Use of CATA to understand consumer preferences
One of the differences between CATA (Figure 1) and methods such as Sorting and Projective mapping, is the way in which wines are presented to tasters. During Sorting and Projective mapping all the wines are served simultaneously, while one wine at a time is served during a CATA tasting. It is therefore important that every taster receives the wines in a different order.
Tasters get a list of words or phrases to describe the characteristics of a wine. All of the words or phrases that apply are marked by the panellists. The words or phrases can include sensory terms, for example, aroma, taste and mouth-feel descriptions, as well as social or emotional aspects that describe their product preferences and liking. “I like it”, “delicious” and “I hate it” are examples of phrases related to the emotional aspects of product preferences and liking. Sensory descriptions can be combined with phrases about product liking in the same CATA list. It is also possible to use a separate question before or after CATA to evaluate liking. Varela and co-authors (2012) suggested that the liking question is asked before the CATA procedure is done, if it is not included as part of the CATA list.
The order of the terms on CATA lists is important. Research has shown that consumers tend to not pay an equal amount of attention to all of the terms. Words at the top end of the list are often given preference and used more often. The order of the terms can be changed for each taster if the list is very short. It can even be changed for different wines tasted by the same taster. The person who sets up these lists and determines the order of the terms, should remember that wine is complex and difficult to taste. Changing the order of terms should not be confusing or frustrating for tasters. If the CATA list is very long, it makes sense to group similar terms, for example group fruity aroma or woody aroma terms together, etcetera. Different lists can even be used for different modalities; colour, aroma, taste and mouth-feel.
The use of CATA to profile wine using experts and trained panellists
Pick-K-attributes is a variation of CATA and is commonly used for sensory profiling of wine, with wine experts and trained panellists as tasters. This variation differs from the standard CATA method, in the sense that tasters are asked to check the most prominent/dominant terms only, instead of marking all the terms that apply. Since CATA does not measure the intensities of the sensory characteristics, it is usually a better idea to use only the most prominent sensory characteristics for profiling, rather than bringing all characteristics into account. This however, depends on the purpose of the sensory tests. In some cases CATA will rather be used for profiling, as a more complex description with more terms that can be obtained than with Pick-K-attributes.
When wine experts are used as the panel during a Pick-K-attributes profiling test, the assumption is made that all the terms are interpreted and understood in the same manner by all the panel members, which is usually the case. However, when a complex set of wines with small differences between them is evaluated by using a long list of terms, the degree of consensus among wine experts without sensory training, is not always good enough. A trained panel can be used for scenarios like this. This variation of the CATA method is not fast and if the panel is paid, it can become a costly exercise. Campo and co-authors (2010) found that the description of the products obtained using this method was indeed excellent. Results compared well to classical Descriptive sensory analysis (DA) results also obtained during the study on the same set of products. More information on DA was given in Part 1 of this series.
Statistical analysis of CATA data
The number of times a description is used for a wine by all the tasters is counted and a table constructed of all these counts (referred to as a frequency matrix). The most common statistical method used to analyse the data is Correspondence analysis (CA). Other methods, for example Multiple factor analysis (MFA) can also be used. The end result is a graph representing all the wines and descriptions used (Figure 1). Wines are positioned close to the terms or phrases that were often used by tasters to describe them. Wines for which the same descriptions were used often will also appear close to each other on the graph. All the wines in a data set are thus represented on a graph.
Graphical representations and descriptions of each individual wine are sometimes useful and word clouds are commonly used to visualise frequency data. Words used by more tasters to describe a wine will appear larger (a larger font) than words used by fewer tasters (Figure 1). These simple tools can emphasise the information captured by CATA and make it easy to understand for the reader.
Possible pitfalls: Keep the following in mind with CATA-type methods
Since wine can be a challenging matrix to taste, especially for consumers, it can be tiring when CATA lists are unnecessarily long. CATA lists can be shortened by focus group tastings prior to the formal CATA experiment.
The reality that CATA-type methods do not measure intensities of sensory attributes, should be kept in mind. This is especially important when the purpose of the tasting is to compare different wines. When the same terms are found at high frequencies for two wines, it does not necessarily mean that the intensity of those sensory properties is high in both wines. It only indicates that both wines have those properties and that it is prominent/intense enough for most of the tasters’ perceptions. When intensity differences between wines are an important aspect that needs to be tested, CATA with intensity rating (Rate-all-that-apply, RATA) or DA should be used.
Benefits and applications of CATA in the wine industry
CATA-type methods are mostly used in the wine industry for profiling, although it is sometimes used for consumer preference (liking) test. Campo and co-authors (2010) pointed out an important benefit that the Pick-K-attributes approach has over the DA method. During DA analysis, the panel members are limited to use a small number of sensory descriptors to describe a set of wines. This is a disadvantage when the data set contains various wines with vastly different sensory properties. In such a case, it is not always possible to describe all the wines with a maximum of 15 to 20 terms (normally used in DA). The Pick-K-attributes list can be much longer than 20 terms, indeed, for some reported tastings, more than 100 terms were used for profiling using a trained panel. Finally, CATA-type methods are quick and simple for both the tasting panel and the analyst who must do the data analysis.
How do DA and the alternative, rapid methods compare?
The different sensory evaluation techniques discussed in the articles in this series is compared briefly in Table 1, to provide readers with practical guidelines on choosing the appropriate method. These guidelines are supported by scientific research done at the IWBT-DVO, SU.
Results obtained with the rapid methods compare well to DA results when product experts or trained panellists are used as tasters. The average time tasters need to finish the tasting, as well as the difficulty of the methods varied. CATA variant Pick-K-attributes was perceived as the easiest method and also took the least time to finish. Mapping was rated by the tasters as the most difficult and time consuming method, and it was the only rapid method that was rated more difficult than DA. The time needed for sensory evaluation and the difficulty of the sensory task are both important considerations. However, the nature of the set of wines to be tasted, the information required from the tasting and expectations about the outcome should be the first priority when choosing a sensory evaluation method.
Pick-K-attributes can provide useful results when the characteristics of each sample are needed and little information is available or needed with regard to similarities between the wines. Sorting is a good alternative to DA if some of the wines in the set are similar, or if the purpose is to specifically investigate the similarities and differences. Mapping could be particularly useful to investigate product similarities when wines present a continuum of characteristics that change gradually, rather than presenting as discrete groups. It is also a useful method to investigate the magnitude of differences between groups of wines or wines within the same group.
The references provided here provide more background. You are also welcome to contact us for more information or assistance with the setup of a rapid sensory evaluation session.
Acknowledgements
This research was made possible by Winetech funding, project IWBT W13/02: “Rapid descriptive sensory methods for wine evaluation – special focus on further optimisation of rapid methods and streamlining of workflow”. Winetech, the Institute for Grape and Wine Sciences (IGWS), Stellenbosch University, NRF and THRIP are acknowledged for financial support.
References
Ares, G. & Jaeger, S.R., 2013. Check-all-that-apply questions: Influence of attribute order on sensory product characterization. Food Quality & Preference 28, 141 – 153.
Campo, E., Ballester, J., Langlois, J., Dacremont, C. & Valentin, D., 2010. Comparison of conventional descriptive analysis and a citation frequency-based descriptive method for odor profiling: An application to Burgundy Pinot noir wines. Food Quality & Preference 21, 44 – 55.
Lawless, H.T. & Heymann, H., 2010. Sensory evaluation of food: Principles & practices. New York: Springer.
Valentin, D., Chollet, S., Lelievre, M. & Abdi, H., 2012. Quick and dirty, but still pretty good: A review of new descriptive methods in food science. International Journal of Food Science & Technology 47, 1563 – 1578.
Varela, P. & Ares, G., 2012. Sensory profiling, the blurred line between sensory and consumer science. A review of novel methods for product characterization. Food Research International 48, 893 – 908.
Consultation services offered by the Wine Sensory Facility, IWBT-DVO, Stellenbosch University:
- Selection of appropriate sensory methods.
- Guidance with selection, training, testing of panellists.
- Experimental design and practical workflow.
- Statistical analysis and interpretation of data.
– For more information, contact Jeanne Brand at jeanne@sun.ac.za, Hélѐne Nieuwoudt at hhn@sun.ac.za or contact the authors at winwynsen@sun.ac.za.