Descriptive Analysis Workbook Attribute Summary

Descriptive Analysis Workbook Attribute Summary

What is it?

Attribute Summary is one of the sheets in the Descriptive Analysis Workbook. It is a summary of attributes across all panelists, reps and samples.

Why would I use it?

It can be helpful to gain insight into the attributes and the effect of the sample and/or panel variation.

Setup options

To generate the Attribute Summary sheet, under 2. Select Options, select Attribute summary.

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 Summary is the first sheet in your Descriptive Analysis Workbook, then you will see a statement similar to the following:
" Analysis performed using analysis of variance (2-way) with samples X panelists interaction and Tukey's HSD at alpha=0.05 ".

Often Product Means Radar Graphs is the first sheet. We provided more details about the above statement in the workflow for the Product Means Radar Graphs sheet.



  1. Attribute . List of attributes, one per row.
  2. N . Total number of responses.
  3. Missing . Number of missing responses.
  4. Mean . Mean scores pooled across reps, panelists and samples.
  5. Std Dev . Average variation about the mean.
  6. Min . The smallest product mean.
  7. Max . The largest product mean.
  8. P Prod . Product p-value. This is the probability of obtaining the observed test statistic or one that is more extreme under the null hypothesis that there is no product effects when considering MSE as the error term.
  9. P Inter . Product/Panelist Interaction p-value. P Inter is the probability of obtaining the observed data under the null hypothesis that there is no interaction between judges and products.
    P Inter = MSE (samples X judges) / MSE (residual)
  10. P Prod Inter .  The probability of obtaining the observed data under the null hypothesis that there is no product effects when considering MSE (samples X judges) as the uncontrolled (i.e. noise or unsystematic) variance term.
    P Prod Inter = MSE(samples) / MSE (samples X judges)
  11. Sig . Indicates whether an attribute is significant and to which degree. The indication is based on P Prod.
    * = 0.1
    ** = 0.05
    *** = 0.01
When P Prod is significant, samples are considered to be discriminated .
When P Inter is significant, panel is considered to be in agreement .
When P Prod Inter is significant, samples are considered to be discriminated and the variation in the panel is not large enough to affect discrimination .


For more details on significant attributes and product discrimination, see the Attribute Breakdown sheet.

Example Conclusion

In the table below we observe that all of the attributes are presumed to have a product effect. This means that at least one of the products is indeed distinguishable. The same conclusion as in the Product Summary sheet.

However, we now have more information on the "noise" in the data. For example, in Sweetness Aroma we see from P Prod that products are being distinguished, but P Inter indicates that there may be concern with panel agreement. Therefore, we look at P Prod Inter and can conclude that the variation in panel response is likely not enough to affect the product discrimination.