When choosing a QBank, I went with USMLE WORLD (UW). It is much cheaper than the competitors Kaplan and USMLE Rx, it gets higher reviews from students that have used more than one QBank, and everything else (interface, online stats) was equal. Something of a no-brainer.
After using it for almost a month, I can say a few non-comparative things.
It’s fair. I haven’t had a question yet where I felt the wording was ambiguous or the answer was a stretch. Every time that I have looked at the options and thought to myself, “I have no idea,” it was because I really didn’t know (instead of knowing the answer and not knowing how to apply it).
It’s interesting. Each question has a full explanation (usually with an illustration or table) for right and wrong choices with a summary “Learning Objective” at the bottom. This is just a smart thing for the company to do, as every wrong answer leads to a new concept learned. I regularly go through my incorrect responses and copy down the new objective into my notes.
It’s hard. While at SGU, my favorite tests were in Pathology. Whoever wrote those tests was a sadist of the fourth order. For example:
- The test stem would have symptoms. [appendicitis with Hx of appendectomy]
- You’d have to figure out the disease. [Crohn’s]
- You’d then have to realize what the appropriate treatment was. [Cortisol]
- You’d have to know the side effects of that treatment. [abd striae, bull neck]
- Finally, you can answer the question: Given this patient’s symptoms, what is the most likely side effect of his treatment? [weight gain]
Everyone complained about how hard those tests were, but I had been waiting for that type of challenge in a course my whole life. I loved those tests, and for similar reasons, I love the questions in the UW Qbank. There have been a few times where, after reading a question and figuring out the answer have thought, “That was the coolest way I have ever seen that asked.” I couldn’t give it higher praise.
If you’ve decided to use UW and are working through the questions, it might help to know how you’re doing. With each question, it will tell you the percentage of people that answer it correctly. With each subject, it will tell you your percentile against other test takers. Useful, right? I’m having a few problems with this.
The stat for “percentage that answer correctly” doesn’t say if that is on the first try or includes all attempts, including repeats. I wish there was a separate statistic for this. Your overall percentile is based on your test average against the mob, but again this can be manipulated by taking the same questions over and over (I’ve tested this myself). For the person going through the questions once without repeating, you may feel that your percentile is a little low (or just hope that it is).
Here are my percentiles on first past through all the available questions in a section. I will expand this list as I continue to cover material. You’ll notice that the scores are very low. As I mentioned before, these aren’t true percentiles as they are not compared against the mob’s first attempt and I list them here just so that people don’t feel so defeated when they take the questions themselves.
- Biostatistics (74th) - I felt very well prepared
- Behavioral Science (waiting to do Psych until later)
- Embryology (71st) - I felt well prepared
- Genetics (48th) - Curse you, Dudek and your horrible book.
- Biochemistry (79th) - I felt very well prepared
- Immunology (70th) - I felt well prepared
- Histology (32nd) - I wrote this off. Maybe I should look at it.
- Anatomy (74th) - Just for fun. I’m an Anatomy geek.
ADDENDUM: I sent an email to the USMLE WORLD team about these questions and I was pleasently surprised to see them respond the next day.
The “percentage that answer correctly” only records the first attempt of the user if the question in taken in the unused mode.
The cumulative performance is based on the entire test percentage and you are correct in assuming that the percentile might be manipulated if a person repeatedly takes the test and answers all the questions correctly. However for the percentile to skew greatly, a large number of users will have to “cheat” the system this way by repeatedly taking the same questions.
However, most of our users take the test first in unused mode and then they use other modes like incorrect or marked questions if they have sufficient time left. This might skew their overall percent by 2-3 % but over a significantly large data set this offset becomes negligible.
Hence, the presented percentile should only be used as a rough indication of where the user stands and preferably should be ignored during the initial tests.
No comments:
Post a Comment