BEAST Software
 
 
 
 
 
 
 
 
 
 
 
BEAST Single Act Sequences  
The sequencing of behavior can have particular power to analyze changes in behavior. Routines that include graphical functions include change-point analysis and log-survivorship interval analysis. 
 
A search over the observation period for abrupt changes in behavior can avoid confounding your results by lumping data from dissimilar causes. 
 
"Profile" graphs show the distribution of acts over time. This graph shows the intervals between player 1's "choice of "weapons" in a Rock-Paper-Scissors game. 
 
Statistical testing for a Change Point indicates that only event 4, 1_Scissors, had a barely significant change (p<0.05) at 67 seconds. There is no reason to suspect that the others were not randomly distributed. this could be the first step in feeling justified in lumping all events from the entire session. 
 
Log survivorship studies of the distributions of bout and interval lengths are well known in the behavioral sciences. 
 
BEAST can test the fit for a single line that is expected if the interval lengths are randomly distributed. It can also fit a 2-part Piecewise Constant Hazard Rate model that assumes that 2 different processes are responsible for the distribution of interval lengths. For a full explanation you should download the BEAST Help file. Suffice it to say that, in the above left curve, except for the one longest interval found, the above curve fitting attempt supports the hypothesis that interval lengths were randomly distributed.