Selection Effects in Online Sharing: Consequences for Peer Adoption

Most models of social contagion take peer exposure to be a corollary of adoption, yet in many settings, the visibility of one’s adoption behavior happens through a separate decision process. In online systems, product designers can define how peer exposure mechanisms work: adoption behaviors can be shared in a passive, automatic fashion, or occur through explicit, active sharing. The consequences of these mechanisms are of substantial practical and theoretical interest: passive sharing may increase total peer exposure but active sharing may expose higher quality products to peers who are more likely to adopt. We examine selection effects in online sharing through a large-scale field experiment on Facebook that randomizes whether or not adopters share Offers (coupons) in a passive manner. We derive and estimate a joint discrete choice model of adopters’ sharing decisions and their peers’ adoption decisions. Our results show that active sharing enables a selection effect that exposes peers who are more likely to adopt than the population exposed under passive sharing. We decompose the selection effect into two distinct mechanisms: active sharers expose peers to higher quality products, and the peers they share with are more likely to adopt independently of product quality. Simulation results show that the user-level mechanism comprises the bulk of the selection effect. The study’s findings are among the first to address downstream peer effects induced by online sharing mechanisms, and can inform design in settings where a surplus of sharing could be viewed as costly.

💡 Research Summary

This paper investigates how the mechanism by which an adopter’s behavior becomes visible to peers—whether automatically (passive sharing) or through an explicit user action (active sharing)—creates selection effects that shape downstream peer adoption. The authors conduct a large‑scale field experiment on Facebook, randomly assigning users who redeem a coupon (an “Offer”) to either a passive‑sharing condition, where redemption is automatically broadcast to all friends, or an active‑sharing condition, where friends only see the redemption if the user deliberately clicks a share button.

To capture the joint decision process, the authors develop a discrete‑choice econometric model. The sharing decision for product k by user i depends on a product‑specific sharing utility μₖ (drawn from a Normal distribution with mean α) and an idiosyncratic term εᵢₖ. In the passive condition sharing occurs with probability one for adopters; in the active condition sharing occurs only if μₖ + εᵢₖ exceeds a threshold. The peer’s adoption decision, observed only when sharing occurs, is modeled similarly with a product‑specific adoption utility λₖ (Normal with mean γ) and an idiosyncratic term νᵢₖ.

Three potential selection mechanisms are embedded in the model: (1) Product selection, captured by the correlation ρ between μₖ and λₖ, implying that higher‑quality products are both more shareable and more adoptable; (2) Dyad (adopter‑peer) selection, captured by the correlation ψ between εᵢₖ and νᵢₖ, implying that users who choose to share tend to have peers who are intrinsically more likely to adopt; and (3) Adopter selection, which is subsumed under dyad selection because the model does not separately observe adopter influence.

Using Bayesian estimation on the experimental data, the authors find strong evidence for dyad selection (ψ > 0) and weaker evidence for product selection (ρ > 0). In concrete terms, peers who receive an active share are exposed to products with a higher expected adoption utility and, more importantly, belong to a segment of the network that is more receptive to adoption regardless of product quality. Simulations based on the estimated parameters show that dyad selection accounts for roughly 70 % of the total selection effect, while product selection contributes only a modest share.

The empirical results reveal that, despite a lower overall exposure rate under active sharing (≈30 % of adopters actually share), the conversion rate among exposed peers is substantially higher—about 12 percentage points—than under passive sharing. This demonstrates that a “quality‑over‑quantity” approach to exposure can be more efficient for viral marketing.

From a design perspective, the findings suggest that platform engineers can manipulate sharing costs to induce selective sharing, thereby reducing unnecessary noise (over‑sharing) while preserving or even enhancing the overall diffusion efficiency. Marketers can leverage this by providing incentives for users to actively share high‑value offers, ensuring that the message reaches a more receptive audience.

The paper contributes to the literature on social contagion by explicitly modeling and empirically identifying selection bias in the exposure process, a factor often omitted in classic diffusion models that assume exposure is a deterministic consequence of adoption. It also bridges the gap between theoretical diffusion frameworks and practical product‑design decisions in online environments.

Limitations include the focus on a single platform (Facebook) and a specific product class (coupons). Future work could extend the framework to other social media, to content‑type products (e.g., news articles, videos), and to dynamic multi‑step sharing processes where the network structure evolves over time. Incorporating richer user‑level heterogeneity and longitudinal data would further illuminate how selection mechanisms operate across different contexts and over longer horizons.