Content area
Full Text
Modern Engineering Statistics by Lawrence L Lapin. Duxbury Press, Pacific Grove, CA, 1997. xi+583 pp. $81.95.
Reviewer: Lloyd S. Nelson, Statistical Consultant, Londonderry, NH 03053-3647.
IF texts such as this were judged on the basis of how many examples and problems they contained, this one (with well over 600) would be a winner. Nevertheless, other considerations turn out to be somewhat detracting. The author's stated goal is to make it easy for engineering students to assimilate statistics.
A great deal of emphasis (two full chapters) is placed on regression without the usual (and necessary) caveats. Students using this book are not dissuaded from thinking that regression can be used to mine plant data usually thought of as "getting the gold out of the mine." What they overlook is that the objective of the plant manager is to hold everything as constant as possible. The usual result is for a regression analysis to find very few, if any, statistically significant effects. Further, despite how important all the regression information from a computer program looks, it does not follow that it has relevance to the real problem or that the data has given up all its information. Experts in this field hand down the following advice about the limitations of regression: For those who know, no warning is necessary; for those who don't, no warning is sufficient!
To deal with the situation in which a manager directs a subordinate to perform a regression analysis, it sometimes suffices to apply a quick nonparametric test to a sample of the data in question. This can often avoid disappointment and embarrassment. Indeed, I think that one of the regression chapters should have been replaced with one on nonparametric procedures. A practicing engineer will find many applications for sign tests, rank tests, etc.
On the plus side, the author writes very well, and the splendid figures make it very easy to follow the line of reasoning. On the negative side, there are some statements that are incomplete, misleading, or wrong. Examples of some of these follow.
The student is told on page 52 that...