USE OF PURCHASE PREFERENCE OPTIONS TO INCREASE "NO PREFERENCE" FREQUENCIES IN PLACEBO PREFERENCE TESTS

Besides responses to the different stimuli being tested in a paired preference test, responses to identical "placebo" stimuli can be used as a screening tool to identify biased consumers. Those consumers who give preference responses to identical stimuli can be assumed to be biased. Accordingly, only the data from unbiased consumers need to be considered for the different stimuli. The problem with this procedure is that the sample size is reduced. The goal of the present research was to see whether using options associated with purchase intent, elicited a greater number of "No Preference" responses to identical "placebo" stimuli. It was found that they did. The increase was large when the preference options implied exclusivity. In conditions where the preference strength options were not so strong, the frequency of "No Preference" responses dropped accordingly.

PRACTICAL APPLICATIONS

A problem with paired preference testing is the tendency of consumers to give false preferences, which produces the seriously misleading overestimation of the proportion of consumers who have preferences for one or other of the products being assessed. The "placebo" condition is an important control for alleviating this problem. The statistical analysis can be improved by finding a protocol that maximizes the proportion of "No Preference" responses in the placebo condition. The key finding here is that using purchase intent questions rather than preference questions may possibly provide a way of achieving this aim.     

PAIRED PREFERENCE TESTS: d' VALUES FROM MEXICAN CONSUMERS WITH VARIOUS RESPONSE OPTIONS

Three groups, each comprising 200 consumers, performed paired preference tests between two flavors of potato chips. For one group, a “No Preference” option was allowed, while for the second group, it was not. For a third group, a “No Preference” option was allowed while the preference responses were differentiated as “Strong Preference” and “Weak Preference.” For all three groups, d′ values representing overall strength of preference did not differ significantly.     

MULTIPLE PREFERENCE TESTS CAN PROVIDE MORE INFORMATION ON CONSUMER PREFERENCES*

Preference testing is used for a variety of reasons. Traditionally, a consumer completes one preference test. However, consumers may not be consistent in their preferences and, therefore, multiple preference tests would be necessary. This research investigated the preference responses of consumers during four preference tests on the same pair of products. Results showed that many of the consumers switched their preference through the series of four tests. Although the basic conclusion drawn from the first test or by any of the four tests was the same in this study, the overall percentages of consumers who preferred product “A” from the four tests were not always similar to one another, nor did the results of the first preference test show the degree of preference that existed for some products. Multiple preference tests may provide more information that is useful to researchers or marketers than a single preference test. Inconsistent responses should be included in data analysis to avoid distortion of preference patterns.     

PAIRED PREFERENCE “PLACEBO” TESTS WITH “IDENTICAL” STIMULI: DOES INTRODUCING GRADED PREFERENCE RESPONSES AFFECT THE FREQUENCY OF “NO PREFERENCE” RESPONSES?

In the "placebo" control condition for paired preference testing, "identical" stimuli are presented to consumers to determine the frequency of preference and "no preference" responses induced by the hidden demand characteristics of the testing condition. As a control for bias, induced by such hidden demand characteristics, these frequencies can be compared with the actual preference frequencies of the nonidentical test stimuli to be assessed for preference. It was hypothesized that the introduction of graded preference response options might reduce the frequency of "no preference" responses in the placebo condition. Using identical yogurt stimuli with related-sample (single-group) and independent-sample (multigroup) designs, this hypothesis was not confirmed.

PRACTICAL APPLICATIONS

The "placebo" condition in paired preference testing provides a way to control the tendency of consumers to give false preference. The statistical analysis can be improved by finding a protocol that maximizes the proportion of "no preference" responses in the placebo condition. This can be done by increasing the number of response options that imply no preference. Yet, there is sometimes a desire to increase the number of preference options by using graded responses for preference strength. Does this alter the dynamics of the ever-important placebo condition by reducing the number of "no preference" responses? This project found that any effect was minimal.     

A THURSTONE-COOMBS MODEL OF CONCURRENT RATINGS WITH SENSORY AND LIKING DIMENSIONS

A popular product testing procedure is to obtain sensory intensity and liking ratings from the same consumers. Consumers are instructed to attend to the sensory attribute, such as sweetness, when generating their liking response. We propose a new model of this concurrent ratings task that conjoins a unidimensional Thurstonian model of the ratings on the sensory dimension with a probabilistic version of Coombs' (1964) unfolding model for the liking dimension. The model assumes that the sensory characteristic of the product has a normal distribution over consumers. An individual consumer selects a sensory rating by comparing the perceived value on the sensory dimension to a set of criteria that partitions the axis into intervals. Each value on the rating scale is associated with a unique interval. To rate liking, the consumer imagines an ideal product, then computes the discrepancy or distance between the product as perceived by the consumer and this imagined ideal. A set of criteria are constructed on this discrepancy dimension that partition the axis into intervals. Each interval is associated with a unique liking rating. The ideal product is assumed to have a univariate normal distribution over consumers on the sensory attribute evaluated. The model is shown to account for 94.2% of the variance in a set of sample data and to fit this data significantly better than a bivariate normal model of the data (concurrent ratings, Thurstonian scaling, Coombs' unfolding model, sensory and liking ratings).     

BETA-BINOMIAL TABLES FOR REPLICATED DIFFERENCE AND PREFERENCE TESTS

Ennis and Bi (1998) discussed the beta-binomial (BB) model for replicated difference and preference tests. Based on the BB model, tables of the minimum number of choice responses to achieve significance at α≦ 0.05 are provided for replicated 2-AFC, Duo-Trio, 3-AFC and Triangular tests. The theory underlying the tables, how to use the tables, as well as some examples to illustrate their use are given. The tables can be used to evaluate the results of replicated difference and preference tests using forced choice methods.     

COMPARISON OF METHODS FOR ANALYZING REPLICATED PREFERENCE TESTS

Preference testing is commonly used in consumer sensory evaluation. Traditionally, it is done without replication, effectively leading to a single 0/1 (binary) measurement on each panelist. However, to understand the nature of the preference, replicated preference tests are a better approach, resulting in binomial counts of preferences on each panelist. Variability among panelists then leads to overdispersion of the counts when the binomial model is used and to an inflated Type I error rate for statistical tests of preference. Overdispersion can be adjusted by Pearson correction or by other models such as correlated binomial or beta‐binomial. Several methods are suggested or reviewed in this study for analyzing replicated preference tests and their Type I error rates and power are compared. Simulation studies show that all methods have reasonable Type I error rates and similar power. Among them, the binomial model with Pearson adjustment is probably the safest way to analyze replicated preference tests, while a normal model in which the binomial distribution is not assumed is the easiest.     

REEXAMINATION OF THE ETHICS OF PLACEBO USE IN CLINICAL PRACTICE

A placebo is a substance or intervention believed to be inactive, but is administered by the healthcare professional as if it was an active medication. Unlike standard treatments, clinical use of placebo usually involves deception and is therefore ethically problematic. Our attitudes toward the clinical use of placebo, which inevitably includes deception or withholding information, have a tremendous effect on our practice regarding truth‐telling and informed consent. A casual attitude towards it weakens the current practice based on shared decision‐making and mutual trust between patients and healthcare professionals. Issues concerning the clinical use of placebo are thus intimately related to patient‐provider relationships, the public's trust in medicine, and medical education. A review of recent survey studies suggests that the clinical use of placebo appears to be fairly well accepted among healthcare professionals and is common in clinical settings in various countries. However, we think that an ethical discussion is urgently needed because of its controversial nature. If judged to be ethically wrong, the practice should end. In the present paper, we discuss the ethicality of the clinical use of placebo with deception and argue against it, concluding that it is unethical and should be banned. We will show that most arguments in favor of the clinical use of placebo can be refuted and are therefore incorrect or weak. These arguments will be presented and examined individually. Finally, we will briefly consider issues relevant to the clinical use of placebo without deception.     

THE POWER OF SENSORY DISCRIMINATION METHODS USED IN REPLICATED DIFFERENCE AND PREFERENCE TESTS

Research on the power of discrimination methods in difference and preference tests has both theoretical and practical significance. Power is important to evaluate the sensitivities of tests and determine sample size. Ennis and Bi (1998, 1999) proposed the beta-binomial distribution to model replicated difference and preference tests with inter-trial variation and analyzed in general the power of the tests. In this paper, the power of discrimination methods for replicated difference and preference tests is discussed further. The equations for calculating power for methods based on the BB model are given. Examples with tables and charts for calculating and comparing the power of the methods are also given.     

MARGARINE: THE DRIVERS OF LIKING AND IMAGE

This paper presents a study of the attributes of margarine, showing the depth of information about consumer perceptions and drivers of liking that emerges from a detailed analysis of relations among attributes. The paper develops three sets of analyses to understand relations among attributes: principal components analysis in order to identify basic dimensions of perception, linear functions relating overall liking to attribute liking or to image ratings in order to identify drivers of liking, and quadratic functions that relate overall liking or image ratings to sensory attribute levels in order to identify optimal sensory levels and to create sensory preference segments. The analyses show how consumer data can generate learning about the consumer perceptions on the one hand, and guidance for product development.     

DIFFICULTIES AND A WAY OUT: A BAYESIAN APPROACH FOR SENSORY DIFFERENCE AND PREFERENCE TESTS

There are some theoretical difficulties in the philosophy and methodology of sensory difference and preference tests. The difficulties lie in the assumption that the subjects have the same response ability, which is in conflict with psychological ideas about the processes of perception and decision. A Bayesian approach may overcome these difficulties by treating the parameter of proportion as a random variable. This paper presents a Bayesian approach for analysis of data from such sensory tests.     

HOW TO ESTIMATE AND USE THE VARIANCE OF d' FROM DIFFERENCE TESTS

d' is an estimate of δ, a measure of the degree of sensory difference between two products, that can be obtained easily using tables, from the proportion of difference tests performed correctly. Tables of δ are available for the 2-AFC, 3-AFC, triangular and duo-trio tests. Tables for calculating the variance of d' for these tests are provided in this paper. They can be used for comparison of d's, especially for those obtained from different difference tests. A simple procedure is described here for computing values for the variance of d'. Having obtained the variance, confidence intervals for d' can be obtained, tests of significance for d' can be made as well as tests of whether two or more d's are significantly different. The formula and tables for the number of judgments required for the estimation of δ are given also in this paper.     

PREFERENCE HIERARCHIES AND THE ORIGIN OF GEOGRAPHIC SPECIALIZATION IN HOST USE IN SWALLOWTAIL BUTTERFLIES

Four allopatric populations of the widely distributed western anise swallowtail butterfly, Papilio zelicaon, use different plant genera as hosts, but simultaneous choice experiments showed that these populations have diverged only slightly in oviposition preference. Of the four populations—two from southeastern Washington State, one from coastal southwestern Washington State, and one from central California—three use hosts that are not available to any of the others. Although variation for the degree of preference for particular plant species occurred within and among populations, all four populations ranked hosts in the same overall order. Monophagy on a local, low-ranking host outside the range of high-ranking hosts did not necessarily lead to the loss of preference for those high-ranking hosts, thereby indicating that the high-ranking hosts would still be accepted, and in some cases even preferred, if a population encountered them again. Hence, the overall preference hierarchy among P. zelicaon populations appears to be evolutionarily conservative. Analyses of differences among families within the California population indicated that increased preference for some hosts is inversely correlated, whereas preference for other hosts may be uncorrelated. Positive correlations may also occur but were not observed among the plant species tested. Overall, the results indicate local monophagy on different plant species in P. zelicaon has not involved major reorganizations in the preference hierarchy of ovipositing females, even in populations that may have fed on a low-ranking host for many generations. Instead, small increases in preference for local hosts have occurred within an evolutionarily conservative preference hierarchy.     

THE BETA-BINOMIAL MODEL: ACCOUNTING FOR INTER-TRIAL VARIATION IN REPLICATED DIFFERENCE AND PREFERENCE TESTS

Binomial tests are commonly used in sensory difference and preference testing under the assumptions that choices are independent and choice probabilities do not vary from trial to trial. This paper addresses violations of the latter assumption (often referred to as overdispersion) and accounts for variation in inter-trial choice probabilities following the Beta distribution. Such variation could arise as a result of differences in test substrate from trial to trial, differences in sensory acuity among subjects or the existence of latent preference segments. In fact, it is likely that overdispersion occurs ubiquitously in product testing. Using the Binomial model for data in which there is inter-trial variation may lead to seriously misleading conclusions from a sensory difference or preference test. A simulation study in this paper based on product testing experience showed that when using a Binomial model for overdispersed Binomial data, Type I error may be 0.44 for a Binomial test specification corresponding to a level of 0.05. Underestimation of Type I error using the Binomial model may seriously undermine legal claims of product superiority in situations where overdispersion occurs. The Beta-Binomial (BB) model, an extension of the Binomial distribution, was developed to fit overdispersed Binomial data. Procedures for estimating and testing the parameters as well as testing for goodness of fit are discussed. Procedures for determining sample size and for calculating estimate precision and test power based on the BB model are given. Numerical examples and simulation results are also given in the paper. The BB model should improve the validity of sensory difference and preference testing.