Tools for Smart Thinking by Richard E. Nisbett
At Altometrics we like to make tools, so the subtitle of this book caught my eye. If you have read Thinking, Fast and Slow this is a good followup book. The scope of the book is broader, but the author is thinking about thinking and using social research to get at how we really think vs. how we like to imagine we think.
The last section in each chapter is called “Summing Up” and restates each of the points made in the chapter. I found the entire book worth reading, but if you were in a hurry, you could just plow through these sections and get a pretty good idea of what you may want to go back and take more time with.
Part I: Thinking about Thought
(1) Everything is an Inference
(2) The Power of the Situation - Context affects our thoughts more than we think. Always remember people can change.
(3) The Rational Unconscious - You know how you often make up explanations for why you did something because you don’t really know why? Other people do that too. Our motivations are often not available to us but we are good at making them up when we need to.
Part II: The Formerly Dismal Science
(4) Should You Think Like an Economist? - Cost/Benefit Analyses are quite useful (one of the only thing Microeconomists agree on.)
(5) Spilt Milk and Free Lunch - Sunk Costs and Opportunity costs are worth paying attention to in your cost/benefit analyses
(6) Foiling Foibles - Loss aversion is powerful
Part III: Coding, Counting, Correlation, and Causality
(7) Odds and Ns - We tend to over-value our ability to spot good candidates in in-person interviews. This brief exposure is a small sample size.
(8) Linked Up - With a plausible description of why a correlation could imply a causation we are very likely to assume it does.
Part IV: Experiments
(9) Ignore the HiPPO - Experiments require random assignment to overcome biases.
(10) Experiments Natural and Experiments Proper - _ Natural experiments, which are sometimes the only possible type, can lead to Randomized-controlled experiments, which are the gold standard. Not carrying out experiments can be quite costly._
(11) Eekonomics - Multiple regression analyses can be useful, but are often misused and tend to be worse than natural experiments.
(12) Don’t Ask, Can’t Tell - Social research is difficult.
Part V: Thinking, Straight and Curved
(13) Logic - Flawless logic leads to validity, but not always to truth.
(14) Dialectical Reasoning - Primarily concerned with context and less analytical than logical reasoning
Part VI: Knowing the World
(15) KISS and Tell - Simplicity is excellent. Falsifiability is a virtue, but confirmability is even more important.
(16) Keeping It Real - Science occasionally progresses through faith and hunches.