Spectrum Descriptive Analysis (SDA) and Quantitative Descriptive Analysis (QDA) are two different approaches to descriptive analysis. Both involved trained panelists who evaluate products for their individual properties.
We're often asked "Should I use Spectrum Method or QDA? Is one worse than the other?"
The truth is that both methods have their strengths and weaknesses. Let's talk about them here. First, they're not entirely different from one another. They share a common goal, to describe materials. The descriptions look similar in that – you get a qualitative set of attributes/terms with each having a quantitative numerical scores/rating associated with each attribute/term. For example, Sweet at a 5.5, Caramelized Sugar at a 3.0, Vanilla at a 2.5, and so on. Where the differences lie is in the amount of training, type of attributes/terms, use of references, use of calibrated scaling, and stability of the method over time.
The Spectrum method is based on the use of precise, descriptive terminology (lexicons) that can be related back to product development at the bench. Detailed physical and chemical descriptors of sensory attributes are used to discriminate among products and document each product’s sensory profile. QDA panels, since they have less training (more on training below) often end up using less precise, more consumer-like attributes, for example, ‘creamy’ when describing a yogurt. A Spectrum panel would break down ‘creamy’ into more specific attributes, such as ‘dairy fat aromatic’, ‘viscosity’, and ‘fatty mouthcoating’ for example. There have been modification to the QDA method over the years which include anchor points on the scale or reference products with specified intensities. These are often added to add some standardization to scoring.
The Spectrum method relies on rigorous training.
Spectrum panelists train for both terminology and intensity scaling and complete a minimum of 100 hours of training for each sensory modality. References are used extensively to aid in clarification of attributes and intensities. After training, Spectrum panels are tested on their abilities with a formal validation study, which is repeated 2-3 times per year to ensure they are performing to a certain standard. Because they are so highly trained and calibrated, Spectrum panels usually only need to complete one replication of data collection. QDA panels generally don’t undergo as much training (sometimes using less than 10 hours of training) and thus more panelists and/or more replications are needed to obtain a robust data set.
SDA Panels use the universal scale rather than category specific scales
During evaluations, each attribute is rated on a standardized, universal scale ranging from 0 = none and 15 = very strong (Table 1). Use of tenths of a point is allowed, resulting in 151 points of differentiation along the scale. The universality of the scale allows comparisons to be made across attributes, across products, across categories, and over time. A rating of 10 in sweetness will mean the same to a Spectrum panel regardless of whether it’s for a soda, a bread product, or a pizza sauce. QDA ratings are typically scaled differently for different attributes and/or products, which makes it difficult to compare data sets over time or across studies.
At Sensory Spectrum, our internal panel works 12 hours per week evaluating a wide variety of products for our clients. Many of the panelists have been working for us for years, and thus have vast experience evaluating different types of products. The products they are most frequently working on these days include meat products, frozen meals/pizzas, fried potatoes, dairy products, salty snacks, coffee, tea, chocolate & confectionery, protein ingredients, and sweeteners, among others. The skills that our training provides allow our panelists to easily train up for and adapt to
new categories. For new projects, we typically conduct 1-2 orientation sessions with the panel to familiarize them with any new attributes/references or evaluation techniques.
Both methods have their benefits, QDA can be faster and cheaper upfront to get relatively robust data, but that data is not usable over time or as precise as working with a properly trained SDA panel.