My proofreading job is one, perhaps two days short of ending. For six months I’ve been reading test items which were developed by ETS but which now will be handled by ACT. (The items were developed for one test in particular; however, even though what follows surely doesn’t break the confidentiality agreement I signed, nevertheless I am withholding its name. Suffice it to say that it’s a big test. Many students take it to gain admission into a certain sort of graduate school.) I don’t know many of the reasons for the shift in acronyms, and those I do know aren’t that interesting. I do know that 26,000 test items were moved from one America’s largest post-secondary test-development centers to the other, and, with a dozen other people, I proofread the transfer.

Here at the end of the project I have been counting. Every test item, see, is separated into a packet. I am counting in order to ensure that every item is accounted for. So my days have been spent sitting in a gray cubicle, wearing latex gloves to ward off paper cuts, counting, one at a time, item after item after item. As I’ve counted, I’ve realized something curious: For any single packet, I rarely know for certain that my count is indeed good.

Why not? There are three ways that the number of items in each packet may be known. Two ways are authoritative. First, there is a spreadsheet which was supposed to be updated continuously to keep track of the status of all items (Number In Revision, Number Approved, etc.) and of the final number of items in each packet. The authority of the spreadsheet is bureaucratic: eyes more permanent at ACT than mine saw the spreadsheet every day. Updated continuously it was not, however, and because at various times items were pulled from packets so they might be used in test administrations, the final counts in the spreadsheet are more often than not wrong. Second, with each packet there is also a tally. Tallies were supposed to be updated when items were pulled—and for the most part they were updated. The tally’s authority is both official and scientific: every packet had x number of items to begin; y were removed; z is the number that remains. But though the tallies are generally more accurate than the spreadsheet, often enough they were not updated, or the number pulled was subtracted erroneously, that it at best was a count to be considered warily.

My count, of course, is the third way to know how many items are in a packet, and it is empirical. There is, of course, a real number of items in each packet: That I would arrive at that real number is likely, even in spite of the fact that a staple might cause two pages to stick together which shouldn’t or that my attention might wane somewhere between 45 and 67. That my attention might wane, however, is why I rarely know whether my counts are good.

After all, when I count there are only two things that happen. Either my count agrees with the tally or the spreadsheet, or it does not. Most often there is no agreement whatsoever with the spreadsheet, but there is often agreement with the tally. But when there is agreement with the tally, then I pack everything up and call it good. I’ve decided that, in this case, a single empirical verification with a source is sufficient for truth. Yet I already know the tally is often wrong: Might not my count be likewise wrong? If so, then I pack too soon.

I suppose the point is that in my counting I’ve instilled much trust in “official” authority even though official authority is often problematic. If I didn’t, I’d never finish counting. As it is, it’s already taken three days, long enough that Kathy thinks I’ve written this in order to demonstrate how bored I’ve been and to seek sympathy. And that may in fact be true; but it’s not entirely true. In something even as simple as counting real numbers of things, empirical observation is still influenced by other authority. Often such influences are unconscious, but they are no less real.

Comments