Facebook does not cause eating disorders: How to read statistics and cut through the media’s crap
{image via pinterest}
Once I learned that as a graduate student in psychology, I would be forced to take at least two semesters of statistics courses, it suddenly became my prerogative to figure out how minimize this trauma. When I was interviewing for various programs, I would ask the current students about their experience in the class, the professor, the rigors and challenges.
“Don’t you want to know how many graduates get jobs or if we feel respected by our faculty?”
“No, I just need to know if I have chance in hell of passing that stats class. Thanks.”
A surprise ending
For all of my whining and foreboding, I turned out to not be a half-bad amateur statistician. And while I would never profess any kind of love for this mathematical science, I can honestly (and with a straight face) say that I’ve come to appreciate my statistic training immensely.
Why? Well, I’ve oddly discovered that I enjoy research. And while conducting studies isn’t currently a major part of my work, I find myself using research constantly. And beyond that, I can now easily see through the loads of statistical crap thrown at me in the media.
Being able to apply a critical eye to what we hear, see, and read makes us smarter consumers and can prevents us from getting totally duped. Or unnecessarily panicked, as is often the case.
Facebook causes eating disorders?
Take for example, the claim that spread like wildfire mid-last year. Headlines around the globe touted, “Facebook causes eating disorders.” In The Register article posted online, the headline was followed by the first line, “A survey carried out in Israel shows that the more time young girls spend on Facebook the more likely they are to develop an eating disorder.”
So what you’re telling me then is that Facebook does NOT cause eating disorders?
This is an example of likely the most common error the media makes in reporting research. They report correlation as causation. Here’s a primer: Correlation means that two things (Facebook usage and eating disorders, in this case) are related in some way. When there is a positive correlation, as the rate of one increases (time on Facebook), the rate of the other increases (eating disorders).
But take this classic example to see why this does not mean that one causes the other. Researchers have found a positive correlation between ice cream sales and murders in a small town (really, I’m serious). Does that mean that ice cream causes murder? Are there enraged lactose-intolerant violent criminals out there who just can’t handle their sundae and turn into predators? Simply, no. In this case, researchers suspect that there’s actually a third variable that contributes to both of these – high temperatures. But if we’re not measuring that third variable, we get lost in believing that our Rocky Road is jail bait.
Chicken or the egg?
For those of you who are curious, the Facebook study looked at 248 Israeli girls’ media habits and eating issues. The problem is that this correlation does not reveal which direction the relationship goes. Meaning, it could be (and it would be my contention, to go out on a limb here) that girls who have or are likely to develop eating disorders spend more time on Facebook, rather than the reverse (that they “catch” eating disorders by being on Facebook). It makes much more intuitive sense that girls who are more focused on image, concerned about body weight and shape, possibly somewhat isolated (i.e. girls with risk factors for eating disorders) would spend more time on social networking sites. And sometimes intuition is just as important as hard data.
Generalizing schmeneralizing
So say a study actually does involve experimental conditions, meaning it can point us to causation. Does that mean that the results are going to be true for all of us? Absolutely not. As you probably know, the majority of studies are conducted using participants from the college campuses where the researchers work, meaning that the sample is quite often college students. Not only does this mean that the participants are usually of a certain age range (18-23), but they also disproportionately represent a certain segment of the population – those that go to college. While some diversity exists, we can reasonable conclude that certain segments are going to be underrepresented, such as the poor, the illiterate, racial and ethnic minorities, and people following Bill Gates lead.
It’s also important to consider where the study is being conducted, meaning what geographic area. If the study took place in Israel or Poland or Texas, it makes a difference. Even subtle things that one wouldn’t assume would depend on location (e.g. genetically determined variables) can be impacted.
The point is, you have to know who exactly the study was looking at and where before assuming that it applies to you.
What are you telling me, really?
Yet another question to ask ourselves in this confusing web of statistics reporting is: Is any of this really meaningful? And, further, is it useful to me personally?
I read an article recently claiming, “Soy doesn’t boost brain power in older women, says study.” Okay… I’m not exactly sure my life was enhanced by knowing this fact. It doesn’t make me want to kick my tofu to the curb (it didn’t say it lessens brain power, after all). You have to consider how meaningful the statistics really are, because before you know it you become that 8%* who spout totally useless information just to sound smart. (*Disclaimer: I made that up.)
And more importantly, statistics often don’t mean a whole lot when it comes down to your individual life. Take the recent study that claims that delivering a baby via cesarean section increases the chances of the child being obese by age three. Last time I asked any woman delivering her child, she wasn’t making the decision to deliver vaginally versus a c-section based on her child’s future penchant for Capri Suns. In fact, she wasn’t basing it on anything other than what her doctor and she decided was best for her and the child (let’s be honest, usually the child) in that moment. No woman I know who’s had a c-section made the decision lightly, and research like this, though potentially valuable in certain ways, isn’t useful when it’s directed at mothers who already feel guilty for just about every little thing they do. Because, after all, mothers are to blame for everything, right?
The bottom line
The bottom line is that you have to be careful when interpreting statistics, and even more careful when deciding how much stock to place in them. Because, honestly, when it comes down to it, when you learn you have a 10% chance of getting an illness, and then you get it, your chance just went to 100%. And that’s all that really matters.





