Morning session (April 9th, 9:00-12:30)
1. Ideal Profile Method - Thierry Worch
The Ideal Profile Method is a sensory methodology in which consumers are asked to profile products. During the test, they are also asked to describe their ideal and to score the products based on their liking.
At the end of the test, each consumer provides the sensory profiles of the products (how do they perceive the products?), the sensory profile of their ideal product (what do they want?) and liking scores (how do they appreciate the products?).
The aim of this tutorial is to propose an overview of the Ideal Profile Method, as well as tools to understand how consumers define their ideals, validate the ideal descriptions, and use these diverse information to improve the products
2. Equivalence testing - Michael Meyners
Equivalence (also called similarity or parity) tests are becoming increasingly popular in many application areas, among which sensory sciences. They can and should be applied whenever the aim of the study is not to show differences, but to conclude similarity. The main objective of the workshop is to educate the participants to identify a problem which needs to be addressed by an equivalence approach rather than a difference test. Subsequently, instead of introducing one approach only and rejecting the others, pros and cons and the implications of different tests will be discussed. The participants will therefore learn to make an appropriate choice for future problems; they should learn that none of the tests is generally superior to all others (and, more importantly that the differences are often enough negligible).
3. Introduction to Bayesian Methods for Sensory & Consumer data – Anne Hasted
Bayesian statistical analysis has been successfully used in other industries for many years now, yet is rarely seen in the sensory and consumer world. This technique offers an approach to formally combine historical data with current data to see results of trials put into context with past research rather than treating each trial in isolation and then making subjective interpretations. Additionally, benefits are seen in both reduction of sample sizes and increased ability to make robust decisions with a move away from reliance on the contentious 'p-value'.
In this tutorial, we give you the background understanding to the key concepts of these analysis methods, that needs to be understood before a Bayesian study can be set-up and we give information on how to interpret and apply results of a Bayesian analysis to improve decision making ability and so maximising use of your data. Practicalities of taking this approach will be also discussed.
4. Preference mapping and conjoint analysis in ConsumerCheck software - Tormod Næs
The objective of the tutorial is to make participants able to run preference mapping and simple conjoint analysis on their own computer.
The tutorial will start with a description of preference mapping and conjoint analysis. Focus will be on both philosophy behind the methods and the methodology used. Estimation of population average effects as well as individual differences will be covered.
The methodology will then be illustrated by the use of the ConsumerCheck software, which is freely available on the website consumercheck.co. In order to be able to participate in hands-on activities, the participants are asked to download the program to their own computer before the tutorial.
Afternoon session (April 9th, 14:00-17:30)
5. The analysis of sensory profiling data revisited by scaling. Univariate and multivariate implications on panel performances and product differences - Pascal Schlich
Classical analysis of sensory profiling data assumes that panelists use a comparable width of the sensory scales. Due to physiological and psychological reasons, this assumption is likely to be wrong. The Mixed Assessor Model (MAM) includes individual scaling coefficients taking into account this panel heterogeneity. We proposed recently MAM-CAP and MAM-CVA as respective extensions of CAP (Control of Assessor performances) system and CVA (Canonical Variate Analysis) taking scaling effect into account. Besides scaling (standard deviation), panelist can also differ on the level (mean) of their scoring and that has been included in these new tools too. We developed two R-packages to support these new techniques, also easily accessible in the TimeSens® software. Thanks to the SensoBase (a database of a thousand of sensory studies), we conducted meta-analysis to investigate whether and to what extent these new methods provide different and eventually “better” results than the classical ones.
The workshop will summarize the 7 papers we published on that topic in the period 2015-2017. Then access to these techniques with the two R-package and TimeSens® will be demonstrated using real data provided by the instructor and/or brought by the participants.
6. Temporal sensory methods: study design & data analysis – Michael Meyners & John Castura
The sensory identity of diverse products – foods, beverages, cosmetics, etc. – has an important temporal component. But until recently, available methods for investigating temporal aspects have required a significant investment in training assessors, have allowed measurement only at discrete time intervals, and/or have produced continuous measurement of only a single attribute. Recently, new temporal sensory methods have been proposed to capture some of the richness and diversity of consumer sensory perceptions that arise from multiple sensations.
The workshop will focus especially on the following two methods: temporal dominance of sensations (TDS; Pineau et al., 2009) temporal check all that apply (TCATA; Castura et al., 2016). We will review study design considerations that are relevant to these and other temporal methods. Workshop attendees will have the opportunity (either during or outside the workshop) to understand the assessor’s task via hands-on web-based data collection using Compusense Cloud software.
7. Insights on Product Opimization – Thierry Worch
Optimizing samples is a critical step in product development as it allows companies to get closer to the consumer’s optimal sample, hence increasing the chances of market success.
Originally, optimization was performed through statistical modelling that combined sensory description of the samples gathered from trained panels to the consumers liking responses. Such approach has been extended with data collection methods that include directly the information regarding optimal product (Ideal Profile Method, Just About Right Scale).
This tutorial will give comprehensive insights on the statistical models and how they can be efficiently used to help you with your product optimization. Original (e.g. Preference Mapping) as well as more recent methods (PrefMFA, PLS based models, etc.) will be presented and discussed.