Psychological Errors in Sensory testing and Ways to Reduce Them
Central Tendency:
Subjects rate samples using the middle point of scale and avoid using the extreme ends.
Reasons: Subjects are afraid to use the ends because these may be a sample that has higher /lower intensity than the sample that was just tested.
Antidotes: Use scales that have less defined endpoints. Use subjects that are familiar with the tested samples.
Expectation Error:
Subjects rate samples according to their expectation based on previous knowledge of the product information.
Reasons: Subjects try to make the right answer from the knowledge about product or the variables that direct interest in the study.
Antidotes: Not giving any information concerning with the test and samples. Order of presentation should be randomized for each subject and assigned unbiased codes for each sample.
Halo effect:
Subjects rate the same attributes when they appear in a series of questions differently from when they were asked separately.
Reasons: This error occurs when there are more than one attributes including in the test, especially with the study that have both acceptance and intensity questions. Subjects try to rate intensity attributes to match with their liking.
Antidotes: Randomize the order of attributes, separate intensity and acceptance attributes, provide training if appropriate.
Error of Habituation:
Subjects provide the same response to a series of products that might be slightly different from time to time. As a result, the test might not be able to capture any different or trend.
Reasons: Subjects become complacent and start evaluating using a routine rather than concentrating on each sample.
Antidotes: Varying the types of samples, make sure panelists know you are tracking performance; add a modified product into the test.
Stimulus Error:
Subjects rate samples according to the other stimulus and not on their perception from the samples.
Reasons: Subjects use prior knowledge about products to “get it right” or because they could expect a particular characteristic or intensity based on a psychological or physical stimulus such as package (color, style, type of package) or defects or characteristic from a different preparation method.
Antidotes: Provide as little information to panelists as possible. Remove packaging or other “clues” whenever possible and appropriate. Never include the person who directly prepares or formulates the test as a subject in the study. Avoid leaving any indication that will lead to product identification or information.
Logical Error:
Subjects rate attributes logically on how the attributes are associated.
Reasons: Subjects relate two or more attributes to each other. For example in case of chocolate cake, subject relates the dark color with chocolate flavor and rate the darker color with the higher intensity of chocolate flavor.
Antidotes: Keeping the sample uniform by masking the samples. Using doctored samples to exercise trained panels.
Contrast Error:
Subjects rate the differences between samples greater (or lesser) than they actually are.
Reasons: Occurs when panelists “compare” (even without realizing they are comparing) samples, which can tend to exaggerate the magnitude of difference in subject mind in the mind of the subject. For example, a poorer quality sample was followed a higher quality sample the score for the higher quality sample may be artificially high.
Antidotes: Error can be minimized by using balanced and randomized sample presentation order and may be reduced to some extent by training.
1 Chambers, E. IV. and Wolf, M.B. 1996. (Eds.) Sensory Testing Methods, 2nd Edition. ASTM, West Conshohocken. PA.
2 Meilgaard, D., Civille, G.V. and Carr, B. 2007 Sensory Evaluation Techniques, 4th edition. CRC Press, Boca Raton, FL.
3 Stone, H., Sidel, J. L. 2004. Sensory Evaluation Practices. 3rd edition. Academic Press, Inc., San Diego, CA.
