Hi Miss Lizzy,
I agree overall with your conclusions: pictures are higher quality because the hardware and software are cheaper and much more powerful, and people will almost certainly choose higher resolutions as better than lower resolutions. Nevertheless, the subjectivity does not detract from the science. Having a sample of people grade a stimulus and then take the mean and standard deviation to do a Student's t-test of the results is a worthwhile activity. I'm not so sure it is for 3rd graders, though.
A more interesting question would be to add "lagging" as an independent variable. Lagging is when a player's game machine slows down or pauses during the action. If other players do not experience lagging at the same time, the lagging player suddenly is at an enormous disadvantage because the action continues with his player frozen. If the higher quality image has more lagging than the lower quality image, then this will be an important consideration in the subjects' choices.
If your students could set up a system with 3 pixel qualities, low medium and high, as well as three lagging conditions, none low and high, then they would be able to present 9 combinations of lagging and image quality. If each subject is only presented with 3 or 4 randomly selected examples from the 9 possibilities, and enough subjects were tested so that overall all 9 combinations were tested the same number of times, an interesting statistical study would come out. I'm not sure how they would do this. I can't think of anything that would be easy for 3rd graders to set up.
Something easier would be to have them push a button to render a 3D scene, e.g. with povray (povray.org). They could be presented with a low-quality, a medium-quality, and a high-quality scene, and decide which they found most interesting (?) to create. It is easy to tell povray to do things that will take much longer to render the image but will add only a modest amount of quality. I would expect that the slowest-rendering images would bore them somewhat and get lower scores, especially if the added quality wasn't great. Again, I don't know if this is for 3rd graders.
Anyway, I think you get the idea, and I hope this will help you figure out something that can work with your students without being too trivial or predictable. If they have a grasp of average and standard deviation (?), that in itself would make it an interesting scientific activity.
Science Buddies Expert Forum