The CLEP normal distribution: CLEPped my first course ever today!

I am working on a degree at the Raritan Valley Community College. Part of the planning required me to try and CLEP one of the computer courses that are a requirement, CISY 103 or 105. Raritan Valley allows the CLEP, which stands for College-Level Examination Program, by the way. (Did you know, these are by the same people who bring you the TOEIC and TOEFL in Japan?)

Well, I have been putting this off, mostly out of nerves. Today, I decided it was the do-or-die day. I made an appointment in Wyomissing, and drove up to the Berks Technical Institute.

Very much like I had prepped for the CPA, I went in, first time, with just what I knew, and my reading. (As I used to say about sitting for the CPA, I just showed up with pencils.) You know, if you don’t pass a CLEP exam–and pass usually means “50” as a scaled score–you have to wait 180 days. That’s a lot like the old CPA exam, too!

As it turned out, I knew a lot more about “Information Systems and Computer Applications” than I realized, and I scored a “72”. Hooray!

I was happy to be over the 50 mark, but a bit bummed that it was 72, which I thought was 72 out of 100. But no! The scale is 20 to 80. So it was 72 out of a scale just going to 80.

I poked around the internet to get a sense of what a 72 is. There is a lot of CLEP talk out there, but no one seems to have a handle on the scoring, other than (like me) they want a score that gets them over the hurdle.

When I took the College Board’s practice examination guide, I got, coincidentally, 65 of the 90 practice questions right, which worked out to 72.2%. But that’s not it, is it?

Then, I read that the score you get is a scaled score, as I said, going from 20 to 80. So I decided to “assume a normal distribution”. This can be dangerous if you are investing money, because not everything is normally distributed! But for [scaled] test scoring, it is probably useful. Up above you see a bell curve to the right. I put 50 as the mean–the average. I gave it a standard deviation of 10, so “one sigma” is 60, “two sigma” is 70 and “three sigma” is 80. I assume the people who are above three sigma also just get an “80”.

Likewise, on the fail side, “negative one sigma” is 40, and so forth. People who don’t manage to answer enough questions right to make even a scaled 20, get given 20 anyway–just a guess. (Like, say, you go in and answer nothing, and then hit exit. You get a 20.)

On this normal curve, or normal distribution, a 72 is about 98.8%–which is great if that’s what it really is. But it doesn’t answer the second question a lot of people would have (after, “did I pass?!”), and that is:

How many did I get right, and how many wrong?

Unfortunately, College Board does not tell you that: the aptly named “raw score”. If you have to answer 70 questions right to get a “50”, and each 7 questions more take you up 10 scaled points, then a “70” might mean answering 84 (70 + 7 + 7) questions right. If you only have to answer 35 questions to get the “50”, then a higher score might mean you were still screwing up on half the test.

My feeling was that I was getting seven or eight of every ten right. But I really don’t know. It was 100 questions in 90 minutes, and I think I was done with 47 minutes to spare. The big question mark is: if I were getting enough right to be 8 in 10, if I dropped to just 7 in 10, would I have been back at 50?

College Board obviously is not going to say what the raw scores for the tests are. They also aren’t going to tell you what you got wrong. They also don’t want you to say what the questions were. Honestly, it all seemed to happen so fast that I can only vaguely remember a handful.

The nice thing about CLEP is that Raritan Valley offers 5 credit hours for this, and it only cost $77 plus the testing center fee of $20. Compare that to the vast fortunes that some schools charge to take a class.