For two decades, as fewer people gripped their beloved BlackBerrys like security blankets, car manufacturers found themselves unable to shift vehicles off forecourts with quite the same enthusiasm, which suggests that either the smartphone and the automobile are locked in some sort of cosmic dance of mutual obsolescence, or humanity has finally discovered a correlation so perfectly inverted that it makes us wonder what we've actually been measuring. The correlation coefficient sits at minus 0.932, which is to say nearly perfect in its wrongness.
What we're almost certainly watching here is the economic cycle of the 2000s and 2010s wrapped around two entirely different technologies. The BlackBerry peaked during the 2008 financial crisis and its aftermath—that glorious moment when everyone who owned one felt simultaneously important and doomed—while new car sales collapsed spectacularly in 2009 and took years to recover as people stopped buying anything that wasn't essential. As smartphones became genuinely universal (and Android ate the BlackBerry's lunch around 2010-2012), the economy had also limped into recovery, and people began buying cars again with something approaching optimism. It's rather like noticing that sales of VHS tapes declined exactly as internet bandwidth improved; both are true, both are connected to the same underlying shift in human behavior, and neither caused the other.
This is what pattern-seeking mammals do when they have access to a spreadsheet and twenty years of historical data—we find the skeleton keys that fit no locks. BlackBerry ownership and new car sales were simply two different animals responding to the same economic weather, moving in opposite directions as it happened, which tells us something important about correlation: it is very good at noticing when two things dance, and very bad at caring whether they're actually dancing together. We saw what we wanted to see.
As an Amazon Associate, getspurious.com earns from qualifying purchases. Learn more.
Want to learn more about why correlations like “Adults who own a Blackberry” vs “New car sales” don't prove causation? Read our guide to statistical thinking.