Comparer des méthodes

	Structural Topic Model	Wordfish Scaling
Domaine	Political Science	Political Science
Famille	Process / pipeline	Latent structure
Année d'origine	2014	2008
Auteur d'origine	Margaret Roberts, Brandon Stewart & Dustin Tingley	Jonathan Slapin and Sven-Oliver Proksch
Type	Mixed-membership topic model with document-level covariates	Unsupervised latent-position model for word-count data
Source fondatrice	Roberts, M. E., Stewart, B. M., Tingley, D., Lucas, C., Leder-Luis, J., Gadarian, S. K., Albertson, B., & Rand, D. G. (2014). Structural Topic Models for Open-Ended Survey Responses. American Journal of Political Science, 58(4), 1064–1082.	Slapin, J. B., & Proksch, S.-O. (2008). A Scaling Model for Estimating Time-Series Party Positions from Texts. American Journal of Political Science, 52(3), 705–722.
Alias	STM, Structural topic modeling, Covariate-aware topic model, Topic model with metadata	Wordfish text scaling, Poisson scaling of texts, Unsupervised text scaling, Wordfish position estimation
Apparentées	5	4
Résumé	The Structural Topic Model (STM) is a text-as-data method that discovers latent themes in a corpus while letting document metadata — party, time, gender, treatment condition — shape those themes. Introduced by Roberts, Stewart, Tingley and colleagues in 2014, it generalizes correlated topic modeling so that topic prevalence (how much a document is about a topic) and topic content (the words used to express a topic) can both depend on covariates. The result is a single model that simultaneously estimates topics and how their use varies across known groups, with uncertainty.	Wordfish scaling is an unsupervised text-as-data method that estimates a single latent position for each political document — a party manifesto, a legislative speech, a press release — directly from its word frequencies, without any reference texts or hand coding. Introduced by Slapin and Proksch in 2008, it models word counts as draws from a Poisson distribution whose rate depends on a document position and word-specific parameters, recovering, for example, a left–right ordering of parties purely from how often each word appears in each text.