Two Right Answers by Ridhi

Two Right Answers by Ridhi - July 2026 Scholarship Essay

The moment that really challenged me was in STA 301 at UT Austin, during the unit on Monte Carlo methods. We were working through a problem that didn't have an easy shortcut, the kind where you're tempted to just simulate it out instead of solving it by hand. A classmate and I started going back and forth in the middle of class because we both thought the other one was wrong. I ran mine as a simulation, generating a bunch of random trials and letting the results settle toward an answer. She solved it the analytical way, working through the actual probability distribution step by step. Our answers matched almost exactly, but the way we got there looked completely different, so neither of us really trusted it until someone told us who was right.

We waited for the professor to tell us. She didn't. She said we basically proved the same thing two different ways. My simulation was just approximating the exact number her math had already found. Neither of us was more right than the other; we just used different tools to get to the same place, and the fact that they matched was kind of the whole point of the lesson.
That messed with me a little, because growing up I always thought there was one right way to do something, usually the most straightforward way, and anything else was either wrong or overcomplicated. That class taught me that being more complicated doesn't mean being less correct. A longer or messier way of solving something can still get you to the same truth as the simple way, and if I had just written off her method because it wasn't the fastest, I would've missed something real.

That idea came back again a semester later when I was working with Texas Fintech Collective on a client's retention problem. A teammate suggested the obvious fix, a tiered loyalty rewards program, which made sense because it's a proven approach and it works. It wasn't a bad idea. But I kept thinking back to that class, and it made me ask a different question. Instead of picking the one obvious solution, what if we looked at the different ways a user could actually churn or stay, kind of like how I looked at different simulated outcomes in stats.

I suggested we build a model that grouped users not just by basic demographics but by simulating how they'd likely behave under different scenarios, like getting a discount, getting a discount plus some kind of outreach, or getting nothing at all. Instead of using one fix for everyone, we could match different responses to different situations based on what the data said was actually most likely to work for that group. It wasn't the simplest idea. It was the more complicated one. But just like in that class, it didn't mean the simple idea was wrong. It just meant there was more than one real way to solve the problem, and putting both approaches together made our recommendation stronger than either one on its own.

That's the thing that's stuck with me since STA 301. I stopped assuming the first clean answer was the only right one, and I got a lot more comfortable sitting with two different approaches instead of needing one to win.