Behavioral Research Lab
The Marketing Department at the University of Florida has one of the best-equipped and most productive laboratories for behavioral research in the nation. Our laboratory consists of 24 lab stations for face-to-face research, a network file server, and a web server and software for conducting complex, multimedia on-line experiments. Experimenters can post lab studies and research participants can sign up for sessions at CB-Central, our state-of-the-art website for scheduling, posting, and tracking research participation. On-line research is posted on the WebLab server and can be accessed through CB-Central when available.
The behavioral lab consists of 24 carrels for the collection of individual data. Carrels each contain a PC running Windows XP (CPU under the desk) and 15 inch flat-screen monitor. Carrels are large enough to allow for presentation of pencil-and-paper surveys and arrays of physical stimulus materials. Carrels also have a noise-dampening construction that minimizes cross-participant interference.
Each computer is connected by MS Active Directory to a dedicated file server from which all studies are run. This design insures that all programs and data files are stored in a central location using a RAID array for redundancy. An experimenter station is positioned on a riser at the back of the room. Experimenters can remotely view and control each participant’s workstation using SMART Technologies SynchronEyes, which allows experimenters to lock the stations pending instructions, block internet access and participants’ access to particular programs, and communicate with individual participants. Study programs can be launched using LaunchPad, a program specifically designed for the randomization and launching of experiments in a lab environment.
The management of experimental sessions is controlled through CB-Central. All aspects of session management are done through the CB-Central interface: Experimenters post their sessions and assign participation credit there, and participants sign up for sessions and manage their participation credits through the interface. Unlike other third-party session management solutions, CB-Central allows experimenters to enforce arbitrary rules (for example, you may only sign up for study 3500 if you have already completed 3400, but not if you have done 3600) at the level of experiments and individual participants. This is especially useful because it allows experimenters to pre-screen subjects for a particular study based on results from an earlier study without the participants’ awareness of this relationship.
To experience CB-Central for yourself, log in as ‘guest’. Note that depending on the time of the semester, there may not be any studies active or available.
WebLab is a system for collecting behavioral data over the Internet. It is designed to run Internet experiments of a variety of forms (e.g., scripting languages, HTML forms, etc.), but is especially good at running Authorware experiments. WebLab has an independent interface for posting and administering studies, but it also interfaces with CB-Central to display active web-studies and award participation credit. The WebLab system allows participants to participate at a time and place of their choosing. It also allows experimenters to collect data from specific on-line samples and reduce the time they spend collecting data in the behavioral lab.
Our department runs studies of many different forms. Typically, our computer-based research is written using Macromedia Authorware, a multimedia computer-based training authoring package. In order to make the experiment authoring process easier, we have developed an Authorware “wrapper” program called LabShell that controls all aspects of interfacing with the control station, the file server and WebLab. Using LabShell, experimenters need simply to create their experiment locally and then embed it in LabShell. Their study can then be run in the lab or over WebLab (or both). To extract and collate data across participant log files, we use Perl, an open-source, cross-platform scripting language for data extraction and reporting.
Obviously, behavioral researchers need to invest time in order to learn any system for experimentation. To further this aim, we provide classes within the department on programming for behavioral experiments, and we maintain a separate online discussion forum and tutorial repository.