Harvard School of Public Health - Yearbook (Cambridge, MA)

Page 21 text:

Pome of the more interesting definitions of statis- tics emphasize the process of making decisions in the presence of uncertainty. A statistician is, therefore, encouraged by the commonly observed phenomenon that the answering of one question in either basic or applied science seems to lead in¬ evitably to the asking of several new questions. This hydra-headed monster is the biostatistician’s friend. There is no need to worry about exhausting the world’s supply of uncertainty. The most urgent decisions to be faced by Public Health during the next few decades will arise out of the increasing size and mobility of human popu¬ lations combined with the increasing aspirations and potentials for the promotion of health through¬ out the world. Challenges for biostatistics will con¬ tinuously arise in terms of demands for more com¬ prehensive and more complex techniques for analyzing the data on which these decisions must be based. But time and again, it appears that the uncer¬ tainty in public health decisions does not come from a lack of analytic techniques. It arises from the lack of adequate and available observations to form a basis for decision. Current advances in the technology of data collection, transmission, storage Margaret Drolette preparing another lucid lecture. Robert B. Reed, A.B., A M., Ph.D., A.M. (hon.) and retrieval make it possible to think of obtain¬ ing these observations with the requisite speed, comprehensiveness and accuracy. Translating this possibility into a reality should be one of the most exciting developments in the future of public health. Robert Reed Dr. Sharratt attacking p and q

Page 20 text:

b a|e sin D B a a eHi Ibid c 5 t n « r- 1 OB ill SM 0! i to aa ilsjfji ‘IiHs T3 □ Tfr ■Ir i in [4:1 Seated —Margaret Drolette, Mary Nolan, Sheila Foley, Claire Wasserboehr. Standing —C. Ralph Bunches Jacob F Lee Slocum, Raymond Neff, Joseph Green, Jane Worcester.

Page 22 text:

...A Constitutional Stroll T iostatistics has arrived! One of its members, sig- nificance testing, has been added to the list of necessities. Everyone talks about significance, the journals insist on it and even doctors are taught tests of it. Every observation is soon challenged by the question, Are you significant?” The only dis¬ senters are the extremely cautious who insist that a real difference must be obvious without a test, an argument that accepts the basic skepticism of sig¬ nificance testing although it rejects the techniques. Acceptance has been accompanied by excessive enthusiasm and significance tests have appeared where they never ought to have. The null hypothe¬ sis that could not be rejected has been offered as proof of equivalence. Elsewhere significance tests have been taken too literally by the gullible who regard them as essential distillates, while the sub¬ stance of the data is relegated to insignificance.” Moreover, respectability imposed a canon of pro¬ priety. Editorial policy in some cases has rigidly prescribed a familiar format rather than one tai¬ lored to the needs of the problem and thereby robbed the techniques of their versatility. Regard¬ less of these imperfections, the utility of analysis of data according to mathematical models has been demonstrated. With computers as draft animals, biostatistics can turn to the revival of relative likelihood argu¬ ments, the development of non-parametric meth¬ ods and, possibly, enjoyment of the fruits of multi- David Heer solving a problem. variate analysis. Will these find their way into Biostatistics lab in the years to come? My guess is probably not. At least not in the sense that 6, t and p or q are found there today. As techniques of analysis become more varied and more demand¬ ing on the user, it becomes less and less justified to present a survey of their rudiments. Putting the £-test in the hands of everyone did not eliminate illogical or otherwise improper comparisons; rather it permitted these errors to be dressed up. The in¬ clusion of new techniques would only enlarge the wardrobe. Statistics is not for everyone, but logical observation is . Therefore the biostatistics course of the future may not be very different from the pres¬ ent one. The emphasis will still be on approach rather than techniques, but it will continue to be coupled with a constitutional stroll through a sig¬ nificance test. James Warram 18