I agree with you about the complete insignificance of a question or two. My point was just that I think because they chose to group 33-35 for purposes of presentation on their website doesn’t mean this is a category of equal scores they are using in making admissions decisions. They may not have wanted to list every possible score and make the web page so long.
I think everyone needs to keep in mind that the kids with the top scores are usually going to be very strong in other areas so it’s hard to say how much influence the test scores have. It would be interesting to see a breakdown of say 800 790 780 770 because I’d like to think that admissions wouldn’t attach any significance to a single question on a timed exam—or even to students who got the exact same number of questions correct but received a score 10 points lower or higher because the exam was curved slightly differently on that day–this often happens. From this data, though, it certainly appears that they do care somewhat about this.
As the scores get lower with more and more incorrect answers, then yes, I do think a real difference is being measured.