Usually, an incumbent who can’t get to 50% in pre-election polling is in trouble.  An incumbent who can’t get to 45% is almost certain to lose, especially with less than three weeks to go, as undecideds are already disinclined to support the candidate they know best.  What does it mean when a poll from a firm linked to Democrats in a key Democratic state can’t produce anything better than 44% for a Democratic incumbent and a virtual tie with a Republican challenger?

Disaster:

Released Wednesday, the poll of 600 likely voters showed Obama leading Romney in Michigan, 44.2 percent to 40.5 percent, but Romney also within the sampling error of 4 percentage points — meaning it’s a tight race. …

The research also revealed a significant swelling of support among Republicans for Romney. In the latest survey, 44 percent of Republicans said they were very excited to vote for the Michigan native, compared to 16 percent in the June survey, before Romney was formally nominated by his party.

By contrast, 48 percent of Democrats reported being very excited to vote for Obama, up from 39 percent in June.

Who conducted the poll?  That’s also an interesting story, emphasis mine:

The poll was conducted by the Michigan polling firm Denno Research and commissioned by Grand Rapids-based Lambert, Edwards & Associates, which also has offices in Lansing. Dennis Denno, president of Denno Research, also is chief of staff to state Sen. Virgil Smith Jr., D-Detroit.

In other words, this isn’t exactly an independent firm.

The gap in this case is about in line with previous polling in Michigan, where Obama has a 4.4-point lead in the Real Clear Politics average.  However, it’s the lowest level of committed support for Obama in the past three weeks.  Prior to that, Obama had a double-digit lead in RCP’s Michigan average for a good part of September.  This could therefore be an outlier — but if so, then the Democrat for whom the pollster works should find himself another chief of staff.

There is another telling point in the data.  Romney leads among independents by six points at 36/30, although a third remain undecided.  Undecideds usually break sharply away from the incumbent in the final weeks of an election, especially when the incumbent isn’t providing any real case for a second term in the midst of economic stagnation.  In 2008, Obama won Michigan by 16 points, 57/41, and won independents by ten, 52/42.  If Republicans pick up most of the independents still undecided and reduce the 12-point turnout advantage Democrats had in 2008, Romney could win this state — and that would all but close out Obama’s hopes for a second term.