Designing Multimedia Internet Experiments using Authorware Attain

Behavioral researchers increasingly require software for designing and conducting Internet experiments. One software package that Internet experimenters are increasingly turning to is Authorware Attain. Authorware Attain provides for the presentation of text, images, sounds, and animation and allows the user to interact with the experiment in complex ways. Experiments can be run remotely over the Internet (or over an Intranet) from a web browser. Communication with the experiment server is possible using a variety of protocols, including HTTP and FTP. I outline the design of Authorware experiments in general and discuss issues related to running Authorware Attain experiments over the Internet. I also provide a compendium of resources for the Authorware designer. Authorware Experiments Behavioral researchers have shown increasing interest in using the Internet to collect data. It is hardly surprising that researchers are tempted by the idea of laboratories without walls. Some researchers do not have ready access to the subject populations needed for their research, or require populations that are distributed over wide geographic areas. Other researchers would like to make research participation easier for the subject. Still others are interested in maintaining a laboratory of less-powerful computers that connect to an experiment server via an Intranet or the Internet. Although many issues related to Internet research have yet to be resolved (c.f. Schmidt, 1997; Birnbaum, in press), the technology for running highly complicated, multimodal experiments over the world-wide-web already exists. In this paper, I will review one software package, Authorware Attain, that is designed for the presentation of multimodal stimuli and the collection of behavioral responses over the Internet. I will provide a brief overview of the software and discuss the issues involved in using it for Internet-based behavioral research. A variety of tools are available for designing computerized experiments. These tools typically vary in ease-of-use and flexibility. Some software packages are developed specifically for experimenters. For example, E-Prime (http://www.pstnet.com/e-prime.htm) allows experimenters to quickly design simple experiments and includes procedures for data analysis. However, it is often difficult or impossible to design experiments having complex designs or involving specialized interaction with the subject using such specialized software. Furthermore, it is not possible to run experiments over the Internet with such software. Authorware Experiments On the other end of the spectrum are programming languages like C, Basic, and Java. These programming languages offer great flexibility, but require substantial training. While it is possible to design experiments that can be run over the Internet with these languages, the process is often quite involved and time-consuming. (See Hecht, et al. 1999 and Stevenson, Francis & Kim, 1999 for examples of Internet experiments using the Java programming language.) Authorware, originally developed for the design and deployment of computerbased training and testing applications, is designed for the middle of this spectrum. Authorware offers a graphical, drag-and-drop interface for the scripting of routine tasks such as presenting stimuli or collecting responses. Authorware also includes a programming language complete with arrays, custom variables, and functions. Authorware has built in various means for communicating across a network and distributes a free Authorware Web Player (formerly called Shockwave for Authorware) that enables subjects to run Authorware experiments on most web browsers. Finally, there is a thriving community of Authorware developers who distribute libraries to perform common or useful tasks. Authoring an Experiment: The Basics Wolfe (1992) reviewed many of the basics of Authorware Professional, an earlier version of the Authorware software. His review focused primarily on using Authorware to design courseware, but many of the issues apply to experiment design. Authorware Attain is similar to Authorware Professional, but many important features have been added. I will first briefly review the basics of the software, noting additions. 1 Authorware Attain is the brand name for version 5.x of the Authorware package. Wolfe (1992) reviewed version 1.0. Authorware Experiments Experimenters design Authorware experiments by placing icons of different types on the program “flowline”. Available icons are summarized in Table 1. Icons control the presentation of stimuli (either text, images, sounds, or animations), the actions of the program (the order in which stimuli are displayed and erased, and how variables are computed), and how the user can interact with the program (typing, pressing buttons, sliding sliders, and so forth). Table 1: Authorware Attain icons and their functions. Authorware Icon Function Display Presentation of static images, movable images, drawings, text, or variables. Motion Moves a screen object along a predetermined path. Erase Erases the contents of one or more display icons. Wait Pauses the program until a user event or for a fixed amount of time. Navigate Directs program flow to a predetermined or subjectselected icon. Framework Contains a hypermedia structure of other functions that subjects can navigate through and search. Decision Allows the program to execute different paths depending on the subjects’ responses. Interaction Allows the subject to react to stimuli by pressing keys, entering text, clicking buttons, etc. Calculation Calculates variables and executes various functions. Map Groups sets of icons into one icon. Digital Movie Presents digital movies. Sound Presents prerecorded sounds. Video Includes still images, sound, or animation from videodisc. Knowledge Object Allow the scripting of design tasks with a “wizard” interface. Experimenters enter parameters that modify the actions of the underlying icons. Birds and Planes: A Simple Experiment As an example, I describe a simple two-condition experiment. The experiment randomly assigns subjects to one of two conditions. Subjects in the first condition see a picture of a bird and the word “bird”. Subjects in the second condition see a picture of a Authorware Experiments plane and the word “plane”. Subjects select either the animate or inanimate button, depending on the stimulus. The condition and response are then written to a text file. Figure 1 shows the Authorware file for this experiment. The first icon (“Condition”) is a decision icon. Double-clicking the icon will open a dialog box of icon properties including options for branching and repeating. We want each subject to only receive a stimulus once, so we set the properties to “Don’t Repeat” and “Branch Randomly to Any Path”. This will cause the program to select one of the two attached map icons (“Birds” or “Planes”) randomly and begin executing that map icon. ----------------------------------------------------------------------Insert Figure 1 about here. ----------------------------------------------------------------------Each of these map icons includes four icons, a display icon, an interaction icon, and two attached calculation icons. The display icon displays the stimulus. Doubleclicking this icon will open the display. Text can be added by using the text tool. (A variety of other simple drawing tools are included.) The picture can be added by using the “insert | image” menu listing. Double-clicking the interaction icon will also open the display. This allows experimenters to enter text or pictures that should only be displayed when the interaction begins, for example, the text of the response question. The interaction icon points to two ovals indicating button interactions. When the subject presses one of the buttons, the code in the corresponding calculation icon is executed. In our case, the code includes three commands: