spurious correlations

correlation is not causation

discover · random · spurious scholar
← previous page · next page →

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is The number of movies Nicolas Cage appeared in and the second variable is The number of transportation security screeners in North Dakota. The chart goes from 2012 to 2022, and the two variables track closely in value over that time.

View details about correlation #5,837

^{Show scatterplot}

What else correlates?
The number of movies Nicolas Cage appeared in · all films & actors
The number of transportation security screeners in North Dakota · all cccupations

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Google searches for 'im not even mad' and the second variable is Popularity of the 'whip nae nae' meme. The chart goes from 2015 to 2023, and the two variables track closely in value over that time.

View details about correlation #5,952

^{Show scatterplot}

What else correlates?
Google searches for 'im not even mad' · all google searches
Popularity of the 'whip nae nae' meme · all memes

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Popularity of the first name Harmony and the second variable is Associates degrees awarded in Agriculture and natural resources. The chart goes from 2011 to 2021, and the two variables track closely in value over that time.

View details about correlation #2,129

^{Show scatterplot}

What else correlates?
Popularity of the first name Harmony · all first names
Associates degrees awarded in Agriculture and natural resources · all education

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Bloomberg Money Stuff articles about insider trading and the second variable is Banco Bilbao Vizcaya Argentaria's stock price (BBVA). The chart goes from 2014 to 2023, and the two variables track closely in value over that time.

View details about correlation #5,892

^{Show scatterplot}

What else correlates?
Bloomberg Money Stuff articles about insider trading · all weird & wacky
Banco Bilbao Vizcaya Argentaria's stock price (BBVA) · all stocks

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Butter consumption and the second variable is Ticket prices at North American movie theaters. The chart goes from 2001 to 2021, and the two variables track closely in value over that time.

View details about correlation #1,604

^{Show scatterplot}

What else correlates?
Butter consumption · all food
Ticket prices at North American movie theaters · all films & actors

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Popularity of the 'like a boss' meme and the second variable is Google searches for 'how to cut own hair'. The chart goes from 2006 to 2023, and the two variables track closely in value over that time.

View details about correlation #4,994

^{Show scatterplot}

What else correlates?
Popularity of the 'like a boss' meme · all memes
Google searches for 'how to cut own hair' · all google searches

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Bachelor's degrees awarded in interdisciplinary studies and the second variable is Electricity generation in Angola. The chart goes from 2012 to 2021, and the two variables track closely in value over that time.

View details about correlation #2,666

^{Show scatterplot}

What else correlates?
Bachelor's degrees awarded in interdisciplinary studies · all education
Electricity generation in Angola · all energy

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Popularity of the first name Ariel and the second variable is The number of fashion designers in Michigan. The chart goes from 2003 to 2020, and the two variables track closely in value over that time.

View details about correlation #2,985

^{Show scatterplot}

What else correlates?
Popularity of the first name Ariel · all first names
The number of fashion designers in Michigan · all cccupations

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Number of competing nations in the Summer Olympics and the second variable is Nielsen Ranking of Smallville Season Finale. The chart goes from 2002 to 2011, and the two variables track closely in value over that time.

View details about correlation #2,648

^{Show scatterplot}

What else correlates?
Number of competing nations in the Summer Olympics · all weird & wacky
Nielsen Ranking of Smallville Season Finale · all weird & wacky

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Air pollution in Appleton, Wisconsin and the second variable is Google searches for 'ice bath'. The chart goes from 2004 to 2023, and the two variables track closely in value over that time.

View details about correlation #2,867

^{Show scatterplot}

What else correlates?
Air pollution in Appleton, Wisconsin · all weather
Google searches for 'ice bath' · all google searches

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Popularity of the 'we live in a society' meme and the second variable is Wind power generated in Namibia. The chart goes from 2006 to 2021, and the two variables track closely in value over that time.

View details about correlation #4,950

^{Show scatterplot}

What else correlates?
Popularity of the 'we live in a society' meme · all memes
Wind power generated in Namibia · all energy

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Robberies in New Mexico and the second variable is Car crashes in the US. The chart goes from 1991 to 2014, and the two variables track closely in value over that time.

View details about correlation #3,981

^{Show scatterplot}

What else correlates?
Robberies in New Mexico · all random state specific
Car crashes in the US · all weird & wacky

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Votes for Democratic Senators in New Jersey and the second variable is Searches for 'never gonna give you up'. The chart goes from 2006 to 2020, and the two variables track closely in value over that time.

View details about correlation #5,223

^{Show scatterplot}

What else correlates?
Votes for Democratic Senators in New Jersey · all elections
Searches for 'never gonna give you up' · all memes

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Air quality in Vineland, New Jersey and the second variable is Warner Bros. Discovery's stock price (WBD). The chart goes from 2006 to 2023, and the two variables track closely in value over that time.

View details about correlation #2,929

^{Show scatterplot}

What else correlates?
Air quality in Vineland, New Jersey · all weather
Warner Bros. Discovery's stock price (WBD) · all stocks

A linear line chart with years as the X-axis and two variables on the Y-axis. The first variable is Popularity of the first name Dexter and the second variable is Google searches for 'bing'. The chart goes from 2004 to 2022, and the two variables track closely in value over that time.

View details about correlation #5,230

^{Show scatterplot}

What else correlates?
Popularity of the first name Dexter · all first names
Google searches for 'bing' · all google searches

← Previous page
Discover a new correlation
View random correlation
View all correlations
View all research papers
Get permission to re-use these charts

Why this works

Data dredging: I have 25,237 variables in my database. I compare all these variables against each other to find ones that randomly match up. That's 636,906,169 correlation calculations! This is called “data dredging.”^Note Fun fact: the chart used on the wikipedia page to demonstrate data dredging is also from me. I've been being naughty with data since 2014.
Instead of starting with a hypothesis and testing it, I instead tossed a bunch of data in a blender to see what correlations would shake out. It’s a dangerous way to go about analysis, because any sufficiently large dataset will yield strong correlations completely at random.
Lack of causal connection: There is probably no direct connection between these variables, despite what the AI says above.^Note Because these pages are automatically generated, it's possible that the two variables you are viewing are in fact causually related. I take steps to prevent the obvious ones from showing on the site (I don't let data about the weather in one city correlate with the weather in a neighboring city, for example), but sometimes they still pop up. If they are related, cool! You found a loophole.
This is exacerbated by the fact that I used "Years" as the base variable. Lots of things happen in a year that are not related to each other! Most studies would use something like "one person" in stead of "one year" to be the "thing" studied.
Observations not independent: For many variables, sequential years are not independent of each other. You will often see trend-lines form. If a population of people is continuously doing something every day, there is no reason to think they would suddenly change how they are doing that thing on January 1. A naive p-value calculation does not take this into account.^Note You will calculate a lower chance of "randomly" achieving the result than represents reality.

To be more specific: p-value tests are probability values, where you are calculating the probability of achieving a result at least as extreme as you found completely by chance. When calculating a p-value, you need to assert how many "degrees of freedom" your variable has. I count each year (minus one) as a "degree of freedom," but this is misleading for continuous variables.

This kind of thing can creep up on you pretty easily when using p-values, which is why it's best to take it as "one of many" inputs that help you assess the results of your analysis.
Y-axes doesn't start at zero: I truncated the Y-axes of the graphs above. I also used a line graph, which makes the visual connection stand out more than it deserves. ^Note Nothing against line graphs. They are great at telling a story when you have linear data! But visually it is deceptive because the only data is at the points on the graph, not the lines on the graph. In between each point, the data could have been doing anything. Like going for a random walk by itself!
Mathematically what I showed is true, but it is intentionally misleading. If you click on any of the charts that abuse this, you can scroll down to see a version that starts at zero.
Confounding variable: Confounding variables (like global pandemics) will cause two variables to look connected when in fact a "sneaky third" variable is influencing both of them behind the scenes.
Outliers: Some datasets here have outliers which drag up the correlation.^Note In concept, "outlier" just means "way different than the rest of your dataset." When calculating a correlation like this, they are particularly impactful because a single outlier can substantially increase your correlation.

Because this page is automatically generated, I don't know whether any of the charts displayed on it have outliers. I'm just a footnote. ¯\_(ツ)_/¯
I intentionally mishandeled outliers, which makes the correlation look extra strong.
Low n: There are not many data points included in some of these charts.^Note You can do analyses with low ns! But you shouldn't data dredge with a low n.
Even if the p-value is high, we should be suspicious of using so few datapoints in a correlation.

Pro-tip: click on any correlation to see:

Detailed data sources
Prompts for the AI-generated content
Explanations of each of the calculations (correlation, p-value)
Python code to calculate it yourself

Project by Tyler Vigen
emailme@tylervigen.com · about · subscribe

CC BY 4.0