Time-Series Analysis: A Comprehensive Introduction for Social Scientists

by John M. Gottman

• ISBN13: 9780521103367
• ISBN10: 0521103363
• Edition: 1st
• Format: Paperback
• Copyright: 2009-03-19
• Publisher: Cambridge University Press

Summary

Since the 1970s social scientists and scientists in a variety of fields - psychology, sociology, education, psychiatry, economics and engineering - have been interested in problems that require the statistical analysis of data over time and there has been in effect a conceptual revolution in ways of thinking about pattern and regularity. This book is a comprehensive introduction to all the major time-series techniques, both time-domain and frequency-domain. It includes work on linear models that simplify the solution of univariate and multivariate problems. The author begins with a non-mathematical overview: throughout, he provides easy-to-understand, fully worked examples drawn from real studies in psychology and sociology. Other, less comprehensive, books on time-series analysis require calculus: this presupposes only a standard introductory statistics course covering analysis of variance and regression. The chapters are short, designed to build concepts (and the reader's confidence) one step at a time. Many illustrations aid visual, intuitive understanding. Without compromising mathematical rigour, the author keeps in mind the reader who does no have an easy time with mathematics: the result is a readily accessible and practical text.

Table of Contents

Preface

Overview

The search for hidden structures

The ubiquitous cycles

How Slutzky created order from chaos

Forecasting: Yule's autoregressive models

Into the black box with white light

Experimentation and change

Time-series models

Models and the problem of correlated data

An introduction to time-series models: stationarity

What if the data are not stationary?

Deterministic and nondeterministic components

Moving-average models

Autoregressive models

The complex behaviour of the second-order autoregressive process

The partial autocorrelation function: completing the duality

The duality of MA and AR processes

Stationary frequency-domain models

The spectral density function

The periodogram

Spectral windows and window carpentry

Explanation of the Slutzky effect

Estimation in the time domain

AR model fitting and estimation

Box-Jenkins model fitting: the ARIMA models

Forecasting

Model fitting: worked example

Bivariate time-series analysis

Bivariate frequency-domain analysis

Bivariate frequency example: mother-infant play

Bivariate time-domain analysis

Other Techniques

The interrupted time-series experiment

Multivariate approaches

Notes

References

Index

Table of Contents provided by Publisher. All Rights Reserved.