|
Stroop Test Programming Notes - Andrew Linn This module was quite a programming challenge. The logic of which color to display was fun to figure out. I could have become bogged down in questions like "Is gray really a neutral color?" (if not, then how would you display a word naming a color? Something has to impart as little meaning as possible). The really big question, as yet not totally unresolved, was the issue of accurate timing. This is best discussed in a separate forum: Computer Based Testing: A Question of Time. Several other sites on the web offer this same test, or some variation of it. To add value, I allow the test to be run in two modes: Demo and Research. In Demo mode, the test runs with default values, 3 instances of each test, progressing from the easiest to the hardest (my opinion). With millions of colors to work with, I decided to limit the number of colors to 5 (red, blue, yellow, black and white) to make sure the colors were vivid and distinct as possible. This was designed to be run over the web, so I needed to eliminate any potential for differences in operating systems or monitor settings. When deciding how to choose the colors to display, I had to work with two items: The color of the block and the color of the ink used on the word. For the block tests, the first thing I would do is find the color for the left block, and remove that color from the possibilities of the right block, then I had the program randomly choose which block was to be the correct one, left or right. Once I figured out which one was the correct answer, I used that block's color to determine the word displayed. The left and right blocks could not have the same color, and depending on the test, the word had to match the correct block, or the color of the ink had to match the color of the correct block. In Director, I used lists to manage this. Start with a list of colors, randomly select one. Remove that choice from the list, and randomly select another. Have a list of words that describe the colors. If the word is supposed to describe the correct choice, select the appropriate word (I used a Case Statement) from the list and display it, if not, select a different word. Choose the color of the ink used to write the word in the same manner. The process is the same, just a little more tedious, in the Word tests. I had a real problem with what to do when a research subject got to the end of a test series. I didn't want them to be able to restart the test (wiping out their data), so I have the program stop at a "Thank You" screen. I was assuming this test would be administered in a controlled environment for research purposes, and I assumed that environment would require a researcher or assistant to act as facilitator. In a HUGE leap of faith, I decided said researcher would read the instructions for this program and remember to press the C key to continue, collect the data, and prepare for the next subject. Only time will tell. I agonized over how to organize the output from this program. Did the researcher just want to know if the subject got the right/wrong answer and how long it took to get it? Could something like this show that people are more likely to select incorrect answers when presented with certain color/word combinations? Finally, I decided to just do a dump of everything: What kind of test it was; what colors were used in which positions; what words/inks were used; was the answer correct, and how long it took to get it. All the data is dumped into a plain text file, each value is separated by a comma for easy importation into a spreadsheet for analysis. The test file is e-mailed in the body of a message via a PERL script Willamette uses on its website. This PERL script is accessible from anywhere on the web. Original Director 8 source code if you have questions, please e-mail me at: alinn@willamette.edu
|
|
||||||||||||||||
