This book is largely about measurement. The claim is that "just plain data analysis" is looking at data without statistics and formulas and math. So, instead, we get a lot on reliability and validity of measurements at the beginning of the book, a chapter on fallacies, with a case study on the crime rate in New York City (I don't know why this particular case study was not put towards the end of the book like the other case studies), a couple of chapters on presenting data (either in tables or in graphs), and three more in-depth case studies on voting, educational achievement, and poverty and inequality) at the end.

As you can see, it's a bit all over the place but the main point is about exploring data, but the main takeaway is that social things can be measured in all sorts of different ways and you need to gather all of these different ways before drawing conclusions.

There are a few things I did not like: (1) the author takes monumental pain in making sure he's "fair and balanced". Every item, every conclusion that might seem to lean on political way is instantly balanced by an item or conclusion going the other way. (2) Good grief, too much reliance on Freakanomics. The chapters on data presentation are actually bad at presenting data. All the graphs were produced in Excel, when there are so many good graphing software available at this point.

So, it's not a bad book, but it's just not all that interesting and new.