Scholar Profile: Olive Jean Dunn (1915–2008)
Olive Jean Dunn was a pioneering American statistician and biomathematician whose work fundamentally changed how researchers interpret data across the biological and social sciences. While the "Bonferroni correction" is a household name in modern research, it was Dunn who formalized and popularized its application in simultaneous inference, ensuring that scientific discoveries were not merely the result of statistical fluke.
1. Biography: Early Life and Academic Trajectory
Olive Jean Dunn was born on September 1, 1915. Her academic journey was characterized by a deep, lifelong affiliation with the University of California, Los Angeles (UCLA). She earned her B.A. from UCLA in 1936, followed by an M.A. in 1951.
The gap between her undergraduate and graduate studies was common for women of her era, often involving teaching or family commitments. However, she returned to rigorous scholarship in the 1950s, a period when the field of statistics was undergoing a massive expansion due to the post-war data boom. She completed her Ph.D. in Mathematics at UCLA in 1956 under the supervision of Paul G. Hoel, a noted mathematical statistician.
Following a brief period as an assistant professor at Iowa State College (1956–1959), Dunn returned to UCLA. She spent the remainder of her career there, holding dual appointments in the School of Public Health and the Department of Biostatistics. She rose to the rank of Professor of Biostatistics and remained an emeritus professor until her death in 2008.
2. Major Contributions: Taming the "Multiple Comparisons" Problem
Dunn’s most significant intellectual contribution lies in the realm of Simultaneous Statistical Inference.
The Bonferroni Correction
In the mid-20th century, researchers faced a "multiplicity" crisis. If a scientist performs 20 different statistical tests, the laws of probability suggest that at least one of them will yield a "significant" result purely by chance (a Type I error). While the Italian mathematician Carlo Emilio Bonferroni had developed the underlying probability inequalities decades earlier, it was Olive Jean Dunn who, in 1961, applied these principles to the problem of multiple comparisons. She provided the mathematical framework and the necessary tables for researchers to adjust their significance levels (alpha) to account for multiple tests, effectively "correcting" for the risk of false positives.
Dunn’s Test
In 1964, she developed what is now known as "Dunn’s Test." This is a non-parametric post-hoc procedure used after a Kruskal-Wallis test has rejected the null hypothesis. It allows researchers to pinpoint exactly which groups in a study differ from one another without making the assumption that the data follows a normal (bell-shaped) distribution.
Discriminant Analysis
Dunn also made significant strides in discriminant analysis—the method of classifying an observation into one of several groups based on specific characteristics. Her work helped refine how researchers determine which variables are the best predictors of a specific outcome, particularly in clinical settings.
3. Notable Publications
Dunn’s bibliography includes seminal papers that bridge the gap between abstract mathematical theory and practical biostatistical application.
- Estimation of the means of dependent variables (1958): Published in the Annals of Mathematical Statistics, this paper laid the groundwork for her work on confidence intervals.
- Confidence intervals for the means of dependent, normally distributed variables (1959): This paper in the Journal of the American Statistical Association (JASA) introduced the application of Bonferroni inequalities to confidence intervals.
- Multiple comparisons among means (1961): Her most cited work, this JASA paper formalized the "Bonferroni correction" for the scientific community.
- Multiple comparisons using rank sums (1964): Introduced the non-parametric "Dunn’s Test."
- Textbooks:
- Basic Statistics: A Primer for the Biomedical Sciences (1964, co-authored with Virginia Clark). This became a standard text for medical students for decades.
- Applied Statistics: Analysis of Variance and Regression (1974, co-authored with Virginia Clark).
4. Awards & Recognition
Though Dunn worked in an era when female mathematicians often received less public acclaim than their male counterparts, her peers recognized her immense value:
- Fellow of the American Statistical Association (ASA): Elected in 1968, a significant honor reserved for those who have made outstanding contributions to statistical science.
- Fellow of the American Association for the Advancement of Science (AAAS): Recognized for her contributions to the application of statistics in the public health sector.
- Fellow of the American Public Health Association.
- UCLA Woman of the Year (1974): A testament to her impact on the university community and her role as a mentor to women in STEM.
5. Impact & Legacy
Olive Jean Dunn’s legacy is embedded in the software and methodologies used by every modern scientist. Whenever a researcher clicks "Adjust for Multiple Comparisons" in a statistical program like SPSS, R, or SAS, they are utilizing the logic Dunn perfected.
Her work was particularly transformative for medicine and public health. Before Dunn, many clinical trials likely reported "breakthroughs" that were actually statistical noise. By providing a rigorous way to handle multiple variables, she brought a new level of integrity to medical research. Furthermore, her textbooks helped demystify statistics for non-mathematicians, training generations of doctors and public health officials to think critically about data.
6. Collaborations
Dunn was a collaborative researcher who understood that biostatistics is, by definition, a team science.
- Virginia A. Clark: Her most enduring partnership was with fellow UCLA professor Virginia Clark. Together, they authored two of the most influential biostatistics textbooks of the 20th century.
- The UCLA School of Public Health: Dunn was instrumental in building the Biostatistics department at UCLA into a world-class center, collaborating with epidemiologists and clinicians to apply her theoretical models to real-world health data.
7. Lesser-Known Facts
- Late-Blooming Career: Dunn did not complete her Ph.D. until she was 41 years old. Her career serves as a powerful example that "late starts" in academia can still lead to world-changing contributions.
- The "Bonferroni" Misnomer: While the method is named after Carlo Bonferroni, he never actually wrote about "multiple comparisons" in the context of statistical testing. It was Dunn’s 1961 paper that popularized the term "Bonferroni" in this context. Some historians of science argue that the procedure should more accurately be called the "Dunn-Bonferroni Correction."
- Table Pioneer: Before computers were ubiquitous, researchers relied on printed tables to find critical values for their tests. Dunn spent years calculating and publishing the extensive tables for the t-distribution used in multiple comparisons, a painstaking task that made her methods accessible to the average researcher.