Sensitivity Analysis for Business, Technology, and Policymaking. Made Easy with Simulation Decomposition (SimDec)

What if your model is trying to tell you something, but all you see is a bland histogram? You run a simulation. You get a distribution. Fine. But why does it look like that? Which inputs are actually driving the weird bump on the right? And which combinations of factors push you into the danger zone? This book began with exactly that frustration: staring at an output distribution that was anything but “normal” and wondering how to make sense of it without poking inputs one by one like you’re testing a stove burner. The answer turned out to be disarmingly simple: take one simulation, decompose it into meaningful scenarios, and colour them directly on the histogram. “It’s just a coloured histogram,” reviewers kept criticizing. And yet, that simple move makes tangled, multidimensional interactions suddenly visible.

So is this a statistics book? A modelling book? A manifesto? A bit of all three. SimDec unifies uncertainty analysis and global sensitivity analysis, but without demanding that you join the priesthood of advanced mathematics. A straightforward binning approach quantifies which variables matter most, and the visualization shows how they interact. Andrea Saltelli, often called the godfather of global sensitivity analysis, described this accessibility as a way to “democratize sensitivity analysis.” That’s the spirit of the book: shake the foundations of how we “shake the foundations of science,” and make powerful tools usable by real decision-makers.

And it doesn’t stay in theory. The chapters travel across finance, entrepreneurship, energy, public policy, engineering, even superconducting magnet design. SimDec has been used to simplify overcomplicated models, to add necessary complexity to oversimplified ones, and sometimes to inspire modelling where none existed before. In the afterword, Sam Savage imagines putting a histogram on a psychiatrist’s couch to ask: what shaped your “personality”? SimDec, he notes, lets us see those multiple personalities—those coloured partitions—clearly. The result? Not just better analysis, but better questions. What drives outcomes? What can we control? And what combinations actually get us where we want to go?

If you’ve ever felt that your model knows more than it’s telling you, this book shows you how to make it speak.