| |
:: Course SummaryThe assessment of the performance of a panel deals with several notions : repeatability, agreement, usage of scales and discriminating power. Differents statistical tools can be used to quantify these measures of panel performance.:: Learning ObjectivesUpon completion of this training course, participants will be able to:
Know what the panel performance criteria are
Differentiate between the statistical tools relevant for measuring each criteria
Understand how to appropriately format the data for each analysis
Conduct a sensory panel performance study
Interpret the results
Extract the pertinent and relevant information and report the results
Take appropriate action to correct any problems detected:: Target AudienceProfessionals using descriptive sensory data in their work and who want to learn practical ways to measure the performance of sensory panelists.:: PrerequisiteKnowledge of basic principles of descriptive analysis is recommended. Moreover, working knowledge of ANOVA and multivariate analysis is also recommended. | | |
:: Topics Covered
- Definition of Performance Measurement Criteria
- Repeatability
- Agreement
- Scale Usage
- Discrimination Ability
- Graphical Exploratory Tools and Descriptive Statistics
- Use of ANOVA to Measure Panel Performance
- Principle
- Why, How and When to Use ANOVA
- Quantifying Performance Criteria
- Use of a Nonparametric Test to Measure Panel Performance: Friedman’s Test
- Principle
- What are the Performance Criteria that Can Be Measured
- Why, How and When to Use Friedman’s Test
- Use of the Principal Component Analysis (PCA) to Visually Assess Performance
- Needed Data and Layout
- Determination of Variability
- Numeric and Graphical Outputs
- What are the Performance Criteria that Can Be Measured
- Why, How and When to Use PCA
- Case Studies: Conducting a Panel Performance Study from Beginning to End
- Choice of Methods
- Interpreting Results
- Communicating Results
- Summary
:: Course ContentAs with any analytical instrument, the descriptive panel should provide precise and accurate data. The sensory community has not formally defined these two notions, although many authors have published papers on the topic, but with some discrepancies as to the interpretation of these terms.
Various statistical approaches can be used to measure the performance of judges. Panel leaders often wish to analyze the data collected at the time the sensory tests are performed in order to evaluate the performance of the judges and to apply any corrective measures if necessary. The three main criteria in evaluating the performance of a sensory panel are: Panel Homogeneity, Discrimination Power, and Panel Repeatability.
The first step in any data analysis is to explore the data by means of descriptive statistics in order to get a "feel" for the data as well as identify any possible errors or problems. Next, the objectives of evaluation and the performance criteria must be specified to determine which statistical method should be used.
This training course reviews the definition of each performance criteria and presents the common tools to explore data, i.e. descriptive statistics, both numeric and graphical. Two methods to evaluate panel performance are detailed: ANOVA (Analysis of Variance) and PCA (Principal Component Analysis). A nonparametric test, the Friedman’s Test, is also explained to offer sensory analysts with an alternative when analyzing rank data or scores. The principles, software output and application of these three statistical methods for panel performance are outlined. Case studies are provided to give participants a concrete application of the material presented.
Moreover, case studies are used to show participants how to perform a panel performance study from beginning to end.
| |
