Cautionary Note
Caution should be used in labeling buttons in this way (e.g., 0-4 for a 5 point scale) unless you are going to make subjects use the mouse.
Remember that the button selections are coded in the data file by which button they select (i.e., 1-12). MediaLab ignores the labels because they could be labeled anything (-5, R, 0, nothing). The rationale for this rule was that no matter what label (if any) on a button, the experimenter would always know which button was selected (i.e., 1-12). This creates a problem though in cases where the label change is also a number from 1-12. Because the labels are then valid keyboard responses, the subject will likely press a button on the keyboard that corresponds to the label.
e.g., if you have reordered the buttons:
How many children do you have?
5 Five
4 Four
3 Three
2 Two
1 One
0 None
Say you have 2 children. You could indicate this by clicking on the fourth option and then a "4" would appear in the data file. However, you could also indicate this by pressing the number "2" on the keyboard. However, MediaLab would see that as selecting button #2 in which case a "2" would appear in the data file. The easiest way around this trap is to not use alternate labels for buttons if those labels are also valid response keys (i.e., 1-12). In the preceding example, this would be much safer:
1 One
2 Two
3 Three
4 Four
5 Five
6 None
Or better yet, add a (k-1) parameter to the scale response. That will disable keyboard input for the item and force subjects to use the mouse (the k-1 function, I believe, has been available since the final release of v2004).