ONLINE - Signal Reconstruction from Multiple Ranked Lists via Convex Optimization applied to World University Ranking Data

07.05.2021 11:15 – 12:15

RESEARCH CENTER FOR STATISTICS SEMINAR / ABSTRACT

The ranking of items is widely used to rate their relative quality or relevance across multiple assessments. Beyond classical rank aggregation, it is of interest to estimate the, usually unobservable, latent signals that inform a consensus ranking.

Under the only assumption of independent assessments, we introduce indirect inference via convex optimization in combination with Poisson Bootstrap. This approach allows us to overcome major numerical limitations of a recent distribution function approach (Svendova and Schimek, 2017, CSDA, 115, 122-135). The relationships between the rankers and the observed item orderings are represented by means of a set of constraints. The estimation strategy is to reduce the noise between these rankings until optimal latent signals can be obtained. Advantages are high-quality parameter estimates as well as error bounds, and a substantial reduction in computational effort in comparison to distribution function but also to probabilistic model approaches. Finally, with our novel signal estimation approach we analyze the consensus of world university rankings (ARWU, QS, and THE) of the year 2020 provided by well-established institutions. On basis of our approach, we point at discrepancies between signal estimation and aggregation results. Shortcomings of rank aggregation concepts, such as the popular Aggregate Ranking of Top Universities (ARTU), are discussed.

Lieu

Online

entrée libre