Researching consumer preferences and perceptions of food products is vital to food manufacturers, retailers and marketing specialists. In a food company, sensory scientists work to determine not only what consumers like and why, but also whether consumers can tell the difference when a manufacturer changes or substitutes one or more ingredients. Triangle testing is a research option with advantages that make it a good research tool and disadvantages that can make using it difficult.
Triangle testing is a discriminative method that uses difference and sensitivity tests. Difference tests function as a gauge to determine the overall differences between two products. Sensitivity testing determines whether changing the manufacturing process or product ingredients significantly changes a food product. For example, budget constraints might cause a food manufacturer to consider substituting expensive ingredients for cheaper ones. If more than one substitution option exists, triangle testing can help determine which option is the closest to the original.
Triangle testing has a simple setup and design, making it both easy and cost-effective for large and small businesses to use. All you need are three to six samples, each consisting of two unchanged products and one changed product, for each panelist. Because there is no right or wrong answer for a difference triangle test, the analysis consists of calculating a simple percentage. For example, perhaps 60 percent of the panelists could not detect any difference in unchanged and changed products. The sensitivity test analysis uses a chi-square distribution to determine whether correct responses -- panelists unable to tell which product is different -- are above or below a predetermined benchmark. A chi-square distribution is a complex statistical probability calculation in which the results are summed and squared. It is best left to a sensory scientist to complete.
Despite its many benefits, triangle testing is prone to biases, errors and effects that can produce inaccurate results. Some of the most common disadvantages, which focus mainly on the testing environment, include positional bias, stimulus errors and the suggestion effect. For example, a sample display order that runs in a straight line can produce positional bias in which panelists most often choose the middle sample as odd. Samples in which items aren’t identical in every way can cause a stimulus error in which panelists assume an item that may look even slightly different is the correct response. The suggestion effect occurs when panelists influence each other by voicing opinions or demonstrating their reactions.
Simple steps can help minimize the disadvantages inherent in triangle testing and improve the accuracy of its results. For example, displaying samples in a triangle rather than a straight line eliminates a middle sample. Taking great care to ensure items are identical to each other eliminates visible differences that stimulate panelists to choose an item just because it looks different. Isolating panelists so they can’t see or speak with each other eliminates any chance that panelists will influence each other. In addition, it’s also important for the test monitor to provide only the instructions necessary to complete the test, as too many facts or hints can cause panelists to make judgments based on expectations.