THE APPLICATION OF THE STATISTICAL METHOD TO PUBLIC HEALTH RESEARCH.

L. Dublin

DOI: https://doi.org/10.2105/AJPH.7.1.14

American Journal of Public Health

Abstract:T HE statistical method is today an indispensable tool for conducting scientific investigations. It is especially important in the fields of public health and applied sociology, where the data are often not quantitative but qualitative in character and where conclusions drawn are more likely to be subject to error than in the field of the more exact sciences. The statistical method, however, is valuable only in so far as it permits certain deductions to be drawn with precision, thus replacing impression or opinion. It is worse than useless if it does not accomplish this end; for its use gives the appearance of accuracy and reliability to the conclusions which are justified only by the soundness of the method itself. Too many fallacies are mistaken for truth because they are stated in statistical terms or are deduced through apparently accurate statistical procedures. This, indeed, has been the cloak of many an error which has obstructed the progress of good public health work in America. I propose in my part of the symposium to discuss the errors which are most important in themselves or are most frequently met with in the literature of the public health worker. In doing this I shall follow the sequence of the usual investigation. I shall consider, first, those errors which mar the investigation at the very outset, that is, those which arise in the planning of the inquiry, the preparation of the schedule and the collection of the data. Second, 1 shall point out those errors which are commonly made in the editing, classifying and tabulating of data. Finally, I shall consider some errors which occur in the analysis or interpretation of tabulated facts. My text and illustrations will be drawn from the current literature of our field.