other_methods.qmd

---
title: "Non Experimental Methods"
share:
  permalink: "https://book.martinez.fyi/other_methods.html"
  description: "Business Data Science: What Does it Mean to Be Data-Driven?"
  linkedin: true
  email: true
  mastodon: true
---

In an ideal world, every causal question in business and technology could be
answered through carefully designed randomized controlled trials (RCTs).
However, the reality of decision-making in these fields often precludes such
luxury. Budget constraints, ethical considerations, logistical challenges, or
simply the rapid pace of technological change frequently render experimental
approaches impractical or impossible. This is where non-experimental methods
come into play, offering powerful tools to infer causality from observational
data.

This section delves into a suite of sophisticated techniques designed to
approximate experimental conditions using data that wasn't generated through
randomized experiments. Each of these methods comes with its own set of
assumptions, strengths, and limitations. We'll discuss not only how to implement
these techniques but also how to critically evaluate their applicability to your
specific business context.

Remember, while these methods can be incredibly useful, they are not magical
solutions. The key to their successful application lies in a deep understanding
of the underlying causal mechanisms at play in your business scenario, careful
consideration of potential confounders, and a healthy dose of skepticism in
interpreting results.

As we navigate through these methods, we'll emphasize the importance of
sensitivity analyses, robustness checks, and transparent reporting of
assumptions. By the end of this section, you'll be equipped with a powerful
toolkit for causal inference in non-experimental settings, enabling you to make
more informed decisions even when randomized experiments are out of reach.

Let's embark on this journey to unlock the causal insights hidden within your
observational data!