President Trump waves while walking toward the White House Feb. 1. (Andrew Harrer/Bloomberg News)

Charles Franklin, director of the Marquette Law School Poll, made an interesting observation over the weekend: The number of Americans who identify as Republican dropped in 2017, and the group that benefited appears to be independents.

There is an interesting connotation to this. President Trump’s low approval ratings (which have improved recently) are often discussed with an asterisk. He may be unpopular, but among Republicans he’s still viewed positively. But if there are fewer Republicans now than there were when he took office, that Trump maintained their support seems less impressive. If six of your 10 friends like your spouse, but then three of the four who don’t like your spouse decide they don’t want to be your friend anymore, it’s not exactly a good sign if six of your remaining seven friends are still supportive.

That analogy, though, is a straw man. Let’s look at the numbers, using data from Gallup, on which Franklin’s analysis was based.

Over the past 15 years, the number of independents has risen in the United States It’s now the biggest partisan group in the country by a wide margin. In Gallup’s most recent data — the dots farthest to the right on the chart below — 44 percent of respondents identify as independent, while 32 percent identify as Democrat and 22 percent as Republicans. (The solid lines are the trends.)

That group of independents, though, includes people who tend to lean to one party or the other. Most independents identify with one party or the other, in fact. Thirteen percent of Americans (and 30 percent of independents) identify as independent without leaning to one party or the other.

So let’s pick out three of the groups above: Republicans (excluding independents who lean Republican), Republican-leaning independents and what we’re calling “pure” independents.

In 2017, the number of people who identify as Republican dropped off. The number of people who identify as Republican-leaning independent stayed about the same, and the number who identified as pure independent went up.

It’s often the case that political transitions happen gradually. In a presidential race, people usually don’t switch from one candidate to the other; they make a pit-stop in “undecided” first. It’s possible, looking at the graph above, that a number of Republicans began to identify as Republican-leaning independent as a number of Republican-leaning independents decided not to lean that way anymore.

Franklin notes that these shifts are more subtle than they may seem.

“[I]f we judge from partisans alone, the Dems have been completely flat from early 2014 to the present, while Republicans gained about 4 points over 2014 through 2016 only to see all of those gains lost in 2017. . . . So if the Dems are mostly flat, and there has been a bit of GOP decline, where have people gone? Pure independents are up recently, by just over 3 percentage points since the 2016 election.”

But let’s go back to the original question. Does this suggest that Trump’s got a smaller, more dense base?

Not really. Even among Republicans, Trump’s approval fell over the course of 2017. (Gallup changed its methodology at the beginning of this year, so the most recent partisan data that’s available is for the end of December.)

Let’s overlay the percentages of partisan identification and approval onto hard numbers — specifically the population of the United States. (Yes, this includes young people who aren’t polled; it’s for demonstration purposes only.) Because Gallup only splits out its approval ratings by Democrat, Republican and independent, we’ve estimated the approval of Democratic and Republican leaners as being halfway between the partisan approval and the independent one.

In January 2017, about 156 million people approved of Trump on average, more than half of them — 81.2 million — Republican. (For each month, we averaged the available values.) That’s out of about 91.2 million Republicans overall.

By December, about 125 million people approved of Trump, with more than half again coming from the Republican Party. There were about as many Republicans in total in December as approved of Trump in January 2017.

In both January 2017 and December, about 52 percent of those who approved of Trump were Republicans. In January 2017, though, about 89 percent of Republicans approved of Trump, while by December, 80 percent did.

There are a lot of margins of error at play here. But if a trend of Republican identity sagging and independents increasing holds, there’s a vitally important question for the Republican Party this November and in 2020.

Trump is president in large part because Republicans skeptical of his candidacy overcame their skepticism to vote for him in 2016 anyway. Some of this was probably distaste for Hillary Clinton, and some of it was probably party loyalty.

Neither this year nor in two years’ time will Clinton be on the ballot (we assume). And if the Republican Party can’t lure those voters back, it’s not clear whether they would necessarily line up to cast a ballot for the party’s candidates.

Trump’s strategy of insisting on boosting his base has appeared to keep his base loyal to him. But to some small extent, his party eroded over the course of 2017, which, given his margin of victory two years ago, bodes poorly for a reelection.