Actual UWorld Correlation with Step 1 Score

[elprezidente]

I've come across a ton of outdated and seemingly opinion-based articles about the correlation between UWorld averages and Step 1 scores, which I feel are useless without sources backing where these stats came from. I'm interested in compiling some data on this; I think it could be helpful for everyone to track their progress and get an idea of where they stand. I'll publish the results on here once I have a solid sample size. If you've taken Step 1 already and received your score, and went through a significant portion of UWorld for prep, please answer the survey below (completely anonymous) with your UWorld average and Step 1 score. Thanks!

https://www.surveymonkey.com/r/W8QPJW7

---

Members don't see this ad.

 

[alpinism]

Have you looked at the step 1 threads from previous years?

http://forums.studentdoctor.net/forums/step-i.27/

Just do a search. Every year literally hundreds of people post their UWorld and Step 1 scores on SDN.

 

[elprezidente]

The purpose of this is to gather data and consolidate it in one place that attempts to find a correlation between UWorld and Step 1 scores. I can look at individual scores and experiences, and that's great for getting advice, providing hope, etc. But I think it would be helpful to look at trends. xx% on UWorld correlated to a score of xxx, with the range of scores from (xxx) to (xxx).

 

[sliceofbread136]

You're gonna get some awfully biased data with this..

 

[PL198]

who cares, information isn't comparable anyway. people use uworld different. some people do questions over. others don't.

some people drive right in and learn from questions and mistakes , other people study a book first and then treat uworld like a test

 

[elprezidente]

I'm sure that there will be biased results, but so far I've gotten 8 or 9 responses and there's a pretty nice correlation between them. I'm hoping that some lower scorers will be willing to post their scores to really even out the distribution. Unfortunately, there will be high-scorer selection bias, but I've gotten a few more "average" scorers to respond as well, and those in the middle seem to fit the line between the highest and the lowest. I'm not looking to publish these results in JAMA, but I do hope to be able to give people an idea regarding what their scores might vaguely translate to.

 

[Psai]

http://usmle-score-correlation.blogspot.com/

My 84% uworld corresponded to a 266 of course

 

[elprezidente]

http://usmle-score-correlation.blogspot.com/

My 84% uworld corresponded to a 266 of course

LOL, this is the exact data that I'm referring to.

 

[Psai]

How do you know it's outdated and opinion based?

It was spot on for me

 

[masaraksh]

... I'm not looking to publish these results in JAMA....

you don't say

 

[Slack3r]

How do you know it's outdated and opinion based?

It was spot on for me

I remember it being pretty close for me, too.

 

[dfib slim]

http://usmle-score-correlation.blogspot.com/

Mine correlated exactly.

 

[elprezidente]

Well then that's good to know. Thanks for the feedback, everyone!

 

[psychMDhopefully]

Does 60% right on uworld get you close to 230 on step 1?

 

[W19]

39% on Kaplan Qbank corresponds to 197... This definitely put my 59% on Kaplan into perspective.

 

[Rekt]

I'm an MS1. Can someone explain what % of UWorld means? I know it's comprised of blocks, but is it you treating it like an exam and that's your score?

 

[pre med 2014]

I'm an MS1. Can someone explain what % of UWorld means? I know it's comprised of blocks, but is it you treating it like an exam and that's your score?

As you go along the question bank your total percent correct is continuously updated. Since most people finish the entire question bank before the exam at least once it can sort of predict performance




Sent from my iPhone using SDN mobile app

 

[pre med 2014]

Does 60% right on uworld get you close to 230 on step 1?

Probably not. I'm at 52% at about 700q's into 2nd pass and still have yet to pass an NBME (last NBME was 164 taken immediately after 40% 1st pass). If 60% was a 230 then I wouldn't feel like I was in such a precarious position at the moment


I'll also add that I'm off the questions for now and reviewing core material so hopefully I'll be around 60% when I get back to it after finishing a megablock of new material/stuff I'm weak on

Sent from my iPhone using SDN mobile app

 

[mymembernames]

I though we were supposed to use it as a learning tool and not focus on the percent and use NBMEs for that?

 

[Psai]

I though we were supposed to use it as a learning tool and not focus on the percent and use NBMEs for that?

Yes but you can't help wondering what the percentages mean and how you will do. I spent the first two weeks obsessing about my percentage which in hindsight was pretty silly

 

[chillaxbro]

Wow %60 = 230? I'm a week into my dedicated study period and havent even looked at half of first aid and I'm getting %60 on all my uworld blocks. I should just move the exam to ASAP and take the next 5 weeks off

 

[Stagg737]

Wow %60 = 230? I'm a week into my dedicated study period and havent even looked at half of first aid and I'm getting %60 on all my uworld blocks. I should just move the exam to ASAP and take the next 5 weeks off

Lol, I'm the same way with Kaplan. My 62% halfway through without any solid review is making me pretty happy right now. I think I need to bomb some questions to bring my confidence back down.

 

[gregoryhouse]

Mine correlate way closer with the NBMEs but my score wasn't too far off the ohter link that was posted

 

[omn]

You're gonna get some awfully biased data with this..

what does this mean? Is this even English.. do you mean skew, or long left or right tails...? so.. for myself.. massively uncorrelated.. 1.75 x through UWorld, so last two weeks, random blocks, fast and always above 70% ... Step 1 was 202 ... no one I know, knows why this happened

 

[ForHumors]

Hate to be that guy, but you can't say "for me, it was uncorrelated". Correlation works on a group level, not an individual level. Instead, say that "UWorld did not predict my score well"

I'm being a dbag I know, but I'll justify it by saying that using stats terms right makes us look smarter to other professions.

 

[ForHumors]

The purpose of this is to gather data and consolidate it in one place that attempts to find a correlation between UWorld and Step 1 scores. I can look at individual scores and experiences, and that's great for getting advice, providing hope, etc. But I think it would be helpful to look at trends. xx% on UWorld correlated to a score of xxx, with the range of scores from (xxx) to (xxx).

I think what @alpinism meant was that you can use old forums to mine data for your correlation.

 

[ZPakEffect]

what does this mean? Is this even English.. do you mean skew, or long left or right tails...? so.. for myself.. massively uncorrelated.. 1.75 x through UWorld, so last two weeks, random blocks, fast and always above 70% ... Step 1 was 202 ... no one I know, knows why this happened

Wait,you were going through UWorld a SECOND time and was only getting >70%? I'm pretty sure that correlation chart linked to above was meant for the first attempt.

 

[PharMed2016]

My UW score correlated with my Step I and for that matter... very closely with Step II.

 

[pre med 2014]

Wait,you were going through UWorld a SECOND time and was only getting >70%? I'm pretty sure that correlation chart linked to above was meant for the first attempt.

That's really scary then for me if that's the case


Sent from my iPhone using SDN mobile app

 

[sliceofbread136]

what does this mean? Is this even English.. do you mean skew, or long left or right tails...? so.. for myself.. massively uncorrelated.. 1.75 x through UWorld, so last two weeks, random blocks, fast and always above 70% ... Step 1 was 202 ... no one I know, knows why this happened

I don't see anything wrong with the English I used? You might want to look at your own grammar before calling out others...

 

[midwestm2]

who cares, information isn't comparable anyway. people use uworld different. some people do questions over. others don't.

some people drive right in and learn from questions and mistakes , other people study a book first and then treat uworld like a test

100% yes. I'm a second year in the midst of it and some of my classmates only go through Uworld questions that have been taught in class, whereas others are going through the entire question bank, including things that haven't been lectured on yet

 

[omn]

Wait,you were going through UWorld a SECOND time and was only getting >70%? I'm pretty sure that correlation chart linked to above was meant for the first attempt.

It was a year later and alot of times those were 90%+ (if for some reason you need an accurate bottom).

I WILL ADD THAT, as has been stated here on UWorld ad nauseum, after one's initial pass on UWorld, one tends to just remember the answer (sometimes even spatially; i.e., knowing it's location in the answer choices) rather than the second pass really being a further "drill down" on the topic.. (which of course is what you want).

I think there IS, let's call it, a vignette skill (grasping) that you have to develop; and it's harder to stay fascinated when you are older.. I recommend mneumonics for the, (allow me) "memorizie" (an adjective) stuff.. I wasn't very adept at using mneumonics, no good exposure to that, and feel that group study can help on that point... but alas, I didn't have that.

Sent from my SM-N900V using SDN mobile

 

[omn]

Hate to be that guy, but you can't say "for me, it was uncorrelated". Correlation works on a group level, not an individual level. Instead, say that "UWorld did not predict my score well"

I'm being a dbag I know, but I'll justify it by saying that using stats terms right makes us look smarter to other professions.

Eh, I disagree, uncorrelated isn't NOT used more colloquially to refer to a connection between two individual parameters (as my use was here in this passage). It's not like skew or tails (about defending the semantic hardiness of statistical terms!!). Face it, even the term "normal distribution" is slightly abhorrent, not to speak of what the other statistical terms are going on about...................................

Sent from my SM-N900V using SDN mobile

 

[CherryRedDracul]

I though we were supposed to use it as a learning tool and not focus on the percent and use NBMEs for that?

Don't forget most med students are Type A personalities.

And to add to the conversation, my Step 1 score was 4 points lower than the score predicted from my UWorld %.

 

[omn]

and just for reference, my MCAT verbal score (my worst) was less than 2/3 of my Physical Sciences score.... and my essay score was below average

 

[dempty]

Eh, I disagree, uncorrelated isn't NOT used more colloquially to refer to a connection between two individual parameters (as my use was here in this passage). It's not like skew or tails (about defending the semantic hardiness of statistical terms!!). Face it, even the term "normal distribution" is slightly abhorrent, not to speak of what the other statistical terms are going on about...................................

Sent from my SM-N900V using SDN mobile

Respectfully, you can disagree, and note what the general public mean when they use a term, but that doesn't mean it's correct. I think he was pointing out the actual meaning and application of correlation in this context (to clarify for anyone who might not know the difference because of the colloquial use). Certainly, you could say your STEP1 score was poorly/highly correlated with your UWorld % correct if you took STEP 1 multiple times and had used UWorld each time as well (then, you'd be able to see the correlation of the two variables).

In terms of biased sample data-- I think what was meant by the comment was that you're asking for data from a source that isn't likely to be representative of the population of interest (one kind of bias). So, he or she wasn't incorrect by stating you're likely to get biased data... https://en.wikipedia.org/wiki/Sampling_bias

 

[omn]

Respectfully, you can disagree, and note what the general public mean when they use a term, but that doesn't mean it's correct. I think he was pointing out the actual meaning and application of correlation in this context (to clarify for anyone who might not know the difference because of the colloquial use). Certainly, you could say your STEP1 score was poorly/highly correlated with your UWorld % correct if you took STEP 1 multiple times and had used UWorld each time as well (then, you'd be able to see the correlation of the two variables).

In terms of biased sample data-- I think what was meant by the comment was that you're asking for data from a source that isn't likely to be representative of the population of interest (one kind of bias). So, he or she wasn't incorrect by stating you're likely to get biased data... https://en.wikipedia.org/wiki/Sampling_bias

The word "correlate" is used far more frequently in ordinary language to discuss two single instances, rather than grouped data as would be one way of finding it in statistics

So, in fact, you are trying to correct me incorrectly and I am powerfully aware that you haven't provided any basis for your implied authority. I don't think you understand word usage correctly, or appreciate that "word use" is one of the principal means of arriving at correct definitions.. and again unclear what your purpose is here of trying to correct me incorrectly, but the means and manner is somewhat revolting.

Sent from my SM-N900V using SDN mobile

 

[dempty]

The word "correlate" is used far more frequently in ordinary language to discuss two single instances, rather than grouped data as would be one way of finding it in statistics

So, in fact, you are trying to correct me incorrectly

I wasn't trying to correct you-- I was pointing out what I thought the other poster intended by his explanation. I was saying that he isn't wrong (neither are you right) just because common usage has a different meaning than this thread (and I said in this context, the OP's context, the statistical meaning is more correct). It's interesting though, that you're trying to argue a point that is counter to the stated intent of this thread (to gather data to estimate a relationship between two variables). Reread the first post. It should be apparent that the OP is, very likely, looking for a correlation between the two variables.

and I am powerfully aware that you haven't provided any basis for your implied authority.

I'm not sure where you found that I was implying authority. I'm sorry if that bothered you. I provided a link to Wikipedia, so you could investigate terminology if you found my explanation to be suspect. It's also curious that you request someone (on an internet message board) supply you with information regarding their authority to post on a topic when it doesn't seem you've provided any evidence of your own authority (but, I couldn't care less about that. This is the internet...).

I don't think you understand word usage correctly, or appreciate that "word use" is one of the principal means of arriving at correct definitions.. and again unclear what your purpose is here of trying to correct me incorrectly, but the means and manner is somewhat revolting.

Sent from my SM-N900V using SDN mobile

I appreciate your concern for my grasp on the topic, and I definitely think there might be a misunderstanding here. I acknowledge that one way of arriving at the definition of a word is its common usage, but we should also consider the circumstance in which the word is used. Again, I'm sorry you're highly offended by someone disagreeing with you, but if you reconsider the context of this thread (OP), it should be more clear to you the intended meaning of correlation.

 

[elprezidente]

Sorry to interrupt the fight going on, just took UWSA1 and got a 232, meanwhile, my block scores have been between 70-75%, consistently. I've heard that UWSA over-predicts - how should I be approaching this?

 

[skarabrae]

Sorry to interrupt the fight going on, just took UWSA1 and got a 232, meanwhile, my block scores have been between 70-75%, consistently. I've heard that UWSA over-predicts - how should I be approaching this?

Fairly certain Uworld readjusted their curve on practice tests to be more reflective. Compare your % correct v predicted score to those of people last year and you'll see there is a difference.

 

[elprezidente]

Fairly certain Uworld readjusted their curve on practice tests to be more reflective. Compare your % correct v predicted score to those of people last year and you'll see there is a difference.

I'm really hoping that that's true. A 232 a few weeks out doesn't sound like a bad place to be - whereas a possible 220 doesn't sound great.

 

[PhillyMed777]

Fairly certain Uworld readjusted their curve on practice tests to be more reflective. Compare your % correct v predicted score to those of people last year and you'll see there is a difference.

This is very funny you mention this. I took them and thought the same thing.

 

[Elevencents]

I never understood the reasoning for obsessing over predictions. You should be working as hard as you can regardless if your "predicted" score is a 230 vs a 240. Are you suddenly going to start working harder if it's not where you want it to be? It's only important IMO for those who are borderline passing and may need to consider postponing. Otherwise you should simply be exposing your weak areas and shoring those up. But that's just me.


Sent from my iPhone using SDN mobile app

 

[pre med 2014]

Are you suddenly going to start working harder if it's not where you want it to be?

Sent from my iPhone using SDN mobile app

Precisely!

...And stop working hard when you are in the safe range.


Sent from my iPhone using SDN mobile app

 

[Psai]

I never understood the reasoning for obsessing over predictions. You should be working as hard as you can regardless if your "predicted" score is a 230 vs a 240. Are you suddenly going to start working harder if it's not where you want it to be? It's only important IMO for those who are borderline passing and may need to consider postponing. Otherwise you should simply be exposing your weak areas and shoring those up. But that's just me.


Sent from my iPhone using SDN mobile app

Because you want to get a sense of where you are at and which areas you need to work on

 

[W19]

Hate to be that guy, but you can't say "for me, it was uncorrelated". Correlation works on a group level, not an individual level. Instead, say that "UWorld did not predict my score well"

I'm being a dbag I know, but I'll justify it by saying that using stats terms right makes us look smarter to other professions.

Seriously!

 

[tvelocity514]

Fairly certain Uworld readjusted their curve on practice tests to be more reflective. Compare your % correct v predicted score to those of people last year and you'll see there is a difference.

Can you share the link showing they changed it please?


Sent from my iPhone using SDN mobile app

 

[skarabrae]

Can you share the link showing they changed it please?


Sent from my iPhone using SDN mobile app

No link. Just compared mine and classmates % corrects/predicted scores to those last year. For example, a 79.5% on UWorld 1 now estimates a score of 252, while a 76.5% on Uworld 1 last year estimated a score of 254.

 

[cyanide12345678]

In my personal n=1 opinion uworld self assessment and uworld itself has a pretty damn good correlation. I think I had around 74 % (something like that). Ended up with a 245. Uworld self assessment put me at 250 for step 1. Had roughly a 85% on uworld for step 2 and got a 267 on the real deal. Scored 265 on the uworld self assessment for step 2.

 

[mymembernames]

In my personal n=1 opinion uworld self assessment and uworld itself has a pretty damn good correlation. I think I had around 74 % (something like that). Ended up with a 245. Uworld self assessment put me at 250 for step 1. Had roughly a 85% on uworld for step 2 and got a 267 on the real deal. Scored 265 on the uworld self assessment for step 2.

Was this all on your first pass through it?