[Undergrad] Pre-UMAT 2017 Discussion Thread

[nira]

Thanks dude. 76 94 92 respectively (This was one of my better tests though). I'll be honest, dont be too hung up over your percentiles. With such a small sample size (especially for the latter exams) any small increase in marks has a large jump in percentiles. If we were to score 91/134 raw (i've scored that) what percentiles do you think we'd be looking at.

yeah no your definitely right, the medentry percentiles don't entirely reflect the actual umat percentile. in fact it is to my knowledge that the percentile you get on ME is much less than the actual umat. I was curious to see the differences in our percentiles, for me my percentiles were 30th , 97th and 85th. But if we were to get around 91/134. thinking optimistically, it would be around the lower spectrum of the 90th percentile range or a very high 80th percentile. this is using A1 advice that 67% of the questions right is 90th percentile and 75% of the questions right being 100 percentile.

 

[frootloop]

This is using A1 advice that 67% of the questions right is 90th percentile and 75% of the questions right being 100 percentile.

@A1 do you have an ACER reference for that, or is this just the percentages you need for those percentiles on MedEntry rather than the actual test?

Also, don't ACER weight question scores by difficulty using some statistical wizardry? So the percentage of questions you'd need to get right for certain percentiles would depend on which questions you got right?

 

[A1]

@A1 do you have an ACER reference for that, or is this just the percentages you need for those percentiles on MedEntry rather than the actual test?

It's for the actual test but the info wasn't from ACER, it was from my expert analysis of... 2 pages worth of data.

Max possible score 300 out of 134 Qs = 2.24 per Q if equally weighted. Since they are not and we are more likely to miss the higher ones I propose to use 2 marks per Q as a guide
90%ile is around 180 marks = 90 Qs = 67% of the test
99%ile is 200 = 100 Qs = 75% of the test (not sure if I said 100%ile earlier but should be corrected to 99%ile).

 

[frootloop]

Where'd you get the data for how many marks are required for each percentile? Like, surely there's no official UMAT data released stating that there's a maximum possible score of 300, or that 90th%ile is around 180 marks, etc?

Your weightings calculations look pretty good as a rough guide (hard to know exactly how much the marks-per-question varies, but the estimate you've used seems fair enough), but again, I'm genuinely interested in the source of your information - even 6 years after sitting it, I'm still way too interested in the UMAT haha

 

[nira]

90th%ile is around 180 marks

if you have a raw score of 180 on your statement of results in the umat you will get 90th percentile
just like if you have a raw score of 181 your percentile in 91st.

 

[nira]

maximum possible score of 300

theoretically the highest raw score you can get for a section is 100
hence there are three sections so theoretically the highest score you can get is 300. which most likely equates to getting 100% of the questions rights, a feat which im pretty sure no one has achieved thus far

 

[A1]

Results | Undergraduate Medicine and Health Sciences Admission Test | UMAT | ACER
"Each of the 3 construct scores is expressed on a scale of 0 to 100."

Score curves
[Undergrad] - 2016-17 Med schools Selection Criteria Y12s & Non-standards | Med Students Online

 

[frootloop]

Results | Undergraduate Medicine and Health Sciences Admission Test | UMAT | ACER
"Each of the 3 construct scores is expressed on a scale of 0 to 100."

Score curves
[Undergrad] - 2016-17 Med schools Selection Criteria Y12s & Non-standards | Med Students Online

There have been raw scores of over 100 before (not that they're common enough to really affect your calculations), so I'd probably take ACER's sweeping statements with a grain of salt.

Your logic all works out pretty well with the information you've got, but I'd be careful about being too confident that you've worked ACER out. The statistical wizardry they do behind the scenes is insanely complicated, so I very much doubt that working out the number of questions a candidate has gotten right is as straightforward as your calculation.

Your conversion calculation is probably a pretty good rough guide for people anxiously wanting to get an idea of where they stand heading into the test. But I'd caution them to use it as a guide only.

As the source you've linked states, 'It is not possible for candidates to replicate this process'. They're super anal about keeping their methods hush-hush, and they aren't dumb enough to give you enough information to replicate their processes on their publically-accessible website haha

 

[A1]

There have been raw scores of over 100 before (not that they're common enough to really affect your calculations), so I'd probably take ACER's sweeping statements with a grain of salt.

The 2010-12 S1 curves did show section scores up to 107-108. The last 2-3 years they stop at 100 or a little lower so there has been a change in that regard.
I wrote this reply on Atarnotes forum, copying it here for your info:

"Five years ago UMAT used to be valid for 2 years, (I think) ACER needed to scale the marks of one year to be comparable with the previous year since they could compete directly against one another in the same applicants pool. The maths resulted in some going over 100, not fair to chop them to 100 while leaving the year cohort alone, but also not fair to scale the whole cohort down as it would advantage the previous year. This is no longer needed".

The other comments from you are true. I provided this guide with the comforting thought no-one is going into the test aiming specifically for said 90 correct Qs. They of course try for as many as they can get, for now just to calm them down that they don't need to get 90% of the test to get 90%ile.

 

[frootloop]

The 2010-12 S1 curves did show section scores up to 107-108. The last 2-3 years they stop at 100 or a little lower so there has been a change.
I wrote this reply on Atarnotes forum, copying it here for your info:

"Five years ago UMAT used to be valid for 2 years, (I think) ACER needed to scale the marks of one year to be comparable with the previous year since they could compete directly against one another in the same applicants pool. The maths resulted in some going over 100, not fair to chop them to 100 while leaving the year cohort alone, but also not fair to scale the whole cohort down as it would advantage the previous year. This is no longer needed".

I'd wondered why I hadn't heard of any >100 raw scores in a while, that definitely does make sense. I hadn't thought about how making UMAT scores valid for only one year instead of two would affect the top end of the scores!

 

[nira]

does any body know what time the am session is held? i'm pretty sure its around 9am but not sure,.

 

[DrDrLMG!]

does any body know what time the am session is held? i'm pretty sure its around 9am but not sure,.

I think it depends on the venue. Some may be slightly earlier than 9am (say, 8.30, or 8.45), and some slightly later (9.15 or 9.30). Hobart is 9am. As is Launceston. I'm less familiar with non-Tasmanian location start times, but I do recall there was some (though minimal) variability.

 

[Sherlock]

You can get the UMAT app on your mobile device. It specifies times for all locations

 

[DrDrLMG!]

You can get the UMAT app on your mobile device. It specifies times for all locations

If those times are true (and I actually doubt they are), then the Hobart and Launceston ones have been brought forward by 45 minutes. Given there are no afternoon sessions in either location, I don't see the point of this. Hence, I don't think the published times are accurate. Happy to eat my hat if my assumption about their incorrectness proves... incorrect!

 

[Dnnysn]

Just did the Acer Prac, Thought it was quite easy til the end, 83/134, a bit overconfident it seems. I haven't really done much study for it though, and didn't time myself.

 

[DrDrLMG!]

I just did a MedEntry exam (after having not even looked at so much as a drill for over a week) and got 103/134, which is my best result ever. YAY! As far as UMAT percentiles go, it's 98th. (S1, 98th, S2, 99th, S3, 88th).

How is everyone else going?

 

[Yamster]

I just did a MedEntry exam (after having not even looked at so much as a drill for over a week) and got 103/134, which is my best result ever. YAY! As far as UMAT percentiles go, it's 98th. (S1, 98th, S2, 99th, S3, 88th).

How is everyone else going?

How'd you improve your s1/s3 so much?

 

[DrDrLMG!]

@Yam, for S3, I suspect it's just been the impact of reacquainting myself with the processes involved over the course of the few drills I've done some weeks ago now (that said, I still didn't do that great - 29/42 - and had to guess a couple).

For S1, I gave myself a little more time than previously for this section specifically, and consciously paid attention to the wording of questions. I know for sure that my downfall in the past has been mis-reading questions, particularly the 'which answer is NOT supported by...' type ones. I got 41/48 so this was definitely my big improver.

Other than that, I honestly have no idea. It might yet have been a fluke! I'll probably aim to do another one this weekend, and at least one a week between now and July 26th.

I also need to register for the exam itself!!

 

[hamp]

I just did a MedEntry exam (after having not even looked at so much as a drill for over a week) and got 103/134, which is my best result ever. YAY! As far as UMAT percentiles go, it's 98th. (S1, 98th, S2, 99th, S3, 88th).

How is everyone else going?

Which exam was this?

 

[DrDrLMG!]

Which exam was this?

Exam two.