Creating a phony health scare with the power of statistical correlation

cell tower.jpg

In the United Kingdom, the more mobile phone towers a county has, the more babies are born there every year. In fact, for every extra cell phone tower beyond the average number, a county will see 17.6 more babies. Is this evidence that cell phone signals have some nefarious baby-making effect on the human body?

Nope. Instead, it's a simple example of why correlation and causation should never be mistaken for the same thing. Writes Matt Parker in the Gaurdian:

This was discovered by taking the publicly available data on the number of mobile phone masts in each county across the United Kingdom and then matching it against the live birth data for the same counties. When a regression line is calculated it has a "correlation coefficient" (a measure of how good the match is) of 98.1 out of 100. To be "statistically significant" a pattern in a dataset needs to be less than 5% likely to be found in random data (known as a "p-value"), and the masts-births correlation only has a 0.00003% probability of occurring by chance.

The match between mobile phone towers and birth rates is an extremely strong correlation and it is highly statistically significant. ... Mobile phone masts, however, have absolutely no bearing on the number of births. There is no causal link between the masts and the births despite the strong correlation. Both the number of mobile phone transmitters and the number of live births are linked to a third, independent factor: the local population size. As the population of an area goes up, so do both the number of mobile phone users and the number people giving birth.

But wait, there's more! To test the media's science savvy, Parker sent out a press release detailing the correlation he found—but without any information on the real cause of both factors. Interesting object lesson or in-real-life trolling? Can't it be both? The press release went out late last week. If you've spotted any stories about the fertility powers of cell phone towers, leave a link in the comments.

(Via Steve Silberman)

Image: Some rights reserved by smith


  1. Statistical correlations are like Zen meditations they make you slow up and think more deeply about the world.

    1. @Anon No2 – Read the BB article and note the name of the columnist mentioned (“Matt Parker in the Guardian”), then look at the article you linked to (hint “Matt Parker, Guardian”) and then read the WHOLE article you linked to (hint, it’s this one being referred to).

      You may now feel somewhat foolish.

  2. The Pastafarian faith (Flying Spagetti Monster) has been thoroughly shaken. “Global Avarage Temperature versus Number of Pirates” will never be taken seriously anymore now.

  3. Let’s correlate ice cream sales vs shark attack! It will show us that ice cream sales causes shark attack. LOL!

    1. You might be right on straight volume of ice cream sold (California being far and away the highest) but Alaskans eat more ice cream per capita than any other state. The more ice cream you eat, therefore, the more likely you are to be attacked by a bear, not a shark. But the more likely you are to be attacked by a bear, the less likely you are to file for bankruptcy, so it’s really a tough choice.

  4. The top 3 google results for me all deal with this test of false science.

    My take on it is that I think this would have been more effective if there was a news conference and witholding the information that this was a test until a few days after the news conference.

  5. I will hold by my theory that each and every cel phone above the statistical norm oversarturates communication factillitation causing an unwanted surplus of intimacy and precipitates out babies!


    Seriously, just want to point out the obverse that without other data to support the stated overpopulation causation is also suspect; any purely correlative association is purely inference no mattter how reasonable and seemingly true.

  6. I’d think that better cell coverage would lead to lower reproduction. Everyone would have something better to do.

  7. The p-value is evil and deserves to be put in the black hole of banned significance tests (with just about all the others).

    1. Nonsense. The p-value is an important tool. Like many tools, its value depends entirely on the user. Knowing whether or not a relationship has arisen out of chance is important. But equally important is understanding the difference between “probably not by chance” and whatever other horrible things it is used for in the media (and even occasionally in peer-reviewed research). This article is a great example – the relationship between cell towers and birthrate is almost certainly not from random chance and that is born out by the p-value being really, really low.

  8. Haven’t seen anything yet, but its bound to show up. News media in many places look for anything to fill space rather it’s been vetted or not.

  9. If you add in Britain’s binge-drinking issues and what drunk-dialing can lead to, cell phone reception may have more causitive power than you may think.

    1. Damn, beat me to it, kinda. Better cell coverage = easier hookups = more babies. Add in the splendid booze-to-your-door operation that is Dial-A-Drink, and there you go.

    2. I too Vote “Causation”.
      Late-night-booty-calls are far more behaviorally probable when all persons involved have personal cell phones
      (instead of shared household landlines)

  10. They have it all backwards: cell phone towers don’t make babies; babies make cell phone towers. It’s like how beavers build dams.

  11. There’s also a strong correlation between plastic eggcup usage and juvenile delinquency. Obviously some kind of chemical leaching into the eggs!

  12. While I love the idea of testing the vetting prowess of the media, I can’t help but feel that feeding news outlets deliberately incomplete information is a bad idea. Of course the science savvy of the media is lacking. The worst part though is that even if furnished with an explanation after the fact, there will be no follow up article. I vote for real world trolling. Everyone knows that major media outlets don’t understand correlation.

  13. I’ve been trying to teach people in my life about this for decades. I’m so glad to see this story. The example I usually use is:

    Successful entrepreneurs all wear wrist watches and talk on cell phones. Therefore, if you wear a watch and talk on a cell phone you’ll be a successful entrepreneur, right? NOT!

    I think the pros call it “sampling on the dependent variable.”

  14. Well done to the mobile phone companies, misuse statistical correlation to point out stupid relationships while burying the fact that while microwave energy does cause cancer. A safe limit will now never be researched as only stupid will believe the research data.

  15. Maggie, stating categorically that there is no causation is much worse, scientifically speaking, than saying there is.

    Saying there is no known mechanism to explain this correlation is good science.

    Saying there is no possibility of the existence of a mechanism to explain this correlation is pseudo-science of the sort that should be roundly mocked. I’m mocking you here, see? Mock, mock, mock! It’s only because I love, though.

    Faith should not try to use the arguments of science, and vice versa. Science says a thing is not false simply because you cannot prove it to be true; that is an argument of faith.

  16. There is also an almost perfect correlation between the number of Storks nesting in Scandinavia in the 1950s/60s to the number of children born in the UK – thus proving that storks do indeed bring the babies.

  17. That wasn’t a phony health scare. You can’t call “more babies!” a health scare.

    What you can do, though, is show a correlation between the presence of phone masts and increased crime. More population, more phone masts, more crime, right? And it’s a negative, which means there’s the opportunity to become hysterical over it.

    I mean, really. More babies?

  18. Mobile phone masts, however, have absolutely no bearing on the number of births. There is no causal link between the masts and the births despite the strong correlation.

    I need to see the study that proves that assertion. Until then I vote ‘Causation’ too. Lack of obvious causation (like arsenic-eating bugs) is not a scientific response to a correlation that strong.

    If that were a psi-testing result it’d have world-changing ramifications. My surmise is that people who communicate A LOT are more likely to be making babies A LOT.

  19. The correlation is adequately explained by a common cause; increased population leads to increase in phone towers and increased number of births.

    However, that merely means that the correlation can be adequately explained in this fashion. The laws of thermodynamics can be adequately explained by phlogiston, too.

    Science does not deal in absolute dogmas, unless you are doing it wrong. A lot of people are doing it wrong, though.

  20. Let’s see. Storks bring babies. Babies build cell telephone towers. If we make the leap of faith that cell telephone towers are desirable places for storks to roost we have a positive feedback loop. The world will soon be overrun with storks, babies, and cell telephone towers.

  21. Science does not deal in absolute dogmas, but when there’s an easy statistical explanation of a correlation and the alternative is some unknown form of fertility particle beams, it’s OK to say there’s no causation and leave out the “almost certainly”. Sheesh, guys. Do you still talk about how magnets probably attract iron?

    1. Thus did the mighty Lavoisier debunk the foolish superstition that metallic rocks sometimes fall from the skies.

      1. Until there was better evidence, he was right to do so. I’ll assume you’ve forgotten the countless people who debunked similar ideas, and so far have turned out to be right.

  22. I think the study has it backwards. Could it be that cell phone companies are choosing to locate towers where they’re less likely to be challenged by residents? And those areas have residents that are renters, less affluent and have larger families?

  23. I’d say it’d be obvious to most punters that it’s misleading to correlate birth rates and the presence of cell phone towers. But I’d like to see more information about the link between cell phone towers and brain cancer, for instance. I’m kind of breaking an NDL here, but I work at a high-profile, multinational advertising agency and we have been told never to use images of people holding cell phones to their ears in any advertising for a well-known Telco in my country.

  24. The BBC needs a good lesson in this. They love doing “The news comes as…” too, to draw links.

    They also do the anti-causation… saying such and such happened DESPITE something else, when the other thing really had no connection anyway but they are a drawing it in.

  25. No statistician in the world would be stupid enough to look at anything other than the number of births per capita.

    Yes, correlation doesn’t imply causation. That’s fine – but using a silly example like this doesn’t add anything beyond the FSM pirate/temperature observation.

Comments are closed.