Identify the causal effect of repeat reposting on cascade growth

Identify the causal effect of repeat reposting on the subsequent growth of cascades, separately from the selection processes that cause certain accounts or posts to receive repeat reposts.

Background

The paper compares cascades with repeat reposts against matched cascades without repeats and also uses controls matched on the estimated reach of the repeating account. These analyses find that growth is associated more with the repeater’s reach and source structure than with the number of repeats.

However, the study is observational, and the authors state that the matching design cannot distinguish the effect of repeating from pre-existing differences in cascade growth or account selection. A causal design, such as a natural experiment, is therefore needed to determine whether repeat reposting itself changes subsequent audience growth.

References

The direction of causality between repeats and growth is not identified. The matched comparison of RQ2 is observational and cannot separate the effect of repeating from selection: low-reach accounts push cascades that are not growing'' andseveral accounts push cascades that are''.

— Attentional DoS: Repeat Reposting, Collective Attention, and Information Diffusion on X  (2609.35385 - Toya et al., 28 Sep 2026) in Section 6.3, “Limitations,” paragraph “Direction of causality and the limits of observational data”