The claim: after a single study (which we reported on back in May), computer scientists now know how people with depression spend time online.
From that knowledge, the researchers suggest we could design some sort of intrusive, spying app on your computer, iPad or smartphone to let you (or Big Brother, in whatever form — college administrators, your parents, or big data mining companies working for advertisers) know when you’re surfing in a “depressive” pattern.
Are the researchers over-generalizing from their data, or do we really know how people use the Internet when they’re depressed?
Let’s find out…
As we explore this article, keep in mind that researchers’ conflict of interest in writing up their results for the mainstream media is a very real one. They will help forward their academic careers and professional reputation by having such a write-up appear in a prestigious newspaper such as the New York Times.1 Such a write-up won’t help as much if the researchers aren’t brazen and absolute in their conclusions.
And yet, we need researchers to explain the complexities of their data and be cautious when generalizing their results. Especially when they’re explaining their results in a regular newspaper (as opposed to a journal article). (Especially when other non-scientists will simply uncritically repeat the finding as though it were fact, because it appeared in the New York Times.)
The researchers found that a small group of college students who scored highly on a single measure of depression — not people who’ve actually ever been diagnosed with depression — appear to like to download more music, movies, and file-sharing, and seem to email others more often than those who didn’t score as highly. “Other characteristic features of “depressive” Internet behavior included increased amounts of video watching, gaming and chatting” and switching between online tasks more often than non-depressed individuals.
Previous research from 11 years ago had found similar results correlating loneliness (not specifically depression, though) with increased email use. It’s also not really surprising to learn that people who are depressed like to watch more TV — or the Internet equivalent of it today, downloading more movies.
People Use Facebook? Smartphones?
But it’s important to also note what this study did not measure — social network and social media use, as well as mobile phone use and texting. After all, I’m sure college students are using Facebook, Twitter and their smartphones to keep in touch with their friends more than they use email.
The lack of specific mention or monitoring of these popular and widely used technology platforms is a significant hole in the researchers’ data. It means the researchers are describing only what they can measure. We’re completely in the dark about technologies they didn’t measure yet are widely used.
Think of it this way… What if researchers only had access to magazine subscriptions of a group of people, but no access to their newspaper subscriptions or TV viewing habits? The researchers could tell us all about their magazine reading habits, but leave out what most people are actually doing — watching TV and reading newspapers.
Convenience Sample — Not a Randomized, Representative Sample
Another problem is that the subjects they used to conduct their study is not randomized nor representative. Taking 216 undergraduate college students from a single university campus is not robust methodology. It’s called a “convenience sample” and is usually done in exploratory or pilot studies in psychology. Worse is that only 28 students of their sample — a tiny 13 percent — were female.
At the onset of the study, a surprising 30 percent of the students met research criteria for depression (specifically, they scored a 16 or higher on the CES-D). That’s a big number, and suggests that their sample had an inordinate amount of depressed students in it. It’s also nearly twice the rate of depression as measured in the 23,000+ students who responded to the National College Health Assessment.2
Big Brother Knows When You Are Sad
The researchers, based upon this single study, are “currently attempting to build a classifier to proactively discover depressive symptoms among students by passive, unobtrusive and run-time monitoring of their Internet usage.”
How “unobtrusive” will it be when someone from the university counseling center comes knocking on your door to inquire about your “depressive” Internet use? What is the rate of false-positives?
And are the researchers really at a stage of development in their research — before it’s been replicated on a single additional college campus — that ensures what they’ve identified is actually a “depressed” pattern of Internet use? What if a dozen other mental disorders exhibit similar Internet patterns? What if it’s college stress, which simply surfaced as higher CES-D scores in this study? What if it’s a male-only phenomenon?
So many questions remain, yet the researchers — computer scientists, not psychologists — feel certain they are on the right path to a new mental health intervention.
In sum, the headline could be more accurately reflected as: How a Small Group of Depressed Male College Students Who Are Not Representative of College Students in General Use the Internet at a Single Campus at a Missouri University.
Not quite as sexy or eye-catching.
- It’s never clear to me why a news organization like the New York Times is okay with letting researchers — who have an intrinsic conflict of interest — write up the findings of their own study and then publish the write-up. I guess they rationalize it by putting it in the Opinion pages, as though people reading the article online will note and appreciate the differentiation. [↩]
- http://www.acha-ncha.org/data/PHYSMENTALF06.html [↩]
- This is a pretty offensive headline, too, although not the writers’ fault. Calling a person who has clinical depression a “depressive” is a depressing reminder of the stigma and short-hand taken by people who don’t appreciate that a person is not defined solely by whatever medical or mental health condition they may have. [↩]