my experiment

statico on 2005-04-01T00:57:16

I'm taking a class on cognition. I have to create, perform and document a formal Psych experiment.

Last night I finished my little testing program (CGI::App, SQLite, Class::DBI, Template Toolkit) and today I tested 14 participants. Without going into too much detail, I essentially tested subjects to see if, when presented with a series of questions, they would produce the expected output after being given a certain stimulus. (Hehe, no, this was a completely visual and audible exercise.)

My test groups appeared as follows:

+-----------------------------------------+
|           | Order 1 | Order 2 | Control |
|-----------+---------+---------+---------|
| Variant 1 | A       | B       | C       |
|-----------+---------+---------+---------|
| Variant 2 | D       | E       | F       |
+-----------------------------------------+


Ideally, groups A, B, D and E would correlate and the control groups would be all over the place. Turns out that wasn't the case: (initials have been changed)

List        Subject     Expected    Total     
----------  ----------  ----------  ----------
A           JR          2           9         
A           QW          6           9         
A           TY          4           9         
B           ER          2           9         
B           AS          6           9         
C           VB          6           9         
C           JK          5           9         
D           GH          5           9         
D           NM          5           9         
E           IO          5           9         
E           UI          5           9         
F           OP          5           7         
F           XC          4           9         
F           AZ          7           9         


...which boils down to:

List        PercentExpected
----------  ---------------
A           44.4           
B           44.4           
C           61.1           
D           55.6           
E           55.6           
F           64.0           


Looking back, I've got a pile of ways I could improve the experiment -- there are a few things that could have been done to make the results more accurate. However, I have to do my best with only the time I can allocate for this class.

Ugh. At least I entertained my subjects with colorful stock photography from sites listed on Blue Vertigo.