Spurious Correlations

by Tyler Vigen

"Spurious Correlations ... is the most fun you'll ever have with graphs." -- Bustle Military intelligence analyst and Harvard Law student Tyler Vigen illustrates the golden rule that "correlation does not equal causation" through hilarious graphs inspired by his viral website.Is there a correlation between Nic Cage films and swimming pool accidents? What about beef consumption and people getting struck by lightning? Absolutely not. But that hasn't stopped millions of people from going to tylervigen.com and asking, "Wait, what?" Vigen has designed software that scours enormous data sets to find unlikely statistical correlations. He began pulling the funniest ones for his website and has since gained millions of views, hundreds of thousands of likes, and tons of media coverage. Subversive and clever, Spurious Correlations is geek humor at its finest, nailing our obsession with data and conspiracy theory.

Genres: Nonfiction, Science, Humor, Mathematics, Sociology, Popular Science, Comedy

197 pages, Kindle Edition

First published May 1, 2015

Reviews

[Lewis · Rating 3 out of 5]

Short but entertaining. By all means, take the time to read the author's comments. It is also useful to expand the graphs and compare the scales for each item. There are often really huge differences in scale on the Y-axis. That reminded me of a class I took where we deconstructed papers to look for scaling errors, inadequate sample sizes, and assumptions which often drew conclusions which at least demanded non-parametric statistics. I enjoyed the book as light reading with some amusing correlations and at least a few which hinted at a higher order causality never explored.

[Pep Bonet · Rating 4 out of 5]

Funny and useless. Fine exercise of data dredging to find spurious correlations between things with nothing in common. Want to teach what correlations are and why correlation doesn't mean causation? Read some pages from the book.
Put otherwise, enjoy what great computing power can do to your ignorant believes.

[Nathan Albright · Rating 5 out of 5]

This is an entertaining book, but it is a book that carries with it a big point for those who are interested in aggregating data trends and taking advantage of the mass of data that is available in our contemporary age [1].  The subtitle of the book gives a truism that we would do well to remember, that correlation does not equal causation, and the book goes out to prove it.  The book is meant to be taken with a certain bit of humor about it, as the author demonstrates when he writes about how he should have been studying for his law school finals when he was writing this book, but many of us can certainly relate to writing when we should be doing something else.  At times the author even gets personal, such as when he pokes at law schools for the way that their professors outsource outlines and the writing of textbooks based on the decisions of judges but still collect the royalties anyway, something that I find funny as a person who has had to buy a great many textbooks for large amounts of money that were not particularly original materials.

The roughly two hundred pages of this book, containing about half as many spurious correlations, are divided thematically into ingestibles (ones relating to food), science-related ones, cultural curiosities, movers, shakers, and moneymakers, and famous folks.  The spurious correlations chosen demonstrate the power of data aggregation and the dangers of using only a few data points, which can lead to very high correlations for entirely unrelated phenomena.  The various correlations chosen reveal some of the interests of the author, from UFO sightings to marriage and divorce rates to films and popular culture.  Each of the spurious correlations chosen is given a rather rudimentary graph that includes two different scales for the axes for the two phenomena that correlate and the years over which the data is taken.  The author makes some witty comment about the correlation and includes notes about where the data come from, some of which have highly entertaining comments of their own on the pages underneath the graphs.  The graphs, rather tellingly, do not tend to be scaled to zero, which makes the apparent correlations greater than they would be normally, and this subtle distortion has a point.

Overall, this is a short book that manages the impressive feat of being both deeply entertaining as well as deeply informative.  These are not qualities that are easy to combine, not least for someone who appears to be a bit of a novice when it comes to writing books, but the author manages this task with aplomb.  By showing a book full of dodgy data dredging in order to find some particularly wacky correlations, the author aims to disabuse readers of the intuition that phenomena that correlate to each other are necessarily connected to each other in some sort of causal fashion, especially because the ease of making connections with data is likely only to increase the amount of spurious connections over few data points that are made.  Since this world is full of people who abuse data and make apparent connections using the same techniques that this author uses humorously in this book, the book therefore serves as a sort of inoculation against bad data techniques, something this author likely does with a high degree of intentionality.  Even as a reader who recognizes the author's agenda, there is a great deal of humor in the book and the author is performing a valuable service to an audience that may not be all that aware of the issues of big data in our present evil age.

[1] See, for example:

https://edgeinducedcohesion.blog/2015...

https://edgeinducedcohesion.blog/2015...

https://edgeinducedcohesion.blog/2015...

https://edgeinducedcohesion.blog/2015...

https://edgeinducedcohesion.blog/2017...

[Nic · Rating 2 out of 5]

On paper, this is exactly the type of book I’d normally love. It just falls incredibly flat from the fun premise.

The concept of showing how people can use garbage statistics to try to skew results is good. What I didn’t care for was the fact that each pair of stats in a chart were presented and a description just sort of vaguely gestured at the chart.

It almost reads like a scrapbook: pithy comment, title of two different stats being presented, chart, personal comment about said chart, website where charts were found. Repeat.

TW: I’m never a fan of making light of su*cide and it happens multiple times in this book. Flippantly pointing at su*cide is pretty gross.

[Chrisanne · Rating 3 out of 5]

Of all the years to read this it had to be the most conspiracy- theory- filled one I can remember. Appropriate.

Thank heavens Chris Evans is rumored to be retiring. 😉

[Sharon · Rating 5 out of 5]

Just the light-hearted read that I needed!

Author Tyler Vigen shows how it's possible to use real statistics to reach completely spurious conclusions (usually by manipulating the Y-axis). The correlations are hilarious ... and you won't want to skip the footnotes.

[jess ~has abandoned GR~ · Rating 5 out of 5]

I. love. this. stuff.

[Danilo Flechaz Muñoz · Rating 3 out of 5]

Un librito curioso que muestra muchos ejemplos de correlación espuria. Didáctico y entretenido, aunque repetitivo.

[Jeff Duda · Rating 2 out of 5]

I'd like to give the book a higher rating based on the innovative topic and approach, but I just can't; I feel the book wasn't done as well as it could have been.

Many of the time series (by the way...what about non-time-series spurious correlations?) had very short records...many were plotted using as few as 6 data points, which really casts doubt on the legitimacy of some of them. I was under the impression we would see nothing but high-sample data sets that featured frivolous connections.

However, some of the particular comparisons made me laugh (like "deaths caused by a cataclysmic storm" vs. "Greatest number of 'fuck's in a movie" 😂), and caused me to look them up myself, which led me down interesting paths of discovery. I also appreciate the author's sense of humor. These aspects definitely improved the book.

[Kam Yung Soh · Rating 3 out of 5]

A hilarious book full of graphs that show that "correlation does not equal causation" by providing graphs filled with rather ridiculous correlations. The book has sections comparing food and drinks, science and technology, culture, personalities and others correlated with rather ridiculous things (most famously, Nicholas Cage film appearances against people drowning in swimming pools).

What the graphs show is that if you dig through enough time related data and compare them with other data, you will most likely discover that they are correlated, even though they should have nothing to do with each other. That is a danger that people should be aware of when shown graphs about apparently correlated data without the required causation. And the ridiculous examples of correlations in this book serve to highlight the danger.

[Allie · Rating 3 out of 5]

This is the kind of thing that's funnier in small doses, and as a book it's kind of overkill. The sections I liked best were cultural curiosities and famous folks because those were often the oddest, least related pairings (i.e. deaths caused by cataclysmic storm vs. greatest number of f*cks in a movie or Bruce Willis film appearances vs. people killed by an exploding boiler). I really liked the spurious correlations that appeared in The Best American Infographics 2015, and there were only about 8 featured there which is a perfectly clever amount.

[Mark · Rating 2 out of 5]

This is cute, but you shouldn't buy it. The concept (or lack of one) is well known, but it isn't really explained here. It's just a collection of funny examples of computer-found data from two things that have nothing to do with each other, but are highly correlated. An example is the number of movies that Nicolas Cage has appeared in each year vs the number of people who drowned in a swimming pool that year. Also, the author only very briefly mentions that the reason he was able to graph these data together is that he used two separate y axes and scaled the data to line the curves up. He doesn't mention it, but there is always a better way to show data than to graph two data sets on one graph with two y axes. When you see graphs like these, you should always suspect ignorance or foul play, even if they seem reasonable.
======
There is an R library that contains a pseudocorrelation function that will let you make these graphs yourself when things are slow at work.

[James]

Military intelligence analyst and Harvard Law student Tyler Vigen illustrates the golden rule that "correlation does not equal causation" through hilarious graphs inspired by his viral website.Is there a correlation between Nic Cage films and swimming pool accidents? What about beef consumption and people getting struck by lightning? Absolutely not. But that hasn't stopped millions of people from going to tylervigen.com and asking, "Wait, what?" Vigen has designed software that scours enormous data sets to find unlikely statistical correlations. He began pulling the funniest ones for his website and has since gained millions of views, hundreds of thousands of likes, and tons of media coverage. Subversive and clever, Spurious Correlations is geek humor at its finest, nailing our obsession with data and conspiracy theory.

[Kaitlin · Rating 3 out of 5]

Actual Rating: 3.5/5

I was mostly reading this as a way to pass a little while, and partially as "research" for my volunteering, to give me examples of ways in which patterns can be manipulated or correlations presented that don't have a causation relationship. It did its job! There are very ridiculous graphs in there, and in the captions or citations Vigen often points out how things were slightly manipulated to create a different picture. He tells you explicitly to look at the y-axis, and when a trend is repeated will suggest that you look back to see how different the graphs look with different y-axes. And then he's pretty funny too.

Overall, I enjoyed it, and it did give me some good correlations to remember.

[Alan Earhart · Rating 3 out of 5]

A nice coffee table book! Set it out and let visitors thumb through it and giggle. All while learning at the same time.

The author has a website where you can go and search through many, many different strange correlations, but the book adds interesting comments on each graph.

If possible, browse the Kindle version if that's the version you want. Kindle versions sometimes barf on books with extensive use of graphics.

[Christopher Avila · Rating 4 out of 5]

Helps understand correlation vs causality

Funny and easy to read, some of the correlations make you think on causality :) ... Good to show students difference between correlation and causality.

[Jo · Rating 4 out of 5]

Entertaining and a great resource for lecturers and teachers to give examples of spurious correlations to students learning about statistics. Not a "sit down and read solidly" sort of a book, better for dipping in and out of.

[Jorge · Rating 4 out of 5]

Terrible simple: bunch of graphs of unrelated variables with high correlation. The silliness is very funny, for instance, number of Nicolas Cage movies correlates very well with deaths by drowning. It is excellent as a source of examples of how correlation is NOT causation.

[Jo Pisa · Rating 4 out of 5]

Illuminante e divertente! Alcuni commenti sono davvero geniali. Si alternano a curiosità carine e schemi veramente inaspettati che trasmettono con chiarezza il messaggio: correlazione non significa causazione.

[Stephen · Rating 4 out of 5]

Two key takeaways from this short but entertaining book:
1. Does correlation imply causation? It's intuitive, but it's not always true.
2. Provided enough data, it is possible to find things that correlate even when they shouldn't.

[Darth JarJar · Rating 3 out of 5]

OK. To be fair, I just skimmed this book. A lot of cool charts, but nothing spectacular. I want to know how he wrote the code to identify these correlations. Getting data from many different sources seems like a nightmare.

[Jocelyn · Rating 2 out of 5]

So-so, some interesting bits of facts scattered around. The text for each comparison chart is not that funny though... I feel that was a wasted opportunity to inject some humor and wit in there.

[Faith · Rating 2 out of 5]

It was ok, but not as funny in this format as it is just seeing a few of the graphs online.

[Sarah · Rating 3 out of 5]

Entertaining, a book to read when you have a few minutes here and there.

[Josh Laird · Rating 3 out of 5]

Correlation does not always equal causation

[Miroslav Nemčok · Rating 3 out of 5]

Quite funny from the beginning, gets boring after a while, becomes annoying if you’re strong enough to make it until the end.

[Hans van der Veeke · Rating 2 out of 5]

At first these random correlations are funny but after a while they get a bit tedious and boring.

[JOHN CHAN · Rating 4 out of 5]

Interesting and quick read

Interesting and quick read. But u may need to cultural context to enjoy it. Good read and recommend for leisure read.

[Alexander Van Leadam · Rating 3 out of 5]

Good fun with statistics and correlations, quite useful for making people think twice about perfect curves and other persuasive graphs.

[Adrain · Rating 5 out of 5]

Very interesting and entertaining. Funny, while educational.