I've contended for a while that people who vote Republican in a general election will vote Republican in the primary, even if their party is unlikely to win. The first factor that should show us this is that in the past minor party candidates have gotten 2-7% in elections, often exceeding that party's registration.
One 2010 race stands out.
This race was supposed to be a nail biter and it turned out to be very close. In addition to the Republican and Democrat, there was an American Independent party candidate. This party is affiliated with the Constitution Party nationally. Many people join it when they find the Republican party not sufficiently conservative. Had David Harmer gotten a quarter of those votes, he would've won. If there was no AI candidate, he would've needed to win those voters 63%-37%. Considering where AI stands on the issues, it's safe to say he would've gotten at least that.
These people likely knew that 1) their candidate had no shot 2) by voting for him, the Republican might lose. They did it anyway. Because people will vote for the candidate they like regardless.
Washington state has had top two in the 2008 and 2010 legislative and congressional elections. Below is how the general election vote compares to the primary vote.
The column on the left covers races where there were only Republicans and Democrats in the primary. The one on the right is for races with Republicans, Democrats, and one or more third party/independent candidate(s). When there's only the two parties, the general election spread is within two points of the primary 87.5% of the time in congressional races and 59.3% in legislative races.
The chart above shows the cumulative primary and general election percentages of the Democrats and Republicans in WA-1-4 in 2008. The numbers closely mirror each other, even though none of these districts was competitive. If you're someone who votes Republican in a general election, you're going to vote Republican in the primary. This happened even though the eventual winner in the general election had challenges from their own in the primary.
If people were voting strategically, you'd see an elevated Democratic total for the challenger in 2nd or 3rd district or the Republican in 4th. Then they could've unseated the incumbent, always attractive to do, in the general election.
When there are 3rd parties involved it gets slightly more complicated. In some cases the 3rd party vote went fairly equally to each candidate in the general, while others had one party taking most of that vote.
This data shows us that people who will vote Republican in the general election are likely to have done so in the primary. It also tells us that a candidate whose party loses the primary 55%-45% isn't going to make that up in the general election.