Data Intelligence

Can an adaptive algorithm quickly improve your LSAT Logical Reasoning score?

Can an adaptive algorithm quickly improve your LSAT Logical Reasoning score?

The takeaways

  • Drilling is always good for LSAT studying
  • AdeptLR's algorithm can help drill efficiently with fewer questions

"Can AdeptLR's adaptive algorithm help improve your LR score faster?" -- This question is frequently asked, and data showed that the answer is YES! Our adaptive algorithm enables efficient drilling by minimizing the number of questions a user needs to drill to achieve a certain improvement in LR score.

In deriving our answer, we compared changes in LR timed sections scores with vs. without AdeptLR drilling. In particular, we asked the question “how many LR questions do users need to drill to achieve a 3+ points improvement with AdeptLR compared to without it?” 

We found that without AdeptLR, a user typically needs to practice 1,053 questions in order to improve at least 3 points in timed sections. In contrast, with AdeptLR drilling, to improve at least 3 points, users on average would need to drill only 486 questions. However, there are two data interpretation caveats to note. First, the number ‘486’ is an average across users–some users drill more and some drill less, so it is not a guaranteed threshold that once you cross 486 questions, your score will automatically jump up by at least 3 points. Second, users' performance in timed sections can fluctuate. On a good day, a user may score 0, and on a bad day a user may score -8. To reduce data noise, we measure improvement by averaging the change in scores from baseline score (i.e. calibration score) over time. This enables us to detect users whose improvement is “stabilized” and consistent.

For someone just starting their studies, we recommend drilling at least 500 (assuming you are practicing across categories). Remember that your practice data is important for our algorithm to personalize the questions fed back to you. The more you practice, the more tailored the algorithm gets. 

Average number of questions users need to answer to increase their LR timed section by +3 points
Marie Ng, Chief Data Scientist
April 20, 2023
Got questions or feedback? Contact us!

You may also like
Adeptbot in mail
Join our newsletter!

Subscribe our newsletter to receive the latest blog posts. No spam.

Subscribe