This is real research. Rigorous, cleanly factorized, unbiased, work shown for others to check.
Real, yes. Open for checking, yes. Rigorous, maybe. Ceanly factorized, not so much. Unbiased, it is to laugh.
Just from the summary I see two classic issues: Selection bias and confusing generational samples with age effects.
Selection bias is cascaded. First, it's sampling only people who joined facebook. Second, it's only sampling the subset who both heard about and chose to download and use the tool and let it watch their activity. I see close to a dozen classes of selection bias here.
Confusing generational samples with aging effects is a classic flaw. Of course when you're first doing a study looking for age effects, about all you CAN do is use generational cohort as a proxy for aging. But people from different generations have a host of differences besides age: Nutrition, nurturing fads, stress from wars and other disasters, disesae exposure, educational variations, and the list goes on.
One of the classic errors that arose from this is the belief among psychologists that intelligence ramps up nearly linearly until early adulthood, knees over, and then slowly drops with age. That lasted until standardized tests had been administered to the same groups over several decades, so the trajectories of the scores from particular individuals and groups could be tracked. It turns out that intelligence does rise and knee-out as described, but the gradual slope with age is UPWARD (even before discounting the higher incidence of specific brain-damaging disease processes with advanced age). The effect had been masked by another: People educated in earlier decades did less well on the things the tests scored.
You can see that this work - or at least those attempting to interpret it - has the same problem:
... and even how peoples' postings tend to evolve as they get older â" as people age, for example, they tend to talk less about video games and more about politics.
Are today's older people more interested in politics because they've aged and have more understanding of them and/or are more affected by them? Or are they more interested because they grew up during or in the aftermath of WW II, Korea, the Cold War, and the mass movements and political suppression surrounding Vietnam and the clampdown on "recreational" drugs.
Are they less interested in video games because they're older or because video games DIDN'T EXIST YET when they had time to practice enough to become skilled?
Conflating age with cohort membership can lead to problems when you try to use the results of such research to predict how people will change with age. For instance: If video game interest is a symptom of low age you can expect people to "grow out of it" and current users to fade out as they find other interests, but if it's a symptom of cohort membership they may become MORE active as they mature further. If political activity is a symptom of age you can expect the young to become more active as they age, but if it's a symptom of life experience you might see new generations becoming active young (as with the Antiwar movement in the '60s and '70s and the Liberty movement today) and people of all ages suddenly becoming politically active after being "radicalized" by the stress of political events.