The article under discussion illustrates the trade-off between optimization and exploration that is fundamental to statistical experimental design. In this discussion, I suggest that the research under discussion could be made even more effective by checking the fit of the model by comparing observed data to replicated data sets simulated from the fitted model.
Standardizing performance expectations across different outlets within a chain, differing in their individual features, their consumers, and the nature of competition they face, can be an onerous t...
In the response to my commentary on his 2007 editorial entitled “It's the Findings, Stupid, Not the Assumptions,” Steven Shugan raises a number of thought-provoking ideas. In this rejoinder, I focus on three issues that Shugan and I hold the most divergent views. First, while Shugan uses the terms “realistic” and “realism” in several different meanings, I define the realism of an assumption as the extent to which it corresponds with the real world. Second, Shugan makes a strong claim that predictions can be objectively evaluated whereas assumptions cannot. I refute his claim by arguing that testing predictions and testing assumptions follow the same research process of checking whether the proposition concerned corresponds with reality. Third, Shugan maintains that given predictive accuracy, assumptions need not be realistic. I hold an opposite view for the obvious reason that the same prediction may be generated by completely different mechanisms, which in turn are based on different assumptions. Last but not least, the example of socialist economic planning shows that unrealistic assumptions can generate dangerous theories.
Website morphing draws on the Expected Gittins’ solution to a partially observable Markov process, on the rapid consumer-segment updating with Bayesian methods, and on matching a website’s look and feel to a visitor’s cognitive style. In each area there are exciting research opportunities including optimality in the presence of switching costs (within a visit), Bayesian updating of cognitive styles across websites, extensions to other segmentation schemes such as cultural styles, morphing of other website characteristics such as advertising, and applications to other media such as smartphones.
Eric Tsang's response makes the legitimate point that prediction and explanation can be different goals. However, his arguments also suffer from several errors in logic, most often the converse error. I do not claim that unrealistic assumptions breed good theories. I only claim that breakthrough theories usually have assumptions deemed unrealistic. Hence, unrealistic assumptions breed both good and bad theories. That is why science tests theories, not assumptions. Moreover, one can easily prove that realistic assumptions are not required. Consider situations when one of two competing theories must be correct. For example, in criminal cases, the prosecution's theory is that the accused committed the crime. The defense's theory is that the accused is innocent. One theory is correct despite the fact that both could make obviously unrealistic assumptions. For example, the prosecution might unrealistically assume that unreliable eyewitness testimony is sufficient to convict. The defense might unrealistically assume that a dubious alibi is sufficient to acquit. Juries decide on all the evidence and not each assumption. I now answer some of Eric Tsang's questions.
The universality of design perception and response is tested using data collected from 10 countries: Argentina, Australia, China, Germany, Great Britain, India, The Netherlands, Russia, Singapore, and the United States. A Bayesian, finite-mixture, structural equation model is developed that identifies latent logo clusters while accounting for heterogeneity in evaluations. The concomitant variable approach allows cluster probabilities to be country specific. Rather than a priori defined clusters, our procedure provides a posteriori cross-national logo clusters based on consumer response similarity. Our model reduces the 10 countries to three cross-national clusters that respond differently to logo design dimensions: the West, Asia, and Russia. The dimensions underlying design are found to be similar across countries, suggesting that elaborateness, naturalness, and harmony are universal design dimensions. Responses (affect, shared meaning, subjective familiarity, and true and false recognition) to logo design dimensions (elaborateness, naturalness, and harmony) and elements (repetition, proportion, and parallelism) are also relatively consistent, although we find minor differences across clusters. Our results suggest that managers can implement a global logo strategy, but they also can optimize logos for specific countries if desired.
We discuss the Salisbury and Feinberg paper [Salisbury, L. C., F. M. Feinberg. 2010. Alleviating the constant stochastic variance assumption in decision research: Theory, measurement, and experimental test. Marketing Sci. 29(1) 1–17], setting their contribution in the historical context of the wider literature on the role of error variability in discrete choice models. We discuss the seminal nature of their contribution and suggest that the paper should be required reading for current and future Ph.D. students.
The subjects of this paper are ethics and professionalism, topics closely linked in contemporary theory, and especially in practice of public relations, whose significance is increasingly coming to the spotlight of experts from this area. Several definitions, classification, the historical development and principles of theories of ethics most frequently used in ethical decision-making within a business environment, have been presented in the first chapter in the endeavor to ascertain the concept of ethics. The next chapter concerns the duties a public relations expert must pay attention to while carrying out his or her activities. Those are: duty towards oneself, towards the organization, society and profession, within which, in the case of a conflict of interest, the duty towards society (so-called social responsibility), or professional duty, must prevail. The chapter that follows concerns ethical problems in the contemporary practice of public relations: the competence of practitioners, possible conflicts of interest and the very sensitive area of media relations. The chapter on models of ethical decision-making involves concrete experts' advice on decision making which are firmly based on ethical principles. Next section concerns professionalism and professional education in public relations. Recommendations concerning topics which should be included in the university education in this area are also presented. The focus is on the following: the absence of standards that would establish who can work in public relations and under which conditions; the lack of a specified educational minimum and expertise which a practitioner should possess; the need for practitioners to be the members of professional associations, as well as to adhere to a required ethical codex. Some of the most significant world public relations associations are mentioned and at the end, and a review of the state of public relations in Serbia is given.