Identify the structural communication-selection process

Identify the structural selection model governing whether public communications become observable as a function of their content or an actor’s latent position, using additional assumptions or data beyond the restricted public-source corpus.

Background

The paper’s corpus contains only communications retrieved under a declared and deliberately narrow public-source protocol. The probability that a communication enters this corpus may depend on its content or on the actor’s latent position, so treating observability as ignorable would require a substantive selection model.

The authors explicitly state that the present corpus cannot identify such a model. They therefore report the risk set and observed counts, examine extremity-dependent visibility in simulation, and limit interpretation to the retrieved public record. A solution would require information or assumptions that characterize the universe and distribution of unobserved communications.

References

If the probability that a communication becomes public depends on its content or the actor’s latent position, a structural selection model would require additional assumptions or data (Rubin, 1976; Carrubba et al., 2006; Hug, 2010). The present corpus cannot identify such a model.

— Public Signals, Concealed Choices: Dynamic Measurement without Behavioral Identification  (2608.21077 - Spruk, 21 Aug 2026) in Section 2.1; Online Supplement Section S13

Private meetings, cabinet deliberations, informal conversations, direct messages, unpublished memoranda, and undocumented exchanges are outside it. Their number, direction, timing, and distribution are unknown.

— Public Signals, Concealed Choices: Dynamic Measurement without Behavioral Identification  (2608.21077 - Spruk, 21 Aug 2026) in Online Supplement Section S3.4, “Public-signal sampling frame and unobserved private communication”