RedJade's Descriptive Analysis output contains detailed information regarding your data and panelists. This file can be daunting at first glance, but is designed to give you an in-depth look at your data. A brief overview of the output is outlined below.


The file should be viewed from left to right by tab, starting with the "CoverSheet" and is organized from summary to more detailed information as you go further through the tabs.



CoverSheet

  • Provides an overview of your entire dataset including number of attributes, panelists and subjects, post hoc method selected, the scale used, number of repetitions, etc.
  • Verify that the data you expected to see is what you see here.


AnalysisSummary

  • Provides an overview of significance across attributes and how our panel performed overall.
  • P Inter measures the contribution of the variability within your panelists - at this point, variability is not yet defined. The closer to 0.000 the more significant the interaction.
  • An explanation of verify and interaction (two terms introduced here) are provided at the bottom of the spreadsheet.


Panel Performance Summary (see Panel Performance Metrics)

  • Provides a very quick summary of how the panel is performing overall and gives us a glimpse as to what attributes and panelists we can look into in more detail.
  • Algorithms are used to determine training recommendations.


Panel Metric Summary (see Panel Performance Metrics)

  • The panel metric summary is in order from MOST important to LEAST important starting with Subject Standard Deviation and ending with Subject Attribute Decision Influence. Each attribute includes an overview of the metric.
  • Subject Standard Deviation - measures consistency of a subject for the same product/attribute across repetitions.  A lower percentage indicates higher consistency across attributes.  
  • Crossover - measures if a subject is scoring significant attributes differently than the panel. A higher score indicates more crossover across attributes. Consider a two sample test where A is higher than B for yellow color. If the subject rates B significantly higher than A, the Crossover score would be 100. 
  • Scale Range Usage - measures the range of means for an individual subject for an attribute and is useful for determining whether a subject's scoring range distorted product means and / or significant differences - a lower percentage is better.
  • Discrimination - measures if the subject p value > 0.5 when the panel has a p value < 0.5 for the attribute - a lower percentage is better.
  • Scale Position - measures the percentage of attributes for each subject that are close to the panel mean - a lower percentage is better.
  • Subject Attribute Decision Influence - measures the percentage of attributes where the subject influenced the attribute significance decision.


Attribute Judgments

  • Works as a panel training guide and helps identify which attributes and/or subjects a panel leader should focus on during their training sessions.


Correlations Table

  • Simple correlations show how attributes are related to one another.


Subject Correlations

  • Shows correlation coefficients between the product coefficients for the panel and the subject - not used by the majority of panel leaders.


Means Table

  • Shows products by attribute listed from most intense to least intense with post hoc lettering. P Value, Interaction as Error and Post Hoc Method are listed.


Means Matrix

  • Shows products by attribute - can be used to easily pasted data into third party applications.


Means & Grouping Matrix

  • Shows attributes with post hoc lettering by attribute.  Products are listed by product number in row 2 and beyond.


Means by Attribute

  • An additional way to look at products, attributes and post hoc lettering.


Product PCA

  • Displays product PCA data.


Product PCA Charts

  • Displays product PCA data in charts.


Attribute PCA

  • Displays attribute PCA data.


Attribute PCA Plot

  • Displays attribute PCA data in charts.  Will display Factor 1 vs. Factor 2, Factor 1 vs. Factor 3 and Factor 2 vs. Factor 3.


Custom Attribute PCA Plot

  • Displays additional attribute PCA plots viewed within RedJade web interface.


Performance Tally

  • Better for Column E percentage to be lower (percentage of attributes with p > 0.50)
  • Better to have more attributes (higher number) in Column I. This means there is less interaction - magnitude, crossover, etc.


Subject Discrimination

  • Displays exact probabilities for each subject for each attribute. Bolded attributes in Column A are significant.
  • P values > 0.5 are in red indicating less significance for that attribute for that subject.


Crossover

  • Higher numbers indicate more crossover. A score of 100 means perfectly backwards from the panel.
  • Only significant attributes are listed.


Subject SD

  • Higher scores indicate higher standard deviation.


Scale Mean Position

  • Values in () represent distance from Panel's Subject mean position.
  • Values in {} represent the number of attributes.
  • Rows 6-8 provide an overall summary by subject.


Scale Range

  • Same as scale mean position, but all numbers are for Scale Range.
  • Rows 7-9 provide an overall summary by subject.


Subject PCA

  • Shows subject PCA by attribute - allows you to identify outliers.


Subject Diagnostic Charts

  • Displays subject information (Subject Score, Range, Crossover, Position, Count in charts.


Product Mean Score Graphs

  • Displays subject and panel means for each product by attribute in charts.


Subject Mean Score Graphs

  • Displays product and panel means for each subject by attribute in charts.


Individual Subject Panel Performance

  • Spans multiple tabs (one per subject)
  • Provides individual panel performance by subject.
  • Provides judgments and actions to improve panel performance along with an explanation on how the judgments and actions were developed. 


Attribute Details

  • This is where you can go to find detailed information about specific attributes.
  • Each attribute has seven detailed tables providing in-depth information to the panel leader. 


Bracket Changes

  • Ranks suggest that there should be significance; however significance was not found - may be due to outlier, crossover, magnitude, etc. Using Attribute Details can help the panel leader to determine if you should overwrite the statistics.
  • It is important to know have detailed stats knowledge prior to overwriting statistics.


All Samples

  • Displays attribute intensity by product by modality.