Introduction

Abstract

The following two papers are written versions of talks given by my father, Sam Greenhouse. It is an honour and a privilege to be able to include them as part of this special issue of Statistics in Medicine dedicated to him. Although these talks were not part of the NIH symposium on 'Perspectives on the Biostatistical Sciences', it seemed ÿtting that in the ÿrst NIH biostatistics symposium in which my father did not participate, he is still able to contribute to the proceedings. The purpose of this note is to provide some background and context for these talks.In late summer 1999, Nan Laird, then Chair of the Harvard University Department of Biostatistics, informed my father that he had been selected for recognition by the Harvard University Institute of Psychiatric Epidemiology and Genetics for his lifetime contributions to psychiatric epidemiology and biostatistics. This was a very meaningful and deeply appreciated recognition. As part of the award ceremony, he was invited to give a lecture on a topic of his choice. Instead of a technical talk, he felt the occasion called for something more re ective and personal. He chose to discuss early developments in psychiatric epidemiology and statistics based on his own experiences at the National Institute of Mental Health during the 1950s and 1960s. Shortly after speaking with Nan, my father learned that he had oesophageal cancer. Although he would not be able to deliver this talk in person, he nevertheless worked almost daily on a written version which he asked me to deliver for him.† The paper that appears here, titled On psychiatry; epidemiology; and statistics: a view from the 1950s and 1960s, was almost entirely written by him. One section, on the role of hypothesis testing in mental health clinical trials, is adapted from some of his earlier writings. This paper nicely complements the papers by Butler [1] and Katz and Berman [2] that also appear in this issue.The second talk that appears here preceded the Harvard talk by nearly 20 years. In 1982, Ted Colton invited my father to present the ENAR Presidential Address. It has been surprising to me the number of people who still remember that talk and encouraged him and later me to publish it. It obviously left a lasting impression on many of those who heard it. The title of that talk was The Growth and Future of Biostatistics and I have taken the liberty here of adding a subtitle, A View from the 1980s, in part, to help set the context but also for the nice parallelism with the title of the Harvard talk. A central theme in this talk is his concern with the problem of selection bias as it arises in observational studies and in 'broken' clinical trials. In e ect, he anticipates the growth and development of research in the area of what is now called causal inference.