This tutorial will help you design and analyze a sensory discrimination triangle test in **Excel** using the XLSTAT statistical software.

## Dataset for running a discrimination triangle test in sensory analysis

An Excel workbook with both the data and the results can be downloaded by clicking here.

A brewer wishes to commercialize a new beer and would like to know if consumers will be able to differentiate the new product from a beer of its own line of products.

To test the hypothesis that both beers are similar, he wishes to use a triangle test.

## Goal of the triangle test

The goal of this analysis is to first generate a design of experiments to taste the beers, then to analyse the result in order to know if the two products are well discriminated.

In the triangle test, three samples are presented to each assessor in different orders. Within these samples, two are similar and the third one is different. Assessors have to identify the sample that is different from the others.

## Setting up a design for a triangle test

Once XLSTAT is activated, select the** XLSTAT-Sensory data analysis**** / Design for sensory discrimination test** command (see below), or click the corresponding button of the XLSTAT-Sensory data analysis toolbar.

Once you have clicked on the button, the dialog box appears. Select the triangle test, the number of judges ins 20 in that study. You can also select the name of the products (in our case beer 1 / beer 2).

After you have clicked on the **OK** button, the computations start and the results are displayed on a new Excel sheet.

## Running the triangle test

The obtained table allows you to prepare your test with your 20 assessors. Each judge taste 3 samples in the designated order. If the judge recognizes the product that is alone, then the last column should be completed with a specific value (we use + in our case), if the judge does not identify the good sample then a specific value is also added in the last column of the table (in our case, we use -). These values can be any value depending on the user choice.

Once the table is completed, you can analyze the obtained tasting results.

## Setting up a triangle test

Select the** XLSTAT-Sensory data analysis**** / Sensory discrimination test** command (see below), or click the corresponding button of the XLSTAT-Sensory data analysis toolbar.

Once you have clicked on the button, the dialog box appears. Select the triangle test, the Thurstonian model will be used. We select the last column of the design in the corresponding box and add + as code for correct. Other data input can be used as the number of correct answers or the proportion of correct answers.

We wish to estimate both d' and probability of discrimination so we choose the estimate option and use an exact binomial distribution. For more details on these concepts please check the help of XLSTAT.

After you have clicked on the **OK** button, the computations start and the results are displayed on a new Excel sheet.

## Interpretation of a triangle test

The first table summarizes the selected options.

Then the test results and interpretation are given. In our case with a p-value smaller than 0.05, we can reject the null hypothesis that d' is equal to zero. That means each beer is not coming from the same normal distribution. We can thus reject the hypothesis that beers are similar. Furthermore, the power of our test is very high which allows us to conclude that our test is robust enough.

The next table gives the parameter estimate including the d'. We can see that all parameters are significantly greater than 0. The d' is almost equal to 3 which is very high meaning there exist a large difference between the two beers (three standard error).

The conclusion of that test is that consumers discriminate the new beer from the older one well. The brewer can assume that this new product will be well identified by consumers.