Book Reviews

Strecker H: Statistische Erhebungen: Methoden und Ergebnisse, Ausgewählte Schriften - Survey Methods of Statistics, Selected Papers. Herausgegeben von (edited by) Beckmann MJ, Wiegert R (Angewandte Statistik und Ökonometrie; Heft 30), Vandenhoeck & Ruprecht, Göttingen 1987, p. VIII+ 341, DM88,-

The selection of papers by Heinrich Strecker appeared on the occasion of his 65th birthday. It concentrates on the two main interests of Heinrich Strecker: the plied theory of sampling and the quotient method in time series analysis. IWo published papers on errors in statistical data and on the determination and estimation of the response variance, both with R. Wiegert as joint author, are cluded.

The selection of published papers does not contain any parts of books by Heinrich Strecker. We find eight articles covering the period 1949 to 1984 together with two chapters of "Stichproben in der amtlichen Statistik" (1960 edited by the Statistische Bundesamt) and a paper delivered at the 43rd meeting of the tional Statistical Institute 1981 in Buenos Aires. IVvo of the articles concern the quotient method, which is closely related to the variate difference method, while all other papers are devoted to the subject of main interest for Strecker, the methodology of statistical surveys. Most of Strecker's appHcations come from agriculture, more exactly live-stock-counting. For many years Heinrich Strecker has cooperated with the Statistical Office of the Kingdom of Belgium, and the experience gathered hereby can be found in this volume. This is true on one hand with respect to comparing the efficiency of different sample models - with equal and with varying probability of selection. On the other hand this concerns surement of the response variance which should be taken into account in all surveys, but in many cases is neglected. Reliable estimates of this component of the variance are only possible if post-enumeration surveys of high accuracy can be made, a very expensive task. Therefore the experience gained from Strecker's work in Belgium is very valuable. The articles collected in this volume together with a book by Strecker et al. (Messung der Antwortvariabilität aufgrund von Erhebungsmodellen mit Wiederholungszählungen, 1983) provide great insight to methodology and results of this important aspect of sampling.

Special attention should be given to the two articles which are published for the first time. "Errors in Statistical Data - Illustration by Examples" describes a classification distinguishing between explicit errors and implicit errors. Explicit errors can be quantitatively determined by comparison with data of higher curacy. Implicit errors are those which "cannot be explicitly presented". Among these errors - by far the most frequent ones - he differentiates between evident errors "visibly manifest in the data themselves" and latent errors, which "do not in any way visibly emerge". Examples for the different types of errors are found not only in agricultural statistics but mainly in population censuses. Here the