Stupid Poll Tricks
Two data points are not a trend.
Memeorandum pointed me to a story headlined “GOP Base Has Shriveled Compared to Last Presidential Election” at the husk of the venerable Newsweek. The clickbait headline did its job but, alas, the report was disappointing.
The number of people who describe themselves as Republican has slightly declined since the last presidential election.
The polling firm Gallup, which tracks party affiliation monthly, says that those identifying as Republicans has shrunk by two percentage points since 2020. In February 2020, 30 percent of those Gallup polled said they were Republicans, while 29 percent considered themselves Democrats.
As of February 2024, when the latest data is available, 28 percent say they are Republicans, while 30 percent say they are Democrats, showing Republican’s base declining. Newsweek contacted the Republican National Committee (RNC) by email to comment on this story.
So, we have two snapshot polls, taken four years apart, showing that Republican party ID is down 2% and Democratic party ID is up 1%. Which, of course, is well within sampling error. That’s . . . not particularly interesting.
This does not prevent them from devoting several paragraphs to explaining what may or may not be a trend, including multiple quotations from scholars.
While the OP is absolutely correct about click-bait coverage of polls, I do have an anecdote of my own. I live in a deep red to dark purple area of Noth Oakland County MI. At this time in 2020 area was carpet-bombed with Trump signs, flags, and other visible signs of support. The overwhelming majority of these stayed up long after the election was over (presumably until the rightful winner of the election was installed). You couldn’t swing a dead cat without hitting one.
Now? Nothing. There are a couple I see that have been up since 2020, but brand new 2024 are nowhere. I just don’t see the broad-based energy around Trump this time. That doesn’t mean he can’t get elected, but the vibes are definitely different.
@Rick DeMent: While Northern Virginia and Fairfax County are both quite blue, my little neck of the woods is very Trumpy. And I, too, note almost no Trump signs these days when they were seemingly everywhere four years ago. Then again, I don’t see a lot of Biden signs when I drive around, either. I think this is likely going to be a very low energy contest, even though the stakes remain quite high.
@Rick DeMent:
I wonder how much of it is that this year, the grifters have their hands firmly on the revenue streams. It may be that the people who would print up all those signs and banners and hand them out are sorely lacking for cash.
Your point about MOE is valid, but let’s take for a moment that the poll numbers are dead on. You can look at is a 2% decline of Republicans in the overall electorate, or that 6% of those who used to identify as Repubs no longer do so. Same thing, but that second number is a bit more interesting.
But as you point out, it’s all within the margin of error.
Of course, Covid took and continues to take significantly more lives among we oldsters that has provided a great part of the R base. Plus those oldsters are dying off for lots of other reasons as well.
@Sleeping Dog:
I’d seen a link to that article yesterday and resisted following it because it was “Newsweek,” thanks for taking a bullet for us @james
@MarkedMan: Another pet peeve of mine. Newsweek stated it properly as a drop of two percentage POINTS, but I would bet most readers read that as a drop of 2%, not a drop of 6%. A 2% drop is negligible noise. 6% at the polls would be enough to decide the election. Newsweek characterized that as both “shriveled” and “slightly declined”. I would say somewhere between, closer to slightly.
James, you said , “at the husk of the venerable Newsweek”. Snarky, accurate, I like it. And you’re quite right. Two data points could be a trend, but may just be noise. At WSJ, and elsewhere, such things are often deliberately cherry picked lies.
It’s just clickbait. It’s not there to argue a real POV, it’s designed to irritate and thus generate ad revenue. Fewer would care about an article entitled “Poll: party ID stable.”
That this post exists justifies the existence of clickbait. It’s