Week after week, month after month, the health (and mental health) news headlines blare with the latest “link” between two things. Take, for instance, a few articles from just this past week we’ve published… Childhood cancer? Less likely to marry. Obese? Depression is more likely. Eat licorice while pregnant? Your child may have a smaller IQ. And my favorite from the past week? Eat candy as a child? You’re going to become a criminal.
Researchers seem content to draw these correlations, knowing full well their data shed little light on the actual problem. Instead, what they manage to do is to shed a whole lot of brain cells. Ours.
I’ll pick on the candy study because it’s low-lying fruit and it’s easy to make fun of. Let’s look at the data reported:
Researchers from Cardiff University in Wales looked at data on 17,415 children born in a single week during April 1970 in the United Kingdom. The data, from the British Cohort Study, included detailed health and lifestyle information on the children at several points during their lifetimes, including ages 5, 10 and throughout adulthood.
Thirty-five of those children went on to report at age 34 that they’d been convicted of a violent crime, the researchers found.
About 69 percent of those who reported having committed violent acts also reported eating candy daily at age 10, compared to 42 percent of those who did not have a violent criminal past, the study authors noted.
So let’s get these numbers straight, just so we have some perspective. Out of 17,415 children, only 35 of them were convicted of a violent crime? That’s astonishing, given that the UK has the highest violent crime rate in Europe. But what’s even more astonishing is that the candy-eating behavior of 17,380 children was not reported. What if 10,000 of those children also reported eating candy at age 10 daily? Wouldn’t that basically nullify the researchers’ findings?
Anyways back to what the researchers did report on… 24 of the 35 said they ate candy at age 10. Gee, I wonder what kind of association we’re now going to make, since things like gender and parenting style showed no significant differences between these two groups of children.
“There appears to be a link between childhood diet and adult violence, although the nature of the mechanism underlying this association needs further scrutiny,” said study author Simon Moore.
Really? Wow, indeed heady research there, Simon Moore. If you can’t say what the nature of the mechanism is underlying this association, I honestly have to wonder at the value of this research. What new information have you imparted to us?
Better yet, I wonder if there’s a more reasonable explanation that could explain this association.
“While it’s an interesting correlation, any scientist will tell you that a correlation never shows causation,” said Melinda Johnson, a spokeswoman for the American Dietetic Association.
“If there is any real link, my instinct is that the daily candy may be indicative of certain lifestyle factors that the researchers did not capture. For example, I do not see that the researchers were able to control for violence in the home. Perhaps children who end up violent as adults also tend to grow up in violent homes, and perhaps candy is used excessively as an ‘ease the pain’ tool.”
Another possibility is that a diet high in sweets is indicative of poor nutrition overall, which could have led to abnormal brain growth during a critical period of development, Johnson added.
Imagine that — something the researchers did not measure might adequately explain this association! Such as the immensely obvious theory that violent households might breed violent children (and candy is simply a byproduct of that relationship).
Researchers, unfortunately, are rewarded to publish, regardless of the merit of what they publish. You cannot be in an academic position for very long if you’re not constantly churning out rubbish like this.
Next time you see one of these sorry headlines or articles about an association between X and Y, know that one of the reasons you’re likely reading about the association on a site like ours or U.S. News & World Report or WebMD is because researchers need to keep their jobs and pay their bills.
As we continue to report on this research, keep a skeptical eye open and take such associations with a grain of salt. We will try and do our best to point out the obvious in such stories — correlation does not equal causation. The association we report on is of likely little value to anyone’s real understanding of the issue.
Now forgive me, as all of this talk about causation and correlation has made me hungry. For some candy.
10 comments
Dear John,
I must admit I clicked on this thinking it was about your wonderful and dedicated editor, Candy.
Regards Sonia
This is great. Keep calling them out!
Great article! I loved reading it the whole way through :D. Keep up the grood work!
You seem to missing out on the structure and purpose of these types of studies. You seem to under the impression that in 1970 they attempted the study to uncover the reasons for adult violent criminal behavior. That isn’t the case. The research was done to periodically take as much data as possible in a random sampling of population. Then, future research could use that sampling to observe long-term behavior and develop correlations. Of what is unknown when the research started – at that is the point.
The research ended up showing a correlation between Candy and Violent Criminal Behavior. What does that mean? Nothing in terms of cause and effect. Why make that correlation then? Once you have the correlation NOW you can do more studies – research if there is a causal and predictive relationship between them, work out mechanisms for such. You have that target ‘Candy as a kid -> Violent behavior as an adult’, and you can now ask the ‘Why’ question you couldn’t have asked before. And that may lead to research that increases detecting violent behavior earlier, or preventing violent behavior, etc…
p.s. You said:
“But what’s even more astonishing is that the candy-eating behavior of 17,380 children was not reported. What if 10,000 of those children also reported eating candy at age 10 daily? Wouldn’t that basically nullify the researchers’ findings?”
The candy eating habits of the non-violent population was reported, and you quoted that report:
“About 69 percent of those who reported having committed violent acts also reported eating candy daily at age 10, compared to 42 percent of those who did not have a violent criminal past, the study authors noted.”
69% of the criminally violent. 42% of non-criminally violent. That is a difference of just more than 25%. Had the two populations behaved the same in terms of candy-eating, then only 15 of them would have eaten candy.
Also, the comparison between the 35/17415 violent criminals, vs. UK’s overall 2,034/100,000 violent crimes is hard to make. One counts the crimes, one the criminals. One is counted by law enforcement the other is self-reporting. One is for a specific year, the other is totaled over a different time period. So the differences appear drastic but it is hard to do it correctly and as is probably isn’t interesting (meaning accurate). My biggest concern with this study is it’s reliance on self-reporting for activities people generally don’t want to self-report.
Apparently Jerry Seinfeld was obsessed with lollies as a boy.
There’s no need to rail against a real or perceived disinformation campaign here. The researchers found a correlation. They did not claim causation. Any dietitian, naturopath, and even a few MDs will tell you that excessive sugar is detrimental to the body. One doctor who had extensive experience with diabetic patients even told me that sugar should be classified as a carcinogen due to its detrimental effects on the circulatory system, which lead to the type of hypoxia commonly associated with some cancers.
From an evolutionary perspective, we rarely had access to sugar, and now that it’s available in literally every place that sells food, we are really going for it as a species. It sets up a certain mild “high” that may become subtly addictive. If you think I’m full of hot air, try going just one week without eating something sweet (no fruits either, which have natural sugars). Everyone in the so-called First World is addicted to sugar.
The correlation that was implied but not stated in this research was that sugar may be associated with addictive behavior, which has already been shown to be clearly correlated with criminal activity.
Dr. Grohol, I understand that this is your website but I don’t understand this vitriol against researchers. Budgets are limited and not every study can take every variable into consideration. The research is one thing, sensational headlines are another. Please learn to separate the two in your mind. This type of reactionary reporting just makes people less likely to critically read and interpret the research on their own, instead of just skimming headlines on CNN and digesting the media’s conclusions without analysis.
One of the things these studies never investigate or control is Subliminal Distraction exposure.
Included in Culture Bound Syndromes are several behaviors involving sudden violence. When you investigate the ethnic groups involved they all have the potential for Subliminal Distraction exposure. (Going Postal – US, Amok – Malaysia, iich’aa – Navajo)
Families traditionally lived in too-small single-room arrangements. Huts, hogans, kivas, and even Tepee’s create the “special circumstances” for SD exposure. Such structures crowd individuals close together so that if one person engages deep mental investment to the point of light dissociation the movement of nearby family members is subliminally detected for SD exposure. (Daydreaming is enough dissociation.)
Recent outbreaks of Grisi Siknis after hurricane Fredrick, when Miskito Indian families would have been confined inside, points to SD exposure for that behavior. It involves delusional fights with invisible foes and attacks on people and things.
For students likely exposure comes from computer use in a busy location with detectable movement in peripheral vision. Too-close side-by-side seating in class rooms replicates those circumstances for someone with sensitivity for SD exposure. Making lecture notes requires full mental investment. Large nearby movements would be subliminally detected.
In seven years investigating I have been unable to find anyone in medicine or psychiatry aware this problem does or even could exist.
Statistical studies produce correlations which have a less than 1/200 chance of being relevant. This is why these investigations turn out wrong. Real scientists will tell you that they are a place to start looking for what to look for. They are not an end in themselves, heck most of the time they are not even the beginning. If you look at many of these studies they are conducted by either MDs or soft subject academics.
An MD, a clinical non-behavioural psychologist, an economist, a mathematician, a physicist, and a logician and Neils Bohr are on a train. They see a cow with brown spots in a field.
MD: “Eating grass gives you brown spots.”
Pysch: “Eating grass gave that cow brown spots.”
Econ.; “The cows in this area have brown spots from eating grass”
Math: “There is a cow in there with brown spots and it is eating grass.”
Physicist: “There is a cow with brown spots on this side which we are observing from, and it appears to be eating grass.”
Logician: “I appear to be observing at least half a cow with brown spots or dirty marks, and it is taking green strands into its mouth.”
Neils Bohr: “If in fact I exist in the form I appear to exist in and if what I know if the world is essentially correct then light is hitting the retinas of my eyes and forming a pattern when interpreted by my brain..”
The Rest: “Shut up Neils!!”
Sensationalism, lack of knowledge, and the need to produce all contribute to research like this, and articles like this, and discussions like this. Of course, everything is correlated with everything else. Correlation is the least informative, and perhaps most misinformative tool in the statistical armamentarium.
Writers should know this, readers should know this, and high school students should be taught that correlation might pique curiosity, and perhaps give guidance to a research direction, but little else.
Much the same could be said about “association.â€