(SMR¼ 0 :445 or 44.5%). This phenomenon, known as the healthy
worker e¤ect, is probably a consequence of a selection factor whereby
workers are necessarily in better health (than people in the general popu-
lation) at the time of their entry into the workforce.- We see an attenuation of the healthy worker e¤ect (i.e., a decreasing
trend) with the passage of time, so that the cancer death rates show a
slight excess after 15 years. (Vinyl chloride exposures are known to in-
duce a rare form of liver cancer and to increase rates of brain cancer.)
Taking the ratio of two standardized mortality ratios is another way of
expressing relative risk. For example, the relative risk of the 15þyears group is
1.58 times the risk of the risk of the 5–9 years group, since the ratio of the two
corresponding mortality ratios is
111 : 8
70 : 6¼ 1 : 58
Similarly, the risk of the 15þyears group is 2.51 times the risk of the 1–4 years
group because the ratio of the two corresponding mortality ratios is
111 : 8
44 : 5¼ 2 : 51
1.4 NOTES ON COMPUTATIONS
Much of this book is concerned with arithmetic procedures for data analysis,
some with rather complicated formulas. In many biomedical investigations,
particularly those involving large quantities of data, the analysis (e.g., regres-
sion analysis of Chapter 8) gives rise to di‰culties in computational imple-
mentation. In these investigations it will be necessary to use statistical software
specially designed to do these jobs. Most of the calculations described in this
book can be carried out readily using statistical packages, and any student or
practitioner of data analysis will find the use of such packages essential.
Methods of survival analysis (first half of Chapter 11), for example, and
nonparametric methods (Sections 2.4 and 7.4), and of multiple regression
analysis (Section 8.2) may best be handled by a specialized package such as
SAS; in these sections are included in our examples where they were used.
However, students and investigators contemplating use of one of these com-
mercial programs should read the specifications for each program before
choosing the options necessary or suitable for any particular procedure. But
these sections are exceptions, many calculations described in this book can be
carried out readily using Microsoft’s Excel, popular software available in every
personal computer. Notes on the use of Excel are included in separate sections
at the end of each chapter.
NOTES ON COMPUTATIONS 31