In summary, panel performance monitoring involves assessing individual panelists' abilities,
agreement within the panel, and the overall reliability of sensory panels using a range of statistical
techniques, including ANOVA and PCA. These methods help ensure the quality and consistency of
sensory evaluations over time.
W4
Descriptive analysis
Clip | Descriptive analysis - Sensory spectrum
Based on the Texture Profile method
Principles:
Standardized attributes and response scales – standardized references are used that
represent specific intensities on the scale – intensity ratings can be treated as absolute scores
and are equal across attributes (25 for sweet taste is equal to 25 for salt taste) – makes the
ratings for products comparable over time and make ratings from different studies
comparable (equal PCA).
High involvement of panel leader
Limitations:
High degree of panelist training
More time-consuming than consensus training using own vocabulary
Clip | Descriptive analysis - Rapid sensory profiling methods
For quick descriptive profiling – relatively new
Verbal-based methods:
Flash profile
Use experiences/expert subjects (n=5-10)
Each receive a total set of products and generate their own individual attributes based on
their perceived differences between the products
, After they made their attribute list, they rank the products based on the intensity of each
attribute
Ordinal data (because of ranking)
Advantage: repetition – the ranking has to be repeated for all the products and attributes
– quality check for the reliability of the data
Disadvantage: interpretation of the attributes can be difficult, because of the large
number of attributes and the lack of definitions in the individual attribute list.
Check al that apply (CATA)
Consumer/expert panel
Prepared list of attributes by the panelleader
Tick attributes that apply, repeat per product
Frequency data
Both:
Allows panelists to use their own set of attributes/vocabulary
No training on attributes/scaling necessary
Similarity-based methods:
Sorting
Group products together with similar characteristics
Free to decide the products per group and number of products per group
Option: add terms to group (labelled sorting) – group 1: sour, group 2: fruity etc.
Data: similarity matrix
Napping
Quick screening of many samples
Place products on a ‘napkin’ according to differences and similarities – products more
similar are placed close together than products that are perceived more different.
Data: x, y coordinates for each product
, Varieties:
o Ultra flash profile (NOT flash profile)
Add terms to product map to describe the differences and similarities between
the products
o Partial napping
Focus on one modality (at the time) – more detail is added, because each
napping session is only focused on one sensory modality at the time. First, the
products are placed based on similarity and differences in appearance, next in
smell, than taste etc.
More time needed to perform the test and analyze the data
Concern on lack of training – lack of information about the differences between the
products, especially when the differences are small. They researched this and trained a
panel in napping – better product discrimination after more training, especially with
producttraining. A choice for the time you want to invest and the accuracy you want in
the results.
Both:
Based on overall differences between products
Comparison of some rapid methods:
Conventional descriptive profiling needs more training – leads to more accurate results, but
more time consuming – time investment leads to more accurate results.
When we look at the differences between global napping and partial napping – simple
adjustments can make already a difference.
,
