Take a look at the graph below.
This is the excerpt from Gizmodo:
You’re very likely to get dumped on Mondays, right before Spring Break, two weeks before Christmas, and at some point before the summer holidays. The good news about the whole mess is that it seems that people feel quite bad about dumping someone right on Christmas Day, so you can breathe a bit easier while unwrapping your presents.
Facebook has quickly become the largest human data set, so yes, it is very interesting to look at the bulk data generated by its users. But from all the sources I’ve read, all of them miss this very important mark:
The data depicted above does not correspond to actual break-up day; rather, the data corresponds to self-reported updates of a break-up made by Facebook users.
Why is this qualification important? Because the way the information is presented above, there are numerous confounding variables. The most notable one is time shift (delay), corresponding to how long it would take for a Facebook user to update their status on Facebook after a break-up.
Here are two scenarios I can think of when time delay is pivotal:
- Monday Break-ups. Suppose someone gets dumped on Friday. He or she may not come to terms with the break-up that Friday, and perhaps try to reconcile the relationship throughout the weekend. So, in fact, most break-ups may occur on a Friday, Saturday, or Sunday; but the status update would come on Monday (when the user couldn’t reconcile things over the weekend, say; or perhaps, coming back to school or work the following Monday morning, Facebook is more accessible, so it’s fitting time to make the status update).
- Christmas Day Break-ups. While it does appear that there are the least number of break-ups occurring on Christmas Day, perhaps it’s because a lot of people are away from computers (and thus Facebook) to update their status. Alternatively, it may be really embarrassing to announce your break-up on Christmas Day, so the user would wait a day or two to make the Facebook update.
That said, I think the data is useful, but it is much more interesting when looking at general trends (more break-ups occur as Spring Break and Christmas approach) rather than pinpointing break-ups on specific days of the week or holidays.
Still unconvinced? Imagine if the data set instead showed specific dates on when Facebook users entered a relationship (girlfriend/boyfriend, engagement, marriage). Would you really believe that if someone changed their status update to “Married” on a Monday morning, they actually got married on Monday? Of course not (unless they’re this couple)!
Bottom line: while the general data presented above is interesting, it’s important not to discount numerous confounding factors (time delay being the most notable one, but also: people untruthfully reporting a break-up).
2) For a superb take on the importance of confounding factors, especially on Twitter, I highly recommend reading “The Confounding Variable of the Retweet.”