Written Justifications are Key to Aggregate Crowdsourced Forecasts

TL;DR

This paper shows that incorporating written justifications from crowdsourced forecasters improves the accuracy of aggregated forecasts, especially during most of the forecast horizon, and analyzes factors affecting justification reliability.

Contribution

It introduces the importance of modeling written justifications in forecast aggregation and provides empirical evidence of their benefits over traditional voting methods.

Findings

Justifications improve forecast accuracy during most of the forecast period.

Majority and weighted vote baselines are competitive with justification-based methods.

Unreliable justifications can be identified through error analysis.

Abstract

This paper demonstrates that aggregating crowdsourced forecasts benefits from modeling the written justifications provided by forecasters. Our experiments show that the majority and weighted vote baselines are competitive, and that the written justifications are beneficial to call a question throughout its life except in the last quarter. We also conduct an error analysis shedding light into the characteristics that make a justification unreliable.