Causal inference is the set of ideas and methods we use to answer questions like: If we change X, what happens to Y? That sounds simple, but most data only shows what happened in the past — not what would have happened under a different choice.

Standard statistics often describes associations: patients on drug A have lower blood sugar on average than patients not on drug A. Causal inference asks a harder question: Did the drug cause the improvement, or would those patients have improved anyway because of age, diet, or severity?

The goal is not perfect certainty. The goal is disciplined reasoning: state the question clearly, make assumptions explicit, choose a method that matches your data and design, and check how fragile the answer is.

The counterfactual idea

At the heart of causality is the counterfactual — the unobserved alternative. For each person or unit, we imagine two worlds: one where they received the treatment and one where they did not. The causal effect is the difference between those worlds.

We never observe both worlds for the same unit. Causal methods use design (randomization), structure (graphs), or modeling (adjustment, matching) to approximate that comparison from observable data.

Why tools like CausalLens exist

Modern causal inference combines statistics, econometrics, and computer science. Methods such as propensity score matching, difference-in-differences, and causal forests each address different data structures.

Software lowers the barrier: you upload a dataset, specify treatment and outcome, and compare multiple methods with balance checks and plain-English summaries — without writing custom code for every study.