: P < 0.05%). It is not known which of the two samples contains more sugar. Question - Which sample do you prefer? Replies: 22 prefer ‘A’

8 prefer ‘B’ From Table 4 it can be concluded that Drink ‘A’ is preferred to Drink ‘B’.


59

Unilateral Test - expect one sample to be preferred. Specimen Answer form for unilateral paired preference test

PRODUCT………………….DATE…………..TIME………ASSESSOR………………….. Do you prefer sample 186 to sample 592? Please examine code 592 first. Please tick the appropriate box.

YES NO

YOU MUST MAKE A CHOICE

Conclusion

no preference the declared sample is preferred

Question — Do you prefer sample ‘A’ to sample ‘B’? Conclude sample A is preferred if number of positive replies is greater or equal to the number shown in Table 3. Example: Two drinks, ‘A’ and ‘B’, are offered to a panel of 30 assessors. The two samples are presented under a random number eg: ‘789’ and ‘379’. The test supervisor accepts a 1% level of significance (i.e.: P < 0.01%). It is known that drink ‘A’ contains more sugar than drink ‘B’. Question - Do you prefer sample ‘A’ to sample ‘B’? Replies: 23 Yes and 7 No. From Table 3 it can be concluded that there is preference for drink ‘A’ over drink ‘B’. N.B. If uncertain always use the bilateral test. Advantages

Simple test to conduct Suitable for children and consumer panels Easy to analyse (for > 100 assessors use t test or CHI squared)

Disadvantages


60

Only suitable for 2 products (note – multiple Comparisons can be used but other

preferences tests are more commonly used. See ASTM manual on sensory testing method, STP 434; 1968)

No magnitude of preference is given i.e. they both may be disliked but one can still be

preferred. Applications

Product Development Product Matching Process Change

RANKING FOR PREFERENCE (Australian Standard 2542.2.6 - 1995) Principle: Judges are asked to rank two or more samples in order or preference i.e.: most preferred sample is ranked first. Ranking is a forced choice procedure i.e. no ties are allowed. Specimen Answer form for ranking for preference.

PRODUCT………………….DATE…………..TIME………ASSESSOR………………….. Please taste the samples in the order presented, moving from left to right and rank them in order of preference. You may retaste the samples to check the ranking. Give the sample that you most prefer a rank of 1 and the sample you prefer next a rank of 2 etc. You must give each sample a different rank. Equal ranks are not allowed.

Samples Rank

Statistical analysis Kramer’s tables, which have been used in the past to analyses differences between rank sums, should not be used due to questions of accuracy and statistical validity. When there is no expectation of a specific rank order being made (eg when ranking preference of new product prototypes) the Friedman Test should be used (see below for details).


61

Example Twelve households were presented with four samples of meat seasoning to be used in cooking. They were asked to use the samples as directed and to rank them in order of preference. The results are shown below: Rankings for the preference of four meat seasonings

Seasoning HOUSEHOLD A B C D

1 1 3 2 4 2 2 1 3 4 3 1 4 2 3 4 1 4 2 3 5 2 3 1 4 6 3 4 2 1 7 3 4 2 1 8 3 4 1 2 9 1 2 3 4 10 1 2 3 4 11 1 2 3 4 12 1 3 2 4

Rank sums 20 36 26 38 The Freidman value F needs to be calculated as follows.

FJP P

T J Pp

12

13 12

( )( )

where J - number of judges P - is the number of samples (products) T1,T2,T3 - are the rank totals for each sample So

)14(123)38263620()14(412

12 2222

F

=190.8-180

=10.8

When the number of panellists is large or the number of samples exceeds 5 then F follows the chi-squared distribution with P-1 degrees of freedom. So the calculated value is compared to the critical chi-squared value in table 7 (7.82 for 3 df). Since 10.8 is greater than 7.82, the experimenter can conclude that there is a significant (p<0.05) difference between the rank sums.


62

Two samples will be significantly different if the absolute value of the difference between the rank sums is greater than or equal to the following critical value:

6

)1( PJPz where z corresponds to a two-tailed normal probability of α

α = 0.05 at the 5% level so z = 1.96 (table 8).

6

)14(412960.1

= 12.4

Sample A B C D Rank Sum 20a 36b 26ab 38b

Rank sums that do not have a common superscript are significantly different (P<0.05)


63

RATING FOR PREFERENCE (Australian Standard 2542.2.3 - 2007) Principle Assessors are asked to evaluate one or more samples and indicate the degree of liking for the product or some characteristic of the product. When performing preference testing it is important to include as many panellists as possible. Personal preferences in foods are being measured which are purely subjective, so the variance in the data is large. This makes it more difficult to obtain statistically significant results. The larger the panel, the more chance there is of obtaining a significant result. Only untrained panellists are used and should be selected at random or from a targeted group related to the product. Pilot consumer panel = 20-25 Consumer panel = 100 Types of response scale Category scale/structured scale The response scale is divided into categories or boxes. The response scale is usually divided into an arbitrary number of categories - usually between 7 and 13 Category scales must be bipolar. Verbal descriptors or facial expressions may be used to identify the levels of acceptance. Hedonic category rating AROMA FLAVOUR TEXTURE Like extremely Like very much Like moderately Like slightly Neither like nor dislike Dislike slightly Dislike moderately Dislike very much Dislike extremely


64

Facial hedonic scale 7 point facial hedonic scale (smiley face)

Appearance Aroma Flavour

Mouth-feel Graphic rating scale

The response is recorded by marking a position on a line Also called visual-analogue scale, line mark scale or unstructured scale Physical lengths 100-150 mm. This scale may also use facial expressions for measurement. Numbers and/or descriptors are usually attached to a rating scale.

Recording and interpretation of results Ratings must be converted to numerical scores for analysis and interpretation. For category scales, successive integers are assigned to successive categories and these are used in analysis, e.g. with a 9-point scale, the integers 1-9 would be used. For graphic scales, the distance, e.g. in mm, between the response mark and one end of the scale serves as the response score. The arithmetic mean and standard deviation, when obtained for each sample, serve as measures of central tendency and variability, respectively. For statistical analysis, the analysis of variance technique is appropriate (or a t-test in the case of one or two samples). Correlation or regression analysis may be used for subjective/objective correlations. Advantages

Test is relatively simple and easily understood; Indicates the degree of preference; Can be used to infer acceptance; Suitable for different age groups; and Can measure > one parameter at a time.

Disadvantages

Statistical analysis is required; and Results may be biased by type of assessors used.


65

Applications

storage trials product development consumer testing quality control subjective/objective correlations research

Example: Three samples of frozen chicken casserole were presented to a 24 member panel that assessed the appearance, flavour, texture and general acceptability of the products using a 13 point scale. The following results were obtained:

CHICKEN CASEROLE A B C

Appearance 10.8 a 8.5 c 10.3 b Flavour 9.9a 9.4 b 9.2 b Texture 10.4 a 9.6 b 9.1 b

General Acceptability 10.3 a 9.2 b 9.4 b Scores within each row that do not have a suffix in common are significantly different.


66

SENSORY EVALUATION IN CONSUMER TESTING Introduction The personal response by current or potential customers of a product, a product concept, or specific characteristics of a product is collectively grouped under what we call consumer testing. However, it is important to define the terms acceptance and preference often associated with consumer testing. Acceptance refers to the degree of liking or disliking for a particular product or the ability of the product to meet expectations of consumers while preference refers to a choice made by panellists among several products on the basis of liking or disliking. Unfortunately ‘preference’ is widely used as a generic term to describe both acceptance and preference judgements. It is important to note for example in paired preference testing that although one product may be preferred to another, neither product may be liked to any degree. The term ‘hedonic’ is an adjective associated with degrees of pleasure or displeasure and is applied to both acceptance and preference testing. Applications of Consumer Testing The reasons for conducting consumer tests usually fall into one of the following categories:

Product maintenance Product improvement/optimization Development of new products Assessment of market potential

Product maintenance Research and development projects may involve cost reduction, substitution of ingredients, process and formulation changes and packaging modifications without affecting the product characteristics and overall acceptance. Usually difference tests would be used to determine whether a difference was perceived or not but it is necessary to take the product out to the consumer to determine if the reformulated product will achieve at least parity with the current product. Product maintenance is also a key issue with quality control/quality assurance and shelf-life/ storage projects. Feedback on consumer response gives important information on those sensory characteristics that are most important to consumer choice and which should therefore be rigorously controlled. A combination of in-house profile testing on the magnitude and type of change over time, condition, production site, raw material sources etc can be used in conjunction with consumer testing to determine how large a difference is sufficient to change the acceptance rating. Product improvement/optimisation The intense competition among consumer products drives companies to constantly improve and optimise products so that they can deliver what the consumer is really looking for and therefore increase market share. In product improvement, prototypes are made, tested by a trained panel to verify that the


67

desired attribute differences are perceptible, and then tested with consumers to determine the degree of perceived product improvement and its effect on overall acceptance or preference scores. For product optimisation, ingredients or process variables are manipulated and a trained panel identifies the key sensory attributes affected and consumer tests are conducted to determine if consumers perceive the change in attributes and if such modifications improve the overall acceptability. Development of new products During the new product development from concept to a range of trial samples to a modified sample range and finally a choice to launch, consumer testing should be used throughout in conjunction with trained panel assessment. Assessment of market potential In addition to the use of sensory evaluation to gather information about key attributes of a product, typical marketing questions such as intent to purchase, purchase price, willingness to pay for added value, current purchase habits, consumer food habits, effects of packaging, advertising and convenience are critical for the acceptance of branded products. Other consumer science areas include drivers and barriers for consumers. It is often convenient for these marketing type questions to be included in a questionnaire presented to consumers when assessing the sensory characteristics of the product. Conducting Consumer Tests There are a number of factors to consider when conducting consumer tests and these are:

Test design Test subjects Test location Test questionnaire

Test Design There are two main types of design; one is qualitative measuring subjective responses while the other is quantitative determining the responses of a large group to a set of questions regarding preference, liking, sensory properties etc. Qualitative Tests include focus groups, focus panels and one-on-one interviews. Each of these has their use in a particular situation depending on what is required and how sensitive the topic is. Essentially small groups are used to uncover as much specific information from as many participants as possible. It is frequently recorded either by video and or audio and a summary is made. Quantitative Tests Essentially all the good practice principles used in sensory evaluation as described in the difference and descriptive testing should be followed here such as 3 digit random codes for product and presentation in a balanced order. Some typical designs used include:


68

Monadic test where only one product is assessed which makes it fast and the least

expensive but is relatively insensitive and requires large numbers of consumers (at least 200).

Sequential monadic where one product is assessed, removed and then replaced by a second product in a balanced design giving it greater sensitivity.

Paired preference testing where two products are assessed simultaneously and a direct comparison is made making it quite sensitive.

Acceptability testing. Usually the nine-point hedonic scale is used to determine consumers liking of a product and if required the relative ratings for liking can be used as a measure of preference.

As with the rating for preference the following scales are suitable for consumer studies. Example of nine-point hedonic category rating AROMA FLAVOUR TEXTURE Like extremely Like very much Like moderately Like slightly Neither like nor dislike Dislike slightly Dislike moderately Dislike very much Dislike extremely

Dislike extremely neither like nor dislike like extremely Example of seven point facial hedonic scale often used for children

Appearance

Aroma

Flavour

Mouth-feel


69

Graphic rating scale - the response is recorded by marking a position on a line (also called visual-analogue scale, line mark scale or unstructured scale) - physical lengths 100-150 mm. This scale may also use facial expressions for measurement.

Attribute testing can be used to gain information on the reasons underlying overall preferences and usually category or line scales are used. These can be hedonic type attributes or sensory attributes in the form of just right scales as shown below.

not sweet enough just right too sweet However need to be careful with “just right“ scales as some groups of people may like hot spicy product while other groups like mild product so two differing products may end up being “just right” depending on which group you ask. On way around this is to ask an intensity question or ask an intensity question and ask the panel to rate their ideal spiciness on the scale. Test Subjects If information on the acceptance of the product by consumers is required, then it is they who should do the tasting. However, this is not always practical in preliminary testing of products, so a compromise can be made by using large numbers staff who assess fairly infrequently. However, it should always be remembered that this is a compromise, and results are best interpreted only in relative, not absolute, terms. Staff cannot be considered to be a representative sample of the target population. When performing consumer testing it is important to include as many panellists as possible. Personal preferences in foods are being measured which are purely subjective, so the variance in the data is large. This makes it more difficult to obtain statistically significant results. The larger the panel, the more chance there is of obtaining a significant result. Recruitment The number of consumers to be tested depends on the purpose of the test, the test design and the precision with which the target population can be identified. In general we require 60 to 120 for most consumer testing. Recruitment and selection of consumers rely on several criteria or demographics such as:

Product usage. It is important to determine if you are looking for low, medium or high users of the product. For speciality products or niche markets, the cost of consumer testing increases as more people must be contacted before the required number are found.

Gender. It is not always necessary to get equal numbers as purchasing or usage habits vary between products. Researchers should use current market information.

Age. If a product has broad age appeal then consumers should be selected by age in proportion to their representation in the user population.

Nationality. Products targeted towards a specific part of the community or for export ideally should be tested in that environment. However, it is possible to use foreign nationals resident here but it depends on how long they have been residing in their


70

adopted country as they can develop the likes and dislikes characteristic of the adopted country.

Social class. This can be based on income or occupation although sometimes it is difficult to get consumers to reveal such information.

Others including race, religion, education level, employment, geographic location, etc. If any of these are important in defining the target audience then the researcher should consider them.

Source of Consumers As mentioned it is important to sample properly from the consuming population but because of cost restraints employees and local residents may be used for things such as product maintenance. However, for new products or product optimisation or improvement the correct audience should be selected. These can come from a database of consumers willing to assess products, recruitment agencies, telephone survey, leaflet drop, shopping centres, embassies, colleges or door to door. Test Location It is possible to conduct consumer testing in a number of locations depending on the resources and the results can vary greatly. Locations include:

Company laboratory facilities, which give good control of the environment and rapid feedback of results but the sensory booths, are clinical and atypical of a real domestic environment.

Central location such as school or church halls or shopping centres are convenient as large numbers can be tested at one time and on a number of products. However the conditions are artificial compared to normal use at home and the number of questions that can be asked may be limited.

Home use tests represent the ultimate in consumer testing as the product is tested under its normal conditions of use. In addition to the product itself, a check on the packaging can also be determined. Generally more information can be gathered as the consumer gets more time and can perform repeated assessments. However it is time consuming and uses a smaller number of people and the possibility of nonresponse is great unless consumers are continually reminded.

Test Questionnaire It is very important that the test questionnaire format is simple, unambiguous, easy to read and understand. You need to consider the objective of the test and any constraints such as time, funding etc. In essence you need to be:

Brief Use simple plain English (provide translation for studies involving foreigners)

technical terms and jargon should be avoided Be specific. Don’t ask a general question when you need an answer on a specific

issue. Multiple choice questions should be mutually exclusive Avoid ambiguity. Be careful in wording a question and selecting the audience who

will answer the question. Watch the effects of wording and leading questions.


71

Don’t ask what they don’t know. This is especially true when you want the respondent to recall information. Critical issues are the length of time since the event occurred and how important the event was to the respondent.

Try and pre-test the questionnaire


72

For example How satisfied or dissatisfied were you with the product? Very satisfied Slightly satisfied Neither satisfied nor dissatisfied Slightly dissatisfied Very dissatisfied The product looks like how it is shown on the package. Agree strongly Agree Neither agree nor disagree Disagree Disagree strongly What did you like about the product? This open-ended question allows for the consumer to add something you may have forgotten but it is sometimes hard to read the answer (handwriting) and some people do not bother with answering. The question order should go from the more general to the more specific and ask overall acceptability first before biasing the consumer with more specific issues. Ask the more sensitive demographic questions last. Data Analysis All quantitative data should be subjected to some form of statistical analysis from simple summary statistics and graphical representation to t-tests and analysis of variance with pairwise comparisons. Further advanced multivariate methods such as principal components analysis and cluster analysis along with regression methods to relate consumer data to other data such as linear regression, partial least squares and preference mapping can also be used.


73

DESCRIPTIVE TESTING Descriptive testing is used to identify and provide a picture or “profile” of the important sensory characteristics of a product. With sensory profiling more than two samples can be assessed simultaneously. This type of test has the advantage of not only being able to tell you whether or not there is a difference between samples but also the nature and magnitude of these differences. Appearance, odour, flavour and texture can all be assessed in this way and the characteristics can be quantified using various techniques and scales as outlined in this section. Applications:

Tracking changes in the sensory characteristics of a product over time for shelf-life evaluation

Examining the sensory properties of a target product for new product development Examining sensory characteristics of different varieties of a product eg to look at

several varieties of apples in order to identify which varieties are sweetest, crunchiest etc.

Sensory diagnostics of ingredient, process or packaging changes Correlations with instrumental methods

The Flavour Profile Method® (Arthur D. Little) This method was developed by Arthur. D. Little in the late 1940’s early 1950’s. It uses a panel of 4-6 trained panellists. Panellists are selected by screening for sensory acuity, interests, attitude and availability. A vocabulary is developed by exposure to a wide range of products from the product category to be assessed. The list is then reviewed and refined and reference standards and definitions applied to each term. The panellists examine the products and the results are reported to the panel leader. Through discussion in an open session lead by the panel leader, a consensus decision is reached for each sample. Aroma, flavour and amplitude, which is the balance or blending of the flavour, is assessed in this way. The scales used with this technique involve the use of numbers and symbols and therefore cannot be analysed statistically. This method is therefore a qualitative descriptive test. The main disadvantage with this type of test is that a dominant panel member or the panel leader could easily influence the panels’ decision.


74

Questionnaire for flavour profile of beer Product Name Date AROMA Characteristic Intensity Hoppy Fruity Sour Yeasty Malty Amplitude(overall aroma) FLAVOUR Characteristic Intensity Tingly Sweet Fruity Bitter Malty Yeasty Metallic Astringent Amplitude(overall flavour) Comments


75

Profile Attribute Analysis®

The Flavour Profile method was renamed the Profile Attribute Analysis with the introduction of numerical scales. Mean scores could then be calculated and the data statistically analysed. However consensus methods are still employed by some people. Again, this runs the risk of a result being skewed by a dominant personality in the group. The Texture Profile Method®

This method was developed at General Foods in the 1960’s. It was based on the principles of the Flavour Profile method to assess the textural characteristics of a product. Textural characteristics are categorised into three groups, mechanical, geometrical and ‘other’ characteristics.

1. Mechanical: relating to the reaction of food to stress eg. hardness, chewiness and adhesiveness

2. Geometrical characteristics: relating to the size, shape and orientation of the particles within the food eg. grainy, fibrous and aerated

3. Other characteristics: relating to the perception of the moisture and fat contents of the food

The order in which the characteristics are assessed is also very important. The order of assessment is first bite, “chewing” or masticatory second phase and residual or third phase. Panellists are selected on their ability to discriminate between textural differences in the product area to be trained. Six to ten panellists are suggested. Standardised terminology and rating scales are used for the assessments and each scale point is anchored with a specific food. Initially the technique used an expanded version of the Flavour Profile scale, however more recently category and line scales have been used. Panellists each make their own individual judgement and then depending on the type of scale used, a consensus decision is reached or statistical analysis is performed on the data. Quantitative Descriptive Analysis (QDA®) This method of descriptive analysis was developed in the 1970’s. Ten to twelve panellists are selected by screening for ability to discriminate between products, their ability to verbalise their perceptions and to work as a group. The first step is to expose the panellists to a wide range of products from the product category to be assessed. Each panellist individually lists as many descriptive words possible that describe differences between the products. Hedonic terms such as nice, good, bad, etc are not allowed. Through a group discussion, the list of descriptive words is narrowed down to remove duplications and redundant terms until a standardised vocabulary is reached. This standardised vocabulary then needs to be defined with verbal definitions or reference standards and anchor points for the scale agreed upon. The panel also decides the order in which the terms are to be assessed. During this process the panel leader only acts to facilitate the discussion and provide references but does not influence or lead the panel. Trial evaluations are then carried out using the agreed vocabulary and refinements may be made until the panel is happy with the terms used. The panel leader evaluates the results from these trial sessions and once confident the results are reliable and repeatable the actual assessment can take place. The assessment and trial sessions are completed in sensory booths following the basic principles of sensory evaluation. An


76

unstructured 6-inch or 15cm line scale is used to measure the intensities of the agreed characteristics. Several replicates (3+) are required to validate the data. Data is then analysed using an analysis of variance. The results are often displayed visually on a spider web or star diagram. Results of ANOVA of orange jelly using QDA Attribute Brand A Brand B SEM Probability

Orange colour 10.2 7.9 0.62 0.011 Orange aroma 7.6 6.9 0.50 0.325 Firmness 9.6 6.6 0.64 0.001 Tartness 8.6 6.9 0.66 0.072 Orange flavour 7.6 6.9 0.72 0.494 Foreign flavour 4.3 4.8 0.48 0.464 Sweetness 7.1 9.6 0.42 <0.001 Rate of breakdown

5.1 6.1 0.60 0.242


77

Other Methods Other methods which you may come across in literature but which will not be discussed in detail are: Spectrum Method This is a descriptive analysis technique developed by Civille to cover any or all of appearance, aroma, flavour, texture or sound characteristics. Panellists use a standardised lexicon of terms to evaluate the products. This method requires extensive training of the panel to use standardised scales anchored with multiple reference points and panellists are trained to use the scale identically. Data is analysed in a similar way to QDA. Example of intensity scale values (0 to 15) for firmness. Scale Reference Sample size value 3.0 Aerosol whipped cream Redi whip 1oz 5.0 Miracle whip Kraft 1oz 8.0 Cheese whiz Kraft 1oz 11.0 Peanut butter CPC Best Foods 1oz 14.0 Cream cheese Kraft/Philadelphia 1oz Time Intensity This is used to track the changes in perception of a particular attribute of a product over time. For example you might rate the intensity of mint flavour perceived in chewing gum over a 3 minute period. This can be measured using pencil and paper or using one of the sensory software packages with time intensity facilities. Free Choice Profiling Unlike other descriptive testing techniques this method does not use an agreed vocabulary to assess the samples. Each panellist generates their own list of terms and scales, although they must use these consistently for all samples. The data from this type of assessment is then analysed using Generalised Procrustes analysis. The main advantage of this technique is the time saved on training a panel; however interpretation of individual attributes can be subjective as the terms are not defined as with other descriptive testing methods.


78

SELECTION, TRAINING AND MOTIVATION OF A PANEL When developing a sensory panel, there are several areas that need to be addressed that include:

The need for a panel in the organisation (R&D, QA/QC) Organisation and management support and commitment (time and money) Resources required

o Sensory staff o Interest and availability of potential panellists o Samples and references for screening and training o Availability of a panel room and booths o Facilities for data collection and statistical analysis

Establishment of a trained sensory panel can be divided into 2 steps:

Selection Training

Selection for Descriptive Testing (Australian Standard 2542.1.3 - 1995) Recruitment Panel members are usually recruited from staff in laboratories, offices and the plant of a company. Some companies test their products at a different company facility. External panellists may also be recruited from the community nearby if the sensory panel work is going to be very time consuming. Talks, circulars, noticeboards or personal invitations may be used to recruit potential panellists. Information should be provided to the prospective panellists concerning the application of sensory evaluation, what will be involved for the panellists and the envisaged work program. Pre-screening questionnaire Potential panellists need to complete a pre-screening questionnaire to obtain background information on their:

interest in participating in the screening and training program as well as ongoing work availability

general good health (note any illnesses or allergies and permanent impairment to the senses) any food idiosyncrasies (strong food dislikes or reactions to foods) other information that might be relevant (age, sex, nationality, cultural and religious

background, previous sensory experience, smoking habits) Panellists should not be asked to assess a food that they dislike. In a company situation, distribute questionnaires for employees to fill in, detailing the above criteria. If you make all the questions optional you will find that the majority of people respond truthfully. Pre-screening questionnaires can also be used to select individuals who can describe sensory concepts. Record all the information you receive in some form of database. Based on the above criteria, decide which prospective panellists are to proceed in the screening process.


79

Interview Individual interviews are required to determine whether prospective panellists will work well in a group situation as well as for the analytical approach required in descriptive testing. An interview is also used to confirm interest and availability. For a descriptive sensory panel, there is a large investment involved in terms of both time and money. It is best to complete a thorough screening process rather than training unsuitable subjects. During the selection process, it is important to make note of both attendance and personalities of panellists. A panellist who is repeatedly late or unavailable can be more trouble than they are worth. Someone who distracts other panellists by talking or making comments, despite repeated requests to remain silent while testing, is a liability, not an asset. However, it is recognised that the best panellists available may need to be used although they may not necessarily meet all the requirements. Sensory screening tests Screening is completed to obtain information on prospective panellists who need to be able to:

Detect differences in attributes present and their intensities Describe the attributes using verbal descriptors and scaling methods for the different

intensities Be able to recall and apply attribute references when required

Prior to the first screening test, a preliminary session is a good idea to set the rules that may need to be enforced politely but firmly. Instructions for panellists

Avoid eating, drinking, smoking or chewing gum for 30 minutes before testing. Do not talk or distract other panellists while testing. Read any instructions on the scoresheet before starting to evaluate samples. Make sure you evaluate the samples in the required order. Don't forget to fill in your name and the date. Do not discuss samples with other panellists until after they have evaluated the samples. Have confidence in your own judgement. Ignore your personal likes and dislikes.

Sensory screening tests also give the prospective panellists an indication of the methods used in sensory analysis. The screening tests used should be chosen with the envisaged sensory program in mind. Basic tastes and odours are commonly used for screening tests as well as materials that illustrate the attributes that may be included in the sensory program. Samples of the actual food products may also be used. A series of triangle or duo-trio tests may be completed to assess the ability of the potential panellists to detect small differences between stimuli at supra-threshold levels. Preferably, potential panellists should respond correctly 100% of the time.


80

Matching tests may be used to evaluate the ability of a prospective panellist to distinguish between different sensory stimuli. In order to evaluate the ability of the panellists to describe sensory responses, a series of products can be presented and potential panellists asked to describe the sensory impression. The products used should be related to those that will be used in the envisaged sensory testing. For example, a range of odours may be presented:

Chemical name Name most commonly associated with the odour Benzaldehyde Bitter almonds, ... Octene-3-ol Mushroom, ...

Phenyl-2 ethyl acetate Floral, ... Diallyl sulfide Garlic, ...

Camphor Camphor, ... Menthol Peppermint, ... Eugenol Clove, ... Anethol Aniseed, ... Vanillin Vanilla, ... Geosmin Musty/mouldy, ...

Beta-ionone Violets, raspberries, ... Butyric acid Rancid butter, ... Acetic acid Vinegar, ...

Isoamyl acetate Fruit, acid drops, ... Dimethylthiophene Grilled onions, ...

Panellists are given these samples to assess one at a time and asked to describe the odour using his/her own words. A system of marking can be devised e.g. 4 points for absolutely correct, 3 points for correct in general terms, 2 points for a vague association, 1 point for a wrong association and 0 points for no response. A satisfactory level for selection of panellists needs to be specified in relation to the materials used. Similar techniques can be applied for taste and texture. The potential panellists may be screened for their ability to rank or rate products for selected attributes using the same technique as the final panel will use. All potential panellists are presented with the samples in the same order. Panellists are chosen if a satisfactory level is attained which will depend on the intensities of the samples used. Also check that they have used most of the scale.


81

Training In this phase, it is important that the panellists develop confidence as well as the skills for product assessment. Panellists must be taught the correct procedures for evaluating samples and ways to reduce or eliminate sensory adaptation. They must also learn to disregard their personal preferences. Between 40 and 120h of training are required for a descriptive sensory panel which will depend on the product, the number of attributes as well as the validity and reliability required. A trained panel usually consists of 10-20 panellists. The initial stage of training involves vocabulary development. The entire range of products is presented to the panellists. They are instructed to individually assess the sensory differences between the samples and record any differences as descriptive words. On completion of this task, the panellists each list the attributes used to describe each sample. At this time, it is very important that the panel leader does not lead or judge the descriptive words generated by the panellist although they can ask for clarification. The panellists themselves will usually start to move towards a general consensus once the total attribute list has been generated. It is then the role of the panel leader to provide reference standards for the attributes that have been previously selected by general panel consensus. The references can be used to help the panellists to identify and remember a sensory attribute found in the sample. The references may be chemicals, ingredients or products. The panellists then assess the samples alongside the references until a consensus is reached regarding the sensory attributes, reference standards and definitions. This process should continue until the panellists are all happy and understand the terms used. Towards the end of training, a scoresheet is created by the panellists. The panellists decide on the order in which the attributes are to be assessed. Generally the panel leader decides on the type of scale used, although the panel decides on the verbal anchors to be used. Once the panellists have become familiar with the samples, references and definitions, panel evaluation sessions are completed that should be similar to the final testing situation. The panellists are presented with coded samples in triplicate and asked to rate them using the scoresheets and attribute scales they have trained with. By statistically analysing the data, the panel leader will be able to determine if further training is required or if the evaluation phase can begin. Like any instrument, the performance of individual panellists as well as the panel as a whole needs to be monitored to check they are producing reliable results. Reliability is checked by completing test replications and the descriptive data obtained is analysed statistically using an analysis of variance. Motivation of panellists is one of the most important factors in maintaining an efficient trained sensory panel. If panellists are motivated and interested they will perform well. For panellists, a sense of completing meaningful work is an important source of motivation. When appropriate on completion of a project, feedback should be given to the panel as to the project objectives and outcomes and the contribution of the sensory results. Individual panellist feedback is also important. They should be made to feel that attendance at sensory evaluation sessions is important. This can be reinforced by running sessions strictly and efficiently to keep their time input to a minimum.


82

Throughout training as well as during ongoing sensory evaluation sessions, it is important to keep the channels of communications open through panel discussion at the completion of a training session or a sensory testing session. Ongoing records of panellists' training and experience are invaluable. In some instances training can occupy more time than the actual experimental testing sessions, especially when you first start. However, if the job is done correctly right from the start, your trained panel will be one of the most valuable resources in the company. Make sure you look after them. An aside: Expert panels Panellists who have a great deal of experience in assessing a particular product are often referred to as "Expert tasters". Commodities that utilise expert tasters include the tea, coffee, wine and dairy industries. These panels usually include only 2 or 3 highly trained tasters. These tasters are particularly sensitive to the nuances of a specific product. They also have the ability to carry the characteristics of standard samples in their sensory memory. It takes a great deal of practice to develop the skill and requires continued tasting to stay "tuned". They are usually responsible for arranging the tasting conditions and samples themselves, in addition to actually tasting and making a final report. This type of panel is most frequently used to assign a quality grade to a finished product, as in butter and cheese grading. In the wine and coffee industries one expert may use these skills to blend individual components to produce a final product with the desired characteristics.


83

REPORTING As with any other scientific experiment your sensory testing needs to be reported in a clear and concise manner. The Australian standards for each test type details what should be included in the report. The results obtained should be interpreted and conclusions drawn using all the information gathered in the experiment. Recommendations may also need to be included depending on the nature of the work. Remember that it is much easier to write the report if you keep a record as you go along!


84

SELECTED BIBLIOGRAPHY American Meat Science Association, “Guidelines for Cookery and Sensory Evaluation of Meat”, AMS, USA, 1978. Amerine, M A, Pangborn, R M and Roessler, E B, “Principles of Sensory Evaluation of Food”, New York: Academic Press, 1965. ASTM, “Manual on Sensory Testing Methods”, STP 434, Am. Soc. Test. Makr., Philadelphia, Pennsylvania, 1968. Aust, L B, Gacula, M C, Beard,S A and Washam, R W. “Degree of Difference Test Method in Sensory Evaluation of Heterogeneous Product Types. Journal of Food Science, 50: 511 – 513, 1985 Bartoshuk, L, “Separate worlds of taste” Psychology Today 14 (9): 48-57, 1980. Bartoshuk, L M, “The biological basis of food perception and acceptance” Food Quality & Preference 4: 21-32, 1993. Bourne, M C, “Food Texture and Viscosity: Concept and Measurement”, Academic Press Inc., California, 1982. Chi-Tang Ho, Manley, C H, “Flavor Measurement”, Marcel Dekker, Inc. 1993. Gacula M C., “Design and analysis of Sensory Optimization”, Food & Nutrition Press. 1993. Gacula, M C and Singh, J, “Statistical Methods in Food and Consumer Research, New York: Academic Press, 1984. Jellinek, G, “Sensory Evaluation of Food: Theory and Practice”, Chichester: Ellis Horwood; 1985. Kemsley, E.K., “Lies, damned lies and chemometrics.” Proceedings of Food Authenticity 96. Norwich, UK. 1-5, 1996. Lawless, H T, “Pepper potency and the forgotten flavour sense” Food Technology 43 (11): 52, 57-58, 1989. Lawless, H T & Heymann, H, “Sensory Evaluation of Food: Principles and Practices”, Chapman & Hall, New York, 1998. Lyman, B, “A Psychology of Food”, Van Nostrand Reinhold Co. Inc., New York, USA, 1989. Lyon, D H, Francombe, M A, Hasdell, T A and Lawson, K, (editors) “Guidelines for Sensory Analysis in Food Product Development and Quality Control”. Chapmann and Hall, London, UK, 1992. Martens, H. and Martens, M., “Multivariate analysis of quality: an introduction”, John Wiley & Sons: West Sussex, England. 445, 2001.


85

Martens, H. and Næs, T., “Multivariate calibration.” John Wiley and Sons: Chichester, UK. 419, 1989.McBride, R L, “The Bliss Point Factor”, Sun Books, Australia, 1990. McBride, R L, (editor), “Psychological Basis of Sensory Evaluation”, Elsevier Applied Science, London, UK, 1990. McRae, R, Robinson, R K & Sadler, M J (eds) “Encyclopedia of Food Science, Food Technology and Nutrition”, Volume 6, Academic Press, London, 1993. Meilgaard, M, Civille, G V and Carr, B T, “Sensory Evaluation Techniques: Boca Raton, Fla: CRC Press, 1999. (3rd Edition) Miflora Minoza-Gatchalian, “Sensory Evaluation Methods with Statistical Analysis (for Research Product Development and Quality Control)”. 1981. Moskowitz, H R, “New Directions for Product Testing and Sensory Analysis of Foods”, Food & Nutrition Press, Inc. 1985. Moskowitz, H, “Applied Sensory Analysis of Food”, Volumes 1 and 2, CRC Press, Florida, USA, 1988. Næs, T., Isaksson, T., Fearn, T., and Davies, T., “A user friendly guide to multivariate calibration and classification.” NIR publications: Chichester, UK. 344, 2002. O’Mahoney, M, “Sensory Evaluation of Food: Statistical Methods and Procedures”, New York: Marcel Dekker, Inc, 1986. O’Mahony, M & Ishii, I “Do you have an umami tooth?” Nutrtion Today May/June, 1985. Pecore, S., Stoer, N., Hooge, S., Holschuh, N., Hulting, F., and Case, F., “Degree of difference testing: A new approach incorporating control lot variability”, Food Quality and Preference 17 (2006) 552 – 555 Piggott, J R, “Sensory Analysis of Food”, London: Elsevier Applied Science, 1988 (2nd edition now available). Piggott, J R, “Statistical Procedures in Food Research”, London: Elsevier Applied Science, 1986. Piggott, J R, Paterson, A “Understanding Natural Flavors”. Blackie Academic & Professional. 1994. Poste, L M, Mackie, D A, Butter, G and Larmond, E, “Laboratory Methods for Sensory Analysis of Food”, Agriculture Canada Publication 1864/E, 1991. Rutledge, K P and Hudson, J M, “Sensory Evaluation: Method for Establishing and Training a Descriptive Flavour Panel, Food Technology 44 (12): 78-84, 1990. Stone, H and Sidel, J L, “Sensory Evaluation Practices”, 2nd edition, New York: Academic Press, 1992. Thomson, D M H, “Food Acceptability”, Elsevier Applied Science, London, UK, 1988.


86

JOURNALS Gacula, M C, “Journal of Sensory Studies”. Food & Nutrition Press, Inc. MacFie, H J., Meiselman, H L., “Food Quality and Preference”. Elsevier Applied Science

STATISTICAL TABLES

Table 1: Probability of X or More Correct Judgments in n Trials (one-tailed, p = 1/3)a n\x 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 5 868 539 210 045 004 6 912 649 320 100 018 001 7 941 737 429 173 045 007 8 961 805 532 259 088 020 003 9 974 857 623 350 145 042 008 001 10 983 896 701 441 213 077 020 003 11 988 925 766 527 289 122 039 009 001 12 992 946 819 607 368 178 066 019 004 001 13 995 961 861 678 448 241 104 035 009 002 14 997 973 895 739 524 310 149 058 017 004 001 15 998 981 921 791 596 382 203 088 031 008 002 16 998 986 941 834 661 453 263 126 050 016 004 001 17 999 990 956 890 719 522 326 172 075 027 008 002 18 999 993 967 898 769 588 391 223 108 043 014 004 001 19 995 976 921 812 648 457 279 146 065 024 007 002 20 997 982 940 848 703 521 339 191 092 038 013 004 001 21 998 987 954 879 751 581 399 240 125 056 021 007 002 22 998 991 965 904 794 638 460 293 163 079 033 012 003 001 23 999 993 974 924 831 690 519 349 206 107 048 019 006 002 24 999 995 980 941 862 737 576 406 254 140 068 028 010 003 001 25 999 996 985 954 888 778 630 462 304 178 092 042 016 006 002 26 997 989 964 910 815 679 518 357 220 121 058 025 009 003 001 27 998 992 972 928 847 725 572 411 266 154 079 036 014 055 002 28 999 994 979 943 874 765 623 464 314 191 104 050 022 008 003 001 29 999 996 984 955 897 801 670 517 364 232 133 068 031 013 005 001 30 999 997 988 965 916 833 714 568 415 276 166 090 043 019 007 002 001 31 998 991 972 932 861 754 617 466 322 203 115 059 027 011 004 001 32 998 993 978 946 885 789 662 516 370 243 144 078 038 016 066 002 001 33 999 995 983 957 905 821 705 565 419 285 177 100 051 023 010 004 001 34 999 996 987 965 922 849 744 612 468 330 213 126 067 033 014 006 002 001 35 999 997 990 973 937 873 779 656 516 376 252 155 087 044 020 009 003 001 36 998 992 978 949 895 810 697 562 422 293 187 109 058 028 012 005 002 001 47 998 994 963 959 913 838 735 607 469 336 223 135 075 038 018 007 003 001 38 999 996 987 967 928 863 769 650 515 381 261 164 095 051 025 011 004 002 001 39 999 997 990 973 941 885 800 689 560 425 301 196 118 066 033 016 007 003 001 40 999 997 992 979 952 903 829 726 603 470 32 231 144 083 044 021 010 004 001 41 998 994 983 961 920 854 761 644 515 385 268 173 104 057 029 014 006 002 001 42 999 995 987 968 933 876 791 683 558 428 307 205 127 073 038 019 008 003 001 43 999 996 990 974 945 895 820 719 600 471 347 239 153 091 050 025 012 005 002 001 44 999 997 992 980 955 912 845 753 639 514 389 275 182 111 063 033 016 007 003 001 45 999 998 994 984 963 926 867 783 677 556 430 313 213 135 079 043 022 010 004 002 001 46 998 995 987 970 938 887 811 713 596 472 352 246 161 098 055 029 014 006 003 001 47 999 996 990 976 949 904 836 745 635 514 392 282 189 119 070 038 019 009 004 002 001 48 999 997 992 980 958 919 859 776 672 554 433 318 220 142 086 048 025 012 006 002 001 49 999 998 994 984 965 932 879 803 706 593 473 356 253 168 105 061 033 017 008 003 001 50 999 998 995 987 972 943 896 829 739 631 513 395 287 196 126 076 042 022 011 005 002 001


87


88

Table 2: Minimum Numbers of Correct Judgments to Establish Significance at Various Probability Levels for the Triangle tests (one tailed, p = 1/3)

Probability Levels

No. of trials (n) 0.05 0.04 0.03 0.02 0.01 0.005 0.001

5 4 5 5 5 5 5

6 5 5 5 5 6 6

7 5 6 6 6 6 7 7

8 6 6 6 6 7 7 8

9 6 7 7 7 7 8 8

10 7 7 7 7 8 8 9

11 7 7 8 8 8 9 10

12 8 8 8 8 9 9 10

13 8 8 9 9 9 10 11

14 9 9 9 9 10 10 11

15 9 9 10 10 10 11 12

16 9 10 10 10 11 11 12

17 10 10 10 11 11 12 13

18 10 11 11 11 12 12 13

19 11 11 11 12 12 13 14

20 11 11 12 12 13 13 14

21 12 12 12 13 13 14 15

22 12 12 13 13 14 14 15

23 12 13 13 13 14 15 16

24 13 13 13 14 15 15 16

25 13 14 14 14 15 16 17

26 14 14 14 15 15 16 17

27 14 14 15 15 16 17 18

28 15 15 15 16 16 17 18

29 15 15 16 16 17 17 19

30 15 16 16 16 17 18 19

31 16 16 16 17 18 18 20

32 16 16 17 17 18 19 20

33 17 17 17 18 18 19 21

34 17 17 18 18 19 20 21

35 17 18 18 19 19 20 22

36 18 18 18 19 20 20 22

37 18 18 19 19 20 21 22

38 19 19 19 20 21 21 23

39 19 19 20 20 21 22 23

40 19 20 20 21 21 22 24

41 20 20 20 21 22 23 24

42 20 20 21 21 22 23 25

43 20 21 21 22 23 24 25

44 21 21 22 22 23 24 26

45 21 22 22 23 24 24 26

46 22 22 22 23 24 25 27

47 22 22 23 23 24 25 27

48 22 23 23 24 25 26 27

49 23 23 24 24 25 26 28

50 23 24 24 25 26 26 28

60 27 27 28 29 30 31 33

70 31 31 32 33 34 35 37

80 35 35 36 36 38 39 41

90 38 39 40 40 42 43 45


89

Table 3: Minimum Numbers of Correct Judgments to Establish Significance at Various Probability Levels for Paired - Comparison and Duo-Trio Tests (one-tailed, p=1/2)

Probability levels

No of trials (N) 0.05 0.04 0.03 0.02 0.01 0.005 0.001

7 7 7 7 7 7

8 7 7 8 8 8 8

9 8 8 8 8 9 9

10 9 9 9 9 10 10 10

11 9 9 10 10 10 11 11

12 10 10 10 10 11 11 12

13 10 11 11 11 12 12 13

14 11 11 11 12 12 13 13

15 12 12 12 12 13 13 14

16 12 12 13 13 14 14 15

17 13 13 13 14 14 15 16

18 13 14 14 14 15 15 16

19 14 14 15 15 15 16 17

20 15 15 15 16 16 17 18

21 15 15 16 16 17 17 18

22 16 16 16 17 17 15 19

23 16 17 17 17 18 19 20

24 17 17 18 18 19 19 20

25 18 15 18 19 19 20 21

26 18 18 19 19 20 20 22

27 19 19 19 20 20 21 22

28 19 20 20 20 21 22 23

29 20 20 21 21 22 22 24

30 20 21 21 22 22 23 24

31 21 21 22 22 23 24 25

32 22 22 22 23 24 24 26

33 22 23 23 23 24 25 26

34 23 23 23 24 25 25 27

35 23 24 24 25 25 26 27

36 24 24 25 25 26 27 28

37 24 25 25 26 26 27 29

38 25 25 26 26 27 28 29

39 26 26 26 27 28 28 30

40 26 27 27 27 28 29 30

41 27 27 27 28 29 30 31

42 27 28 28 29 29 30 32

43 28 28 29 29 30 31 32

44 28 29 29 30 31 31 33

45 29 29 30 30 31 32 34

46 30 30 30 31 32 33 34

47 30 30 31 31 32 33 35

48 31 31 31 32 33 34 36

49 31 32 32 33 34 34 36

50 32 32 33 33 34 35 37

60 37 38 38 39 40 41 43

70 43 43 44 45 46 47 49

80 48 49 49 50 51 52 55

90 54 54 55 56 57 58 61

100 59 60 60 61 63 64 66

Source : .E.B .Roessler et al.. Journal of Food Science, 1978, 43, 940-947


90

Table 4: Minimum Numbers of Agreeing Judgements Necessary to Establish Significance at Various Probability Levels for the Paired – Preference Tests and Difference (two tailed, p=1/2).

Probability Levels

No. of trials (n) 0.05 0.04 0.03 0.02 0.01 0.005 0.001

7 7 7 7 7

8 8 8 8 8 8

9 8 8 9 9 9 9

10 9 9 9 10 10 10

11 10 10 10 10 11 11 11

12 10 10 11 11 11 12 12

13 11 11 11 12 12 12 13

14 12 12 12 12 13 13 14

15 12 12 13 13 13 14 14

16 13 13 13 14 14 14 15

17 13 14 14 14 15 15 16

18 14 14 15 15 15 16 17

19 15 15 15 15 16 16 17

20 15 16 16 16 17 17 18

21 16 16 16 17 17 18 19

22 17 17 17 17 18 18 19

23 17 17 18 18 19 19 20

24 18 18 18 19 19 20 21

25 18 19 19 19 20 20 21

26 19 19 19 20 20 21 22

27 20 20 20 20 21 22 23

28 20 20 21 21 22 22 23

29 21 21 21 22 22 23 24

30 21 22 22 22 23 24 25

31 22 22 22 23 24 24 25

32 23 23 23 23 24 25 26

33 23 23 24 24 25 25 27

34 24 24 24 25 25 26 27

35 24 25 25 25 26 27 28

36 25 25 25 26 27 27 29

37 25 26 26 26 27 28 29

38 26 26 27 27 28 29 30

39 27 27 27 28 28 29 31

40 27 27 28 28 29 30 31

41 28 28 28 29 30 30 32

42 28 29 29 29 30 31 32

43 29 29 30 30 31 32 33

44 29 30 30 30 31 32 34

45 30 30 31 31 32 33 34

46 31 31 31 32 33 33 35

47 31 31 32 32 33 34 36

48 32 32 32 33 34 35 36

49 32 33 33 34 34 35 37

50 33 33 34 34 35 36 37

60 39 39 39 40 41 42 44

70 44 45 45 46 47 48 50

80 50 50 51 51 52 53 56

90 55 56 56 57 58 59 61

100 61 61 62 63 64 65 67


91

Table 5a 5 % Points for the Distribution of F n2\n1 1 2 3 4 5 6 8 12 24 8 1 161.40 199.50 215.70 224.60 230.20 234.00 238.90 243.90 249.00 254.30 2 18.51 19.00 19.16 19.25 19.30 19.33 19.37 19.41 19.45 19.50 3 10.13 9.55 9.28 9.12 9.01 8.94 8.84 8.74 8.64 8.53 4 7.71 6.94 6.59 6.39 6.26 6.16 6.04 5.91 5.77 5.63 5 6.61 5.79 5.41 5.19 5.05 4.95 4.82 4.68 4.53 4.36 6 5.99 5.14 4.76 4.53 4.39 4.28 4.15 4.00 3.84 3.67 7 5.59 4.74 4.35 4.12 3.97 3.87 3.73 3.57 3.41 3.23 8 5.32 4.46 4.07 3.84 3.69 3.58 3.44 3.28 3.12 2.93 9 5.12 4.26 3.86 3.63 3.48 3.37 3.23 3.07 2.90 2.71 10 4.96 4.10 3.71 3.48 3.33 3.22 3.07 2.91 2.74 2.54 11 4.84 3.98 3.59 3.36 3.20 3.09 2.95 2.79 2.61 2.40 12 4.75 3.88 3.49 3.26 3.11 3.00 2.85 2.69 2.50 2.30 13 4.67 3.80 3.41 3.18 3.02 2.92 2.77 2.60 2.42 2.21 14 4.60 3.74 3.34 3.11 2.96 2.85 2.70 2.53 2.35 2.13 15 4.54 3.68 3.29 3.06 2.90 2.79 2.64 2.48 2.29 2.07 16 4.49 3.63 3.24 3.01 2.85 2.74 2.59 2.42 2.24 2.01 17 4.45 3.59 3.20 2.96 2.81 2.70 2.55 2.38 2.19 1.96 18 4.41 3.55 3.16 2.93 2.77 2.66 2.51 2.34 2.15 1.92 19 4.38 3.52 3.13 2.90 2.74 2.63 2.48 2.31 2.11 1.88 20 4.35 3.49 3.10 2.87 2.71 2.60 2.45 2.28 2.08 1.84 21 4.32 3.47 3.07 2.84 2.68 2.57 2.42 2.25 2.05 1.81 22 4.30 3.44 3.05 2.82 2.66 2.55 2.40 2.23 2.06 1.78 23 4.28 3.42 3.03 2.80 2.64 2.53 2.38 2.20 2.00 1.76 24 4.26 3.40 3.01 2.78 2.62 2.51 2.36 2.18 1.98 1.73 25 4.24 3.38 2.99 2.76 2.60 2.49 2.34 2.16 1.96 1.71 26 4.22 3.37 2.98 2.74 2.59 2.47 2.32 2.15 1.95 1.69 27 4.21 3.35 2.96 2.73 2.57 2.46 2.30 2.13 1.93 1.67 28 4.20 3.34 2.95 2.71 2.56 2.44 2.29 2.12 1.91 1.65 29 4.18 3.33 2.93 2.70 2.54 2.43 2.28 2.40 1.90 1.64 30 4.17 3.32 2.92 2.69 2.53 2.42 2.27 2.09 1.89 1.62 40 4.08 3.23 2.84 2.61 2.45 2.34 2.18 2.00 1.79 1.51 60 4.00 3.15 2.76 2.52 2.37 2.25 2.10 1.92 1.70 1.39 120 3.92 3.07 2.68 2.45 2.29 2.17 2.02 1.83 1.61 1.25 8 3.84 2.99 2.60 2.37 2.21 2.09 1.94 1.75 1.52 1.00

Source : Table 9 is taken from Table V of Fisher and Yates : 1974 Statistical Tables for Biological, Agricultural and Medical Research published by Longman Group UK Ltd. London (previously published by Oliver and Boyd Ltd. Edinburgh) and by permission of the authors and publishers.


92

Table 5b 1 % Points for the Distribution of F n2\n1 1 2 3 4 5 6 8 12 24 8 1 4052 4999 5403 5625 5764 5859 5981 6106 6234 6366 2 98.49 99.00 99.17 99.25 99.30 99.33 99.36 99.42 99.46 99.50 3 34.12 30.81 29.46 28.71 28.24 27.91 27.49 27.05 26.60 26.12 4 21.20 18.00 16.69 15.98 15.52 15.21 14.80 14.37 13.93 13.46 5 16.46 13.27 12.06 11.39 10.97 10.67 10.29 9.89 9.47 9.02 6 13.74 10.92 9.78 9.15 8.75 8.47 8.10 7.72 7.31 6.88 7 12.25 9.55 8.45 7.85 7.46 7.19 6.84 6.47 6.07 5.65 8 11.26 8.65 7.59 7.01 6.63 6.37 6.03 5.67 5.28 4.86 9 10.56 8.02 6.99 6.42 6.06 5.80 4.47 5.11 4.73 4.31 10 10.04 7.56 6.55 5.99 5.64 5.39 5.06 4.71 4.33 3.91 11 9.65 7.20 6.22 5.67 5.32 5.07 4.74 4.40 4.02 3.60 12 9.33 6.93 5.95 5.41 5.06 4.82 4.50 4.16 3.78 3.36 13 9.07 6.70 5.74 5.20 4.86 4.62 4.30 3.96 3.59 3.16 14 8.86 6.51 5.56 5.03 4.69 4.46 4.14 3.80 3.43 3.00 15 8.68 6.36 5.42 4.89 4.56 4.32 4.00 3.67 3.29 2.87 16 8.53 6.23 5.29 4.77 4.44 4.20 3.89 3.55 3.18 2.75 17 8.40 6.11 5.18 4.67 4.34 4.10 3.79 3.45 3.08 2.65 18 8.28 6.01 5.09 4.58 4.25 4.01 3.71 3.37 3.00 2.57 19 8.18 5.93 5.01 4.50 4.17 3.94 3.63 3.30 2.92 2.49 20 8.10 5.85 4.94 4.43 4.10 3.87 3.56 3.23 2.86 2.42 21 8.02 5.78 4.87 4.37 4.04 3.81 3.51 3.17 2.80 2.36 22 7.94 5.72 4.82 4.31 3.99 3.76 3.45 3.12 2.75 2.31 23 7.88 5.66 4.76 4.26 3.94 3.71 3.41 3.07 2.70 2.26 24 7.82 5.61 4.72 4.22 3.90 3.67 3.36 3.03 2.66 2.21 25 7.77 5.57 4.68 4.18 3.86 3.63 3.32 2.99 2.62 2.17 26 7.72 5.53 4.64 4.14 3.82 3.59 3.29 2.96 2.58 2.13 27 7.68 5.49 4.60 4.11 3.78 3.56 3.26 2.93 2.55 2.10 28 7.64 5.45 4.57 4.07 3.75 3.53 3.23 2.90 2.52 2.06 29 7.60 5.42 4.54 4.04 3.73 3.50 3.20 2.87 2.49 2.06 30 7.56 5.39 4.51 4.02 3.70 3.47 3.17 2.84 2.47 2.01 40 7.31 5.18 4.31 3.83 3.51 3.29 2.99 2.66 2.29 1.80 60 7.08 4.98 4.13 3.65 3.34 3.12 2.82 2.50 2.12 1.60 120 6.85 4.79 3.95 3.48 3.17 2.96 2.66 2.34 1.95 1.38 8 6.64 4.60 3.78 3.32 3.02 2.80 2.51 2.18 1.79 1.00 Source : Table 9 is taken from Table V of Fisher and Yates : 1974 Statistical Tables for Biological, Agricultural and Medical Research published by Longman Group UK Ltd. London (previously published by Oliver and Boyd Ltd. Edinburgh) and by permission of the authors and publishers.


93

Table 6: Critical Value of ta Level of significance for one-tailed test 0.1 0.05 0.025 0.01 0.005 0.0005 Level of significance for two-tailed test df 0.2 0.1 0.05 0.02 0.01 0.001 1 3.078 6.314 12.706 31.821 63.657 636.619 2 1.886 2.92 4.303 6.965 9.925 31.598 3 1.638 2.353 3.182 4.541 5.841 12.941 4 1.533 2.132 2.776 3.747 4.604 8.61 5 1.476 2.015 2.571 3.365 4.032 6.859 6 1.44 1.943 2.447 3.143 3.707 5.959 7 1.415 1.895 2.365 2.998 3.499 5.405 8 1.397 1.86 2.306 2.896 3.355 5.041 9 1.383 1.833 2.262 2.821 3.25 4.781 10 1.372 1.812 2.228 2.764 3.169 4.587 11 1.363 1.796 2.201 2.718 3.106 4.437 12 1.356 1.782 2.179 2.681 3.055 4.318 13 1.35 1.771 2.16 2.63 3.012 4.221 14 1.345 1.761 2.145 2.624 2.977 4.14 15 1.341 1.753 2.131 2.602 2.947 4.073 16 1.337 1.746 2.12 2.583 2.921 4.015 17 1.333 1.74 2.11 2.567 2.898 3.965 18 1.33 1.734 2.101 2.552 2.878 3.922 19 1.328 1.729 2.093 2.539 2.861 3.883 20 1.325 1.725 2.086 2.528 2.845 3.85 21 1.323 1.721 2.08 2.518 2.831 3.819 22 1.321 1.717 2.074 2.508 2.819 3.792 23 1.319 1.714 2.069 2.5 2.807 3.767 24 1.318 1.711 2.064 2.492 2.797 3.745 25 1.316 1.708 2.06 2.485 2.787 3.725 26 1.315 1.706 2.056 2.479 2.779 3.707 27 1.314 1.703 2.052 2.473 2.771 3.69 28 1.313 1.701 2.048 2.467 2.763 3.674 29 1.311 1.699 2.045 2.462 2.756 3.659 30 1.31 1.697 2.042 2.457 2.75 3.646 40 1.303 1.684 2.021 2.423 2.704 3.551 60 1.296 1.671 2 2.39 2.66 3.46 120 1.289 1.658 1.98 2.358 2.617 3.373 00 1.282 1.645 1.96 2.326 2.576 3.2 aThe value listed in the table is the critical value of t for the number of degrees of freedom listed in the left column for a one- or two-tailed test at the significance level indicated at the top of each column. If the observed t is greater than or equal to the tables value, reject Ho. Source: Table III of Fisher and Yates, Statistical Tables for Biological, Agricultural and Medical Research, published by Longman Group Ltd, London (previously published by Oliver and Boyd Ltd, Edinburgh) and by permission of the authors and publishers.


94

Table 7: Critical Values of Chi-Squarea Level of significance for one-tailed test 0.10 0.05 0.025 0.01 0.005 0.0005 Level of significance for two-tailed test df 0.2 0.1 0.05 0.02 0.01 0.001 1 1.64 2.71 3.84 5.41 6.64 10.83 2 3.22 4.6 5.99 7.82 9.21 13.82 3 4.64 6.25 7.82 9.84 11.34 16.27 4 5.99 7.78 9.49 11.67 13.28 18.46 5 7.29 9.24 11.07 13.39 15.09 20.52 6 8.56 10.64 12.59 15.03 16.81 22.46 7 9.8 12.02 14.07 16.62 18.48 24.32 8 11.03 13.36 15.51 18.17 20.09 26.12 9 12.24 14.68 16.92 19.68 21.67 27.88 10 13.44 15.99 18.31 21.16 23.21 29.59 11 14.63 17.28 19.68 22.62 24.72 31.26 12 15.81 18.55 21.03 24.05 26.22 32.91 13 16.98 19.81 22.36 25.47 27.69 34.53 14 18.15 21.06 23.68 26.87 29.14 36.12 15 19.31 22.31 25 28.26 30.58 37.7 16 20.46 23.54 26.3 29.63 32 39.29 17 21.62 24.77 27.59 31 33.41 40.75 18 22.76 25.99 28.87 32.35 34.8 42.31 19 23.9 27.2 30.14 33.69 36.19 43.82 20 25.04 28.41 31.41 35.02 37.57 45.32 21 26.17 29.62 32.67 36.34 38.93 46.8 22 27.3 30.81 33.92 37.66 40.29 48.27 23 28.43 32.01 35.17 38.97 41.64 49.73 24 29.55 33.2 36.42 40.27 42.98 51.18 25 30.68 34.38 37.65 41.57 44.31 62.62 26 31.8 35.56 38.88 42.86 45.64 54.05 27 32.91 36.74 40.11 44.14 46.96 55.48 28 34.03 37.92 41.34 45.42 48.28 56.89 29 35.14 39.09 42.69 46.69 49.59 58.3 30 36.25 40.26 43.77 47.96 50.89 59.7 32 38.47 42.59 46.19 50.49 53.49 62.49 34 40.68 44.9 48.6 53 56.06 65.25 36 42.88 47.21 51 55.49 58.62 67.99 38 45.08 49.51 53.38 57.97 61.16 70.7 40 47.27 51.81 55.76 60.44 63.69 73.4 44 51.64 56.37 60.48 65.34 68.71 78.75 48 55.99 60.91 65.17 70.2 73.68 84.04 52 60.33 65.42 69.83 75.02 78.62 89.27 56 64.66 69.92 74.47 79.82 83.51 94.46 60 68.97 74.4 79.08 84.58 88.38 99.61

aThe table lists the critical values of chi square for the degrees of freedom shown at the left for tests corresponding to those significance levels heading each column. If the observed value of xobs

2 is greater than or equal to the tabled value, reject Ho. Source: Table IV of Fisher and Yates, Statistical Tables for Biological, Agricultural and Medical Research, published by Longman Group Ltd, London (previously published by Oliver and Boyd Ltd, Edinburgh) and by permission of the authors and publishers.

Table 8: Areas in the Tail of the Normal Distribution

x

.00

.01

.02

.03

.04

.05

.06

.07

.08

.09

0.0 .5000 .4960 .4920 .4880 .4840 .4801 .4761 .4721 .4681 .4641 0.1 .4602 .4562 .4522 .4483 .4443 .4404 .4364 .4325 .4286 .4247 0.2 .4207 .4168 .4129 .4090 .4052 .4013 .3974 .3936 .3897 .3859 0.3 .3821 .3783 .3745 .3707 .3669 .3632 .3594 .3557 .3520 .3483 0.4 .3446 .3409 .3372 .3336 .3300 .3264 .3228 .3192 .3156 .3121

0.5

.3085

.3050

.3015

.2981

.2946

.2912

.2877

.2843

.2810

.2776

0.6 .2743 .2709 .2676 .2643 .2611 .2578 .2546 .2514 .2483 .2451 0.7 .2420 .2389 .2358 .2327 .2296 .2266 .2236 .2206 .2177 .2148 0.8 .2119 .2090 .2061 .2033 .2005 .1977 .1949 .1922 .1894 .1867 0.9 .1841 .1814 .1788 .1762 .1736 .1711 .1685 .1660 .1635 .1611

1.0

.1587

.1562

.1539

.1515

.1492

.1469

.1446

.1423

.1401

.1379

1.1 .1357 .1335 .1314 .1292 .1271 .1251 .1230 .1210 .1190 .1170 1.2 .1151 .1131 .1112 .1093 .1075 .1056 .1038 .1020 .1003 .0985 1.3 .0968 .0951 .0934 .0918 .0901 .0885 .0869 .0853 .0838 .0823 1.4 .0808 .0793 .0778 .0764 .0749 .0735 .0721 .0708 .0694 .0681

1.5

.0668

.0655

.0643

.0630

.0618

.0606

.0594

.0582

.0571

.0559

1.6 .0548 .0537 .0526 .0516 .0505 .0495 .0485 .0475 .0465 .0455 1.7 .0446 .0436 .0427 .0418 .0409 .0401 .0392 .0384 .0375 .0367 1.8 .0359 .0351 .0344 .0336 .0329 .0322 .0314 .0307 .0301 .0294 1.9 .0287 .0281 .0274 .0268 .0262 .0256 .0250 .0244 .0239 .0233

2.0

.02275

.02222

.02169

.02118

.02068

.02018

.01970

.01923

.01876

.01831

2.1 .01786 .01743 .01700 .01659 .01618 .01578 .01539 .01500 .01463 .01426 2.2 .01390 .01355 .01321 .01287 .01255 .01222 .01191 .01160 .01130 .01101 2.3 .01072 .01044 .01017 .00990 .00964 .00939 .00914 .00889 .00866 .00842 2.4 .00820 .00798 .00776 .00755 .00734 .00714 .00695 .00676 .00657 .00639

2.5

.00621

.00604

.00587

.00570

.00554

.00539

.00523

.00508

.00494

.00480

2.6 .00466 .00453 .00440 .00427 .00415 .00402 .00391 .00379 .00368 .00357 2.7 .00347 .00336 .00326 .00317 .00307 .00298 .00289 .00280 .00272 .00264 2.8 .00256 .00248 .00240 .00233 .00226 .00219 .00212 .00205 .00199 .00193 2.9 .00187 .00181 .00175 .00169 .00164 .00159 .00154 .00149 .00144 .00139

3.0

.00135

3.1 .00097 3.2 .00069 3.3 .00048 3.4 .00034

3.5

.00023

3.6 .00016 3.7 .00011 3.8 .00007 3.9 .00005

4.0

.00003

Source: Table III of Murdoch and Barnes, Statistical Tables for Science, Engineering, Management and Business Studies, published by Macmillan Press Ltd and by permission of the authors and publishers.


95