Java Data Grinders - Applets/Applications for the Analysis of Behavioral Data

Read the introduction to the basic concepts behind this library. The JavaGrinders Application package includes components for automated video tracking, implementations for a range of statistical techniques used in behavioral analysis, and interfaces for the control of mechanical and sensor systems as part of dynamic, experimental designs.

Quantitative methodology figures prominently in the behavioral sciences. It is somewhat surprising, therefore, that few statistics packages and biometry books cover some of the most useful techniques in our field. The analysis of transition matrices, or of interaction frequencies within a group of individuals, are commonly dealt with in behavior, and yet none of the common statistics packages even mention the possibility of a contingency table where the diagonal might be missing. Many researchers in behavior have written their own programs for such analyses, little is being shared, and a lot of effort has been duplicated. This site is therefore dedicated to providing a clearinghouse for common as well as specialized techniques in the behavioral sciences. Feel free to use, contribute, discuss, and disseminate this project.

A summary of the concepts behind this site is available as sets of lecture notes for Advanced Statistics in Biology and Animal Behavior.

The capabilities provided through this initiative are available in different forms depending on your level of expertise and need for specifying options flexibly. This document introduces a set of applets which are commonly run through a webbrowser. This option is coded to represent the most common uses for this type of analysis but it is rather inflexible with respect to its options - Frequency Matrix Applet, Transition Matrix Applet, and the Dominance Structure Applet.

For those who wish to gain direct control of the options provided for these procedures, you may use the classes provided in the JavaGrindersLibrary (a statistical package with procedures for the analysis of contingency tables, dominance and transition matrices, and spatial data) to develop your own analysis programs using the Java language. Classes that implement these features are provided in form of a library which you can include in your own Java programming. The JavaGrinders Application page is dedicated to guide you through that process. Moreover, a videotracking application with an analysis of spatial structure is now available as well - a new version of the library is in the works and will be released shortly. Email me if you are interested to test the beta version.

Experimental Design Road Map

Frequency data To test for the independence of two variables use the Frequency Matrix Applet: This applet uses contingency tables and tests whether the occurrence of two variables is independent (e.g. whether eye color is independent of sex). Chi-square and G-Tests are performed along with Williams' correction, Goodman's Test for quasi-independence if the diagonal is missing, and a cell-wise examination of the matrix using Freeman-Tukey deviates.
Ordinal data
Continuous data

Research Question Road Map

Temporal Structure of Behavior. Q: Are the occurrences of subsequent behavior patterns independent of previous ones? - use the Transition Matrix Applet. Download Contingency.sit below and run the applet locally according to instructions.
Dominance Hierarchy Structure: Q: Are dominance relationships within a group of animals significantly linear? - use the Dominance Structure Applet. Download Contingency.sit below and run the applet locally according to instructions.
Dominance Hierarchy Structure: Q: Are individuals who often perform a particular behavior also more likely to be the recepients of this behavior from others? - use the Dominance Structure Applet. Download Contingency.sit below and run the applet locally according to instructions.

Java Applets

Java Applets download automatically as part of webpages and begin to execute their "untrusted" code. In this case, it makes good sense to only allow actions which pose no danger to the local environment. For security reasons then, an Applet cannot access or modify files on the local computer where it is running. This necessitates that data are entered manually and precludes the use of any local data files provided by the user.

You can gain direct control of the computations and can thus change a variety of options which are otherwise preset on the Java Grinders WebPage. You might for example like to increase the size of the matrix beyond the options available here. The simplest way to do this is to download an archive of HTML and compiled code files for the FrequencyMatrix and HierarchyMatrix Applets.

Download as:

zipped folder - JavaGrindersApplets.zip

Extract content. Open the file "FrequencyMatrix.html" or "HierarchyMatrix.html" using a text editor, and change values of the various fields as desired (i.e. number of rows, columns, or values of individual cells). The types of parameters and their possible values are:

Parameter_Name, Value, Description
{"title", "string", "graph title"},
{"rows", "int", "# rows"},
{"columns", "int", "# columns"},
{"shape", "int", "1 - full matrix, 2 - upper triangle, 3 - lower triangle"},
{"diagonal", "int", "1 - with diagonal, 0 - without diagonal"},
{"Cell_0_1", "double", "hardcode Value for cell at location 0 and 1"}.

You will also encounter the line

which loads the respective applet from the enclosed library as long as you keep the file "JavaGrindersLib_small.jar" in the same location as the accompanying html files. You can increase the size of the applet by providing larger values for width and height (in pixels).

Save your changes for the .html file and open it with your favorite webbrowser. The browser will load and run the applet locally using your custom settings.

You might also want to check some of the following links:

JWatcher - is a powerful tool for the quantitative analysis of behaviour. JWatcher can be used simply as an event recorder which logs the time at which keys are pressed. It was designed primarily for focal sampling, in which a single subject is observed continuously. The real power of JWatcher is, however, in its analysis routines. These calculate time budgets and provide statistics about the duration of behavioural states and the intervals between them.

You can help expand and improve the content of this site by submitting links, bug-reports, source code, feature requests, articles, or general feedback directly to the Java Grinders Editor <lobsterman.bgsu@gmail.com>. Back to my personal homepage, or my professional homepage.

last updated 1/6/09