The Qi Blog

What has Qi Statistics been up to? Read our blog including statistical suggestions, quick tips and reports from the world of statistical consultancy here...

ASA Statement on p-values

Friday 15th December 2017

Well over a year ago now...but still a hot topic! AMERICAN STATISTICAL ASSOCIATION (ASA) STATEMENT ON STATISTICAL SIGNIFICANCE AND P-VALUES The ASA is the world's largest community of statisticians in the United States. Members serve in industry, government and academia in more than 90 countries, advancing research and promoting sound statistical practice to inform public policy and improve human welfare. In March 2016, the American Statistical Association (ASA) released a "Statement on Statistical Significanceand P-Values" with six principles underlying the proper use and interpretation of the p-value. "What we hope will follow is a broad discussion across the scientific community that leads to a more nuanced approach to interpreting, communicating, and using the results of statistical methods in research." This guidance on p-values was…

view more »

Tutorial Day at Sensometrics 2018

Wednesday 6th December 2017

Qi Statistics are proud to be running 3 Tutorials at the upcoming 14th Sensometrics meeting entitled "Going Further" - April 9-12, Montevideo, Uruguay 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?) More 2. Insights on Product Opimization - Thierry Worch Optimizing samples is a critical step in product development as it…

view more »

Discrimination tests

Tuesday 16th May 2017

Discrimination tests (e.g. triangle test, m-AFC, duo-trio, and recently the tetrads) are useful methodologies to detect whether process or recipe changes will be detected by consumers. The statistical analysis of such data can be done in two-ways: (1) estimating the proportion PD of discriminators who can detect the differences (Guessing model), or (2) estimating the perceptual distance δ between pairs of products (Thurstonian approach). In both cases, the results obtained (whether it is PD or d' observed) is compared to a threshold value define by the user. If the value observed is larger than the threshold, one would conclude that the difference is large enough to be detected by consumers. These two methods have their own advantages and inconveniences. Although…

view more »

Bayesian networks in a sensory context

Sunday 16th April 2017

A Bayesian network is a probabilistic model that represents graphically the conditional dependencies between variables. Researchers often use consumer and expert panels to collect various quantitative measures of a product's appeal. These measures may include contextual, sensory, emotional, usage and analytical product scores as well as measures of consumer acceptability. A Bayesian network of this data gives a graphical representation of both the key drivers of acceptability and also the inter-relationships and hierarchies between the descriptor variables, identifying both linear and non-linear relationships. We can use the resulting network for dynamic product optimisation or as a framework for further structured model building. The example (graphic) uses data on consumer liking of different flavour fragrances…

view more »

Quicknotes: 2 sample tests

Friday 17th March 2017

I have 60 consumers who have all rated the same 2 food samples. Should I use a 2 sample t-test to check for a difference between the 2 foods? The word 'sample' can have 2 different meanings. Market researchers and product developers will often refer to the products that they give to consumers as samples. In statistics, and in the context of our example, a sample refers to the group of consumers who are taking part in your study; as the consumers have been selected, or 'sampled', from the wider population of consumers. So the 60 consumers taking part in your study constitute a single 'statistical sample', and the answer to the original question is 'no'. …

view more »

Cosmetics in space, statistically speaking…

Saturday 28th May 2016

Talk at Society of Cosmetic Scientist Annual Conference, National Space Centre, 11 May '16 In May, Gemma was invited to speak at the Society of Cosmetic Scientist Annual Conference, National Space Centre, on the use of statistics in cosmetic science, entitle 'Adding Value with Statistics - Pleasure or Pain'. The venue was amazing, with all the talks being projected onto the planetarium roof! One certain way to liven up statistics... The talk touched on why statistics is becoming more and more necessary and useful, in the rapidly changing and vast landscape scientist and researchers now find themselves in, opening with an example from a well known advert from the 70's involving cats and their food choices! The talk focussed on the key…

view more »

JMP Summit April

Friday 20th May 2016

In March this year Anne was invited to speak at the JMP discovery summit in Amsterdam, on the use of JMP software for sensory analysis. They were keen to hear how a well known sensory expert would use JMP, to possibly extend the capabilities of the current software to make it a software of choice for sensory and consumer professionals. The talk went down extremely well and the title was a big draw - it was standing room only at the back! To find out more about Anne's view on JMP for sensory and consumer analysis and read the talk to see all the visualisations available in JMP, please click here. If you would like any discussion about JMP training or demo's please…

view more »

How is a mean ‘Adjusted’? And why would I bother?

Friday 18th March 2016

Using adjusted means to correct for different sample replication in sensory tests. Have you ever wondered why sometimes the mean scores are presented in results tables under the heading 'Adjusted Means'...? Well if so, there is a good reason that means are adjusted! Think of the following scenario: Assessor 2 has a car that often breaks down, meaning she can't make the testing session. But she is a good assessor, so we still want her data. In the picture below, owing to these missed sessions, Assessor 2 has only provided a single score for products A & B although all 3 sessions worth of data for product C. Assessor 1 however, is much more reliable and has provided all data. So what can we…

view more »

Standard Deviation versus Standard Error

Tuesday 8th March 2016

Most scientists and engineers quote a Standard Deviation when they measure a mean value, despite the fact that they will go on to compare their mean with another mean (or standard value). Examining the Standard Deviations of the samples helps you to understand how much variability you are seeing but can't give you any confidence in whether or not the means are significantly different from each other - which is often what you are after. On the other hand, tests for statistical differences are based on SEs, so plotting these as error bars gives a visual indication of whether or not the means are significantly different. If you don't get the point then maybe you should attend one of our Fundamentals courses?

view more »

Latest News

First Qi Conference planned for Thur 16th May…

Qi are planning a 1-day conference to be held in…

Newsletters - Spring 2018 edition out now!

All our current and past newsletters can be accessed and…

This site uses cookies to help make your experience the best we can. You can find out more about the cookies we use by reading our cookie policy. Dismiss