Why the “hmmm”? Two reasons. One: With one dubious exception, 36 percent is the lowest haul Trump has had in any poll of Michigan this month. In three of the previous five surveys, his lead has been 19 points or better. I’m skeptical that Trump loses this state given that it’s an open primary, but if you’re looking for evidence that the race is tightening before the big vote tomorrow, here you go.

Two: It’s not just the topline numbers that suggest Trump’s margin is narrowing. From Monmouth’s crosstabs:

The GOP race is closer among evangelical Christians – 31% Trump, 31% Cruz, 15% Kasich, 13% Rubio – than it is among non-evangelicals – 39% Trump, 27% Kasich, 17% Cruz, 13% Rubio. However, evangelicals make up less than half of the likely electorate. Trump trails among voters who describe themselves as very conservative. This group, which makes up just over one-third of the electorate, go for Cruz (42%) over Trump (26%), but Trump leads among both somewhat conservative (44%) and moderate (36%) voters. Kasich is in second place with these groups (21% somewhat conservative and 31% moderate)…

In interviews conducted Thursday and Friday, Trump held a solid lead with 39% support compared to 22% for Cruz, 17% for Kasich, and 14% for Rubio. The race was much tighter in Saturday and Sunday interviews at 32% for Trump, 26% for Kasich, 25% for Cruz, and 12% for Rubio.

With all the discussion about hands in this election, Monmouth decided to test how Rubio and Cruz would do mano a mano against Trump. Although Rubio is struggling to make the delegate threshold in a four-way race, he would virtually tie Trump in a hypothetical two-person contest – 46% for Rubio and 45% for Trump. Cruz, though, does better, beating Trump 48% to 41% head to head. In the Rubio-Trump contest, both Cruz and Kasich voters would split at more than 2-to-1 for Rubio. However, in the Cruz-Trump matchup, Rubio voters would go 3-to-1 for Cruz, while Kasich voters would split at half for Cruz and less than 1-in-5 for Trump with another 1-in-4 saying they would not vote at all.

From a 17-point Trump lead on Friday to a six-point Trump lead on Sunday, huh? I’m tempted to chalk that up to Cruz’s showing in Saturday’s primaries, but that doesn’t add up. For one thing, the Saturday/Sunday poll numbers here wouldn’t have fully priced that information in yet. For another thing, it’s not Cruz who surged sharply over the weekend in this poll, it’s Kasich. This looks more like late deciders swinging towards the local guy, possibly due in part to Trump’s terrible debate on Thursday and Kasich’s much better performance. (A majority of Frank Luntz’s focus group after the debate had Kasich winning.) And 23 percent of Michigan Republicans told Monmouth they’re either completely undecided or had only a slight preference as of yesterday, so there’s still room for Kasich and Cruz to grow. Frankly, if you’re a Rubio voter in Michigan, you’d be a sucker not to abandon him at this point and vote for either Cruz or Kasich: Rubio’s actually below the 15 percent threshold needed to qualify for delegates in the state, which means you’re at risk of having your vote for him go completely to waste. If Saturday’s results are any indication, you may see conservatives there dump Rubio en masse at the last minute and switch to Cruz as the last conservative standing against Trump. Which means, between the post-debate trend towards Kasich in this survey and the post-Saturday Cruzmentum on the right, we may be looking at a photo finish among the top three candidates in Michigan tomorrow night.

Or we might not. FiveThirtyEight continues to have Trump as a, er, 93 percent favorite to win the state. Granted, they had him as a 96 percent favorite to win Louisiana, where he barely held off Cruz, and a 51 percent favorite to win Kansas, where Cruz destroyed him in a landslide, but those were closed elections and Kansas was a caucus. Michigan will be a heavy lift, especially since Cruz and Kasich are bedeviled by the same problem that’s been bedeviling Cruz and Rubio, namely, splitting the anti-Trump vote evenly enough that it allows Trump to sail through to victory. Even if Trump holds on tomorrow night, though, the outcome may be useful for anti-Trumpers. Beating him in Ohio (and Florida) is crucially important to forcing a brokered convention, and a strong showing from Kasich tomorrow will give him further momentum to make that happen next Tuesday.

As for Cruz fans, note the bit in the excerpt about how Cruz fares head to head against Trump versus how Rubio fares. Supposedly, Cruz is weaker against Trump going forward than Rubio would be because more primaries are being held outside the south, in more moderate states. And you see that reflected in the evangelical versus non-evangelical numbers here: Cruz ties Trump among the former but gets pasted among the latter, which explains why Trump is leading overall. If that dynamic replicates itself across other states to come, Cruz is in deep trouble. And yet, despite Trump’s advantage with non-evangelicals, Cruz beats him handily in a true two-man race in Michigan whereas Rubio struggles to win narrowly. That makes me wonder if Kasich winning Ohio really is the best hope for anti-Trumpers. If it’s true that the party’s set to align behind Cruz, even in bluish states, provided he can make it a true Trump vs. Anti-Trump choice, then the right’s top priority should be pushing Rubio and Kasich out ASAP. Take that as some cold comfort next Tuesday night while you’re weeping over Trump having just won Ohio and Florida.