Planning an Experiment and Generating a Hypothesis
With a research question in mind, scientists work to choose the right behavioral theory
and then must think through how it applies to the case in question.
Predictions from the chosen theory are estimated
qualitatively by the researcher, who must think through the psychological processes of the subject as they go through
the experiment.
Predictions from the chosen theory are examined
quantitatively by creating simulations based on existing models of the chosen theory.
Modifying Existing Theory
We propose that existing theory is inadequate to explain phenomena observed in our experiment,
and we provide a modification to the theory which explains the data.
You must read through my paper and think through the experiment from the subject's perspective in
order to understand how to use it.
Here is the new equation.
I must make a compelling argument for this contribution's importance and
relevance outside of this experiment.
Model Evaluation Using Experimental Data
Our model is better than the status quo. This paper explains in detail why we think it is
better and how it allows you to make predictive estimates more accurately. The p-value provides
quantifiable evidence that some effect exists.
Our model is 5% more accurate than the status quo in it's estimation of our experimental data.
The improved goodness of fit provides a quantifiable measure of how much the theory is improved
through inclusion of this effect.
Tools
Free and easy tools to modernize your research.
Model Builder
Participate in our Online Usability Study
Agent Simulator
Sneak Peak; Development Postponed
Data Reconsiliation Tool
Development Not Yet Started
BehaviorSim tools help researchers formulate and test models using simulations, guide experimental design and theory development.
Contact Us
Help us help you help others.
Model Builder Usablity Study
Help Shape our Tools
The behaviorSim Model Building Tool is a GUI web application designed to help users create models. The application is entirely browser-based, meaning that you do not have to install any software to use it.
With this tool we aim to:
Make formulation of dynamical models of human behavior more accessible.
Facilitate discovery and learning of new modeling techniques.
The behaviorSim agent simulator uses models (defined using the Model Builder or otherwise) to recreate real-world experiments in simulation. In this tool simulations for individual or a population of users can be created, customized, and explored. Simulation agents can be configured to use different existing models, and personality constants can be set for each agent. Contextual inputs (aka exogenous inflows) can be defined to match your experimental conditions, and target measures can be output for comparison to real-world data.
A proof-of-concept implementation of the agent simulator has been written in Python and is available on the University of South Florida's PIE Lab Github Page, but it is in a very rough state. A more polished, more user-friendly application is next on the to-do list after successful evaluation of the Model Builder. Please contact us if you are interested in contributing to this open-source effort.
Data Reconsiliation Tool
Evaluate Your Models Using Data
The Data Reconsiliation Tool is a general purpose tool for allowing side-by-side comparison of real and simulation data. Development on this tool is currently un-funded and thus nearly non-existent. Luckily, existing tools from other research domains may be of use here, but further investigation on the ideal software(s) and development in order to interface with existing work is needed.