Raw Score

[rleon82]

Now how difficult or expensive would it be for Elses to give us a Raw Score as we leave the test? Stick the test in a ScanTron machine and give us a raw score.

A coworker took his GC and was given a Raw Score when he turned his test in.

Any thoughts on this?

 

[RIP - VTEnviro]

Since the passing raw score varies by exam, this wouldn't answer your burning question, unless you aboslutely bombed it or aced it.

 

[rleon82]

That is all I want to know.

 

[Brimstone]

I don't see any reason why something like this couldn't be implemented. Sure, the results wouldn't be official until NCEES completed the review, but if you got 56 problems correct (70%), then chances are you passed.

Although, knowing the raw score wouldn't allow for NCEES to control the "unknown" passing score.

What if you were the guy whose raw score was 58/80 (72%), and then you received the letter saying you failed in the event of a test deemed easy by the council? Whether that scenario could happen is speculation, but I'd probably be pissed if it were me.

 

[Road Guy]

NCEES run by engineers = Not Likely to change easily B)

But I would be all for it. from what I hear the 10 weeks notice is an improvement compared to the older (hand graded) test

 

[Mike in Gastonia]

I don't see any reason why something like this couldn't be implemented. Sure, the results wouldn't be official until NCEES completed the review, but if you got 56 problems correct (70%), then chances are you passed.

I keep seeing people talk about 56 out of 80 and 70%. Where does this come from? And what does it mean?

If you are saying you have to get 70% right to pass, I don't think that's true.

 

[JoeysVee]

I don't see any reason why something like this couldn't be implemented. Sure, the results wouldn't be official until NCEES completed the review, but if you got 56 problems correct (70%), then chances are you passed.

I keep seeing people talk about 56 out of 80 and 70%. Where does this come from? And what does it mean?

If you are saying you have to get 70% right to pass, I don't think that's true.

So I am a Gigantic DoucheBags the accurate number?

I think they are just saying if you get 70% correct you passed....no doubt. Although you could pass if you got less than that depending on the cut score. There point is the cut score has never and will never be above 70%.

 

[Mike in Gastonia]

So I am a Gigantic DoucheBags the accurate number?
I think they are just saying if you get 70% correct you passed....no doubt. Although you could pass if you got less than that depending on the cut score. There point is the cut score has never and will never be above 70%.

I called NCEES back before they switched to pass/fail was told that they standardize the score and make a "70" passing. That wasn't 70%, but whatever the raw score that was required to pass became a 70. So for example if it was decided that a 46 raw score out of 80 was passing, that became a 70. 80 out of 80 was a 100 and they then had an equation with 80 raw equal 100 and 46 equal 70 to tell people their standardized scores. So if someone failed and was told they had a "68" that really meant they didn't miss by 2 questions but by something more than that.

It's confusing and I think that's why they just went to pass/fail.

Bottom line though, you are probably right. If you get 56 out of 80 you'll probably pass.........

 

[RIP - VTEnviro]

I don't see any reason why something like this couldn't be implemented. Sure, the results wouldn't be official until NCEES completed the review, but if you got 56 problems correct (70%), then chances are you passed.

You would need a answer key at each site for each version of each exam being offered. That's how scantron machines work. You feed the key through, then feed the exam sheets.

So you'd either need a separate machine for every version of every exam, or have to keep refeeding through keys and different sheets for everyone on 1 or 2 machines.

Seeing how NCEES looks at question security, that doesn't seem likely. I'd bet the correct answers aren't anywhere to be found in the building on test day.

 

[DVINNY]

Seeing how NCEES looks at question security, that doesn't seem likely. I'd bet the correct answers aren't anywhere to be found in the building on test day.

I agree.

 

[RIP - VTEnviro]

There isn't any reaosn to have the answers at the site. The proctors aren't grading them. And it's a security risk.

 

[EdinNO]

A friend of mine did the MCSE and the tests are taken on the computer.

When you hit the final button, it tells you right there whether you passed or not- immediate resolution!

I've known people who take contractor's license exams who get results within a couple of weeks, tops.

We just have to suffer, it seems. Engineers can't engineer a better system for themselves?

Ed

 

[EdinNO]

I also feel pretty confident that some programmer (like one I used to work with) could certainly write some code that a scantron or other computer could use to immediately grade exams.

Of course, NCEES would have to, before-hand confer on which questions to "throw out" for which disciplines and which morning and evening sections. If they came to a conclusion ahead of the test (which I guess could somehow be construed as a security risk if things leaked), the grading machines could be pre-programmed and ready to give immediate results.

Maybe this is the process that takes place, but all the "normalizing" takes place based on exam results and not just NCEES' take on "easy" or "hard" exams. I mean, how can they determine a question or even an exam was particularly difficult without the stats on how the examanees did?

Ed

 

[benbo]

The cut score is a huge mystery that is discussed on the infamous other board every time the test is taken. They cannot just curve the exam because they might have a group of particularly poorly prepared examinees. So they try to figure out how good the test takers are compared to all historical test takers. It is done by a process called equating. I think the method is called the "Modified Angoff" (you can look it up). Basically, there is a large pool of questions that are repeated on several exams. By doing this, the examiners can get an idea of what percentage of people answer these "sample" problems correctly. That way, they can tell if a particular exam is more or less difficult than the norm, or if the group of test takers are better or worse prepared. For example, on the test you take there may be 10 of these repeated questions, and 70 basically original questions. Assume that statistically, the typical examinee historically has answered 5 of these 10 questions correctly. So if on this test the examinees only answer 3 correctly on average, they know the problem is with the test takers, not the test. I hope this has sufficiently confused everyone.

 

[Road Guy]

that actually makes some sense

 

[EdinNO]

... and if they get 8 out of 10, then again the answer must be that the test takers are better prepared.

No metter what, its the test takers fault. The test has to be right since the test takers didn't score the average on the guage questions. The test is always right! Heck with it all, throw away the normalization and just give us our scores.

Just kidding. I guess you are saying that if the group gets the average score on the guage questions, but does poorly overall (ie, on the rest of the questions), then the test is considered difficult and cut score adjusted downward. So there are maany itterations:

poor on guage q's vs poor, avg, good overall

average on guage q's vs poor, avg, good overall

good on guage q's vs poor, avg, good overall

Yikes, seems like a lot of work.

But I still think it can be done pretty quickly, almost instantaneously, with the right programmer. Unless there is much more to it than that.

Ed

 

[Road Guy]

:suicide:

 

[RIP - VTEnviro]

I'm with benbo. That makes sense. :claps:

I really hate all the conspiracy theorists and their wild cut score ideas. One of the things that bothered me about "the other board"