Violations of transitivity: Implications for a theory of contextual choice

Violations of strong stochastic transitivity in concurrent-chains choice were first reported by Navarick and Fantino. In a series of articles, Navarick and Fantino concluded that neither a unidimensional model capable of predicting exact choice probabilities nor a fixed-variable equivalence rule was possible for the concurrent-chains procedure. I show that when choice is modeled contextually (i.e., when preference for a schedule is affected by factors other than the schedule itself, e.g., aspects of the alternative schedule), a unidimensional, exact-choice probability model is possible that both predicts the intransitivities reported by Navarick and Fantino and provides a fixed-variable equivalence rule for the concurrent-chains procedure. The contextual model is an extension of the generalized matching law and violates a key assumption underlying traditional choice models—simple scalability—because of (a) schedule interdependence and (b) bias from procedural contingencies. Therefore, strong stochastic transitivity cannot be expected to hold. Contextual scalability is analyzed to reveal a hierarchy of context effects in choice. Navarick and Fantino's intransitivities can be satisfactorily explained by bias. If attribute sensitivity is context dependent, however, and if there are similarity structures among choice alternatives, the contextual model is shown to be able to predict violations of ordinal preference. Therefore, it may be possible to formulate a deterministic, general psychophysical model of choice as a behavioral alternative to probabilistic, multidimensional choice theories.     

Expected risk and single-peaked preferences

Eight ordinal conditions necessary for Huang's (1971a) theory of expected risk were tested in empirical risk orders over 31 two-outcome money gambles elicited from 20 subjects. Additionally, the single peakedness assumption of Coombs' (1969) portfolio theory was tested with preference orders over a subset of 15 of these gambles with equal expected values. The theory of expected risk was well corroborated; the portfolio theory was not.     

Decision theoretic foundations for axioms of rational preference

Rationality postulates for preferences are developed from two basic decision theoretic principles, namely: (1) the logic of preference is determined by paradigmatic cases in which preferences are choice-guiding, and (2) excessive comparison costs should be avoided. It is shown how the logical requirements on preferences depend on the structure of comparison costs. The preference postulates necessary for choice guidance in a single decision problem are much weaker than completeness and transitivity. Stronger postulates, such as completeness and transitivity, can be derived under the further assumption that the original preference relation should also be capable of guiding choice after any restriction of the original set of alternatives.     

Testing mixture models of transitive preference: comment on Regenwetter, Dana, and Davis-Stober (2011)

This article contrasts 2 approaches to analyzing transitivity of preference and other behavioral properties in choice data. The approach of Regenwetter, Dana, and Davis-Stober (see record 2011-00732-003) assumes that on each choice, a decision maker samples randomly from a mixture of preference orders to determine whether A is preferred to B. In contrast, Birnbaum and Gutierrez (2007) assumed that within each block of trials, the decision maker has a true set of preferences and that random errors generate variability of response. In this latter approach, preferences are allowed to differ between people; within-person, they might differ between repetition blocks. Both approaches allow mixtures of preferences, both assume a type of independence, and both yield statistical tests. They differ with respect to the locus of independence in the data. The approaches also differ in the criterion for assessing the success of the models. Regenwetter et al. fitted only marginal choice proportions and assumed that choices are independent, which means that a mixture cannot be identified from the data. Birnbaum and Gutierrez fitted choice combinations with replications; their approach allows estimation of the probabilities in the mixture. It is suggested that researchers should separate tests of the stochastic model from the test of transitivity. Evidence testing independence and stationarity assumptions is presented. Available data appear to fit the assumption that errors are independent better than they fit the assumption that choices are independent.     

Evidence against Rank-Dependent Utility Theories: Tests of Cumulative Independence,Interval Independence,Stochastic Dominance,and Transitivity

This study tests between two modern theories of decision making. Rank- and sign-dependent utility (RSDU) models, including cumulative prospect theory (CPT), imply stochastic dominance and two cumulative independence conditions. Configural weight models, with parameters estimated in previous research, predict systematic violations of these properties for certain choices. Experimental data systematically violate all three properties, contrary to RSDU but consistent with configural weight models. This study also tests whether violations of stochastic dominance can be explained by violations of transitivity. Violations of transitivity may be evidence of a dominance detecting mechanism. Although some transitivity violations were observed, most choice triads violated stochastic dominance without violating transitivity. Judged differences between gambles were not consistent with the CPT model. Data were not consistent with the editing principles of cancellation and combination. The main findings are interpreted in terms of coalescing, the principle that equal outcomes can be combined in a gamble by adding their probabilities. RSDU models imply coalescing but configural weight models violate it, allowing configural weighting to explain violations of stochastic dominance and cumulative independence.     

The wandering ideal point model: A probabilistic multidimensional unfolding model for paired comparisons data

In analogy with the wandering vector model, a probabilistic multidimensional unfolding model is proposed for representing paired comparisons data. It is shown that contrary to other stochastic multidimensional unfolding models, the present model does not imply strong stochastic transitivity, only moderate stochastic transitivity. A maximum likelihood parameter estimation procedure is developed and an illustrative application is presented.     

A method of weighing qualitative preference axioms

A method is developed for determining the absolute and relative strengths of qualitative preference axioms in normative Bayesian decision theory. These strengths are calculated for the three most common qualitative axioms; transitivity, the sure-thing principle, and dominance. The relative strength of the latter two axioms with respect to transitivity is calculated for special cases, and a bound is derived which is applicable to a larger class of decision problems. Possible implications of this theoretical work for decision heuristics are discussed.     

Laterality effects in the method of triads

The method of triads for psychophysical judgments was analyzed in terms of the unidimensional unfolding model of Coombs (1964). This model predicts laterality effects on the stochastic transitivity of triples for which existing models have been unable to account. An experiment was conducted in which triadic judgments were made on circles differing in area. The results confirmed the stochastic transitivity predictions, but a quantitative formulation of the model proved unsatisfactory.     

When mixed options are preferred in multiple‐trial decisions

We report three studies showing that in prospective multiple‐trial decisions people often select a mix of sure and risky options over pure bundles of either option. Such a preference is not ‘rational’ because a mixed option cannot be the EV‐maximizing choice. Experiment 1 confirmed a mixed‐option preference for gains but not for losses. Showing a graph of the multiple‐trial outcome distribution reduced but did not eliminate this effect, suggesting that it is not due purely to a failure to aggregate correctly over the multiple trials. Experiment 2 replicated the mixed option preference using a wider range of problems. Experiment 3 compared choices in the trinary choice conditions used in Experiments 1 and 2 with binary choices between pairs of the multiple‐trial sure, mixed, and risky options. In the binary choice condition the mixed option was no longer the modal choice, suggesting that the strong mixed option preference found in the trinary choice conditions is mainly due to a compromise effect. However, the binary choice probabilities did show violations of strong stochastic transitivity in a pattern that suggested a slight bias toward the mixed option. Copyright © 2006 John Wiley & Sons, Ltd.     

On the importance of distinguishing between within- and between-subject effects in intransitive intertemporal choice

Most probabilistic paired comparison models treat inconsistent choices as caused by independent and random errors in the pairwise judgments. In this paper, we argue that this assumption is too restrictive for the analysis of paired comparison data obtained from multiple judges when transitivity violations are systematic. We present a new framework that contains the random error assumption as a special case but also allows for systematic changes in an option's utility assessments over the pairwise comparisons. Accounting for both between- and within-judge sources of variability, we demonstrate in an application on intertemporal choice that the proposed framework can capture systematic transitivity violations as well as individual taste differences.     

Weak Consistency for Ensuring Priority Vectors Reliability

Consistency and transitivity are important and leading research topics in the study of decision‐making in terms of pairwise comparison matrices. In this paper, we search for conditions that, in case of inconsistency, guarantee ordinal compatibility between ordinal ranking (actual ranking) derived from a transitive matrix and cardinal rankings provided by the most usual priority vectors proposed in the scientific literature. We provide the notion of weak consistency; it is a condition weaker than consistency and stronger than transitivity and ensures that vectors associated with a matrix, by means of a strictly increasing synthesis functional, provide a preference order, on the related set of decision elements, equal to the actual ranking. This notion extends, to the case in which the decision‐maker can be indifferent between two or more alternatives/criteria, weak consistency introduced in previous papers under constraint of no indifference. Finally, we introduce an order relation on the rows of the matrix, that is, a simple order if and only if weak consistency is satisfied; this simple order allows us to easily determine the actual ranking on the set of decision elements. Copyright © 2015 John Wiley & Sons, Ltd.     

The transitivity of choices between different response requirements

This experiment tested the transitivity of hens' choices between response requirements differing in both form and number. In a concurrent second-order schedule procedure, 6 hens chose between two alternatives by making either key-peck or door-push responses. The reinforcement rates on the two alternatives remained constant and equal throughout conditions, but the number of responses (i.e., key pecks or door pushes) required on each alternative was varied by changing the second-order (fixed-ratio) requirements. The preferences obtained from two pairings of response requirements allowed prediction of the preferences expected in a third pairing. No intransitivities were found, implying that the response requirements lie on a common unitary scale of value. For response-based measures, the obtained preferences varied evenly around perfect, multiplicative prediction, and all satisfied strong transitivity, implying an underlying interval scale of value. For time-based measures, only moderate transitivity was satisfied, implying only an ordinal scale of value. Time-based measures were confounded with the differing times taken to complete each response requirement. The existence of such scales indicates that direct comparisons of different response requirements may be possible.     

Violations of stochastic transitivity on concurrent chains: Implications for theories of choice

The concurrent-chains procedure has been used to measure how choice depends on various aspects of reinforcement, such as its delay and its magnitude. Navarick and Fantino (1972, 1974, 1975) have found that choice in this procedure can violate the condition of stochastic transitivity that is required if a unidimensional scale for reinforcements is to be possible. It is shown in this paper that two simple unidimensional models of choice on concurrent chains can produce violations of stochastic transitivity. It is argued that such violations may result from the complex contingencies of the concurrent-chains procedure.     

The transitivity of work values: Hierarchical preference ordering of socially desirable stimuli

This research investigated whether or not subjects could logically preference order work values. It is important to a hierarchical conception of work values that individuals exhibit transitivity in their expressed preferences, especially as all work values are considered to be highly socially desirable, and thus, importance ratings may lack variance between values. In Study 1, 97 subjects pair-compared eight values. In Study 2, 86 subjects pair-compared different behaviors that represented four work values. In both cases, subjects exhibited total transitivity in their choices more often than could be predicted by chance, and most intransitivities met at a minimum the criteria of weak Stochastic transitivity.     

Testable conditions for triads of paired comparison choices

Forty-two choice models, each representing stimuli by one-dimensional probability distributions, are obtained by relaxing the assumptions of Thurstone's Case V Law of Comparative Judgment. The models which imply or fail to imply each of nine testable probabilistic conditions are determined. Stochastic transitivity is vulnerable in most of these models. The results suggest discarding weak stochastic transitivity, and in its place using the conjunction of weak stochastic transitivity and the triangular condition. However, unless it is possible to predict which stimuli will produce violations of the conditions, none of the conditions can be rejected on the basis of too frequent intransitive triads of choices.I am grateful to the referees, and to J. H. Griesmer and C. H. Coombs for a number of helpful criticisms and comments. J. H. Griesmer suggested a proof of Theorem 9, and a referee suggested a more direct proof of Theorem 5. A proof of Theorem 5 and the result concerning a maximally intransitive judge were given in [18].     

The exchangeable multinomial model as an approach to testing deterministic axioms of choice and measurement

The multinomial (Dirichlet) model, derived from de Finetti's concept of exchangeability, is proposed as a general Bayesian framework to test axioms on data, in particular, deterministic axioms characterizing theories of choice or measurement. For testing, the proposed framework does not require a deterministic axiom to be cast in a probabilistic form (e.g., casting deterministic transitivity as weak stochastic transitivity). The generality of this framework is demonstrated through empirical tests of 16 different axioms, including transitivity, consequence monotonicity, segregation, additivity of joint receipt, stochastic dominance, coalescing, restricted branch independence, double cancellation, triple cancellation, and the Thomsen condition. The model generalizes many previously proposed methods of axiom testing under measurement error, is analytically tractable, and provides a Bayesian framework for the random relation approach to probabilistic measurement (J. Math. Psychol. 40 (1996) 219). A hierarchical and nonparametric generalization of the model is discussed.     

Ambiguous comparative judgment: Fuzzy set model and data analysis

Abstract:  This paper proposed two types of fuzzy set models for ambiguous comparative judgments, which did not always hold transitivity and comparability properties. The first type of model was a fuzzy theoretical extension of the additive difference model for preference that was used to explain ambiguous preference strength. The second was a fuzzy logic model for explaining ambiguous preference in which preference strength was bounded, such as a probability measure. In both models, multi-attribute weighting parameters and all attribute values were assumed to be asymmetric fuzzy L-R numbers. For each model, a method of parameter estimation using fuzzy regression analysis was proposed. Numerical examples were also provided for comparison. Finally, the theoretical and practical implications of the proposed models were discussed.     

Choice theories for differentially comparable alternatives

Several binary choice theories represent choice probabilities as a function of a utility difference divided by a comparability index which is a distance metric. All such theories have the property of moderate stochastic transitivity.