Accuracy: Perceived versus Actual

A common CAT aspirant question: Should I target 80% or 100% accuracy?

My answer: If you target 100%, you end up at around 80%. If you’d target 80, you’d end up much lower.

This is what this blog is about ?

What else you can do inside qs leap ?

2500+ Free
Practice Questions

Get Free Access to 2500+ GMAT/GRE Questions

30 Min
Prep Classes

Attend Free GMAT/GRE Prep Classes Everyday

Virtual One-to-One

On-demand online meetings with Admissions Teams for free

I’ll also address another common (and relevant) query. The query is well presented by this question asked on Quora: How important is accuracy on the CAT? Specifically, which is a better result:

  1. Answered 15, Correct 15. Score → 15 x 3 = 45; or
  2. Answered 25, Correct 20. Score → 20×3 – 5 = 55

I paraphrase.

My answer: CAT only looks at the final scores in each section. The algorithm is not convoluted to also look at the number of corrects/ incorrects etc. Simply put, a higher score is a better score.

So, a score of 55 is better than a score of 45. There is no extra penalization for incorrect answers apart from the +3/ -1 already declared.

This much is easy to understand.

The trouble comes when students train their minds for the fact that lower accuracy percentage with more number of attempts is a good thing, and as a consequence, start guessing on the CAT. I concede, that those are not random guesses. Typically students only guess on questions they were able to eliminate at least 1 or 2 incorrect options. Yet, this is a slippery slope.

To explain this, I introduce 2 new terms:

  • Perceived accuracy: The % of questions you perceived you had answered correctly (while attempting during the test)
  • Actual accuracy: the % of questions you had actually answered correctly (when you compare your answers with solutions after the test)

Perceived accuracy v/s Actual accuracy

Now, let me ask you this: when you attempt practice questions, do you get them all correct? If you are human, I’m sure you make mistakes. In other words, even when you answer questions with complete confidence, in all probability, you will get a few wrong.


perceived accuracy: 100%

actual accuracy : <100%

  • If you’d like, try to do this exercise
    • Take a practice test with proper time limits
    • Answer only those questions that you are sure of
    • After the test, verify:
      • did you have a significant number of attempts? (> half the number of questions at least)
      • did you get all your attempts correct?
  • In general, an actual accuracy of 80–90% would be considered really good.

In my experience, under regular test conditions, actual accuracy < perceived accuracy … always.

Alright. Now, let us consider those numbers again. 15/15 and 20/25.

Say you start answering questions with a mindset that it is great to be at 80% accuracy. What this means is, you are not answering only those questions that you are sure of, but also questions on which you feel you have a 70–80% probability of getting the answer correct.

Therefore, you are making your own perceived accuracy lower ~80%. We know for sure the actual accuracy will be lower than perceived accuracy. So, your actual accuracy would drop further down. And while you thought you’d get 20/25 correct, you end up getting 16/25 correct, taking your score to 39.

Of course, these numbers are just to illustrate my point. My strategy while taking the CAT (and to all my students subsequently) has been to answer only those questions that you are sure of. It served me well, and has been serving my students well too.

All the best!


Anish Passi is the founder of Test Cafe. With 99th percentiles in both the GMAT (760/ 800) and CAT (99.55 percentile), Anish has a keen understanding of how aptitude tests work. He has shared his conceptual and test taking expertise with students for over a decade, and has helped them master their tests, and shape their careers.

Channel Name