Extent to which pre-retraction Twitter mentions indicate problems leading to retraction

Determine, using large-scale datasets comprising both retracted and non-retracted scholarly articles, the extent to which pre-retraction Twitter mentions contain recognizable evidence of problems within the articles that may lead to eventual retraction.

Background

Prior studies have shown that retracted articles attract substantial attention on Twitter and sometimes receive criticism or negative sentiment before formal retraction. However, most evidence has been limited to specific cases or smaller samples, leaving uncertainty about how broadly and reliably Twitter discussions signal problems across large, heterogeneous datasets.

This problem asks for a systematic quantification of the predictive signal within pre-retraction tweets when analyzing matched sets of retracted and non-retracted articles, thereby clarifying the usefulness of social media as an early warning system for research integrity.