PPT On CHOOSING AND TRAINING PANELISTS

Download

CHOOSING AND TRAINING PANELISTS Presentation Transcript:
1.CHOOSING AND TRAINING PANELISTS
The panel – analytical “tool”
Need to assess the ability of panelist to reproduce judgements
Inconsistent judgements –
Interest
Motivation
General attitude
Emotional state of panelist

2.Panelist – office, plant or research staff
Part of work routine for personnel
Full cooperation of supervisors of panelist
No need to evaluate if objects
The higher no. semitrained panel, higher individual variations balance out
Small, highly trained panel give reliable result than large untrained panel
Min 10 – 20 persons with three or four replications by each judge for each treatment

3.Panelist – good health, absent if conditions might interfere with normal function of taste and smell
Emotional factor, interest, and motivation more important than age,sex
General guideline – refrain from smoking, chewing gum, eating or drinking for at least 30 min
Motivation of panelist affects response –
Interested – more efficient
Make it an important activity, their contribution important as well
How?? Running the test in a controlled, efficient manner 

4.Selection of panelist based on treshold tests is unreliable (Amerine et al. 1965)
Better – selection based on ability to detect differences in the food to be tasted
Start with a large group
Rank them according to their ability to discriminate among samples prepared
Panelist – inherent sensitivity to the characteristic being evaluated and should be able to duplicate judgments
New panel – each product, because person who discriminate well on some products often and not on others
Disregard their personal preferences

5.DESIGNING EXPERIMENTS AND CHOOSING METHODS OF ANALYZING DATA
All experiment – simple mathematical model
Variability in results of sensory tests – interpretation by statistical methods
Results expressed in degrees of significance, which is the probability that the results are caused by chance.

6.FACTORS INFLUENCING SENSORY MEASUREMENT
Expectation error
Stimulus error
Logical error
Halo effect
Suggestion
Motivation
Contrast effect
Positional bias

8.EXPECTATION ERROR
Any information the panelists receive about the test will influence the results
Preconceived impression
Samples –coded so that panelist can’t identify
Code – should not introduce bias
Use 3 digit random numbers

9.STIMULUS ERROR
Desire to be right – judgment of panelist may be influenced by irrelevant characteristics of samples
Because of stimulus error, all samples must be as uniform as possible
Mask unwanted differences

10.LOGICAL ERROR
Closely associated with stimulus error
Causes the panelist to assign ratings to particular characteristics because they appear to him to be logically associated with other characteristics
Eg – slight yellow colour in dehydrated potatoes may indicate oxidation to the panelist and he will find different flavour in sample
Control by uniformity and mask differences

11.HALO EFFECT
When more than one factor in a sample is evaluated
The panelist often forms a general impression of a product and if asked to evaluate it for odour, texture, colour and taste at the same time, results may be different than if evaluate individually
Eliminate this effect by evaluating one characteristics at a time

12.SUGGESTION
The response of a panelist can influence reaction of other panelists
Separate panelists – no conversation and discussion
Testing area – free from noise and distraction and separate from the preparation area

13.MOTIVATION
Affects his sensory perception
Interested – more efficient
How to maintain – giving them reports of their results
Trained panelist – more motivated
Panelists – made to feel important
How – running the tests in a controlled efficient manner

14.CONTRAST EFFECT
The presentation of a sample of good quality just before one of poor quality cause the panelist to rate the second sample lower than it would normally be rated, vice versa.
The order of presentation of the samples should be randomized for each panelist so that contrast effect will be equalized

15.POSITIONAL BIAS
Particularly in the triangle test
When very small differences are found, panelist have a tendency to choose the middle sample as being different.
Can be eliminated by random presentation