In a post last month, detailing the stats behind this year’s Rattle Poetry Prize, I made the following statement:

[W]hen you read the winners in December, if the honorable mentions seem no stronger than the rest of the poems in the magazine, and there are a few in the open section that seem even better than the $5,000 prize winner, don’t worry — that’s exactly what should happen, statistically (if we assume the quality of a contest entry is the same as that of a regular submission, which is probably the case). [...] The contest winners won’t necessarily be the best poems in an issue — they’re simply typical of what we always publish.

My reasoning was that publishing 11 out of 6,000 poems gives us the same rate of publication that we see with standard submissions (about 0.2%).  The only unknown variable was the relative quality of the poem-set.  Are contest entries better poems, on average, than regular submissions?

It might seem easy to answer this question at first — the contest has an entry fee, and there’s a lot of money at stake, so of course entrants are going to send their best poems to the contest, right?  But while that certainly would drive quality higher, there are other mitigating factors.  Regular submissions may be sent to other journals simultaneously, whereas in the contest that’s forbidden.  Poems sent to our contest can only be sent to our contest and none other — we tie them up for at least six weeks, and since other magazines’ reply times are slow and irregular, this often means poems people would like to submit have to be held, because they’re still under consideration elsewhere.  What’s more, a big prize draws the weaker writers out of the woodwork — if you’re willing to go through all the hassle and postage for the meager payment of a couple review copies, you probably take this poetry thing pretty seriously.  If it was the $5,000 that caught your eye, that might not always be the case.  It certainly seems like the absolute worst poems we every read are some of the contest entries. All poems are scored 1-10, and there are a disproportionately high number of 1’s.

So that’s why in last month’s post I felt confident saying that the quality of contest entries was probably similar to that of regular submissions.  Contests cut both ways.

Now that we’re (finally) done with all aspects of the 2009 contest, I have to correct the record.

We always have to pick 11 poems to publish, the big winner, and then ten honorable mentions.  No matter what happens, we’re going to publish at least 11 poems from the contest.  So the best way to judge the quality of work received is by looking at the number of additional poems we offered publication — each of these appear in the following summer’s issue.  The idea is that we follow through and offer publication to all the poems that we normally would, regardless of how they were submitted.  After publishing zero additional poems in 2008, we found 19 we wanted to publish this year.  Here are the historical numbers:

Year / additional poems / total publication %
2006 / 7 / 0.42%
2007 / 12 / 0.51%
2008 / 0 / 0.25%
2009 / 19 / 0.49%

So in terms of total publication percentage, the anomalous year was actually 2008 — I have no idea why the entries were so weak last year, but honestly, they were.  We love the winning poem, and there was a solid plateau at the top, but the drop-off was steep — I remember being worried that we might not be able to find a full 11.

With three stable plot points, even with one inexplicable dip, the pattern is clear:  contest entries are indeed a higher quality at the top than regular submissions.  Even in the worst year, we published at a slightly higher rate than our typical 0.2%. Our tastes don’t change; the poems must be better.

So when I said last month that being published in the open section of Rattle was as difficult as winning money in the contest, I was wrong.  The competition in the contest is more fierce than I assumed.  If you crunch the numbers, the 11 winning poems aren’t just typical of what we publish — they’re firmly in the upper third.