How do NBME scores correlate w number of questions wrong??

[nope80]

When we get a raw score, for example 450, how does that correlate with the number of questions incorrect? Can you also figure out how many more questions you needed to get correct to bump up your score a certain number of points or is it nonlinear and dependent on the form???

---

Members don't see this ad.

 

[pseudoknot]

It varies for each NBME. I wouldn't worry about stuff like that. Just try to learn as much as you can and focus on your weak areas.

 

[studybug]

I am not sure. I have always thought that since there are 200 total questions, you just multiply by 4 to get to 800. It's a bit surprising because 75% correct will be equivalent to 600 which is like 242.
I like to know what others have to say about this.

 

[FutureChiDoc]

. It's a bit surprising because 75% correct will be equivalent to 600 which is like 242.
I like to know what others have to say about this.

Why do people make up stats like this?

It seems like everyone has an opinion of what 75% converts to.

 

[pseudoknot]

I am not sure. I have always thought that since there are 200 total questions, you just multiply by 4 to get to 800. It's a bit surprising because 75% correct will be equivalent to 600 which is like 242.
I like to know what others have to say about this.

No. They are statistically adjusted for difficulty. There is no single conversion that is valid for all exams.

 

[studybug]

That's a starting point. I believe it's close to the actual scoring particularly if you take the harder forms. 75% is 150 and if you multiply that by 4, you get 600. I was told a while back that if you took an easier form, the number will be a bit lower. Remember, 100% will equal exactly 800.
Hope my thinking makes sense to you.

 

[metallo]

That's a starting point. I believe it's close to the actual scoring particularly if you take the harder forms. 75% is 150 and if you multiply that by 4, you get 600. I was told a while back that if you took an easier form, the number will be a bit lower. Remember, 100% will equal exactly 800.
Hope my thinking makes sense to you.

It makes absolutely no sense at all. There is clearly no linear relationship between the raw number of questions answered correctly and the score assigned by the NBME. Given that the scoring is, in fact, adjusted for each exam based on its relative difficulty, there cannot be a simple linear equation that accurately predicts scores. It really amazes me that someone could make it through the first two years of med school and think otherwise.

No. They are statistically adjusted for difficulty. There is no single conversion that is valid for all exams.

Exactly!

 

[pseudoknot]

It really amazes me that someone could make it through the first two years of med school and think otherwise.

Quantitative skills are not valued in medical school. There is no better illustration of this than many of the postings in the Step I forum here.

 

[Sinfekl]

I took the new NBME form six with feedback. I got slightly more than 75% of the questions correct, got a 210. I'm still a ways out so not flipping out or anything, and I'm only aiming for a 220 or 230, but still, 75% is only 210!?!? I thought we had a lot more leeway than that.

Maybe I missed the easy ones that were weighted more? Or is it vice versa?

 

[metallo]

I took the new NBME form six with feedback. I got slightly more than 75% of the questions correct, got a 210. I'm still a ways out so not flipping out or anything, and I'm only aiming for a 220 or 230, but still, 75% is only 210!?!? I thought we had a lot more leeway than that.

Maybe I missed the easy ones that were weighted more? Or is it vice versa?

I think it's probably safe to assume that missing an easy one will hurt you more than missing an average or difficult question, while getting a tough one right will help you more. Therefore, I would assume that if you had two people, one who scores, say, 75% and another who scores 70%, they could actually receive the similar 3 digit scores if the one with 70% answered several difficult questions correctly and the one with 75% missed several easy ones.

I'd love to know exactly how they score the exam, but that seems to be top secret information.

 

[humabot]

I'd love to know exactly how they score the exam, but that seems to be top secret information.

They score it based on how much income your parents make like every other standardized exam.

 

[metallo]

Quantitative skills are not valued in medical school. There is no better illustration of this than many of the postings in the Step I forum here.

You're right, and it's unfortunate, because there are many instances in which it's helpful to have a good understanding of math/statistics. I've noticed that some people on my current rotation still don't fully understand basic concepts like sensitivity and specificity... even after taking Step 1! Then there are those who struggle to calculate I&Os and med dosages. The thing is, these are otherwise really bright people who just seem to have trouble manipulating numbers.

 

[McGillGrad]

I think it's probably safe to assume that missing an easy one will hurt you more than missing an average or difficult question, while getting a tough one right will help you more. Therefore, I would assume that if you had two people, one who scores, say, 75% and another who scores 70%, they could actually receive the similar 3 digit scores if the one with 70% answered several difficult questions correctly and the one with 75% missed several easy ones.

That's exactly right. There's a reason why they can offer a totally different exam to each person but still maintain a statistically valid conversion to a standardized score.

I bet is it something similar to:
Once each question reaches a statictically valid number of responses, they consider your right/wrong answer depending on how everyone else did on the question. Thus, getting something right that most get wrong raises your score more than getting a question right that most get right.

If the stats are too esoteric to digest, then just do a set of question on USMLEworld and look at your final percent and compare it to how others did in the same block. If you manage to get 50% right (which seems a bit low by itself), you will still be above average (i.e. 220+) if everyone else scored a 47% for the same set. So your score of 50% has no meaning until it is scaled against how everyone else scored...high/low/equal.

 

[Noeljan]

I hope it doesnt hurt me too much. I always miss the easy questions (esp since i couldnt sleep not one minute before step 1) and get the difficult questions right. Who knows, their grading will always remain a secret mystery