What is it?

Setup options

To generate the Attribute Breakdown sheet, under 2. Select Options , select Attribute breakdown .
The Analysis Options from Defaults area indicates what your default selections for ANOVA model and post hoc test are. If a change is necessary, click Change Advanced Analysis Options .

Analysis Overview

If the Attribute Breakdown is the first sheet in your Descriptive Analysis Workbook, then you will see a statement similar to the following:

1. DF . Degrees of Freedom is the minimum number of independent coordinates that can specify the position of the system completely.
   

   DF = N - 1
   
2. SS . Sum of Squares is the sum of all evaluations of the squared differences of each observation from the overall mean. It helps us see variance in the data.
   
3. MS . Mean Square Error (variance) indicates whether there is noise in the data.
   
4. F Ratio .  Signal to noise ratio. A higher F ratio indicates that the p Ratio (the p-value) is more likely significant.
   
   F Ratio = systematic error (understanding for possible error) / unsystematic error (noise that cannot be accounted for)
   
5. p Ratio . The p-value
   
6. F Interaction . The ratio of MSInteraction (samples X judges) / MSE (residual).
   
7. p Interaction .  Samples X Judges Interaction p-value. The probability of obtaining the observed data under the null hypothesis that there is no interaction between judges and products.
   
   Probability of F Interaction given numerator (interaction) and denominator (error) degrees of freedom. 
   
   p Interaction = MSE (samples X judges) / MSE (residual)
   

F Interaction , p Interaction and Samples X Panelist Int only appear when ANOVA with interactions is selected in the Defaults.

Example Conclusion from the ANOVA Table

From the MS column, we can confirm that samples account for the majority of the variation. We expect to see this because we know from previous sheets that our samples are significantly different.

Since we have a trained panel, we would also hope that our panelists and sample x panelists interactions are not significant. Here we see that it is significant (p Ratio = 0.03) and might consider further review of panel performance.

1. Mean . Averaged scores across panelists and reps. The means are sorted from highest to lowest.
   
2. Mean Diff . The mean above the sample subtracted from the sample mean. The first Mean Diff will always be 0.
   
   Example: Apple Juice 2 - Apple Juice 1 = 85.03 - 86.08 = -1.05
   
3. SD . Standard Deviation on the mean.
   
4. Tukey's HSD . The column header reflects the post hoc and its level selected in the Defaults . The post hoc lettering in this column indicates whether there is a significant difference between samples.
   
5. Rank . The rank is determined by the intensity. The sample with the highest intensity will be ranked 1st.
   
   The rank value in this column is the average for each sample intensity ranking for the given attribute for each panelist and rep.
   
6. Rank Diff . The rank above the sample subtracted from the sample rank. The first sample Rank Diff will always be 0.
   
   Example: Apple Juice 2 - Apple Juice 1 = 1.83 - 1.38 = 0.45
   
7. Sample Range . The sample range row is given for the Mean and SD.
   
   Sample Range = Max - Min.
   

Example Conclusion from the Mean Scores and Ranks Table

From the above table we can see that for Sweetness Flavour, Apple Juice 4 has the lowest mean score and is consistently ranked 5th across panelists and reps. We also see from the post hoc lettering that it is different from all other samples.

Mean & Rank Differences Table

1. Mean Diff . The mean of the column sample subtracted from the mean of the row sample. A negative score indicates the mean of the column sample is larger than the mean of the row sample.
   
2. Rank Diff . The rank of the column sample subtracted from the rank of the row sample. A negative score indicates the rank of the column sample is larger than the rank of the row sample.
   
   
   

Example Conclusion from the Mean & Rank Differences Table

You may wish to compare the means and/or rank of two particular products of interest.

Rank Direction - Panelist Counts Table

1. Lower . The number of panelists who scored the column sample lower than the row sample.
   
2. Higher . The number of panelists who scored the column sample higher than the row sample.
   

Example Conclusion from the Rank Direction - Panelist Counts Table

In the Rank Direction - Panelist Counts table above we can compare all product rank orders based on the panelist count.

Here we confirm our observation that Apple Juice 4 was ranked 5th by all panelists.