“The logic of science is the logic of business and life” -- quote from philosopher John Stuart Mill -- is the opening salvo of Richard Nisbett's new book, "Mindware: Tools for Smart Thinking"
Alas, the human species is generally not very “logic of science” friendly, hence the motivation for professor Nisbett's approachable tome to advance a clear-thinking, evidence-responsive populace.
Mindware isn't the first social science book to address human weaknesses in scientific thinking. Indeed, in the ten years I've been writing for Information Management, the topic of sound and unsound thinking is right at the top of my blog list. Among the outstanding books on smart and not-so-smart cogitation I've covered over the years are the following. I'm sure there are probably a few I've overlooked.
Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets
The Black Swan: The Impact of the Highly Improbable
Thinking Fast and Slow
Misbehaving: The Makings of Behavioral Economics
Everything Is Obvious: *Once You Know the Answer
The Halo Effect...and the Eight Other Business Delusions That Deceive Managers
The Drunkard's Walk: How Randomness Rules Our Lives
Superforecasting: The Art and Science of Prediction
Super Crunchers: Why Thinking-By-Numbers is the New Way to Be Smart
Stumbling on Happiness
A recurring caveat: we're often guilty of the fundamental attribution error in which our assessments of others' behavior are overly influenced by small-N, dispositional factors to the detriment of critical situational variables. An example – job hiring, where a 30 minute interview might influence a go/no-go decision more than 10 years of demonstrated, pertinent employment success.
With the exception of perhaps only the brilliant Thinking Fast and Slow, Mindware is the most ambitious of my smart thinking reads. In his quest to make readers better thinkers, psychologist Nisbett searches far and wide, finding help in psychoanalytic, developmental, social, experimental, and cognitive psychology; classical and behavioral economics; statistical methods that include correlation, regression, and designed experiments; and the philosophies surrounding causality, logic, dialectics, and knowledge.
The chapters on economics and statistics are well tread. Like many psychologists, Nisbett seems pleased with the emergence of psychology-founded behavioral economics at the expense of traditional utility-maximizing micro driven by homo economicus. And he pulls no punches, unfavorably contrasting the observational, regression-driven methods of microeconomics with the randomized experimental approach of cognitive and social psychology, taking Freaknomics author Steven Levitt to task for his reliance on correlational, observation methods.
One nugget of statistical wisdom with experiments especially hit home. 'When you assign each case to all of the possible treatments, your design is more sensitive. That is to say, a difference of a given magnitude found by a “within design” is more likely to be statistically significant when tested in a “between” design. That's because all the possible differences between any two cases have been controlled away, leaving only the treatment difference as the possible cause of the relationship.' Kind of reminds me of the “experiment” on the prowess of legendary Alabama football coach Bear Bryant. “He'll take his'n and beat your'n, then take your'n and beat his'n”
I have a few minor differences with Mindware on statistical terminology. For Nisbett, the experiment with random assignment to treatment/control is the gold standard design methodology, much preferred over observational methods where “selection” to treatment groups is biased. His natural experiments are what I call quasi-experiments – not quite the cachet of randomized experiments, but higher design validity than simple observational studies. My natural experiment is often referred to as an accidental experiment, where randomization is introduced unobtrusively. Lottery-based assignment to charter schools is an example.
Also, I see multiple regression analysis (MRA) as a statistical technique that's used for both experimental and non-experimental investigations. The design behind the investigation – randomized experiment, quasi-experiment, simple observational study – determines how strongly the results of the MRA can be interpreted.
Not as provocative as Fooled By Randomness, not as purely entertaining as The Drunkard's Walk, not as awe-inspiring as Thinking Fast and Slow, and not as spot-on business pertinent as The Halo Effect, Mindware: Tools for Smart Thinking is nonetheless an important thinking book which has earned a spot on my platinum list. I recommend it to both business analytics professionals and data scientists.