Controlling for the Effects of Information in a Public Goods Discrete Choice Model

Abstract

This paper develops a reduced form method of controlling for differences in information sets of subjects in public good discrete choice models, using stated preference data. The main contribution of our method comes from accounting for the effect of information provided during a survey on the mean and the variance of individual-specific scale parameters. In this way we incorporate both scale heterogeneity as well as observed and unobserved preference heterogeneity to investigate differences across and within information treatments. Our approach will also be useful to researchers who want to combine stated preference data sets while controlling for scale differences. We illustrate our approach using the data from a discrete choice experiment study of a biodiversity conservation program and find that the mean of individual-specific scale parameters and its variance in the sample is sensitive to the information set provided to the respondents.

Are Fast Responses More Random? Testing the Effect of Response Time on Scale in an Online Choice Experiment

“…Another route to specify cut-off points is to look at confidence intervals of WTP estimates. Variance of estimated WTP is driven by the relative magnitude of the random term of utility, which in turn is determined by scale (Czajkowski et al 2014a). If for some (groups of) respondents scale is found to be so low that resulting confidence intervals of WTP estimates are too large to be meaningful, these responses could be discarded.…”

“…Response time is likely to be just one of such systematic influences on choice randomness. Familiarity of with the good to be valued (Czajkowski et al 2015), level of education (Czajkowski et al 2014b) or choice certainty (Beck et al 2013) have been shown to be other potential factors. Beyond that, other variables evaluating the choice task from the perspective of the respondent, such as attribute attendance, protest attitudes or previous experience as a survey participant should be accounted for in such a model.…”

“…z n is modelled to include certain choice-related variables such as response time and choice certainty. Applications of the GMNL model with this interpretation include Czajkowski et al (2014bCzajkowski et al ( , 2015, Kragt (2013), Juutinen et al (2012) and Christie and Gibbons (2011). While the conditional logit and ECLC models can be computed using a traditional maximum likelihood approach, simulated maximum likelihood estimation with 1000 Halton draws is required for all other models.…”

“…To test this hypothesis the generalised multinomial logit (GMNL) model (Fiebig et al 2010;Gu et al 2013) is employed on a dataset of an online choice experiment survey valuing different impacts of the installation of an offshore windfarm in the UK. The GMNL model has recently been used in similar choice experiment settings to investigate the effect of familiarity with the good to be valued (Czajkowski et al 2015) or level of education (Czajkowski et al 2014b) on scale. Response time is measured and modelled as a determinant of systematic inter-respondent differences in scale.…”

Theory and practice of controlling at enterprises in international business

“…Nevertheless, it should be noted that today there is no universal tool of controlling. It is believed that benchmarking, potential analysis, SWOT analysis, moderation method, analysis of strategic deviations, system of balanced parameters, strategy maps, system of distribution of indirect costs, standard quality, Six Sigma, policy of open reporting, theory of constraints, scenario development, system of accounting and cost analysis are the tools of strategic controlling which are most adapted to the use at industrial enterprises in modern conditions [13][14][15][16].…”