User:HamiltonRoberts/sandbox

Robert Sinclair Rodger is most well-known for developing the statistical methodology that bears his name - Rodger's method. References to his most important published articles are given in the Wikipedia entry on Rodger's method. He and his family have been Canadian citizens since the late 1970's, and he lives most of the time in Halifax, Nova Scotia, Canada. He can be contacted at: rodgermethod@gmail.com.

Robert S. Rodger was born in Glasgow, Scotland and raised in the county of Fife. During World War 2 he volunteered and joined the county regiment, the Black Watch – and still holds a King's Honorary Commission in that regiment. He served in Africa and Germany. Upon returning home, Robert was persuaded to become a school teacher, and taught in numerous Fife schools full time for three years (and part time for an additional three and a half more years). He was admitted to the University of Edinburgh, where he studied philosophy, psychology, mathematics, and statistics and graduated with a First-class Honours M.A. in psychology.

He was then offered an academic appointment in psychology at the Queen's University of Belfast. Robert had married his wife, Margaret, three years earlier, and their two children (Mark and Susan) were born in Belfast. He also took his Ph.D. at Queen's. Probably as a result of his first book, Statistical Reasoning in Psychology (first edition published in 1961), Robert was offered a more senior academic appointment at the University of Sydney, New South Wales. The family relocated to Australia, where Robert's concept of a new statistical criterion was discovered.

Suppose an imaginary researcher (Dr. Norm Tee) always planned and carried out his research very carefully. And every research report used only two samples, which were always independent of his previous samples. If every test of the null hypothesis was a 5% t-test, then Bernoulli’s theorem says that the probability of Tee having rejected one or more of 31 independent nulls, given that they are all true, must be x = 1-(0.95)31 = 0.80. That mathematics is beyond doubt, and what Dr. Norm Tee does, paper by paper and overall, is beyond reproach (although it would have been more efficient to use more than two samples at a time). Why then should Tee's cousin, the equally imaginary Dr. Post Hoc, use F0.05;31,128 if he analyzes all 32 of his own samples (each with N = 5) altogether? Should he not use F0.80;31,128 = 0.769? These ideas led to the publication of tables of Fx;ν1,ν2 for x = 1-(1-α)ν1 for α = 0.05 and 0.01 in Robert's second book, Intermediate Statistics (published in 1964), and two subsequent articles in 1967 in the British Journal of Mathematical and Statistical Psychology (the first of these was on type I error rate and the second one was on type II error rate). In 1966, the University of Melbourne awarded Robert the John Smyth Memorial Medal “for contributions to education,” so he flew to Melbourne to give the John Smyth Memorial Lecture on ‘Statistical Data Snooping.’ It was while flying over the Australian Alps (drinking a small glass of Australian sherry), that he realized that he had been evaluating his new method by computing the average rate (i.e., expectation) at which his method rejected true null contrasts. He recognized that the expectation was a far better criterion than x = 1-(1-α)ν1. This was so because, first, it addresses the matter of primary interest (number of nulls rejected) rather than a less relevant, side issue (probability of rejecting one or more nulls). Second, since the expectation is an average, it is likely to be far more stable (i.e., 'robust') than a quantity in the tail of the distribution.

In 1969 Robert received a sabbatical leave and a Killam Senior Fellowship to do research at Dalhousie University in Halifax, Nova Scotia. He took ill while in Halifax; so was unable to complete the research he had planned. When his request for an (unpaid) extension to his leave was denied, he resigned from Sydney and joined the permanent Dalhousie academic staff. Statistical articles continued to appear, including one on a non-parametric application of Rodger's method (1969), a re-evaluation of the analysis of factorial design data (1974), two papers with tables of the new, expectation-based F[Eα];ν1,ν2 and Δ[Eβ];ν1,ν2 (1975), and two (in 1976 and 1978) with tables that permit researchers to actually use the (σ-free) method of two-stage sampling developed by Charles Stein under a cloak of United States military secrecy by the end of World War II. During his tenure at Dalhousie University Robert continued to teach full time, served for six years as chairperson of the psychology department, and played an instrumental role in the formation of a faculty union under the Trade Union Act of Nova Scotia. Robert was persuaded to head up the certification process, which was successful in 1978. He was the second president of that association (i.e., union), served as a negotiator, and wrote every word of the first collective agreement. Following that he continued to advise and assist the Dalhousie Faculty Association for about ten years, and was also recruited at the national level to participate in collective bargaining conferences. In addition, Robert became the first president of the Canadian Association of University Teachers Collective Bargaining Cooperative, that assisted faculty unions across the country.

Canada's 1982 Constitution says there shall be no discrimination against people based on their age, so the issue of compulsory retirement had been brought to the courts. Following the Supreme Court of Canada’s ruling that forced retirement was a reasonable limitation on this particular freedom in a free and democratic society, Dalhousie University was able to compel Robert’s retirement from his academic career when he turned 65 years of age (he remains an adjunct professor there, however ). Following this retirement, Robert became a part time ‘consultant statistician’ to a department of the federal government of Canada, and he stayed in that post until he voluntarily retired five years later. Robert S. Rodger continues his interest in statistical matters. He is the primary author of the Wikiversity entry on Rodger’s method (http://en.wikiversity.org/wiki/Rodger%27s_Method), and his most recent statistics-related article, on the 'power' superiority of Rodger's method for multiple comparisons, has been accepted for publication in the Journal of Methods and Measurement in the Social Sciences Vol.4(1) in 2013.