Surveys research is a type of research where data is collected by asking questions to a sample of persons from a population. On the basis of the collected data, conclusions are drawn about the population as a whole. The question is whether this always is a scientifically sound research method.

The presentation gives an historic overview of how surveys indeed became a reliable research instrument. However, not every survey is a good survey. If a survey is not properly designed, wrong conclusions may be drawn. This presentation is about some of the methodological problems and their consequences.

The fast development of the Internet has led to a new form of survey, and this is the web survey. Almost everybody can do a web survey. There are many examples of badly designed web surveys. They suffer from methodological problems like under-coverage and self-selection.

Unfortunately, the media are not able to distinguish the good from the bad. Therefore, this presentation can be seen as a warning not to believe everything that is reported about survey results.

Increasing cooperation between different parties such as bar owners, municipalities, doormen and police force was proposed as a policy measure. The tasks and duties of the different parties were described in an agreement. The effectiveness of the agreement was studied using police records. From the police databases records of violence in a nightlife area were extracted and aggregated on a monthly basis. Collecting these data for a four year period we obtained a series of count data. To answer the question whether the introduction of an agreement was effective, we used Poisson regression.

A key element in many school performance feedback practices is the concept of "value added", also known as "residual gain score". It refers to the notion that schools should not be compared with other schools based on the raw performance scores of their students. To make fair comparisons, the influence of student and family background characteristics and eventually school context should be filtered out. What is then left – the net school effect – is considered to reflect the "real" impact of the school on students’ learning.

There are two ways to explain the concept of value added in school feedback reports: either as the difference between a school’s observed mean and its expected mean, or as the difference between the school’s adjusted mean and the grand mean for the reference category. From a pure mathematical point of view, both ways are strictly equivalent. Following the output provided by the statistical software they use, many researchers are inclined to use the latter way. Research on school practitioners’ understanding of school performance feedback however shows that the concept of value added is better understood when the first way is used. Since software such as MLWin does not provide school level predictions including fixed effects other than the intercept (expected means and standard errors), putting this finding into school feedback practice becomes quite challenging. The Flemish School Feedback Project has tried to tackle this problem, taking its basic principles into consideration: (1) the school feedback is based on a comparison of new school data with an available representative data set, and (2) the school feedback should be provided in a fully automated way, which includes automation of all processes involved: data collection, data processing and the production and distribution of school tailored feedback reports. The present paper focuses on school feedback based on one measurement occasion, but the findings can be extended to feedback on more measurement occasions.

On a daily basis surgeons, nurses, and managers face cancellation of surgery, peak demands on wards, and overtime in operating rooms. Moreover, the lack of an integral planning approach for operating rooms, wards, and intensive care units causes low resource utilization and makes patient flows unpredictable. An ageing population and advances in medicine are putting the available healthcare budget under great pressure. Under these circumstances, hospitals are seeking innovative ways of providing optimal quality at the lowest costs.

I developed during my PhD research instruments for optimizing surgical patient planning on basis of a cyclic and integrated operating room planning approach, called master surgical scheduling. One of the stages in this approach is clustering individual surgical cases into standardized surgical cases that can be scheduled in a repetitive manner. Moreover, I studied additional models to deal for instance efficiently with emergency operations. Application of these instruments enables the simultaneous optimization of the utilization of operating rooms, ward and intensive care units. Moreover, iteratively executing a master schedule of surgical case types provides steady and thus more predictable patient flows in hospitals.

The approach is generic and so can be implemented taking account of specific characteristics of individual hospitals. Prerequisites for successful implementation of logistical models in hospitals comprise sufficient room for last-minute changes as well as keeping the ultimate responsibility for individual patient scheduling with medical specialists. Both are satisfied in the master surgical scheduling approach which has already been successfully implemented in hospitals.

When two competitive treatments, A and B, are available, some subgroup of patients may display a better outcome with treatment A than with B, whereas for another subgroup the reverse may be true. If this is the case, a disordinal (i.e., a qualitative) treatment-subgroup interaction is present. Such interactions imply that some subgroups of patients should be treated differently, and are therefore most relevant for policy with regard to assignment of patients to programs (i.e., treatment plans). In case of data from randomized clinical trials with many patient characteristics that could interact with treatment in a complex way, a suitable statistical approach to detect disordinal treatment-subgroup interactions is not yet available. In this presentation, we introduce a new method for this purpose, called Treatment INteraction Trees (TINT). TINT results in a binary tree that subdivides the patients into terminal nodes on the basis of patient characteristics; these nodes are further assigned to one of three classes: a first one for which A is better than B, a second one for which B is better than A, and an optional third one for which type of treatment makes no difference. The tree can be used to develop rules for patient-tailored treatment assignment. Results of an application of TINT to real data from the Breast Cancer Recovery Project will be shown. A short demonstration of the software will be given.

Elise Dusseldorp studied psychology at the university of Leiden, where she also obtained her PhD in 2001. Elise is a statistician at the Quality of Life division of TNO (the Netherlands Centre for Applied Scientific Research) in Leiden, and a post-doctoral researcher in the Quantitative Psychology and Individual Differences Research Group of the Catholic University Leuven, Belgium. Her main area of interest in the field of statistics is the modelling of interaction effects in prediction problems. While many statistical models focus on the unique effects of predictive factors on one or more outcome variables, her challenge is to explore the way in which the combined influences of factors on an outcome can be assessed.