The Expectation Gap: A Behavioral Lever in Algorithmic Recommendations

Yikang Yan (2024) notes consumers care less about “personalization” per se than about the gap between recommended products and their psychological expectations, alongside perceived efficiency and privacy. Building on this, the project proposes randomized field experiments on e-commerce and social media platforms that manipulate expectation-setting (e.g., pre-views of anticipated recommendation accuracy, rank-expectation cues, confidence intervals) to create exogenous shifts in expected fit. We then measure how expectation gaps causally affect spending, opt-outs from data sharing, and brand trust. To go beyond click/purchase outcomes, we follow Gopnarayan, Aru, and Gluth (2023) by collecting process data—response times, scrolling, dwell time, and eye-tracking where feasible—and estimating evidence accumulation models to infer latent preference conflict when expectations are violated. We also elicit norms about “appropriate personalization” (Edirneligil & Tanhan, 2024) to see whether perceived norm violations mediate privacy backlash. Finally, we train hybrid DNNs on both outcome and process features (see Aoujil et al., 2023 on BE+AI trends) to predict which users are most sensitive to expectation gaps. Novelty: shifts the focus from algorithmic accuracy to psychologically constructed expectations; treats the expectation gap itself as a manipulable lever. Impact: guidance for platform design (how to set expectations ethically), privacy policy (anticipating opt-out dynamics), and a generalizable process-informed modeling framework (Gomes, 2022) that links micro-level deviations to macro spending and trust.

References:

1. Social Norms And Norm Elicitation In Behavioral Economics. A. Edirneligil, Esra Tanhan (2024). Sosyal Mucit Academic Review.
2. The Impact of E-commerce and Social Media Personalized Recommendations on Consumer Behavior in the Digital Era from the Perspective of Behavioral Economics. Yikang Yan (2024). Advances in Economics, Management and Political Sciences.
3. Behavioral economics and finance: a selective review of models, methods and tools. O. Gomes (2022). Studies in Economics and Finance.
4. Artificial Intelligence and Behavioral Economics: A Bibliographic Analysis of Research Field. Zakaria Aoujil, Mohamed Hanine, E. Flores, Md Abdus Samad, Imran Ashraf (2023). IEEE Access.
5. From DDMs to DNNs: Using process data and models of decision-making to improve human-AI interactions. Mrugsen Nagsen Gopnarayan, Jaan Aru, S. Gluth (2023). Decision.