Skip to main content Skip to navigation

David Firth, software: Llama

Log Linear and Multiplicative Analysis

What is it?

Llama is a small package which works in the Xlisp-Stat statistical computing environment. Llama exists to specify and fit log-linear and log-multiplicative models to data in the form of a multi-way contingency table (that is, a cross-classification or cross-tabulation by two or more categorical variables).

PC, Mac or Unix?

Yes! The required Xlisp-Stat environment runs on a PC under MS Windows (including 95 and NT), or on a Macintosh, or on many Unix/X11 systems.

What does it cost?

Both Llama and Xlisp-Stat are distributed free. Of course, that doesn't mean there's no cost: you'll still have to spend some time getting, installing and finding out how to use the software. Experience with students and colleagues suggests that Llama is not too hard to learn, though: it is a small, special-purpose system, and it has a rudimentary menu/dialog interface.

How do I get it?

You'll need to download up-to-date versions of:

 Xlisp-Stat    Llama      

And the manual?

Documentation for the Xlisp-Stat system is provided by the excellent book LISP-STAT: An Object-Oriented Environment for Statistical Computing and Dynamic Graphics by Luke Tierney (Wiley, 1990). The Book is not essential for basic use of Llama but will facilitate more general analyses and graphs.

Find specific documentation for Llama, including some examples, here. An overview is given in the published paper "LLAMA: An object-oriented system for log multiplicative models", in COMPSTAT 1998, Proceedings in Computational Statistics (eds. R. Payne & P. Green), pp305-310, Heidelberg: Physica-Verlag.


Llama is still useful, but is no longer actively supported. Its functionality is included entirely in the new gnm package for the R statistical computing environment.

Related links:

Completely unrelated link: