The Basic Idea
You need three things to conduct a MediaLab experiment. First, you need the materials or stimuli you want to show participants. In many cases, you may simply have a number of instructions and questions you want to give participants. Although MediaLab has the capability to present many file types, some people just want to administer a basic questionnaire. If this is the case, then you should find MediaLab an extremely easy way to do this. However, if you want to show images, play sound files or movie clips, display Word documents, PowerPoint shows, HTML pages, or execute other stand-alone programs, you will find MediaLab to be extremely cooperative in this respect as well.
Second, you need dependent measures. Typically, these will consist of questions you have regarding participants' responses to your instructions or materials. These questions may consist of lengthy open-ended responses, short fill-in-the-blank type responses, multiple choice and scale responses, thought or recall listings, or even ratings of subjects' own responses. MediaLab will allow you to ask your questions using a wide variety of question formats.
Finally, you need an experimental design. For this, you need to know which of your materials and questions are assigned to each of your experimental conditions, and the order in which they will presented. You may have just a single condition with a single questionnaire, or a complicated factorial design with many questionnaires and randomization. Either way, MediaLab is very flexible in terms of experimental design.
Once you have your design planned, your questions in mind and your materials prepared, MediaLab plays the simple role of executing your experiment and preparing your data for analysis. To get MediaLab to do what you want, you need to understand two new types of files:
The Experiment File
The experiment file (i.e., files with the .exp extension) deals with things on a very general level. Here, you simply specify which files are going to be presented in each of your conditions. You basically tell MediaLab, "In condition 1, present file1, file2, file3, etc., and in condition 2, present file1, file4, file5, etc." These files can be Word documents, PowerPoint shows, HTML files, movies, sounds, images, executables, and MediaLab Questionnaires. To learn more about experiment files, see the page about Key Concepts of MediaLab Experiment Files.
The MediaLab Questionnaire
This is where MediaLab becomes extremely flexible. The questionnaire (.que) is a file that can be presented just like any other in your experiment. In the experiment file, you identify the files to be presented in your various conditions—including questionnaire files. Questionnaires differ from the other file types in that they are a single self-contained file full of instructions for MediaLab. These instructions can tell MediaLab to present various files (e.g., images, sounds, videos, executables, Word documents, PowerPoint shows, HTML pages), but questionnaires also contain the instructions for MediaLab to ask questions and gather data.
MediaLab questionnaires provide instructions for essay-type open ended responses, multiple choice responses (with single or multiple response options), multiple response questions (e.g., select all that apply), fill-in-the-blank answers, continuous on-line ratings, thought and recall listings, and thought ratings (where subjects rate there own open ended responses on dimensions you define). To learn more about questionnaire files, see the page about Key Concepts of MediaLab Questionnaire Files.
Running the Experiment
1.Once you have defined your experimental conditions, and created the files which it is going to present, simply click the Select and Run Experiment command from the Run menu of MediaLab. After finding and double clicking on your experiment, you will be asked which condition you want to run, and what subject ID you want written to the data file. The rest is easy. MediaLab runs the experiment, gathers the data and writes it to all to a single data file that can be read directly by Excel or SPSS. MediaLab also creates a data input list for your data set in both Excel and SPSS formats so that you can start analyzing your data immediately. MediaLab also provides a data merging utility so that you can easily combine data gathered on different computers before running your analysis. See Data for more details.
What MediaLab is NOT
Before planning a study with MediaLab it critical to know its limitations. Most importantly, we advise serious caution about using any program written for the Windows environment for studies requiring extremely accurate presentation times and response times involving visual images. MediaLab does utilize a timer with a resolution of less than one millisecond and was originally written with the functionality to conduct a wide variety of standard reaction time experiments such as priming studies. For a number of technical reasons, this capability has been removed from the main program and has been developed as a stand alone program called DirectRT that will work cooperatively with MediaLab (although we still use the high-res timer). DirectRT provides high precision timing routines for the presentation of images and measurement of response times in the Windows environment and is recommended for any application where high precision timing is critical.
The second limitation of MediaLab to keep in mind is that it works on the principle of a static experimental design. Dynamic changes in design are not currently possible. Basically this means that a subject needs to be assigned to a condition at the start of the experiment, and what occurs in that condition is fixed. Although fairly complex skip patterns and dynamic stimuli can be set up in MediaLab questionnaires, a subject can not be re-assigned to a different condition based on a response contingency. This is not usually a big deal, but it's something to keep in mind.