XLSTAT: RATA data analysis in Excel

This tutorial will show you how to run and interpret a RATA data analysis in Excel using the XLSTAT statistical software.

 

Background

RATA is a sensory evaluation technique where panelists select all applicable attributes for a sample and then rate their intensity. It is widely used in food, beverage, and consumer product research.

 

Dataset for Example

The data used to illustrate the RATA feature relates to a Rate-All-That-Apply (RATA) experiment in which 9 assessors evaluated 12 chocolates using 13 attributes over 3 sessions.

 

Setting Up

• Open XLSTAT.

• Click on the Sensory/ CATA data menu and select the RATA data analysis feature.

The CATA data dialog box appears. 

In the General tab:

• Select the RATA data corresponding to the attributes.

• Select the data corresponding to the Product labels.

• Select the data corresponding to the Assessor labels.

• Select the data corresponding to the Session.

In the Outputs tab:

• Choose between CA or PCA for Dimension Reduction.

• Click on the OK button to launch the computations.

• The results are displayed on a new Excel sheet.

 

Interpreting the Results

The first important result of the RATA feature is the table summarising the ANOVAs for each attribute. In this table, we can see that for all the attributes except Roasted Coffee, the Product variable is significant at the 5% threshold, which means that these attributes are discriminating.

Another important result of the RATA feature is the graphical representation of products and attributes. On the following chart, we can see that we have 88.14% inertia reported by our first two axes, so we can be satisfied with these two. The chart shows an opposition between bitter flavours (roasted cocoa, sour...) and sweet flavours (cocoa butter, sweet...). Choco72 is seen as a smooth chocolate, while Choco86 is more bitter.

The last important result is the weight given to each assessor. Indeed, the lower the weight of a judge, the more atypical his point of view is. For example, here the weights are very similar from one assessor to another, so there are no atypical individuals in this dataset. This conclusion is confirmed by the homogeneity value, whose maximum value is 1, and which here is 0.92. 

In the RATA analysis, it is also possible to use the CLUSCATA algorithm to construct groups of subjects with different points of view automatically.