Subscription pricing
Admin Note: These stats were updated on 2022-01-08.
That question came up in a recent discussion thread and we were curious. So, we ran the numbers using data from PT 58 through current. Results:
Logical Reasoning, all questions (1 omitted)
A 337 18.95%
B 375 21.09%
C 365 20.53%
D 367 20.64%
E 334 18.79%
Total 1778
Logic Games, all questions (1 omitted)
A 179 20.55%
B 180 20.67%
C 173 19.86%
D 172 19.75%
E 167 19.17%
Total 871
Reading Comprehension, all questions
A 181 18.64%
B 223 22.97%
C 185 19.05%
D 201 20.70%
E 181 18.64%
Total 971
Logical Reasoning, last five questions
A 68 16.19%
B 88 20.95%
C 79 18.81%
D 89 21.19%
E 96 22.86%
Total 420
Logic Games, last five questions
A 47 20.61%
B 47 20.61%
C 46 20.18%
D 46 20.18%
E 42 18.42%
Total 228
Reading Comprehension, last five questions
A 31 14.35%
B 59 27.32%
C 35 16.20%
D 47 21.76%
E 44 20.37%
Total 216
Now that you have this information, what do you do with it? Not much. It's mostly to satisfy your curiosity. Some of you might be tempted to use this information to guess on future LSATs. Hopefully, you won't be in a situation where you have to blindly guess across all five answers. If you end up in that situation however, then sure, this is probably as good a guide as any to which answer to guess. As I've said numerous times before though, the LSAT writers are very crafty.
51
23 comments
@juliet7sage
Hi! I mean the Answer Choice of the last question of each section. Apologize for any misunderstanding.
Instructions unclear, selecting B on LSAT
Just to clear things up. This is for curiosity only. It's not meant to be a guide to guessing because there isn't a statistically reliable way to guess better. The LSAC would never allow that to happen.
But more importantly, thinking about the test in those terms misses the point. You're far more likely to edge out an additional question by not focusing on "how to guess better." Rather, focus your attention on the things that have a causal impact on improving your score: logic, grammar, etc. in other words, focus on the reasons why right answers are right and wrong ones are wrong.
Hi @harmeet96357
I gather you are looking for data per question type. Please let me know if I misunderstood. Sorry, we do not have the data on each question type per section.
The numbers we ran are for all questions in each section using data from PT 58 through current.
Hi! really helpful information. Also would like to know do you have data on the LAST Question of each section??? I think this might be much more helpful.
Hi @mweinberg97466,
Thank you for your question! I will check this with my team, and I will get back to you as soon as possible.
@juliet7sage
Hi Juliet, so I am confused on what to guess because 7Sage and PowerScore use different Data for what to guess on....
For example, Dave Killoran says to choose "D" on 1-20 on LR when in doubt, and "E" on the last 5. JY says to choose "B" when in doubt on 1-20 and "E" on the last 5 for LR.
Note, JY uses data from PREPTEST 58 (Sept 2009 ) onwards, while Dave uses data from Nov 1991. Which one is more reliable?
Here is the PowerScore:
Powerscore: https://www.powerscore.com/lsat/help/guessing.cfm
Hope you guys can address this! That would super helpful.
Thanks,
Matthew
And RC last 5 is the only "borderline" answer of any value, and that is, in all likihood worthless as well.
> @stephencarpinello514 said:
> Out of curiosity, do we have updated stats?
Hi there,
Yup, these stats have just been updated using the data from PT 58 through PT 92.
Let me know if you have any questions!
https://www.powerscore.com/lsat/help/guessing.cfm
Out of curiosity, do we have updated stats?
For what it's worth: PT 58 through what was current when the original post was made, PT 77, is 21 PTs --- 20 numbered ones plus the C2. The "last five question" sample sizes are 126 for each section, which is 21 times 6.
Oops, you said sample SIZE, I'm not too sure. Lol. Good question. @steve898
@steve898 I'm pretty sure it's because the frequencies of the letter choices have been converted into percentages ... That's my assumption anyhow. :)
For the "last five question" results, why aren't the sample sizes a multiple of 5?
Agreed. If the true chance of B were 20%, we'd expect to see some serious mean regression in the larger data set (all RC questions). Yet even in the set with 500+ questions, B is ~5% more likely than any other letter.
@jy-ping can you please post an updated version of this for folks taking the December test this weekend? Would be massively helpful!
Wow the stat on RC last 5 is surprising. That's borderline statistically meaningful.
They do that all the time! It happened twice on one PT (can't recall which). Honestly, it left such a bitter taste in my mouth that I now read ACs A and E first on all parallel questions.
While JY is constantly plotting, he's also in a very different timezone than those of us in the US :D
HAAAAHHHAAAAA. This is so true though.
The LR data for the last 5 questions is pretty much in line with my anecdotal evidence. For the last 5-6 questions on LR, I usually find myself reading answer choice E first and reading my way up to A. I think there is a discontinuity with E due to the parallel reasoning question typically falling in these last 5 questions, and I "feel" like the answer to those (with no actual hard data to test my hypothesis) is usually D or E and not A. For parallel reasoning questions, I always think about the poor guy who had to write 4 lengthy wrong answer choices, and if it were me writing the question, I'd make the test taker have to sift through A-D first.
I am convinced JY never sleeps. He stays up all day and night plotting how he can further conquer the LSAT.