Descriptive Analysis: When Evaluating Spirits, Use Your Words

As the name implies, descriptive analysis—also sometimes called qualitative analysis—is sensory testing that is designed to quantify qualitative aspects of either a product or a group of products.

For example: If you find that a particular batch of whiskey has a slightly different aroma compared to previous batches, descriptive analysis could be used to help you identify the components of that aroma and quantify their intensities in relationship to previous batches. In turn, that information can help you determine what may have caused the difference.

Descriptive analysis is different from discriminative testing in that it seeks to determine the exact components and their intensities in a sample, rather than to simply determine whether there is a perceptible difference between samples. That makes descriptive analysis an incredibly important tool when you’re trying to better understand the interplay of flavors in your products, and how your distillery’s processes might be affecting those flavors. You can also use descriptive analysis to track flavor drift in products over time and to analyze competitors’ products in relation to your own.

In previous articles we’ve touched on evaluator recruitment, the statistics of sensory science, and both discriminative and hedonic testing. Here, we’ll learn more about descriptive-analysis testing and its uses in the beverage industry.

[PAYWALL]

Selecting a Panel

There are many different ways to use descriptive analysis, but they all follow roughly the same steps.

The first stage is to decide what kind of panel you want to conduct the analysis. Panels can be made up of distilling professionals, consumers, or a mix of both. Different panel compositions affect the results of analysis, so it’s important to determine what configuration will get you the most accurate and repeatable data.

Once you’ve determined the best panel composition, it’s time to recruit panelists. It’s critical to consider only individuals suited for the task—for instance, if you plan to evaluate products rich in smoky compounds, you should ensure that all potential panelists can differentiate between different kinds of smoke, such as peat or mesquite. Panelists also should be able to differentiate between intensities of various stimuli, and they should be able to consistently describe the flavors and aromas they perceive. Ideally, candidates have completed some level of recruitment training before they’re considered eligible to be on a panel.

Finally, you need to screen prospective panelists for motivation. Motivation is often overlooked when selecting panelists, but it’s arguably one of the most important factors. Even if a candidate passes all the standard selection criteria with flying colors, if he or she is unmotivated to accomplish the tasks given, then the data they provide will be invalid. Worse still, a single panelist’s lack of motivation can spread throughout an entire panel, tainting the data given by the entire panel.

Establishing the Lexicon

Once you’ve selected your panelists, it’s important to train them for the task at hand.

Humans all have their own unique senses of smell and taste; what tastes like caramel to one may taste like toffee to another. For descriptive analysis to be accurate, it’s important for panelists to be aligned so that they describe stimuli consistently. To do this, panels should adopt a unified lexicon of terms and concepts that allow them to describe their taste experiences in a shared way.

One common way to create a lexicon is for panelists to taste representative samples of the type of product that they’ll be analyzing, meanwhile having each panelist generate a list of terms they would use to describe the samples. The panelists then share their lists with each other; guided by experiment designers, the panel can slowly combine the lists, eliminating duplicates and similar words to create a single, agreed-upon lexicon.

Then, the panelists should practice evaluating samples using only this agreed-upon lexicon. This helps to ensure that they’re not leaving any aromas or flavors out of the lexicon and that all viewpoints are being considered equally.

Clearly, the process of creating a common lexicon can be time- and resource-intensive. For those without the time, it’s common to use a pre-existing vocabulary. For distilled spirits there are many such established lexicons, which often come in the form of flavor wheels such as this one:

Source: Scotch Whisky Research Institute

Using a flavor wheel to align panelists’ vocabulary can be an effective way to establish a unified lexicon—however, flavor wheels do have limitations. First and foremost, it’s vitally important for experiment designers to select the correct flavor wheel for the products to be evaluated. There are often many flavor wheels for each category of product, and some may not be suited for specific genres.

For example, a flavor wheel designed for evaluating whiskey may in fact be skewed for a particular type of whiskey. If someone created the flavor wheel to evaluate bourbon, it probably won’t include a section for peaty notes. Likewise, a wheel created to evaluate Scotch will no doubt undervalue the aroma and flavor of corn. Thus, before adopting the vocabulary of any flavor wheel, it’s important to carefully evaluate the listed descriptors for their relevance to the products being tested.

The other drawback of using an established flavor wheel is that it takes some additional training to get all panel members fully aligned with its terminology because panelists may not be familiar with all the terms on the wheel. Therefore, test coordinators should spend extra time ensuring that each panelist understands the terms and can consistently associate all sensory experiences with the available descriptors.

There are many other ways to create a unified lexicon and train panelists on using it. The chosen method can depend on the descriptive technique being used, the goals of the experiment, and the overall design of the experiment. Regardless how you get there, it’s crucial that all panelists understand the terminology and agree to use the accepted lexicon when describing samples. That will help ensure that any data collected by descriptive analysis are valid and useful.

Evaluation and Upkeep

Once you’ve selected your panel, generated the lexicon, and trained them on using it, the next step is to test the panel as a whole for consonance.

Consonance is a measure of a panel’s ability to give repeatable and consistent results over multiple evaluations. There are many ways to evaluate consonance—and the test designer should choose the exact method—but no panel should begin evaluating samples until they have reached a predetermined, minimum level of consonance. Without consonance, the results from any descriptive analysis will be inconsistent and unusable.

Once a panel has met the required level of consonance, they are ready to begin evaluating samples—however, that doesn’t mean they’re free from supervision. Many descriptive-analysis panels evaluate products over periods of months, years, or even decades; during this time, an individual panelist’s ability or perceptions of taste and aroma can change dramatically. Therefore, individual panelists as well as the panel as a whole need to be routinely retrained and retested to ensure consistency. That will help to make sure that your descriptive analysis runs efficiently, and that any data collected are usable and comparable.

Costly Yet Powerful

Descriptive analysis gives distillers the ability to learn about the interplay of flavors and aromas of their own products as well as those of competitors. They can then use that knowledge to make important process and quality decisions, and it can also be helpful when developing new products.

However, properly running descriptive analysis is time-consuming, complex, and often expensive. Even very large companies sometimes have trouble running and maintaining consistent descriptive-analysis panels, and many companies choose to outsource the work to third parties.

Nonetheless, the information yielded by descriptive analysis has been crucial to the success of many products, old and new, and it remains an important tool in the belt of any distiller.