I would further emphasize that the ruler may not be in all of the videos. I would focus on using the box measurements for the conversion rates because we know there will be some sort of arena bounds in each video (they aren't just gonna let the chicks roam free and out of the video capture range).
I agree with Chris that I would think that the conversion would be a little more accurate when using a larger number because they are clicking to determine the pixel to cm ratio.