Causal Estimation of User Learning in Personalized Systems

TL;DR

This paper develops a non-parametric causal model to measure how user learning and system personalization affect outcomes over time in online platforms, proposing new experimental designs to accurately identify these effects.

Contribution

It introduces a novel causal modeling framework and experimental designs that account for personalization, enabling accurate measurement of user learning effects in personalized systems.

Findings

New experimental designs successfully identify causal effects.

Parametric assumptions enable long-term effect estimation.

Simulations validate the effectiveness of proposed methods.

Abstract

In online platforms, the impact of a treatment on an observed outcome may change over time as 1) users learn about the intervention, and 2) the system personalization, such as individualized recommendations, change over time. We introduce a non-parametric causal model of user actions in a personalized system. We show that the Cookie-Cookie-Day (CCD) experiment, designed for the measurement of the user learning effect, is biased when there is personalization. We derive new experimental designs that intervene in the personalization system to generate the variation necessary to separately identify the causal effect mediated through user learning and personalization. Making parametric assumptions allows for the estimation of long-term causal effects based on medium-term experiments. In simulations, we show that our new designs successfully recover the dynamic causal effects of interest.