I’m not a psychometrician. Or a test security expert. Or a statistician. Or an economist.

I’m just a guy with a Wikispace and a question.
And I’m going to ask anyone who understands these things better to set me straight.

Thesis: There are features of the Minnesota Teacher Licensing exam (the MTLE) that might make it more likely for candidates to cheat…
1. The Incentive
To get a first-year license, teachers are now required to pass all five of the MTLE tests. At least that’s my reading of the law.
Also, since teacher salaries are generally calculated on a “step and lane” system, the sooner you start, the sooner your salary starts to go up.
2. The Opportunity
I may be wrong about this, but I think the MTLE tests are administered in four-week windows. (They used to be six-week windows.) (The test windows date are linked to the basic skills info page – under “score reporting”.)

And it’s possible that everyone who takes the test during a single window gets the same test. So, for example, right now we’re in the March 18 to April 14 window, which would mean that if I took the test tomorrow, I’d get the exact same test (i.e., the same questions) as my buddy got back on March 18.
3. How This Cheating Would Work
Pretty straightforward here … I’d ask all my friends to tell me what questions they remember from the test.

Or maybe I’d ask my smartest non-teacher friend – the guy with the freak memory – to go sign up and take the test with the goal of remembering as much as he could. Okay, it’s not perfect, but think of what you could get:
  • the writing topics,
  • the general topics of the reading passages,
  • specific memories of harder questions

Clearly, having “insider information” about the contents of a test creates an advantage …
4. The Disincentive (What protects against cheating?)
Honor? Your statement when you take the test that you’re not going to cheat?
The risk of getting caught?

The problem is … there’s not a huge amount of risk of this. (In fact, the n on many of these tests is so small, a statistical analysis of improbability of certain patterns of correct and incorrect answers would be harder to do.)
5. What Easily-Obtained Data would speak to this?
If students are getting inside information about the contents of the test from previous test takers, then we’d expect to see the pass rate INCREASE over the four weeks of the testing window.
Maybe not steadily increase … but there should be a bump up in the scores after the first week … or maybe it wouldn’t show up until the third or fourth week.

If there’s hardly any cheating, then we’d expect essentially the same pass rates every week.

So … to lay this question to rest, all MDE would have to do is to publish the pass rates by week of testing. Simple.
6. What’s wrong with this analysis?
Is there something wrong with this analysis? Maybe some of my facts are wrong.
The MTLE is not the most … transparent … test in the world. Because it’s such a small operation, reporting out takes place on a smaller scale. And it’s hard to figure out who to ask for information like this. But I figure somebody knows….

Final thought … If there is cheating going on, then we have a teacher licensing test that – at least partially -- selects against people with integrity in favor of people with less integrity …