Willem Saris


Willem Egbert Saris is a Dutch sociologist and Emeritus Professor of Statistics and Methodology, especially known for his work on "Causal modelling in non-experimental research" and measurement errors.

Biography

Saris was born in Leiden, South Holland, Netherlands, in 1943. He finished his study of Sociology at the Utrecht University in 1968 and earned his PhD from the University of Amsterdam in 1979. He became full professor in political science, especially the methodology of the social sciences in 1983. Till 2001, he was working at the University of Amsterdam. In 1984, he created the Sociometric Research Foundation in order to improve social science research by the application of statistics. In 1998, he became member of the methodology group that facilitated the creation of the European Social Survey. As a consequence, he was also member of the Central Coordinating Team of the ESS from 2000 until 2012.
In 2001, he moved to Barcelona where he was granted a position as ICREA professor at ESADE. From 2009 to 2012, he was also director of the Research and Expertise Centre for Survey Methodology at the Pompeu Fabra University. He was also one of the founders and first chairman of the European Survey Research Association.
Saris has advised about 25 doctoral students. Among them were Nelly Kalfs, Annette Scherpenzeel, Jack Menke, Gilbert Cijntje, Akos Munnich, Emiel Kaper, Jaap Kamps, Ivar Vermeulen, Robert Voogt, William van der Veld all at the University of Amsterdam, Christiana Martini at the University of Padua, Alberto Gimeno and Antoni Dorse, Laura Guillén Ramo, all at ESADE, Barcelona, Daniel Oberski, Tilburg University, and Mélanie Révilla, Wiebke Weber, Paolo Moncagatta, André Pirralha, and Diana Zavala-Rojas at RECSM- Universitat Pompeu Fabra.

Academic career

Studies

Political decision making

Over a long period, Saris was involved with Irmtraud Gallhofer in an applied research project studying political decision making on the basis of governmental meeting minutes and notes of advisers. After developing reliable instruments for analysis, the team applied their approach on varied decisions of the Dutch government as well as major decisions in world history, such as decisions concerning the start of the First and Second World War, the Cuban missile crisis and the use of the atomic bombs in 1945. The first phase of the study was directed on the argumentation of individual decision makers. In a later phase, the research was extended to the study of collective decision making by the same governmental groups. The result was that the decision makers were making relatively simple arguments with respect to serious and far-reaching outcomes of war and peace.

Statistical aspects of structural equation models (SEM)

Worried about the testing procedures of structural equation modelling, Saris worked together with Albert Satorra to improve these procedures. They together developed different procedures to evaluate structural equation models. The final product was a procedure for detecting misspecification in these models taking into account the power of the tests. A program was developed for these tests by William van der Veld.

Improvement of measurement in survey research

Application of SEM showed how large the errors were in survey data. Therefore, he directed his research on procedures to improve the measurement instrument such as the application of continuous response scales instead of category scales. In doing so, he detected that people could provide responses in very precise but different ways. This variation in response functions would be perceived as measurement error if it was not detected. In order to prevent this problem he suggested to use two fixed reference points on a scales points where there could be no doubt what they meant.
Also studies were undertaken to improve measurement by use of computer assisted data collection, especially in order to remove the effect of the interviewer in the data collection. The latter research led to the development in 1986 of the Telepanel a procedure comparable with the Web survey procedure but at a time before the web existed.
Given the well known fact that most people do not know much about many problems discussed and inspired by the decision theoretical approach discussed in a previous project a research line was developed to develop a decision aid for people to participate in decision making with respect to complex issues. In this context a decision aid, the Choice questionnaire, was developen by him in cooperation with Peter Neijens and Jan de Ridder. In this procedure the respondents had to evaluate necessary information about the issue before they were asked to make their choice. Because of the lack of validity of the proposals for measurement instruments specified even by experts in the different fields, he created together with Irmtraud Gallhofer the three steps procedure for designing survey questions. The suggestion is that use of this procedure guarantees the validity of the proposed questions.

Evaluation of survey measurement instruments with respect to reliability and validity

When he realized that one could indeed improve survey measures but one could never reduce the errors to zero, he decided to change his research in the direction of estimating the size of the systematic and random errors in the measures in order to be able to correct for measurement error. Inspired by the work of Frank Andrews a research program has been developed in cooperation with 11 different research groups in Europe from different countries. First the approach to evaluate measurement instruments, the so-called Multitrait-Multimethod experiment has been evaluated and a new model for MTMM data was developed in cooperation with Frank Andrews. This so-called True Score model allows for the separation of random and systematic errors. Also a new design, The Split-ballot MTMM design, for MTMM experiments was developed in order to reduce the number of repetitions of the same questions in cooperation with Albert Satorra and Germa Coenders. Next, data have been collected with this design within the telepanel and the ESS and the data were analysed by members of his research group in RECSM to determine the reliability and validity of measures.

Prediction of the quality of surveys

Probably the most important development in his research took place when he realized that one could never study the reliability and validity of all questions. This would be too expensive and time consuming. The alternative he suggested was to develop a coding system for characteristics of the questions and do a meta analysis trying to predict, on the basis of the coded characteristics of the questions, the reliability and validity of these questions. If these predictions were good enough the prediction procedure could also be used for other questions that were not studied so far but of which the same characteristics were coded. It turned out that the prediction of the quality of question, determined in the MTMM experiments, on the basis of the characteristics of the questions, was quite good. This led him to the idea to design a computer assisted expert system which uses all available information of data quality to predict the quality of new questions. This program, called Survey Quality Predictor or SQP was first developed by him in MS-DOS and later transformed in a Windows version. At present there is a new version made by Daniel Oberski based on 3700 questions evaluated in MTMM experiments. Users can add their own questions in the data base and receive evaluations of the quality of the questions with respect to reliability and validity. Users get also information about possible improvements of the questions. The importance of this program is that the evaluation is not limited to the questions that have been evaluated but can be used for any questions for which the characterics are coded in the system. All these questions and quality predictions are available to all users free of charge. In this way the knowledge base of questions and quality predictions is growing day by day. Momentarily the program contains already 67.000 questions introduced by more than 2000 unique users of the program.

Awards

As a member of the CCT of the European Social Survey he became laureate of the Descartes Prize 2005, for the best collaborative research. In 2009, he received the Helen Dinerman award by the World Association for Public Opinion Research in recognition for his lifelong contributions to the methodology of public opinion research.
In 2011, he received a doctor honoris cause of the University of Debrecen.
In 2013, he received the important service to survey research prize of the ESRA.
In 2014, he was awarded together with Daniel Oberski by the American Association for Public Opinion Research with the Warren J. Mitofsky Innovators Award for the Survey Quality Predictor and his contribution to the improving questionnaire design.

Notable contributions

Saris has authored and coauthered numerous publications. Some of his main relevant publications are listed below, by main topics.
; Books
;Political decision making
; Structural Equation models
; Improvement of measurement
; Multi trait Multimethod research
;The development of a procedure to predict the quality of survey questions