Each semester I teach a large section of BIOL 1101, our college's intro course. I separate the class into 3-5 sections, based on the number of TAs I have. We do a standard placement test (designed by me and my colleagues) and distribute students evenly through the lab and discussion sections.
This past semester I had 4 TAs, all good kids who I've had in class before: Ann, Bob, Carl, and Dee. I meet with them extensively before the semester, but they have "lord's rule" over the grading and progress of their sections during the semester.
After a mandatory seminar we took on grade inflation last year, I got it in my head to plot the students' grades along with the standard TA evaluations we do each semester. Both are (happily) on a 4 point scale.
I've just now finished this chart in Excel, but couldn't figure out the export program. So I've displayed the data as near as I can in MS Paint. It looks awful, but the trend is clear.
Those TAs who had the highest student evaluations also gave the highest grades. It is definitely not statistically significant - my brethern would string me up - but I think it reveals a real truth about something.
The blue line shows the average grade earned, the red line shows the evaluation average for the TA, and the black line is the average final exam score for students in that section...the black line is a number I give. That it's so straight across suggests to me that our placement test did a good job sorting students fairly evenly, and that my TAs rise and fall in their evaluations based on the grades they give.
Tell me I'm wrong.