Every few months an article comes out challenging the long-observed trend of people growing more conservative as they age. Usually it’s around an election. This time an article out of the Financial Times by John Burn-Murdoch made the rounds, going viral across TikTok and Twitter.
I’m not here to criticize Burn-Murdoch — he’s a good statistician with a talent for digging insights out of datasets. However the presentation of his data in his Financial Times article, Millennials are shattering the oldest rule in politics, seems a little odd to me.
At first glance that seems extremely compelling. There’s a noticeable shift away from supporting conservatives as millennials get older — at least that’s what you’d immediately glean from reading that chart.
A lot of people want this to be true, which you can probably guess by reading the comments on anything talking about this.
But here’s my issue with this chart before I disagree with its conclusion:
Who are we tracking here? The broader categories are clear — the generations — but are we tracking the same people over time, just pulling results-by-age from various elections?
If we’re taking a snapshot of election histories then yes, the result is clear — millennials who are 40 now are voting more liberal than they did at the end of their 20’s and similarly to how they voted in their early 20’s. If that trend continues then we’d have balked the red-shift trend.
But here’s the thing — either we’re tracking only a small subset of millennials with that graph and that’s our conclusion, or we’re tracking a current snapshot of the generation, which reads entirely differently. If that’s a snapshot of all millennials then the conclusion is that millennials from the center of the cohort grew noticeably more conservative, which would also be an anomaly.
I don’t think that’s what he’s doing here, since there are no 40 year old boomers, for example, anymore — something this chart would require given its presentation. So this is probably tracking electoral history and shifting the age boundaries up to track some subset of the generation, not the entire generation.
There are still issues with that, and the methodology he used isn’t open, but here’s an example of why that’s not comprehensive:
Millennials were born 1981–1996. The oldest of them started voting in 2000. The youngest started voting in the 2015. That 15-year spread is important. Your oldest millennials are 41, your youngest are now 26. So, again, this chart can’t be tracking any more than the oldest millennials — or maybe he’s mixing samples. Either way his presentation of data and the data you’d need to generalize millennials as a group doesn’t add up at face value.
It’s that pesky x-axis he used: age. It’s not particularly useful unless everyone is the same age.
So, let’s use a more viable metric.
With the above information you can pretty much say that everyone in the 18–29 demographic in 2008 was a millennial, save the last year of that cohort, 29-year-olds. In the 2008 house elections that demographic went blue with a 65 to 35 Democrat to Republican split. In the Presidential election they voted 66 to 32.
So how did those same people (Millennials) vote in the 2020 Presidential election? That’s 12 years later when 100% of those 2008 voters were now in the 30–44 demographic.
This cohort is a little more broad, though, with 41–44 year old being Gen X’ers, who are probably leveling the data a little more than before.
Regardless, in 12 years Millenials shifted somewhere around 10 points away from Democrats and at least as much toward Republicans. In 2022, now occupying the entire cohort except for two years, they voted in the house slightly more conservatively at 51 to 47. Unless that small minority of Gen X’ers is hyperconservative — they aren’t — those numbers are somewhat representative of Millennials as a group.
Does that sound like they’re balking the trend?
Because it sure doesn’t read like it.
Let’s also keep in mind that 2020 and 2022 featured some of the worst candidates for Republicans in history, which we shouldn’t ignore when it comes to voting.
Let me reiterate —Millennials in the United States have voted:
2008: 65D to 35R
2020: 52D to 44R
2022: 51D to 47R
That’s a red shift. It’s still happening, for now, in the voting trends. It’s still happening as recently as 2 months ago.
I’d love to see the code and methodology Burns-Murdoch ran to generate his charts. There’s something missing here. I’m not accusing him of doing the journalistic equivalent of p-hacking, but I get the feeling there’s a reason this is published as news to generate clicks rather than as an academic paper. It doesn’t pass a sniff-test, even this rudimentary one.
