Reviews
This book is written for scientists, which I'm not. However, I found it really interesting and applicable to what I've done with statistics and data analysis in the marketing world. Software is making it a lot easier for a marketer to become an armchair statistician, and there are dangers lurking in that space. It's really easy to get cynical about all data analysis after reading this. To me, it reiterated that data analysis can not stand alone outside of business sense and subject matter expertise. My biggest takeaways are: - Bring plenty of humility into any serious data analysis effort. Nobody wants to think that their hard work was all for naught, and we can be pretty innovative at finding positive conclusions where non really exist - Statistics are not necessarily objective. Fair people with the right incentives can disagree on methods and therefore results. - A good statistical analysis requires a lot of subject matter expertise in whatever you're studying in addition to statistics. This seems obvious, but you can just "look at the data" and figure things out. - Most of the businesses I've worked with don't have nearly enough data to support any sort of rigorous statistical analyses, especially when you start slicing and dicing the data. - Stop reading news articles about scientific studies. Not sure I needed a reminder on this one. Specific things I liked: List of relevant questions for any analysis: 1) What do I measure? 2) Which variables do I adjust for? 3) Which cases do I exclude? 4) How do I define groups? 5) What about missing data? 6) How much data should I collect? Reminder about Hanlon's Razor: "Never attribute to malice that which is adequately explained by incompetence," "Misconceptions are like cockroaches: you have no idea where they came from, but they're everywhere - often where you don't expect them - and they're impervious to nuclear weapons."
