Roselinde Kessels

A comparison of criteria to design efficient choice experiments

To date, no attempt has been made to design efficient choice experiments by means of the G- and V-optimality criteria. These criteria are known to make precise response predictions, which is exactly what choice experiments aim to do. In this article, the authors elaborate on the G- and V-optimality criteria for the multinomial logit model and compare their prediction performances with those of the D- and A-optimality criteria. They make use of Bayesian design methods that integrate the optimality criteria over a prior distribution of likely parameter values. They employ a modified Fedorov algorithm to generate the optimal choice designs. They also discuss other aspects of the designs, such as level overlap, utility balance, estimation performance, and computational effectiveness.

An Efficient Algorithm for Constructing Bayesian Optimal Choice Designs

Recently, Kessels et al. (2006) developed a way to produce Bayesian G- and V-optimal designs for the multinomial logitmodel. These designs allow for precise response predictions which is the goal of conjoint choice experiments. The authors showed that the G- and V- optimality criteria outperform the D- and A-optimality criteria for prediction. However, their G- and V-optimal design algorithm is computationally intensive, which is a barrier to its use in practice. In this paper, we present an efficient algorithm for calculating Bayesian optimal designs by means of the different criteria. Particularly, the speed of computation for the V-optimality criterion has improved dramatically.The new algorithm makes it possible to use Bayesian D-, A-, G- and V-optimal designs that are tailored to individual respondents in computerized conjoint choice studies.

Recommendations on the use of Bayesian Optimal Designs for Choice Experiments

In this paper, we argue that some of the prior parameter distributions used in the literature for the construction of Bayesian optimal designs are internally inconsistent. We rectify this error and provide practical advice on how to properly specify the prior parameter distribution. Also, we present two pertinent examples to illustrate that Bayesian optimal designs generally outperform utility-neutral optimal designs that are based on linear design principles.

Bayesian optimal designs for discrete choice experiments with partial profiles

In a discrete choice experiment, each respondent chooses the best product or service sequentially from many groups or choice sets of alternative goods. The alternatives are described by levels of a set of predefined attributes and are also referred to as profiles. Respondents often find it difficult to trade off prospective goods when every attribute of the offering changes in each comparison. Especially in studies involving many attributes, respondents get overloaded by the complexity of the choice task. To overcome respondent fatigue, it is better to simplify the choice tasks by holding the levels of some of the attributes constant in every choice set. The resulting designs are called partial profile designs. In this paper, we construct D-optimal partial profile designs for estimating main-effects models. We use a Bayesian design algorithm that integrates the D-optimality criterion over a prior distribution of likely parameter values. To determine the constant attributes in each choice set, we generalize the approach that makes use of balanced incomplete block designs. Our algorithm is very flexible because it produces partial profile designs of any choice set size and allows for attributes with any number of levels and any number of constant attributes. We provide an illustration in which we make recommendations that balance the loss of statistical information and the burden imposed on the respondents.

Optimal designs for conjoint experiments

In conjoint experiments, each respondent receives a set of profiles to rate. Sometimes, the profiles are expensive prototypes that respondents have to test before rating them. Designing these experiments involves determining how many and which profiles each respondent has to rate and how many respondents are needed. To that end, the set of profiles offered to a respondent is treated as a separate block in the design and a random respondent effect is used in the model because profile ratings from the same respondent are correlated. Optimal conjoint designs are then obtained by means of an adapted version of an algorithm for finding D-optimal split-plot designs. A key feature of the design construction algorithm is that it returns the optimal number of respondents and the optimal number of profiles each respondent has to evaluate for a given number of profiles. The properties of the optimal designs are described in detail and some practical recommendations are given.

Public preferences for prioritizing preventive and curative health care interventions: A discrete choice experiment

BACKGROUND Setting fair health care priorities counts among the most difficult ethical challenges our societies are facing. OBJECTIVE To elicit through a discrete choice experiment the Belgian adult populations (18-75 years; N = 750) preferences for prioritizing health care and investigate whether these preferences are different for prevention versus cure. METHODS We used a Bayesian D-efficient design with partial profiles, which enables considering a large number of attributes and interaction effects. We included the following attributes: 1) type of intervention (cure vs. prevention), 2) effectiveness, 3) risk of adverse effects, 4) severity of illness, 5) link between the illness and patients health-related lifestyle, 6) time span between intervention and effect, and 7) patients age group. RESULTS All attributes were statistically significant contributors to the social value of a health care program, with patients lifestyle and age being the most influential ones. Interaction effects were found, showing that prevention was preferred to cure for disease in young adults, as well as for severe and lethal disease in people of any age. However, substantial differences were found in the preferences of respondents from different age groups, with different lifestyles and different health states. CONCLUSIONS Our study suggests that according to the Belgian public, contextual factors of health gains such as patients age and health-related lifestyle should be considered in priority setting decisions. The studies, however, revealed substantial disagreement in opinion between different population subgroups.

Regression-Based Decompositions of Rank-Dependent Indicators of Socioeconomic Inequality of Health

In this paper we explore different ways to obtain decompositions of rank-dependent indices of socioeconomic inequality of health, such as the Concentration Index. Our focus is on the regression-based type of decomposition. Depending on whether the regression explains the health variable, or the socioeconomic variable, or both, a different decomposition formula is generated. We illustrate the differences using data from the Ethiopia 2011 Demographic and Health Survey (DHS).

How to reform western care payment systems according to physicians, policy makers, healthcare executives and researchers: a discrete choice experiment

BackgroundMany developed countries are reforming healthcare payment systems in order to limit costs and improve clinical outcomes. Knowledge on how different groups of professional stakeholders trade off the merits and downsides of healthcare payment systems is limited.MethodsUsing a discrete choice experiment we asked a sample of physicians, policy makers, healthcare executives and researchers from Canada, Europe, Oceania, and the United States to choose between profiles of hypothetical outcomes on eleven healthcare performance objectives which may arise from a healthcare payment system reform. We used a Bayesian D-optimal design with partial profiles, which enables studying a large number of attributes, i.e. the eleven performance objectives, in the experiment.ResultsOur findings suggest that (a) moving from current payment systems to a value-based system is supported by physicians, despite an income trade-off, if effectiveness and long term cost containment improve. (b) Physicians would gain in terms of overall objective fulfillment in Eastern Europe and the US, but not in Canada, Oceania and Western Europe. Finally, (c) such payment reform more closely aligns the overall fulfillment of objectives between stakeholders such as physicians versus healthcare executives.ConclusionsAlthough the findings should be interpreted with caution due to the potential selection effects of participants, it seems that the value driven nature of newly proposed and/or introduced care payment reforms is more closely aligned with what stakeholders favor in some health systems, but not in others. Future studies, including the use of random samples, should examine the contextual factors that explain such differences in values and buy-in.JEL classificationC90, C99, E61, I11, I18, O57

Structural equation modeling for decomposing rank-dependent indicators of socioeconomic inequality of health: an empirical study

We present a flexible structural equation modeling (SEM) framework for the regression-based decomposition of rank-dependent indicators of socioeconomic inequality of health and compare it with simple ordinary least squares (OLS) regression. The SEM framework forms the basis for a proper use of the most prominent one- and two-dimensional decompositions and provides an argument for using the bivariate multiple regression model for two-dimensional decomposition. Within the SEM framework, the two-dimensional decomposition integrates the feedback mechanism between health and socioeconomic status and allows for different sets of determinants of these variables. We illustrate the SEM approach and its outperformance of OLS using data from the 2011 Ethiopian Demographic and Health Survey.

Socioeconomic status and health : a new approach to the measurement of bivariate inequality

We suggest an alternative way to construct a family of indices of socioeconomic inequality of health. Our indices belong to the broad category of linear indices. In contrast to rank-dependent indices, which are defined in terms of the ranks of the socioeconomic variable and the levels of the health variable, our indices are based on the levels of both the socioeconomic and the health variable. We also indicate how the indices can be modified in order to introduce sensitivity to inequality in the socioeconomic distribution and to inequality in the health distribution. As an empirical illustration, we make a comparative study of the relation between income and well-being in 16 European countries using data from the Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 4.