Showing papers in "Sociological Methodology in 1973"

Problems of Statistical Estimation and Causal Inference in Time-Series Regression Models

Abstract: I am grateful to Arthur Goldberger and Robert Hall for comments on an earlier draft, to Franklin Fisher for stimulating discussions on many of the topics treated, to Samuel Popkin and Lawrence McCray for valuable editorial suggestions, and to Raisa Deber, Robert Eccles, and Takashi Inoguchi for research assistance. A timely grant from the office of Provost Walter Rosenblith covered production expenses.

Approaches to the Interpretation of Relationships in the Multitrait-Multimethod Matrix

Abstract: The author was supported by a National Institute of General Medical Sciences Training Program in Methodology and Statistics (GMO-1526) and by postdoctoral support from Professors William H. Sewell and Robert M. Hauser during the writing of this chapter. Computer analyses were supported by a grant to Professor Sewell from the National Institute of Mental Health (M-6275). Computer facilities were provided by the Madison Academic Computing Center. The author wishes to thank Robert M. Hauser, Arthur S. Goldberger, and Robert P. Althauser for making invaluable comnments on an earlier version of this chapter. The responsibility for the contents rests entirely with the author.

Questions about attitude survey questions

Abstract: Survey investigators are aware that there are many ways to word the same question and that these various wordings can produce different response distributions. Stouffer (1955) notes that "even slight variations in the wording of a question can produce variations in response, just as is the case with questions used by the Census to ascertain whether a person is in the labor market and seeking work. In this book we have tried to pin a finding not upon one question alone but upon a variety of ques-

Construction of Composite Measures by the Canonical-Factor-Regression Method

Abstract: This is an extended version of a paper presented to the Methodology Section of the American Sociological Association at the 1973 annual meeting in New York. The author is indebted to David R. Heise, Herbert L. Costner, Arthur S. Goldberger, Hubert M. Blalock, and Lois MacGillivray for their critical and constructive comments on earlier versions of this paper. Any errors, of course, are the sole responsibility of the author.

Spectral Analysis and the Study of Social Change

Abstract: Can rigorous quantitative methods be used to study historically significant processes of social change? If such a possibility exists it must surely involve the statistical analysis of time-series data. For many important social processes, highly aggregated time series are virtually the only quantitative indicators presently available. Fundamentally, any quantitative record of social change is a time series of sorts.

Some issues in sociological measurement

Abstract: For a discipline that ignored errors of measurement too long, suffering the consequences of low predicti-ve capability and spurious theoretical debate, sociology has recently shown a remarkable turnabout, investing an appreciable proportion of its scholarly pages to analyses of the measurement problem. Of course the concern is not entirely unprecedented: the team that produced the monumental The American Soldier showed a keen and sophisticated interest in measurement problems with eventual fruition in Paul Lazarsfeld's latent structure analysis and Louis Guttman's scalogram procedures (Guttman's later work seems less associated with sociometrics than with psychometrics). Yet in the years since World War II the potential impact of latent structure analysis was little realized because only a few passing through Columbia University were exposed to it (a situation remedied recently by Henry and Lazarsfeld, 1968). Guttman scales suffered another fate, becoming for a decade a fetish among American sociologists, applied so often where the underlying model was inappropriate that some metricians turned to discrediting the technique generally (Nunnally, 1967).

Inferring Validity from the Multitrait-Multimethod Matrix: Another Assessment

Abstract: Campbell and Fiske's (1959) elaboration of discriminant validity and the multitrait-multimethod (MTMM) matrix creatively joined two among many conceptualizations of valid measurement. Variables purporting to measure one underlying trait should not be highly correlated with measures of supposedly different traits. Measures should not vary because of the methods of measurement used. These two conceptions were brought together by the device of confounding one trait and one method in each of several measures.

Alternative Approaches to Analysis-of-Variance Tables

Abstract: It goes without saying, although it is usually said, that any errors in this chapter are the sole responsibility of the authors; the merits of the chapter, on the other hand, are the partial responsibility of the following persons who graciously supplied brief comments on a preliminary draft: T. A. Bancroft, 0. D. Duncan, Holly Fuchs, and David Knoke. Our work was carried out under the auspices of an NIMH grant for training social scientists (MH10577), although that grant provided no direct financial support.