Judith M. Tanur

Cognitive psychology meets the national survey.

Because of their mutual interests, cognitive psychologists and survey researchers might have been expected to have a long history of collaboration. In fact, only recently have attempts been made to foster a dialogue between the two disciplines. This article describes some of the collaborative efforts that have been initiated and shows how mutually beneficial work might be conducted in this fascinating interdisciplinary area. In many national surveys, respondents are asked to recall events from their lives. For example, in the National Health Interview Survey, a major government-sponsored sample survey designed to obtain information on the health of Americans, respondents are asked, During the past 12 months, about how many times did you see or talk to a medical doctor? or During the past 12 months, did anyone in the family have an ulcer? In this and other surveys conducted by the National Center for Health Statistics, people have been asked to recall the occurrence of health-related conditions and their consequences, such as days lost from work and/or school. The resulting survey estimates of health problems and the associated utilization of health care services were used in the formulation of the legislation for both the Medicare and Medicaid programs. In the National Crime Survey, a major statistical series designed to obtain information on the incidence of crime and its impact on society, respondents are asked about victimizations that may have occurred in the last 6 months. Questions such as these are asked: In the last 6 months, did anyone try to rob you by using force or threatening to harm you? and In the last 6 months, did anyone beat you up, attack you, or hit you with something, such as a rock or bottle? This effort told us that, in 1981, 41 million crime victimizations occurred and almost a third of all households were victimized by violence or theft (U.S. Department of Justice, 1983). In election day exit polls, voters are asked to report the timing of an internal event--deciding for whom to vote. For example, during the 1984 primary season NBC News asked voters in primaries, Voters choose their candidates at different times during an University of Washington Carnegie-Mellon University State University of New York at Stony Brook

From the inside out and the outside in: Combining experimental and sampling structures

The three basic tenets of experimental design (randomization, replication, local control) find parallels in sampling design. While the ways these parallel structures are applied differ across the two areas, the commonalities suggest ways of strengthening work in both. We describe examples of embedding sampling within experiments, the use of experimental design structures within sample surveys (including interpenetrating networks of samples), and the generalization from experiments to populations through random selection and sampling. A more complicated connecting structure between experiments and surveys is illustrated by the U.S. Census Bureau model for nonsampling errors. We conclude with a description of alternative approaches to basic inference in complicated embedded structures.

A Note on the Partial Correlation Coefficient

At the dividing point a = 1-the non-modal uniform distribution-k = 1.2; for a oo. This example clearly confirms Darlingtons point. However, consider a symmetric two-tailed gamma distribution:

The Design and Analysis of Longitudinal Surveys: Controversies and Issues of Cost and Continuity

Longitudinal survey data can arise in many different settings, e.g., from rotating panel surveys, in cohort studies, and in the context of field experiments that involve economic and social phenomena that change over time. In all of these settings the longitudinal feature implies repeated interviews of respondents from nonstationary populations, and both panel attrition and missing data present special concerns. The issues here are ones involving both design and analysis. Among the design issues in a longitudinal survey is how to achieve a high degree of data continuity by following movers, when the cost of such continuity is high. If the sampling units of interest are groups as opposed to individuals, there is often a critical need for operational definitions of “family” and “household”, because the concepts are dynamic and change over time. Among the analysis issues addressed in the paper are (i) the use of longitudinal vs. cross-sectional methods of imputation and adjustment for missing values, and (ii) the use of weights in longitudinal analyses to adjust for unequal probabilities of selection and nonresponse.

Transactional sex and prevalence of STIs: a cross-sectional study of MSM and transwomen screened for an HIV prevention trial.

Few studies have characterised the degree of engagement in transactional sex among men and transgender women who have sex with men and explored its association with sexually transmitted infections and human immunodeficiency virus in Ecuador. We screened 642 men who have sex with men and transgender women for a pre-exposure prophylaxis clinical trial (iPrEx) in Guayaquil, Ecuador, 2007–2009. We analysed the association of degree of engagement in transactional sex and prevalence of sexually transmitted infections including human immunodeficiency virus using chi-square and analysis of variance tests. Although just 6.2% of those who screened self-identified as sex workers, 52.1% reported having engaged in transactional sex. Compared to those who had never been paid for sex, those who had been paid were more likely to have a sexually transmitted infection (56.6% vs. 45.0%, pu2009=u20090.007) and trended towards a higher human immunodeficiency virus prevalence (16.6% vs. 10.4%, pu2009=u20090.082) at screening. Transgender women compared to other men who have sex with men were more likely to have sexually transmitted infections diagnosed at screening (75.6% vs. 50.0%, pu2009=u20090.001). Transactional sex is practiced widely but occasionally among the men who have sex with men and transgender women in Guayaquil who screened for the iPrEx study; however, engaging in transactional sex may not lead to a sex worker self-identification. Both transactional sex and being a transgender woman are associated with sexually transmitted infections prevalence.

THE POLLS—A REVIEW: MEASURING UNEMPLOYMENT IN THE NINETIES

Devant le changement de la societe et les nouvelles methodes denquete, le Bureau des statistisques du travail a modifie son questionnaire relatif au chomage. Ces modifications ont eu un effet sur les nouveaux resultats au niveau du pays et au niveau des etats : le taux de chomage est plus eleve

Quality of Statistical Education: Should ASA Assist or Assess?

Quality statistical education-like quality in education in general-is a Good Thing. I am sure we all agree to that and agree too that quality statistical education is something to be striven for. But this may be as far as full agreement can be guaranteed-though we might go a little further and agree that a minimum criterion of quality education is that students be given no false information. Beyond that, we would probably find disagreement on how to define and hence evaluate the quality of statistical education. Are we concerned with evaluating curriculum-with determining whether the content of a program is appropriate? And appropriate for whom? For graduate students? For undergraduate majors? For students taking service courses on all levels and from all disciplines? What is the appropriate statistical curriculum? Or are we concerned with evaluating the formal qualifications of those teaching the courses? Is a PhD in statistics necessary for teaching every course in statistics? Is it sufficient? Or are we concerned with evaluating the institutional arrangements for the teaching of statistics across a campus? Minton and Freund (1977) have explored different sorts of institutional arrangements. While they come down in favor of a centralized arrangement, they also remind us that circumstances alter cases. On some campuses a decentralized model is more appropriate, as long as the faculty teaching statistics in various departments communicate with one another. Or are we concerned with evaluating what goes on within the classroom itself? Is our primary concern the accuracy of the information presented or the efficiency of the transmission of knowledge between teacher and students? It seems to me that these are judgment calls-the sort of operationalizing that each of us as a statistician is supposedly trained to do and the sort of question that each of us answers frequently in our professional life. But are they questions that we want our association to answer, to take an official position on? I think not, for I think the diversity in our educational system fostered by not having hard-and-fast, institutionalized answers to such questions is valuable enough to be nurtured, even at the risk of permitting what some of us might consider inadequate departments, curricula, courses, or teaching to exist. But let us assume for a moment, and for the sake of the argument, that we can agree on the dimensions of quality education. We are then still faced with the problem of how to go about measuring that quality. One way to measure quality is to ask the consumers of the education. Students course and teacher evaluations, using any one of several carefully constructed standard instruments now available, seem to be reasonably reliable when aggregated over a class of moderate size (more than 20 or so, c.f. Feldman 1977). They have been further aggregated to provide evaluations of entire departments (Stumpf 1979; Smock and Brandenburg 1978). It seems to me that such aggregated student ratings are useless in evaluating the closeness of a curriculum to some ideal, however defined. Aggregated student ratings seem useful, however, in giving a department an idea of how well it is managing to communicate with its students. ASA could provide a service to departments wishing to evaluate their communication in this way by providing a bibliography of student rating instruments and of the empirical work that has been done to validate them. A second way to assess quality is through the use of reputational measures. One simply asks members of a discipline (those members chosen randomly or in some deliberate manner) to rate each department offering degrees in that discipline (usually on a scale ranging from inadequate to outstanding). Reputational measures have been used in several large studies (Cartter 1966; Roose and Andersen 1970) over the last fifteen years. Another such study is currently underway, and it, unlike its predecessors, does include the discipline of statistics. This is a nationwide assessment of researchdoctorate programs. It is being conducted by the Conference Board of Associated Research Councils with administrative and analytic functions to be carried out by the National Research Council of the National Academy of Sciences. It differs from earlier studies in that it provides information to raters about the departments being rated-primarily a list of faculty members. Thus the quality of statistical education within statistics departments offering research doctorates is being evaluated by this reputational system, whether we like it or not. It has been claimed that the purpose of these largescale reputational studies is neither to afflict the comfortable nor to comfort the afflicted (Logan Wilson in the forward to Roose and Andersen 1970), but rather to offer guidance to students (in particular, graduate students) seeking training and to suggest remediation to those departments found wanting. I can understand the consumers guide aspect of these studies. Clearly the reputation of the department in which one receives ones training has real effects on ones later career (though it is not necessarily true that these effects are mediated by the quality of the training received). I fail to understand, however, how such studies can lead to remediation within departments, unless they signal the need to hire new faculty of high repute-but this begs the question of whether such repute is related to teaching skill. The results of these reputational studies have been validated against such objective measures of faculty and student quality as number of Nobel Laureates, members of the National Academy, and winners of fel*Judith M. Tanur is Associate Professor, Sociology Department, State University of New York at Stony Brook, Stony Brook, NY 11794.

The William Kruskal Legacy: 1919–2005

William Kruskal (Bill) was a distinguished statistician who spent virtually his entire professional career at the University of Chicago, and who had a lasting impact on the Institute of Mathematical Statistics and on the field of statistics more broadly, as well as on many who came in contact with him. Bill passed away last April following an extended illness, and on May 19, 2005, the University of Chicago held a memorial service at which several of Bills colleagues and collaborators spoke along with members of his family and other friends. This biography and the accompanying commentaries derive in part from brief presentations on that occasion, along with recollections and input from several others. Bill was known personally to most of an older generation of statisticians as an editor and as an intellectual and professional leader. In 1994, Statistical Science published an interview by Sandy Zabell (Vol. 9, 285--303) in which Bill looked back on selected events in his professional life. One of the purposes of the present biography and accompanying commentaries is to reintroduce him to old friends and to introduce him for the first time to new generations of statisticians who never had an opportunity to interact with him and to fall under his influence.

Sample Survey Methodology, History of

Sampling methods have their roots in attempts to measure the characteristics of a nations population by using only a part instead of the whole population, i.e., a census. The movement from the exclusive use of censuses to the at least occasional use of samples was slow and laborious. A watershed was the debate over representative methods and the rise of probability sampling, including the seminal 1934 paper of Jerzy Neyman. This article describes these developments as well as the impact in the United States of the results of this work on sample surveys, the contributions of statisticians working in government statistical agencies, and the subsequent establishment of random sampling and survey methodology more broadly.

William H. Kruskal, Mentor and Friend

Discussion of ``The William Kruskal Legacy: 1919--2005 by Stephen E. Fienberg, Stephen M. Stigler and Judith M. Tanur [arXiv:0710.5063]