Tisane is an interactive system for authoring generalized linear models (GLMs) and generalized linear mixed-effects models (GLMMs). It provides a study design specification language to author conceptually founded linear models that are optimized for generalizability.

Tisane’s study design specification language allows analysts to specify conceptual and data relationships between variables. Analysts can query Tisane for a linear model involving specific variables of interest. Tisane checks the query is well-formed based on the expressed relationships and automatically generates a space of possible linear models that respect the stated relationships. Tisane asks analysts disambiguating questions about their analysis intents through a graphical user interface (GUI). Based on the specified study design and additional responses, Tisane outputs code for fitting and visualizing a statistical model. The output model is guaranteed to be conceptually consistent with the specified relationships and optimized for generalizability.

Tisane is designed for analysts who have domain expertise, are not statistical experts, and are comfortable with minimal programming (e.g., many researchers).

The Tisane project is currently active at the University of Washington’s Paul G. Allen School of Computer Science.


Eunice Jun and Audrey Seo lead development of Tisane with advisement from Jeffrey Heer and Rene Just.

Shreyash Nigam has also contributed to the project.