How is the PCAT graded?

[Forgiven4889]

How is the PCAT graded? Do you get graded for the number you get right or is like a thing like the gre where the first 5 to 10 questions are weighed more than say the next 20 or something.

 

[apteryx]

They multiply the amount incorrect by how many days you have left to live. Or something just as evil.

But hey, when you figure it out, let us know. Nobody knows how they do it. Shrouded in mystery, Harcourt is.

 

[mug3n]

Each question 1-48 per subsection is weighted equally. 8 questions do not count as they're for experimental purposes.

PCATs are scored based on how the previous round of PCAT takers did. So if you wrote in 2008, your scores are compared against the 2007 PCATers and that's how you arrive at your percentiles... meaning x% of people did worse or equal to you in that particular subsection. So it could be a double edged sword if the 2007 people did really well in one subsection which could drop your percentile but on the other hand if they did crap then your %tile will shoot up.

Outside of that, we really don't know how Harcourt does its scoring. They don't release any data about the PCAT scores (mean, median, stdev, and such) nor do they release answer keys/questions so we'd never know the exact scores (as in x # right out of 48). And more curiously, we have no idea how Harcourt compiles the "composite" score. We've seen people that did crap in a lot of sections (<60) and one good section (>90) and still got an 80 something composite. 2 good sections over 90 and the rest hovering at about high 70-low 80 is enough to get a massive (>94) composite.

 

[Transformer]

From my experience of searching through old SDN forums, some say our scores are compared to first time test takers from 1998 to 2003. Some say our scores are compared to scores from the previous test cycle. Honestly, no one really knows which group we are compared against or how our composite and raw scores are compiled.

What we do know is one of the two essays is graded and the other is experimental. As stated by mug3n, 8 out of the 48 questions are experimental and only 40 questions are counted in each of the 5 subsections.

I'm not sure if anyone who has taken the PCAT multiple times have noticed that percentile rank composite scores over 50% gets inflated and composite scores under 50% gets deflated.


Lets take two sets of scores:

Example #1)

Verbal
439 94
Biology
450 97
Reading
452 98
Quant
426 87
Chem
415 75

Composite
436 96

The average percentile rank score for 5 sections = (94+97+98+87+75)/5 = 90.2
yet the percentile rank composite = 96

Note that the average of the 5 scaled scores equal the scaled composite score but this is not the case for composite scores. (e.g. (439 + 450 + 452 + 426 + 415)/5 = 436.4)



Example #2)

Verbal
403 56
Biology
387 32
Reading
404 56
Quant
409 68
Chem
384 27

Composite
397 45

The average percentile rank score for 5 sections = (56+32+56+68+27)/5 = 47.8
yet the percentile rank composite = 45

Note that the average of the 5 scaled scores equal the scaled composite score but this is not the case for composite scores. (e.g. (403 + 387 + 404 + 409 + 384)/5 = 397.4)

A common analogy for the above composite score relationship is "The rich gets richer and the poor gets poorer"



If you check people's PCAT scores from the following forum,

http://forums.studentdoctor.net/showthread.php?t=427578

you will notice the above relationship holds true for all PCAT composite scores.


P.s. If composite score is exactly at 50% or slightly above 50%, then the average percentile rank score for 5 sections equal the percentile rank composite.

 

[xhamburgersamx]

Wouldnt they just rank everyone that took the same test in august and then assign a percentile, seems pretty simple

 

[fenixtnlfan]

It's in the SDN FAQs. Here's the direct link.

http://forums.studentdoctor.net/showpost.php?p=2724746&postcount=6