Discovering experimental design: An interactive teaching exercise using Fisher's tea‐tasting experiment

Abstract

An appreciation of experimental design is an important aspect of introductory statistics teaching in a wide range of applied disciplines, including medical statistics. Understanding the impact of design decisions on the choice of analysis method and subsequent interpretation of results can help to embed the importance of statistical thinking in the experimental process. I discuss an interactive exercise, based on R.A. Fisher's celebrated “Lady Tasting Tea” experiment, that is intended to raise awareness of design issues as part of an undergraduate statistics module. The exercise used a discovery approach, with students encouraged to identify design issues and agree on solutions themselves via small group discussion, with only low‐level prompting from the instructor. The value of this teaching style and possible extensions of the tea‐tasting experiment to other related topics suitable for more widespread use are also discussed.

A Systematic Review on Teaching Strategies for Fostering Students’ Statistical Thinking

“…Since statistics is a science that necessitates a certain degree of creative thinking, particularly in analytical endeavours, educators need to use active learning strategies. For instance, by using causal inference, experimental design, and simulation-based methodologies, students are given an opportunity to investigate statistical problems, topics, or scenarios (Cummiskey et al, 2020;English & Watson, 2015;Fanshawe, 2021;Tintle et al, 2015). This assignment allows students to examine statistical problems or events within a broader framework rather than just focusing on a particular instance.…”

“…According to Fanshawe (2021), a fundamental component of beginning statistics education in several applicable fields, such as medical statistics, is recognising and understanding experimental design. Using statistical thinking in the experimental process may help by understanding how design choices affect the choice of analytical methods and the subsequent interpretation of results.…”

Combinatorial Thinking as Key for Introducing Hypothesis Testing—Evaluation of the Planned Secondary-School Reform in Hungary

“…Students would need to consider different options: eight cups or more, known or unknown number of cups with "tea first," all guessed correctly or not. In different experimental settings, the possible cases are different, which would lead to a different evaluation of the evidence from the actual experiment to judge the hypothesis that the Lady has a special skill of recognising how the tea was prepared (see Fanshawe, 2021). The probability of classifying all cups correctly may vary from 0.00024 to 0.0143 depending on whether eight or twelve cups are involved and whether the Lady is told the number of each preparation.…”

Artificial Neural Network Based Intelligent Tea Taster-Review

“…The tea is then evaluated for its appearance, aroma, and taste, with attention paid to factors such as color, strength, and any nuances or complexities in the flavor . Tea tasting can be used to identify the best tea leaves to use in blends, to evaluate teas for sale or to check the quality of tea after it has been packaged and stored (Fanshawe, 2021).…”