Everybody Lies

✍️ Seth Stephens-Davidowitz

Tags: seth-stephens-davidowitz , data-science , lang-en

This is a 3.5 star review (rounded up to 4 because of all the fascinating examples)

Recently a lot of books about Big Data have been published and “Everybody Lies” is one of them. First disclaimer is that the title is misleading. Only a small subset of the book is devoted to that specific topic. The book itself is full of interesting stories from horse race analytics and predicting election results based on Google searches to trying to infer how the economy is doing based on satellite photos.

The book has two interesting takeaways. First, with Big Data we can zoom in into the data and analyze it from multiple points of views and gather insights that were not possible with a small sample sizes. One example is the Raj Chetty study based on IRS tax records (Here, I really don’t know how is it possible to get every American tax record, even if they are ‘anonymized’) . Second, is how the internet searches can be seen as a digital truth serum (the complete opposite happens with our public social media posts). We share our inner thoughts without fear of judgement. The author shows that this kind of data sometimes can be more useful and truthful than surveys.

The cons:
- I haven’t read the scientific papers to check for the technical soundness of the methods, but from the way the book is written sometimes it gives the impression that the author has all the conclusions beforehand and tries to dig the data searching for the evidence to back them up.
- The author has a ‘Freakonomics-snarky’ writing style which can be a turn off for some readers.