IR Scoring

[SanjayG915]

The following is not a duplicate question: there is a thread from 2012 discussing this topic. At the time though, the answer was: "we don't really know b/c IR is so new."

Based on what we know now, are there any rough rules-of-thumb for how # correct on IR maps to your IR scaled score?

(i.e.: if I get 7/12 correct, that equals a ?, etc, etc)

[tim]

Not exactly. Different problems seem to have been assigned different difficulties, so your score appears to be more of a weighted average of the problems you get correct.

[RonPurewal]

It's not an exact correspondence, because there's more than one set of IR questions in existence.

Whenever there are two or more forms of a standardized test, those forms must be "normalized" against each other. (Short version: Lots of complicated statistical calculations must be performed, ultimately producing correspondence tables between any two forms of the test/section.)
It's impossible, of course, to create two different tests/sections that have exactly the same student performance profile. So, the normalization will always produce weird tables, and there's never going to be a neat conversion table with uniform/integer step sizes. Not gonna happen.

[RonPurewal]

The one thing we've noticed in general, though, is that, in order to achieve the highest score (8) on the IR, you only need to get 8 or 9 of the 12 questions correct.
In other words, like the rest of the exam, IR doesn't require perfection (or anything close). Even though it's non-adaptive.

[SanjayG915]

Ron & Tim: awesome answers, very helpful.

In slogging through a number of the official practice Q's, I've noticed that there are a few that seem borderline impossible to finish in <3min, much less 2.5min. I'm thinking there's an analogue here to the rest of the test: some questions are simply going to take EVERYONE (including crack Manhattan teachers) more than the average. The key is that there are some that will take less & it's a game of "banking time" on some so that you can use that "banked time" on others. True?

[RonPurewal]

Ron & Tim: awesome answers, very helpful.

In slogging through a number of the official practice Q's, I've noticed that there are a few that seem borderline impossible to finish in <3min, much less 2.5min. I'm thinking there's an analogue here to the rest of the test: some questions are simply going to take EVERYONE (including crack Manhattan teachers) more than the average. The key is that there are some that will take less & it's a game of "banking time" on some so that you can use that "banked time" on others. True?

Don't over-think this. It works like everything else in life.
This is how life works: If you have to do a bunch of ___s, then some ___s will take longer than other ___s.

What IS weird and maladaptive, on the other hand, is to think "I should finish EVERY ___ in EXACTLY the same amount of time."
Huh? That's just not how life works. Common sense should reject all such thoughts right out of the gate.

[RonPurewal]

In IR, the situation is even more pronounced, because some problems are based on large blocks of text, while others are based on tables and graphs.

• Blocks of text are slow.
Large blocks of text are primarily useful for things like storytelling, in which it's normally not essential to capture too many tiny details—as long as you have the "gist" of any particular paragraph, you're fine.
For ANYTHING in which details matter—especially quantitative details—paragraphs are horrible. And slow. Horrible and slow.

• Graphs and tables are fast.
"Fast" is the reason why mankind invented graphs and tables in the first place. (Think of how long it would take you to deduce something from a graph or table, vs. to make the same deduction from the same information in paragraph form.)

As a result, you'll sometimes see extremes.
It's not altogether uncommon for text-based problems to take 3-4 minutes but for (some) table- and graph-based problems to take no more than 20-30 seconds. It happens all the time, actually.

[RonPurewal]

More generally, good time management involves NO thinking about numbers of minutes and seconds.

Good time management is ...

1/
... being completely and brutally honest with yourself about when you are stuck,

2/
... IMMEDIATELY stopping what you're doing if you are stuck (and trying to come up with an alternate approach),

3/
... guessing and moving on, if you can't think of anything else.

Once you've developed these instincts, you won't need to look at the timer at all anymore. You can still check it, say, 1/4, 1/2, and 3/4 of the way through the section, but even that shouldn't really be necessary—it's just quality control.

#1 is the one that most people don't do. If you do not know EXACTLY WHY you are doing what you're doing, then STOP DOING IT.
NEVER do random steps of math.
NEVER "browse" text without a specific goal and/or a VERY specific question whose answer you're looking for.

[sw001]

Hello Ron,

I really liked your last post on this thread. You have very nicely specified the biggest problem everyone face and its solution!

[RonPurewal]

Well, not everyone. Some of us aren't cursed with tenacity and perseverance. (:

But lots and lots of people, sure.

[sw001]

:)

Ron - I do have a question on IR section -

I saw on the web (please confirm this) - that some of the question you will see on IR will be experimental. And those experimental questions could range from 2 to 4. And even if I get them right, they won't count towards my score. Is that correct? I read it in a post published by Economist.

[tim]

I haven't heard anything of the sort, although it would make perfect sense for them to include some experimental questions. Here's what to keep in mind though:

(1) It doesn't matter at all. This will have nothing to do with your strategy, except possibly to underscore the importance of not getting hung up on any particular question. However, this is good advice regardless of whether there are experimental questions.

(2) Any time you start a sentence with "I saw on the web", most people will tune you out - and you should tune yourself out as well. The mere existence of something on the web is of no more consequence than just dreaming it up, unless you have a reputable source that you can quote with a citation so that people can determine how much credibility to attribute to it.

[sw001]

(2) Any time you start a sentence with "I saw on the web", most people will tune you out - and you should tune yourself out as well. The mere existence of something on the web is of no more consequence than just dreaming it up, unless you have a reputable source that you can quote with a citation so that people can determine how much credibility to attribute to it.[/quote]


Thank you Tim for your reply and advise! Yes, I saw the post on Economist (link attached), for that reason - got concerned. I did think about the legitimacy factor this phrase "I saw on the web" may attribute to - while writing :)

https://gmat.economist.com/blog/test-ta ... ion-scored

[jnelson0612]

Great discussion, everyone. :-)

[tim]

Thanks for the citation! That sort of thing definitely helps. Ultimately though, don't worry about experimental questions because there's nothing there that you can use to your advantage.