Riddle me this...

[tomslawsky]

something I've been confused about since I started my quest for GMAT greatness. When a question difficulty is referred to, it is usually within the framework of a "total score". For example- people say that the quant of Kaplan 800 is at about 650 level. Or that the last 50 questions of the official guide quant are 700 level.

Here is my confusion. Quoting that a "quant" question is of about "700" level tells me nothing. There are many combination to get a 700 on the GMAT. If you score a 99th percentile in verbal and a 69th percentile in quant, you're in the 700+ club. If you score a 99th percentile in quant and a 60th percentile in verbal, you're in the 700 club. To these test takers, a "700" level quant question is radically different, as a 51 quant question is miles away from a 42 level question. I think of it in terms of the SAT. I can theoretically look at a math question and make a good guess as to whether it is a 750 level question or a 550 level question. But since the verbal is in no way integrated with the math, there is no way I could ever tell you whether the same question is a "1550" level question or an "1100" level question.

In essence, why aren't questions referred to by sub-score percentile, ie that is a 85-95th percentile QUANT question. Even better would be to say that a question is estimated to be in the range of a sub score. For example- question 175 in the official guide problem solving is about a 43-45 level question. What am I missing here? Why aren't questions estimated by sub-score, rather than composite difficulty? Any insights? Thanks.

[StaceyKoprince]

It's really just to make it easier for students to understand.

In terms of the actual algorithm, questions aren't classified as a simple number or even a range of numbers. (If you're interested in how questions are actually rated, take a look at our new e-book, The GMAT Uncovered - and then you'll also understand why it would be difficult to use the actual rating system to provide an easy "difficulty level" number or range of numbers.)

If we called a question a 43-45 level question, the vast majority of students would have no idea what we were talking about. Also, given that the two-digit scales are different for quant and verbal, there would be even more confusion, as a 43-45 level verbal question is significantly more difficult than a 43-45 level quant question.

On the other hand, most people are familiar with the 3-digit overall scores. As a result, most companies use that as a proxy to give a rough idea of difficulty level. 700 is about the 90th percentile (a tiny bit higher, actually), so a "700 level" question is supposed to be equivalent to a roughly 90th percentile question.

[tomslawsky]

Makes sense, sort of. To me, it is more confusing to refer to a question composite difficulty rather than sub-section difficulty or percentile difficulty. I guess my brain is hard wired different because I've been confused about this for over a year. I would never say that someone's college schedule is "easy" because they were taking psych 100 and history 100, while ignoring the calculus 3 and particle physics classes.

The ironic thing is that, assigning a quant or verbal question as a composite score difficulty level, I think would never pass the sniff test were this method to be presented as a GMAT critical reasoning passage. Someone can get a 700 score with a 51 or a 44 quant score- too wide of a range.

I know, I know, sometimes my brain marches to the beat of its own drum :)

[StaceyKoprince]

Don't think of it as a composite score. When someone says something is a "700-level question" they just mean that it's around the 90th percentile. That's all. :)