Abstract
<p>Adding an option to a choice set can alter how existing options are compared and systematically shift choice. The attraction, similarity, and compromise effects demonstrate such context dependence, but the empirical record is mixed. We conducted a meta-analysis of 83 studies comprising 964,766 individual choices, distinguishing changes in the target’s standing relative to a competitor from changes in its marginal choice probability. Primary hierarchical analyses yielded relative-choice estimates in the predicted direction for all three effects and, for absolute choice, increased target choice for attraction but little overall change for compromise. Sensitivity analyses accounting for heterogeneity, precision-related effect-size inflation, and possible publication selection provided the strongest evidence for an average attraction effect in both relative and absolute choices but limited evidence for an average similarity and compromise effect. All three effects exhibited substantial heterogeneity across experiments. Moderation analyses showed that effect magnitudes varied with decoy structure, presentation format, attribute type, and decision criterion. Together, these results indicate that context effects are conditional rather than uniform: their magnitude and direction depend on how choice problems are constructed, represented, and measured. The findings challenge choice theories to explain how the canonical effects arise, why their magnitude and direction vary across experimental conditions, and which aspects of choice they alter. For interventions that seek to shape choice via context effects, reconfiguration of the choice set alone is insufficient. An effective intervention will depend on aligning the intended context effect with how the options are represented and compared.</p>