Adrian Raftery will succeed Peter Marsden as editor of Sociological Methodology beginning with the 1996 issue of the annual. Members of the ASA Section on Methodology and ASA members generally thank Peter for the outstanding job he has done during his five-year tenure at Sociological Methodology. When I interviewed Adrian about his new duties he said, "I am very excited about editing Sociological Methodology. It is a successful journal which has been developing rapidly under the last three editors" [Nancy Tuma, Clifford Clogg, and Peter Marsden]. He reminded me of a study that the Section of Methodology did a few years ago that showed that Sociological Methodology has the third highest citation rate of any serial publication in sociology, right behind American Sociological Review and American Journal of Sociology.

Sociological Methodology has fostered the development and dissemination of methodological developments from path analysis to structural equation models to loglinear models to event history analysis. Adrian hopes to continue that role for Sociological Methodology during his tenure. He see qualitative methods, data collection, and computer intensive methodologies as potentially fruitful new sources of articles and innovation. In my interview with him, Adrian also said, "Recent statistical research has led to algorithms that allow us to exploit the richness of modern data sets by analyzing individual, contextual, and societal effects at the same time. This started with the EM algorithm and the bootstrap and more recent advances include the Gibbs sampler, Markov chain Monte Carlo methods, neural networks, and wavelets. No one knows yet if these are useful for sociology, and I hope that Sociological Methodology will take the lead in finding out and in extending them to fit the needs of sociologists." He said he also hopes that applications articles will cover a broader range of the field, "For example, medical sociology is a rapidly growing area with its own needs, and there are exciting methodological developments in macrosociology."

A little background on the new Sociological Methodology editor: Adrian is a native of Dublin, Ireland, and retains his Irish citizenship. He is Professor of Statistics and Sociology at the University of Washington. He began teaching at the University of Washington in 1985 while on sabbatical from Trinity College, Dublin. He was appointed Associate Professor of Statistics and Sociology the following year and promoted to full Professor in 1990. He studied at Trinity College, Dublin (BA, 1976; MSc, 1977), and the Universite de Paris VI (Docteur De Troisieme Cycle in probability and mathematical statistics, 1980). He is married and has a daughter and a son.

Adrian is a prolific writer. He has authored or co-authored 51 refereed publications, 11 more articles that are under review, and 27 other items. His sociological writings are on social mobility, educational stratification, model selection, and the environment.

Adrian is best known in sociology for his Bayesian Information Coefficient (BIC), described in the ASR in a two-page comment that might have the highest ratio of citations to pages in the annals of sociology. The appeal of Raftery's BIC is how it provides a simple solution to a vexing problem. Most sociologists use statistical models to analyze data in order to describe the main features of the topic at hand. When they use the classical tools of statistical inference to test their models, they use the wrong criterion. Classical methods, e.g., the likelihood-ratio test for a loglinear model, "are designed to detect any discrepancies between the model and the observed data. With a large enough sample, the likelihood-ratio test will find them and reject even a good model...The problem is we should be comparing the models, not just looking for possibly minor discrepancies between one of them and the data" (Raftery, 1986, p. 145). Raftery's BIC is the solution, and his approximation to BIC for the case of choosing a loglinear model was so simple [L2 - dfln(N)] that it has been applied hundreds of times in the eight years since its publication. Adrian continues to work on BIC and will publish two significant articles next year in the Journal of the American Statistical Association. The goal of this work is to get sociologists to recognize the uncertainty that surrounds the choice of any model.

I met Adrian at the 1984 meetings of the International Sociological Association Research Committee on Stratification and Mobility (RC28). His presentation, subsequently published in Quality and Quantity as "Social mobility measures for cross-national comparisons" was so clear that I could tell in minutes not only that I disagreed with him but why. Clarity has proved to be a lasting trait; disagreement has not. That winter we began a collaboration that resulted in an National Science Foundation grant, three published articles, and several more conference papers, mostly on educational stratification. Our most recent publication (Sociology of Education, February 1993) advances the hypothesis that class barriers to educational attainment are a function of the supply and demand at each level of education. From the premise that privileged parents have a strong interest in the advancement of their own children but almost none in the perpetuation of any existing class barrier, we derive four hypotheses that we give the collective label, "Maximally Maintained Inequality" or "MMI." Our work is on Ireland, but the application of MMI to 13 other countries by the contributors to Persisting Inequalities, edited by Hans-Peter Blossfeld and Yossi Shavit (Westview, 1993), and recent work on Russia, shows it to be a very promising perspective on educational stratification.

Adrian's work on environmental issues is perhaps less well-known to sociologists. That is our loss. He has done important work in the field of alternative energy, including solar and wind power. He has criticized estimates of how safe nuclear power plants are (more exactly he has pointed out that the human population is not as safe as calculations about the probability of an accident at any one nuclear plant suggest because the risk accumulates across plants, adding up to a dangerously high probability of a major accident in any given ten-year period). In 1991 he published an important article on industrial accidents.

Adrian's most important work on environmental issues is his contribution to estimating the population of bowhead whales (Balaena Mysticetus). This work has resulted in articles in life sciences and environmental journals as well as in the Report of the International Whaling Commission.

Adrian has been an exemplar of the developing tradition of making software available to the research community immediately upon the publication of new methods. He archived his software for the analysis of mobility tables as early as 1983, and he has made BIC software available through StatLib, an archive for free statistical software on the Internet. In my interview with him, Adrian said that as Sociological Methodology editor, he intends to impress authors with the need for this kind of dissemination. "I will encourage authors to make software that implements their methods easily available, for example by anonymous ftp or via StatLib."

Adrian will be happy to talk with prospective authors about possible articles. The best way to reach him, though, is through the new node he has set up on the Internet. Just address e-mail to: socmeth@stat.washington.edu