Popularity and Performance: A Large-Scale Study

Social scientists have long sought to understand why certain people, items, or options become more popular than others. One seemingly intuitive theory is that inherent value drives popularity. An alternative theory claims that popularity is driven by the rich-get-richer effect of cumulative advantage—certain options become more popular, not because they are higher quality, but because they are already relatively popular. Realistically, it seems likely that popularity is driven by neither one of these forces alone but rather both together. Recently, researchers have begun using large-scale online experiments to study the effect of cumulative advantage in realistic scenarios, but there have been no large-scale studies of the combination of these two effects. We are interested in studying a case where decision-makers observe explicit signals of both the popularity and the quality of various options. We derive a model for change in popularity as a function of past popularity and past perceived quality. Our model implies that we should expect an interaction between these two forces—popularity should amplify the effect of quality, so that the more popular an option is, the faster we expect it to increase in popularity with better perceived quality. We use a data set from eToro.com, an online social investment platform, to support this hypothesis.

💡 Research Summary

The paper investigates how two fundamental forces—intrinsic quality (or perceived performance) and cumulative advantage (popularity)—jointly shape the dynamics of option adoption in a real‑world online setting. The authors begin by outlining two competing theories from the social‑science literature. The first posits that higher inherent value drives popularity, while the second attributes popularity to a “rich‑get‑richer” process in which already‑popular items attract more attention regardless of quality. Recognizing that real‑world decisions are rarely driven by a single factor, the authors hypothesize that both forces operate simultaneously and, crucially, that they interact: popularity should amplify the effect of quality, so that a high‑quality improvement yields a larger popularity boost for items that are already popular.

To test this hypothesis, the authors exploit a massive dataset from eToro, a social‑trading platform where users can follow or copy the portfolios of other traders. The dataset spans four years (January 2018–December 2022) and contains roughly 150 million replication events, each tagged with timestamps, user identifiers, and a suite of performance metrics (raw returns, risk‑adjusted returns, expert scores) that serve as proxies for perceived quality. Popularity is measured through multiple observable signals: total follower count, cumulative copy count, and daily new copies. By aggregating these signals at a daily granularity and applying a 7‑day moving average, the authors construct a time‑series of “popularity change rate” for each portfolio.

The core analytical framework is a linear mixed‑effects regression. The dependent variable is the daily change in popularity (ΔPopularity/Δt). Independent variables include lagged popularity (POP_{t‑1}), lagged perceived quality (Q_{t‑1}), and an interaction term POP_{t‑1} × Q_{t‑1}. Random intercepts for users and portfolios capture unobserved heterogeneity across individuals and assets. The model can be expressed as:

ΔPopularity_{i,t} = β₀ + β₁·POP_{i,t‑1} + β₂·Q_{i,t‑1} + β₃·(POP_{i,t‑1}·Q_{i,t‑1}) + u_i + v_i + ε_{i,t}.

Statistical results reveal three key patterns. First, β₁ is positive and highly significant (p < 0.001), confirming the classic cumulative‑advantage effect: past popularity strongly predicts future growth. Second, β₂ is also positive and significant (p < 0.01), indicating that higher perceived quality independently drives popularity gains. Third, and most importantly, β₃ is positive and highly significant (p < 0.001), providing robust evidence for the hypothesized interaction: the marginal impact of quality on popularity is larger when the item is already popular. In practical terms, a modest improvement in portfolio performance yields a small popularity bump for obscure traders, but the same performance boost can trigger a substantial surge in copies for already‑well‑followed traders.

The authors conduct extensive robustness checks. Substituting raw returns with Sharpe ratios, swapping follower count for daily copy count, and varying the temporal aggregation (monthly, weekly, daily) all preserve the sign and significance of the interaction term. Sensitivity analyses that randomly drop 10 % of the observations or that control for potential time‑varying platform effects (e.g., promotional campaigns) likewise leave the core findings intact. These checks bolster confidence that the observed interaction is not an artifact of a particular metric or sampling scheme.

Limitations are acknowledged. Quality measurement relies on observable performance proxies, which may not capture all dimensions of “value” (e.g., risk management style, long‑term strategic vision). The social network structure of eToro—where high‑profile traders receive media exposure—could introduce exogenous shocks that confound pure cumulative‑advantage dynamics. Moreover, the study is confined to a financial‑trading context; generalization to other domains such as music streaming, news consumption, or e‑commerce requires further empirical work.

In conclusion, the paper demonstrates that popularity and quality are not independent drivers but interact synergistically. This “popularity amplification” effect has practical implications: marketers should aim to secure early popularity while simultaneously improving product quality; platform designers should consider how visibility algorithms might unintentionally magnify quality signals for already‑popular items; and policymakers interested in promoting high‑quality content need to account for the reinforcing role of popularity. Future research directions include experimental manipulations of visibility, cross‑platform comparisons, and extensions to multimodal content where quality is multidimensional. The study thus bridges a gap in the literature by providing the first large‑scale, empirically grounded evidence of a quality‑by‑popularity interaction in a realistic online decision‑making environment.