Gottman, John M. Time-series analysis. A comprehensive introduction for social scientists. Paperback reprint of the 1982 original. (English) Zbl 1183.00005 Cambridge: Cambridge University Press (ISBN 978-0-521-10336-7/pbk). xvi, 420 p. (2009). This book is a comprehensive introduction to the linear time-series techniques, both time-domain and frequency-domain. It is designed for people who are not necessary very familiar with mathematics but need to use statistical analysis of data in their research or professional life. The goal of the book is to softly introduce a reader into the world of time-series analysis. The construction of the text is a natural consequence of the desire to make reading easier for an unprepared reader. The chapters are short and usually focus only on one new concept and end by examples and a summary. Most technical considerations as well as some mathematical primers are thrown to appendices. There are many illustrations helping the readers to build their intuition. The text is written in a lively and accessible way, and surely can be useful for people beginning their work in the field. The first part of the book includes 6 chapters. It is, according to the author’s own words, “an advertisement for time-series analysis”. It introduces some concepts and develops the reader intuition by means of a sequence of well selected and accessible examples. The second part of the book deals mainly with the concept of stationarity. Its last chapter takes up the problem of modeling non-stationary data and suggests some possible solutions - the use of general models with deterministic and stochastic trends. In part III, stationary time-domain models are discussed. The author shows here how the time series dynamics becomes more complicated, introducing moving average (MA), autoregressive (AR), and ARMA models. The duality of the MA and AR specifications is also discussed. The next four chapters form part IV which is devoted to stationary frequency-domain models. The notions of the spectral density function, the periodogram, and the spectral windows are discussed. The part ends with an attempt to integrate time and frequency domains by deriving the spectral density functions of moving average and autoregressive processes. In part V, estimating the parameters of time-domain models is concerned. In chapter 19, the Mann-Wald, Box-Jenkins and Wu-Pandit approaches are presented. Chapter 20 introduces the Box-Jenkins ARIMA models. Next, linear least-squares forecasts are discussed, and the last chapter of this part contains a worked example of model fitting. The last two parts VI and VII deal with multivariate time-series analysis. In part VI, the bivariate frequency-domain and time-domain analysis are discussed and illustrated by a worked example on mother-infant play. The last two chapters 26 and 27 deal, respectively, with the interrupted time series experiment, and multivariate time-series regression analysis. All the topics considered in the book form an interesting and well-prepared practical handbook. It should be noticed, however, that this book was first published in 1981 and remains unchanged since then. In the meantime, there has been an explosion of new ideas in time-series analysis, including cointegration and conditional heteroskedasticity, which have been intensively developed for almost three decades. Thus, the word “Comprehensive” in the subtitle of the book is quite unfounded nowadays. Nevertheless, the book can still be considered as a good starting point for all those who want to become acquainted with the methods and models of time series analysis. Reviewer: Malgorzata Doman (Poznań) Cited in 1 Review MSC: 00A06 Mathematics for nonmathematicians (engineering, social sciences, etc.) 62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH) 62-01 Introductory exposition (textbooks, tutorial papers, etc.) pertaining to statistics 91B84 Economic time series analysis 91F99 Other social and behavioral sciences (mathematical treatment) Keywords:time series; autoregressive; moving average; ARMA; ARIMA; stationarity; Slutzky effect; spectral density function; periodogram; spectral window; OLS; Yule-Walker equations; forecasting; multivariate linear models Citations:Zbl 1167.00300 PDFBibTeX XMLCite \textit{J. M. Gottman}, Time-series analysis. A comprehensive introduction for social scientists. Paperback reprint of the 1982 original. Cambridge: Cambridge University Press (2009; Zbl 1183.00005)