US News Updated Step 1 scores?

[SteyrFWB]

---

Members don't see this ad.

UCF 243 FIU 241 FAU 239 UM 233 UF 230

What advantages do new schools have over old schools?

Old schools with pedantic professors in first two years, who have been there too long, can't teach USMLE stuff?

Or new schools with no professors, just let students loose; students studying from their bedrooms the first two years?

 

[efle]

^ I would expect a bunch of students sitting at home studying Boards&Beyond/UFAP, to do a lot better on step1 than a similar bunch of students that did a school-designed curriculum. If new schools are giving more self study or just basing their own material more on the universal step1 materials, that could be what you're seeing.

 

[Dr. Doctor MD]

Just because this same topic comes up every couple months...

1. These are averages with very large standard deviations (like 10-15 pts minimum). Theres no difference between a 245 at 1 school and a 235 at the next.

The bolded in blatantly false.

You can make an argument that schools with step scores 2-3 points away have no difference. Maybe even up to 5 points if you want, but any higher than that and it's a real difference. You're misunderstanding standard deviation and standard error of the mean. I promise you any school with over 100 students and a step 1 score difference greater than 5 are likely statistically different.

 

[Med Ed]

^ I would expect a bunch of students sitting at home studying Boards&Beyond/UFAP, to do a lot better on step1 than a similar bunch of students that did a school-designed curriculum. If new schools are giving more self study or just basing their own material more on the universal step1 materials, that could be what you're seeing.

This analysis of Step 1 study behaviors and outcomes made a big splash last year. While Michigan's student population is not exactly representative of medical students as a whole, the bottom line is consistent with a lot of other data shown at national meetings and collected for internal use. The most noteworthy predictors of higher Step 1 scores included:

- MCAT
- Undergraduate GPA
- Performance in pre-clinical curriculum
- Number of practice questions completed

The length of dedicated study had no discernible effect (nor did the number of hours spent studying each day), although higher scoring students tended to start some type of prep before dedicated time commenced. Correlation versus causation is always the problem with these analyses, but they are intriguing nonetheless.

 

[efle]

This analysis of Step 1 study behaviors and outcomes made a big splash last year. While Michigan's student population is not exactly representative of medical students as a whole, the bottom line is consistent with a lot of other data shown at national meetings and collected for internal use. The most noteworthy predictors of higher Step 1 scores included:

- MCAT
- Undergraduate GPA
- Performance in pre-clinical curriculum
- Number of practice questions completed

The length of dedicated study had no discernible effect (nor did the number of hours spent studying each day), although higher scoring students tended to start some type of prep before dedicated time commenced. Correlation versus causation is always the problem with these analyses, but they are intriguing nonetheless.

WOW! An r-squared of 0.6 after controlling for MCAT and preclinical grades is HUGE. I pretty much consider this proof that doing step 1 review materials throughout M1-M2 + a boatload of practice questions, in lieu of the school curriculum, is absolutely the way to go.

 

[Goro]

This analysis of Step 1 study behaviors and outcomes made a big splash last year. While Michigan's student population is not exactly representative of medical students as a whole, the bottom line is consistent with a lot of other data shown at national meetings and collected for internal use. The most noteworthy predictors of higher Step 1 scores included:

- MCAT
- Undergraduate GPA
- Performance in pre-clinical curriculum
- Number of practice questions completed

The length of dedicated study had no discernible effect (nor did the number of hours spent studying each day), although higher scoring students tended to start some type of prep before dedicated time commenced. Correlation versus causation is always the problem with these analyses, but they are intriguing nonetheless.

This is pretty much what we've seen at our school, except that MCAT is a weak predictor at best, and preclinical GPA is the best predictor. The number of COMBANK items done before COMLEX also is an excellent predictor for that exam. We've found no correlation between UG GPA and Board performance.

I'm a big fan of practice questions.

 

[ndafife]

I promise you any school with over 100 students and a step 1 score difference greater than 5 are likely statistically different.

You are 100% unequivocally wrong here.

A difference of 5 with a standard deviation of 10-15 for each? Good luck getting statistical significance there

 

[Med Ed]

WOW! An r-squared of 0.6 after controlling for MCAT and preclinical grades is HUGE. I pretty much consider this proof that doing step 1 review materials throughout M1-M2 + a boatload of practice questions, in lieu of the school curriculum, is absolutely the way to go.

Be careful. This paper is tantalizing, but it's just one institution. I definitely agree that practice questions are an invaluable tool from day 1 of medical school.

I hesitate to recommend abandoning a given school's curriculum unless there have been issues with Step 1 failures in recent years. Board prep materials are somewhat backwards-looking, everyone has access to the same ones, and the Steps are norm-referenced tests, which means a high score is a combination of knowing information across the yield spectrum.

 

[efle]

Be careful. This paper is tantalizing, but it's just one institution. I definitely agree that practice questions are an invaluable tool from day 1 of medical school.

I hesitate to recommend abandoning a given school's curriculum unless there have been issues with Step 1 failures in recent years. Board prep materials are somewhat backwards-looking, everyone has access to the same ones, and the Steps are norm-referenced tests, which means a high score is a combination of knowing information across the yield spectrum.

That's fair. I'm a little biased to love these results because I've abandoned my school's lectures entirely and now only study from the universal step 1 materials. The content overlaps more than enough to still pass the school exams. And it is so much easier to learn the content when it's taught as a cohesive unit from one instructor (e.g. Boards and Beyond, Pathoma) instead of a bunch of different professors giving their versions of what is important on a variety of topics. I end up learning more, and understanding it better, in half the time. In the rare cases where the school covers information that is absent from Boards&Beyond/UFAP, it's usually because the professor is spending a bunch of time talking about their extremely specific research interests that would never show up as an exam question.

My school also has recently been under-performing on step 1 relative to similar student bodies, so the above might not be as applicable to everyone. To any current applicants/freshly minted M1s that might see this, though: if you are noticing that your school lectures feel low-yield and there's tons of stuff in First Aid or Qbanks that you never learned in lectures, don't hesitate to switch to learning from Boards/UFAP. I wish I had started sooner.

 

[efle]

You are 100% unequivocally wrong here.

A difference of 5 with a standard deviation of 10-15 for each? Good luck getting statistical significance there

Here's an example set

Harvard: 180 students, reported mean 245, std.dev assumed 15
Hopkins: 120 students, reported mean 235, std.dev assumed 15

t test: t value 5.7, p < 0.0001

you hit significance very, very quickly when your samples are 100-200 people. The above example remains very significant even if comparing only a 5 point gap of 240 vs 245 (P < 0.01)

 

[deleted938090]

The standard deviation of the population is unknown so we are approximating that with the sample standard deviation.

We are assuming scores are normally distributed (this may not be true, it’s probably skewed right so perhaps it could follow a gamma distribution). Sample size n>30 so it’s a good set of data; as regards to self reported, there may be non response bias towards lower end.

The data we are given are samples drawn from the population of students who take the test. Our statistical tests are used to determine how confident that the sample we obtained will contain the true mean in the interval we generate. As we choose larger and larger samples, the data will tend to fall closer and closer (smaller spread), going towards 1/sqrt( n ). So although the standard deviation may be around 10-15, as n approaches 100, the standard error of the mean will be around 1-1.5. The inverse t score for n-1 degree of freedom and a critical t value can be looked up on a table to create a confidence interval; if the two confidence intervals overlap, we cannot conclude a difference.

Hence this is why @Dr. Doctor MD stated that scores around 3-5 points can be said to have no difference. A 95 confidence t approximation for n = 100 is around 2 since t approaches z distribution as n increases. 2 times 15/10 is about 3 points above or below. It is possible for two intervals to overlap at about a 5 point difference and conclude that there is not a significant difference.

However the data itself isn’t a good measure for the school because factors around step score affect its result. Taking step score later will obviously boost scores; taking it earlier will lower scores. Schools with research oriented or primary care oriented will have less emphasis on the score. The step scores are more of a reflection on the students’ goals and time of taking more so than the prestige of the institution (although there will definitely be contribution, the exact amount of contribution by the institution is difficult to say.) Approach this data with caution: going to a school with high step won’t guarantee You will do well like that, nor will going to a school with lower step average guarantee you will have lower scores.

 

[DubbiDoctor]

That's fair. I'm a little biased to love these results because I've abandoned my school's lectures entirely and now only study from the universal step 1 materials. The content overlaps more than enough to still pass the school exams. And it is so much easier to learn the content when it's taught as a cohesive unit from one instructor (e.g. Boards and Beyond, Pathoma) instead of a bunch of different professors giving their versions of what is important on a variety of topics. I end up learning more, and understanding it better, in half the time. In the rare cases where the school covers information that is absent from Boards&Beyond/UFAP, it's usually because the professor is spending a bunch of time talking about their extremely specific research interests that would never show up as an exam question.

My school also has recently been under-performing on step 1 relative to similar student bodies, so the above might not be as applicable to everyone. To any current applicants/freshly minted M1s that might see this, though: if you are noticing that your school lectures feel low-yield and there's tons of stuff in First Aid or Qbanks that you never learned in lectures, don't hesitate to switch to learning from Boards/UFAP. I wish I had started sooner.

I find the large predictive power of preclinical grades to be really interesting. The top 23.4% of the class had an average step 1 score above the 85th percentile for US MD students. That's higher than the average Umich student with an MCAT score at or above 37 - probably within the top 5% of all US MD students. How, if at all, has abandoning school resources affected your preclinical grades?

 

[XxTargaryenxX]

UCF 243 FIU 241 FAU 239 UM 233 UF 230

What advantages do new schools have over old schools?

Old schools with pedantic professors in first two years, who have been there too long, can't teach USMLE stuff?

Or new schools with no professors, just let students loose; students studying from their bedrooms the first two years?

nothing with either, the only difference is STUDY TIME before step 1 and the clinical year for the Step 1. For example a lot of the top performing schools don't take step 1 until THIRD YEAR IS OVER, that means they get 1 full year of clinical experience which helps them do so well on step 1. I laugh when I see FIU stats which is a 240+ average, they get an extra year to study for step 1 and its a year of clinical experience. Did you see their average for step 2 ck? when they also have to take it in third year AND they get the same amount of time as other us medical schools? their step 2 ck average is actually below the national average.

 

[The.Supinator]

Does anyone know how I can get the average of each individual school, specifically I need to know what the average is for NEOMED, Northeast Ohio Medical University.