Escaping p-value land: The future of lean decision-making in sensory evaluation

Sara King/ August 20, 2017/ Workshop/ 0 comments

This workshop explores current topics and trends in sensometrics, and includes engaging presentations, arguments, and a panel discussion. Focus is on general concepts rather than technical details, making the workshop of interest to a general audience with an introductory knowledge of statistics.

In Part 1, we present the 2016 American Statistical Association’s (ASA) statement on significance testing and p-values, and explore its implications for analysis of sensory evaluation and consumer data. Pros and cons of particular statistical approaches are laid out, touching on concepts such as p-values vs. confidence intervals, hypothesis testing, and implications in product development and claims substantiation applications.

In Part 2, we ask whether Rev. Thomas Bayes (1701-1761) can enable leaner product development. We present a brief introduction to the concept of Bayesian statistics, and discuss where to obtain prior information (e.g. historical data or expert knowledge). Benefits of Bayesian approaches (leveraging prior knowledge, contextualizing data in terms of business risk/reward, reduced dependence on p-value / strict sample sizes) are discussed and contrasted with some limitations (sensitivity to the prior, reliability or applicability of the prior knowledge, and appropriateness in a claims substantiation context).

In Part 3, the discussion broadens to touch on related issues in sensory evaluation. When conducting multiple statistical tests, adjustments for multiplicity to control Type I error can inflate Type II error rates, but the false discovery rate (FDR) provides a different way of framing this problem. The role of adaptive designs is then explored. These designs are relevant in various contexts, one example being slow enrollment in consumer studies, e.g., for a product category that consumers might use only briefly at a particular life stage.

In Part 4 the workshop concludes with a moderated panel discussion on benefits and barriers to adoption, and includes an open Q&A session.

Meyners, M., Castura, J. C., Carr, B. T., Berget, I., Grzeda, A., Hasted, A., Hodgson, G. (2017). Escaping p-value land: The future of lean decision-making in sensory evaluation. 12th Pangborn Sensory Science Symposium. 20-24 August. Providence, RI, USA.

Share this Post

Leave a Comment

Your email address will not be published. Required fields are marked *