Redefining Digital Access
Paper Session
Saturday, Jan. 3, 2026 2:30 PM - 4:30 PM (EST)
-
Chairs:
Luisa Blanco, Pepperdine University - Mary Lopez, Occidental College
Algorithmic Bias and Historical Injustice: Race and Digital Profiling
Abstract
This paper studies the implications of attempts at "ethnic-affinity" profiling on Facebook that reflects users' engagement with content on Facebook. Profiling by ethnic-affinity is highly correlated with Census estimates of population race by geography. However, more users were profiled as African-American in former slave states relative to the baseline population. This occurs because the targeting algorithm was better at identifying Black users through differentiated engagement with cultural content in these states. This implies that policies restricting the collection of racial identity data will be unsuccessful due to the existence of proxies, and that relying on proxies may introduce troubling biases.The New Digital Divide
Abstract
We build and analyze new metrics of digital usage that leverage telemetry data collected by Microsoft during operating system updates across forty million Windows devices in U.S. households. These measures of US household digital usage are much more comprehensive than those made available through any existing commercial or government survey. We construct representations of devices in ZIP codes and find evidence of significant variation in usage reflecting an urban-rural divide. We also show the existence of substantial disparities in usage even within narrowly defined Metropolitan Statistical Areas. Income and education correlate with these observed differences. These effects are large and suggest digital literacy gaps that extend beyond the availability of essential IT infrastructure at the local level. These findings call for interventions beyond the traditional focus on infrastructure access and address usage and skills development. The indices are made publicly available to support future research.Digital Incentives in Surveys: Evidence from an RCT on Response Rates, Vendor Choice, and Sociodemographic Effects
Abstract
Digital incentives have become a popular strategy to increase survey participation, but questions remain about their effectiveness across subgroups and the design features that shape individual behavior. This study draws on a multi-year randomized controlled trial called Texts4Teens, a text-based intervention delivered across Texas school districts. From 2021-22 to 2023-24, over 14,000 middle school parents were randomized to receive a digital gift card incentive upon completing an end-of-program survey. While the availability of incentives was randomized within each year, other design features—such as incentive amount ($5 vs. $10) and whether reward options were disclosed in advance—varied across years.We examine five key research questions: (1) Who responds to digital incentives? (2) Among those offered incentives, who completes the steps to claim them? (3) How do participants’ reward selections vary by sociodemographic background? (4) What is the effect of increasing the incentive amount on response rates and claims? (5) What is the effect of providing advance information about incentive options? The analysis combines randomized comparisons with difference-in-differences models to estimate both average and subgroup effects by race/ethnicity and socioeconomic status. Outcome measures include survey completion, reward claim behavior, and choice of vendor (e.g., broad retail vs. specialized). By integrating administrative, survey, and fulfillment data, the study offers new insights into how incentive strategies can be tailored to improve participation, reduce nonresponse bias, and promote equity in research and evaluation settings.
Discussant(s)
Seema Jayachandran
,
Princeton University
Alex Albright
,
Federal Reserve Bank of Minneapolis
Jeff Prince
,
Indiana University
Pamela Jakiela
,
Williams College
JEL Classifications
- J1 - Demographic Economics
- L8 - Industry Studies: Services