pMSE Plot Analysis

pMSE Plot Analysis

Purpose

To visualise the relationships between panellist discrimination, repeatability and agreement.

Data Format

  1. See the profiling dataset.
  2. The attributes should be of scale or interval type.

Background

This uses panel level ANOVA models:

Attribute = Product + Judge + Product:Judge + Residuals

And panellist level ANOVA models for each panellist:

Attribute = Product + Residuals

Options

  1. Number of Decimals for P-Values: The number of decimals places to round p-values to.
  2. Anonymise Assessors? Choose to replace the assessor names or not. There are options for randomly generated names or names from the assessor metadata.
  3. Anonymise Products? Choose to replace the product names or not. There are options for randomly generated names or names from the product metadata.
  4. Anonymise Attributes? Choose to replace the attribute names or not. There are options for randomly generated names or names from the attribute metadata.

Results and Interpretation

  1. The panellist level discrimination p-values are the product effect p-values in the associated panellist level ANOVA.
  2. The panellist level repeatability p-values are calculated from F-tests of the panellist residual mean square in the associated ANOVA against the residual mean square from the panel level ANOVA.
  3. The panellist level agreement is calculated from F-tests. We decompose the interaction mean square in the panel level ANOVA by assessor, then test for the appropriate assessor against the residual mean square from the panel level ANOVA.
  4. P VALUES: This is a list of tables, one for each attribute where each table shows the discrimination, repeatability and agreement p-values for each assessor.
  5. Plots: All 3 combinations of plots for discrimination, repeatability and agreement are included, with the positions of each panellist plotted. The regions which are commonly considered “Good” are highlighted in green, while regions commonly considered “Bad” are highlighted in red.
  6. Information: Warnings or information on the analysis, for example if there was not enough data to calculate the terms for an assessor then this will be noted here. 

Technical Information

  1. R packages: this uses the car package for ANOVA with type II sums of squares.