Reply #33: I run experiments.

[Igel]

Sometimes the results don't match the prediction my hypothesis would make. Ha. "Sometimes"? Frequently.

I consider the experiment. I consider my hypothesis. I consider how I took my measurements.

Sometimes just one is wrong; sometimes two are wrong; and, yes, sometimes all three are wrong.

If my hypothesis presumes that people would look at any one of 4 pictures on the computer screen (I do eyetracking experiments) starting from a center point with equal likelihood, anything that violates that assumption makes the experiment invalid. If some pictures are more brightly colored, or larger, or there's too long a time between when I "force" them to look at the center point and when they hear the stimuli so that their gaze has moved ... my data's toast. Once the data said I was wrong; then analysis showed they strongly preferred the lower right hand corner, regardless of the stimuli or picture. Then I remembered the was a window with cars and kids to the right of the computer. Lower the blinds ... voila!--good data.

Or if I measure things differently (one part of the data is when their eye moves to the quadrant, another part is when they actually gaze at the object), my data's toast.

Getting experiments right is damned difficult. Making sure you're measuring what you think you're measuring is damned difficult.