BEAST Software
 
 
 
 
 
 
 
 
 
 
 
BEAST Multiple Act Sequences and Communication 
The sequencing of behavior either within or between individuals can have particular power to analyze changes in behavior. Routines suitable for study of communication and similar problems all include graphical functions and include log-linear models of sequencing of a table of preceding-following events and Continuous Time Markov Chain models for a table of events. Each routine has powerful statistical tools to explore departures from randomness that range from standard nonparametric and parametric tests to Randomization and Monte Carlo analysis. "Circles and Arrows" ethogram diagrams are drawn automatically to visualize the frequency of every possible sequence pair or the degree to which each pair sequence departs from a random-expected frequency of transition. 
 
Here is one example in which 5 events are used as both preceding and following acts to examine sequencing within an individual. One half second has been chosen as an intrabout gap for Bite to avoid counting a series of closely spaced bites at a partition. The act might be more aptly named "Biting".  
 
 
The first Ethogram drawn is for the raw counts (below)  - how many times each act followed another. 
 
 
You can quickly redraw the ethogram based on even more informative statistics such as the normalized residuals (below) that indicate the degree to which the transitions were greater than those expected based on a random distribution of acts. 
 
 
Statistical functions are too numerous to present here and include 2 and 2 dimensional information theoretical estimates, log-linear model fitting and extensive curve-fitting functions to fit continuous time Markov chain models for a detailed examination of transition rates between events. Piecewise constant hazard rate models are applied when the data fail to fit a simple Markov chain model.