Are Voided Scores part of the Curve?

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

Compass

Squishy
Moderator Emeritus
10+ Year Member
15+ Year Member
Joined
Apr 15, 2006
Messages
3,366
Reaction score
3
Out of curiosity, if you void your scores, do they still get counted as part of the curve for scoring purposes, even if you don't get graded? I mean, after all, it is DIGITAL data and it's not like it costs them money to grade it for the curve, as opposed to the writing section which they probably actually don't grade.

Curiosity killed the cat :(

Members don't see this ad.
 
I've been told by many that the curves and therefore the scores are predetermined. Thus it's up to what MCAT thinks is hard not how people do. This eliminates variables like if a group of people take the test that are very well prepared vs a group that is completely unprepared - it would be unfair to penalize the smarter group for simply being at the wrong place at the wrong time.
 
How would you predetermine a curve for a test? Do they send all the captured pre-meds who never made it into med school and had nothing left to live for into a little room, force them to take their 41st MCAT, and then determine a curve from those?

(intense sarcasm implied, please don't hurt me)

And how would the well-prepared group be penalized? The curve is not diluted by weaker students, it is expanded.
 
from my knowledge, the curve is produced by analyzing the scores from the PREVIOUS year.

But this does sound flawed. Correct me if I'm wrong.
 
Members don't see this ad :)
But each test is different... how does one determine whether one test is harder than another? Let's say there's a killer exam on June 18th morning, and a breezy one on June 18th afternoon. Applying the same curve on those could be traumatic to the morning and breezy for the afternoon! o.o
 
But each test is different... how does one determine whether one test is harder than another? Let's say there's a killer exam on June 18th morning, and a breezy one on June 18th afternoon. Applying the same curve on those could be traumatic to the morning and breezy for the afternoon! o.o

that's exactly what I was thinking. But that fact that I quoted was straight out of AAMC website if I recall correctly.. somebody else showed it to me and I was like "wtf?"

I guess in a way, it makes it fair so that people who write during the school year does not get a significant advantage. But then again, it may also be really inaccurate since you'll be basing your score against a different test.
 
thats y the scales are predetermined, the percentiles however are not. If the scaled scores were purely based on how people did during that testing session, the scale would be identical to the scaled scores. For example have any of you noticed that while in verbal a 13 is liekly to be a 99.8 % tile that same 13 in physical is only likely to be around 97 % tile. The reason once again is because the scales are predetermined.


also the people that make the mcat determine which test they think is hard, or rather which test has more hard questions and make the scale accordingly.
 
thats y the scales are predetermined, the percentiles however are not. If the scaled scores were purely based on how people did during that testing session, the scale would be identical to the scaled scores. For example have any of you noticed that while in verbal a 13 is liekly to be a 99.8 % tile that same 13 in physical is only likely to be around 97 % tile. The reason once again is because the scales are predetermined.


also the people that make the mcat determine which test they think is hard, or rather which test has more hard questions and make the scale accordingly.

The issue I have with this explanation, is that the scaled scores are finite because the questions you get right and wrong are finite. Each question has the same weight. As such, the following issue arises, mostly in verbal:

0 wrong = 15
1 wrong = 14
2 wrong = 13

The percentile scores are based on the scaled scores, of course, but is it always 0,1,2 wrong, you get x,y,z, or does the scaled score fluctuate as well to account for say, no one got a perfect?
 
so basically the way i think it works is, lets say your the only person that cycle to get 2 wrong in verbal (everyone else got more then 2 wrong), you would still get a 13, however your percentile would now be 99.9, on the other extreme, if half of the people got only 2 wrong, you would still get a 13 however the percentil would now be 50 %
 
so basically the way i think it works is, lets say your the only person that cycle to get 2 wrong in verbal (everyone else got more then 2 wrong), you would still get a 13, however your percentile would now be 99.9, on the other extreme, if half of the people got only 2 wrong, you would still get a 13 however the percentil would now be 50 %


That's stupid... I'm not saying you are, but rather the idea it's just scaling the percentile.

I think they actually run test questions during different administrations and then mix and match questions of known difficulty. The difficulty overall determines how many questions you are able to get wrong for a certain scaled score. This makes more sense a percentile adjustment and it's what the official guide says.
 
Oh, and to Compass' original question, I've wondered the same thing.

I am going to guess that they do use voided data, because it is just as relevant to them as unvoided data. I am not sure though. It's not the voided data for that administration, but rather the voided data for the questions that appeared on your exam from previous administrations.


I still wonder why it takes so long to score the exams though if it truly is a predetermined scale as AAMC states. I also wonder why when some people get scores it shows 13-14 on a section until it is resolved to one of the two (This happened for the 3/28 exam I believe).

The whole system blows me away. I feel like they think they're being straightforward, but then I'm not sure why that can't state why it takes so long to grade them and a score doesn't pop up as soon as you submit for the 3 numerical sections... Some written exams like the LSAT that aren't on the computer are back in less time than it takes for the MCAT to get scored.

Oh, and people can say it's because they check for cheating, but I'm sure they could come up with a simple algorithm that analyzes similarities in answers at given testing centers, then if they find something, they can reference the centers tapes. It shouldn't take 30 days though.
 
ye i prob wasn't that clear, i think i agree with everything you said

my main pt was that the scales are pre determined
 
Oh, and to Compass' original question, I've wondered the same thing.

I am going to guess that they do use voided data, because it is just as relevant to them as unvoided data. I am not sure though. It's not the voided data for that administration, but rather the voided data for the questions that appeared on your exam from previous administrations.


I still wonder why it takes so long to score the exams though if it truly is a predetermined scale as AAMC states. I also wonder why when some people get scores it shows 13-14 on a section until it is resolved to one of the two (This happened for the 3/28 exam I believe).

The whole system blows me away. I feel like they think they're being straightforward, but then I'm not sure why that can't state why it takes so long to grade them and a score doesn't pop up as soon as you submit for the 3 numerical sections... Some written exams like the LSAT that aren't on the computer are back in less time than it takes for the MCAT to get scored.

Oh, and people can say it's because they check for cheating, but I'm sure they could come up with a simple algorithm that analyzes similarities in answers at given testing centers, then if they find something, they can reference the centers tapes. It shouldn't take 30 days though.

Writing Sample Insurance
 
As stated on the AAMC's website, paraphrasing:

There is no "curve", the raw score to scaled score conversion is preset, as the questions on the test have all been used before (with the exception of experimental questions) so they know the degree of difficulty, and how test takers typically do on those questions/passages. This in turn allows them to predetermine the scale. This is so no test is more 'difficult' or 'easier' than another, or a more 'beneficial' test taking date, and that way a 35 in 2009 is the same as a 35 in 2007. The percentiles are based on the slew of test takers for previous dates and years, so the percentiles shouldn't change. A 13 should never be a 99%tile one year and a 50%tile another year, it will always be the same percentile because the test in made in such a way that you're scaled score can be compared to any other test taker, and your scaled score reflects the percentile for the massive sample of test takers they have. A 13 is a 13 is a 13, it will never be the equivalent of a 7 one year because then the test wouldn't be standardized or valid.
 
Top