Partisan identification in the US, polls only weakly correlated to voter registration

Question

This article by Nate Cohn says:

Among those characterized as Democrats based on party registration or primary vote history, 69 percent identified as Democrats in the poll; similarly, 65 percent of those characterized as Republicans identified as Republicans.

This is a very weak correlation. What is the interpretation of this fact? Conventional wisdom is that the US these days is very partisan and tribal, and that most people who say they're independents actually lean one way or the other -- that, to coin a phrase, most independents are independents in name only (IINOs).

Can it be explained because there are a lot of people who have changed their party preference, but didn't bother to change their voter registration? If so, then they presumably aren't voting in primaries (unless their state has jungle primaries).

Is there evidence that many people register with the party they don't like, so that they can have an influence in that party's primary? (I used to do this.)

Is the conventional wisdom about IINOs just wrong or oversimplified?

Are there just a lot of people who don't pay much attention to politics, so that just as they can't tell you who their congressional representative is, they can't tell you their own voter registration? With motor voter registration, are many people registering to vote quickly and carelessly, not paying much attention to what they put down as their party? (A few years ago, I gathered signatures on a college campus for a city council recall. When I approached students and asked "Are you registered to vote in Fullerton?," the most common response was "I don't know." Many answered yes but then put down an address in a different, nearby city, suggesting that they didn't understand how voter registration worked.)

Answer 1

This really isn't much of a surprise to me, as the motivations for registering for one party or the other, or voting in one primary over the other, can be more pragmatic than ideological.

Voting in a party's primary doesn't compel one to vote for that party's candidate in the general, and as far as I know, there's no real consequences or responsibilities in registering for a given party on the voter registration, other than to limit what primary you can vote in. And in some places, your voter registration doesn't even include party affiliation at all. In my state you don't, and you can vote in either the Republican or Democratic primary in each election cycle (but not both, and you can't switch if there's a runoff).

Thus, it is rational for an independent voter (whether IINO, truly independent, or third-party supporter) to vote in one of the major parties primaries to have some say in the candidates that make it to the general. This is especially true if one party is dominant and almost certainly will win the general. In that case, the party primary is the de facto election. In the extreme case, there may even only be one name on the ballot for some races. I see this in some races in my area (judges, local representatives, mostly), where Party B doesn't even field a candidate. Thus, even many people who will vote straight-ticket Party B will vote in Party A's primary to have their vote mean something.

Of course, areas that have competitive general electrons, or "jungle primaries", there's less incentive to register for the opposing party, but still an incentive for independents. It'd be interesting to see if the correlation with "official" party affiliation and voting behavior is higher in those areas.

Additionally, as @divibisan notes in a comment, many voters shy away from calling themselves Republicans or Democrats, even if they solidly vote one way or the other, and prefer to identify as "Independent".