“I know just enough statistics to be dangerous.”
That’s what numerous engineers tell me. But of course the truly dangerous engineers would never say this. The dangerous ones blaze ahead and never look back. The engineers who say they know just enough to be dangerous are actually the ones who know enough not to be dangerous.
One way I help engineers is to analyze data. Given a set of measurements, what can we conclude? Are there any subtle traps to avoid? How much confidence can we have in our final answer? Sometimes engineers ask me to sign off on an analysis they’ve already done, just as they’d have another engineer review their engineering work.
Another way I help is to save money by determining what data to collect and how best to collect it. You don’t want to run an experiment only to find out you collected more data than necessary, or that you collected the wrong data. Getting the experimental design right saves resources.
Backing up even further, I sometimes help clients think about what kinds things they may be able to learn from their data. Here’s this expensive thing we measure. Is there some way to accurately infer it from things that are less expensive to measure? Is there something useful we could do with data we’re collecting as a side-effect of something else we’re interested in?