Student and the Lanarkshire milk experiment.

Area of Science:

• Statistics
• Epidemiology
• Experimental Design

Background:

• The 1930 Lanarkshire Milk Experiment (LME) investigated the effects of raw vs. pasteurized milk.
• William Sealy Gossett ("Student") provided a statistical critique in 1931.
• The LME involved a complex design with school and pupil-level allocation.

Purpose of the Study:

• To re-examine the 1930 Lanarkshire Milk Experiment from a modern statistical perspective.
• To evaluate "Student's" criticisms in light of contemporary experimental design.
• To draw lessons for observational studies in epidemiology.

Main Methods:

• Re-analysis of the 1930 Lanarkshire Milk Experiment's design and statistical approach.
• Examination of "Student's" criticisms concerning incomplete block structures.
• Analogy drawn with modern clinical trials, such as in osteoarthritis research.

Main Results:

• "Student's" criticisms are assessed based on advancements in experimental design and analysis.
• The importance of appropriate standard error estimation in complex experimental structures is highlighted.
• Potential explanations for variability in observational studies, including hidden clustering, are discussed.

Conclusions:

• The LME's complex design offers enduring lessons for statistical analysis.
• Modern statistical methods can provide deeper insights into historical experiments.
• Hidden clustering may explain inconsistencies in observational epidemiological studies.