Lesson 6 - Sensory Analysis
This lesson deals with the sensory analysis of beer. It details how to set up the elements of a sensory panel, such as panel selection, training, and the design of the sensory facilities, and then covers how to run the panel and collect the results. It discusses the various types of sensory test and their benefits before going on to cover the use of statistics to convert sensory data into information about the sensory quality of beer. Finally, it looks at the presentation of results from sensory tests and lists key flavour compounds in beer.
Difference Tests
Consumers rightfully expect their favourite brand to taste the same today as it did yesterday. Difference, or discrimination, tests help us to identify and qualify differences and ascertain where they occur, so are key to the management of beer quality.
We use difference tests to answer two questions:
Question 1. Are the samples different?
This can identify a change or lack of consistency.
Question 2. Are the samples different with respect to a particular attribute?
An example would be ‘Is one of the samples more bitter or less sweet?’ and answers to this question help us identify the nature of the inconsistency.
We can also use difference tests to screen panellists; we introduce a beer with a known difference, monitor their performance, and check their responses against statistically significant results from the panel.
Difference tests are comparatively rapid and can be undertaken by either trained or untrained tasters but never both at the same time. If trained tasters are used, fewer are required to produce a statistically sound result.
In difference testing there are two possible risks or errors:
The α-risk is the probability of concluding that a perceptible difference exists when one does not. This is also known as a Type I error.
The β-risk is the probability of concluding that no perceptible difference exists when one does. This is also known as a Type II error.
To analyse the data from difference tests, we usually use a statistical table. The statistical table gives significance levels for numbers of correct answers from panels of different sizes. We compare the number of correct responses from our panel to the number required for there to be a significant difference for a panel of the same size at a predetermined level of significance (normally 5%). If the number of correct answers is equal to or greater than that on the table then we can say the sample is different at the confidence level stated (95%). We can also use a software package to calculate the significance of the difference. We will consider confidence and significance in more detail later in this lesson.
Next we learn about the most commonly used difference tests.