If you were conducting a study that was looking at how social media use potentially affected or was affected by mood, for instance, you would likely rely on self-reported statistics for both measures.
And one would think that those people would have a pretty decent grasp on how long they spent scrolling through Facebook, Instagram, or Twitter.
Everyone understands that these self-reported numbers will have error, and some studies have demonstrated it, but this meta-study from Facebook, comparing self-reporting with actual server logs, shows that the connection is possibly not reliable enough to use for serious scientific work.
Responses to some questions, when compared to internal data, showed that people overestimated their time on site by hours on average.
As you can see, despite what they thought, very few people actually spent more than three hours on the site per day, with the vast majority spending about one. And the opposite was true of logins: Comparably few people thought they opened the app ten times a day or more, yet that was extremely common.Younger people especially were prone to error, which given that these studies tend to have more of that demographic, only emphasizes the problem.
None of this is to say that Facebook is not a site people spend a lot of perhaps too much time on, even by their own estimate.
In other words, instead of saying teens who spend 2 hours or more on Facebook were more likely to you might say, users in the top 10 percent of self-reported Facebook use were more likely to or some such.
Original article