1. Take research findings in psychology, sociology, etc., all fields which are having big replication problems.
2. Pick out findings that are surprising enough to write a pop-science book about. These are less likely to be true -- that's why they're surprising.
3. Oversimplify for popular consumption, sound very confident, and omit things that don't fit the story. (This makes for a much better read, as epistemically sinful as it may be.)
Each step makes good commercial sense, and I can't hate him for it, but the resulting books are unlikely to be Actually True.