Your Toughest Data Questions Answered: Live Session on Selection on Observables & RDD Basics.
- Maria Alice Maia
- Dec 2, 2024
- 2 min read
Over the last few weeks, we've assembled a powerful toolkit for causal inference from observational data: Covariate Adjustment, Propensity Scores, and Doubly Robust Estimation.
But having a toolkit and knowing which tool to use are two very different things. This is where theory meets the messy reality of your day-to-day work, and it's where many data initiatives stall.
The persistent confusion I see is not about the complexity of the math, but about the strategy of the application. It’s the ultimate form of "Doing Data Wrong" for a data professional: having the right tools but choosing the wrong one for the job.
So, let's move from theory to practice. This week, I’m hosting another live Q&A session to tackle your toughest questions on method selection.
Bring me your real-world analytical dilemmas. Let's workshop them. To get the discussion started, here are the kinds of questions we should be wrestling with:
"I have a dataset with many potential confounders. How do I decide whether to use propensity score matching or a direct covariate adjustment in a regression model? What are the trade-offs?"
"My propensity score model has some poor overlap (scores close to 0 or 1), making my weighted estimates unstable. Is it safer to trust a well-specified outcome regression, or is this the exact scenario where Doubly Robust Estimation is critical?"
"We give a performance bonus to every employee who scores exactly a 95 or higher on their annual review. How can I measure the true causal impact of that bonus on their engagement in the following quarter?"
It's your turn now.
Bring your most challenging question to this week's call, on December, 5th on 1p.m. EST / 10a.m. PST. If you are not subscribed to our newsletter, subscribe by filling the box in the footer and receive the link to the call.
I will answer all the questions raised during the 1-hour call, where we'll also start to touch on the basics of Regression Discontinuity Design (RDD).
My entire career, from my academic research to my executive roles at companies like Itaú and Stone, has been about one thing: choosing the right tool for the right problem to drive a real business outcome. My mission is to help you build that same decision-making muscle. This knowledge is not mine to keep.
Let's build a community that doesn't just know the methods but understands the strategy behind them.
If you’re ready to move from a list of techniques to a real analytical framework, the best way to continue this journey is by joining my email list.
