The important thing to remember about Senate elections is that every single one counts. It's not like the House, where for the next two years it doesn't matter all that much whether the Republicans have a 15 seat or 25 seat or 35 seat advantage...it matters some, but not all that much. Party "control" -- majority, that is -- simply overwhelms everything else.
That's not true in the Senate. Sure, it matters quite a bit which party is in the majority. But the size of the majority matters, and the identity of individual Senators matters far more than the identity of individual Members of the House matters.
(Just to begin with: the House has nearly absolute majority party control of what comes to the floor for a vote. In the Senate, any determined individual Senator can probably bring anything she wants to the floor and force at least a procedural vote on it).
And then there's the issue of term length. Yes, thanks to incumbency advantages elections in the House for the 113th Congress beginning in January 2013 certainly do affect what will happen in the cycles which elect the 114th and 115th Houses. But in the Senate, we're directly, right now, electing one third of the 114th and 115th Senates. Or, to put it in practical terms: suppose that Barack Obama wins, but that no Supreme Court opening shows up in the next four years. That's okay; we're voting this November on a third of the Senators who will, in that case, confirm or reject SCOTUS nominees from the next president, not the one that (in that hypothetical) we're re-electing now. Or course, there are also incumbency factors on the Senate side...if Tommy Thompson wins he may well remain in the Senate for 12 or 18 years, but if he loses there's no way he's going to be elected again.
So how does it look? If you're the Democrats, just shockingly good (and I'm writing this one from the Dem's perspective -- easier than switching back and forth). Nate Silver right now is projecting that Democrats either hold their current 53 seat majority or drop down one (actual forecast is 52.4 Democratic Senators). Over at Pollster, Democrats (counting King in Maine as a Democrat, which it's worth pointing out isn't 100% certain) have solid leads that would give them 46 seats plus five more leaning Democratic, with six remaining tossups. Republicans have only 43 solid seats, with no states leaning towards them.
That leaves a total of 11 seats in play, five leaning towards the Democrats. Going through those eleven, starting with the five leaning Democratic:
* Murphy leads by 4.7% in Pollster's average in CT; Nate Silver's prediction model calls it 78% for the Democrats.
* Warren up by (Pollster) 3.5% in MA; Silver calls it 89%.
* Brown up 4.5% in OH; 538: 93%.
* Casey up only 3.8% in PA; 538 has a larger current lead, and says 93%.
* Baldwin up 3.8% in WI; 538: 85%.
I think the 538 odds sound about right; it wouldn't be a shock if one of these went to a currently-losing Republican, but a lot more likely that Democrats will sweep these five than they'll drop two or more of them -- although a late GOP surge could do just that. The only think I see here would be the possibility of a hidden vote against Baldwin; there's of course no way of knowing if that might happen.
In two toss-ups, the Democrat leads in the polls; Democrats reach the current 53 by retaining the leaning D and these two:
* Kaine up 3.0% in Virginia; 538: 82%.
* Tester up 1.6% in Montana; 538: 43%.
Virginia looks to me (and 538) a lot more like Wisconsin than like a true toss up. Montana, on the other hand, is better for Republicans. The difference between Pollster and Silver there are the state factors built into the 538 prediction model; the two basically agree on the current very slim Tester lead.
Four toss-ups with Republicans in front:
* Heitkamp down by 2.7 in ND; 538: 18%.
* Carmona down 1.2% in AZ; 538: 31%.
* Berkley down 2.6% in NV; 538: 26%.
* Donnelly down 4.9% in IN; 538: 36%.
Obviously the big one here this week is Indiana, where there are no public post-gaffe polls. Silver has the current Mourdock lead (and the Berg lead in ND) smaller than what Pollster makes it.
My initial reaction is that I'll be surprised if Mourdock can survive...it's not a very big lead, and judging from the way both parties are responding there's general agreement that it was a major gaffe...which means that opinion leaders are going to treat it as one, which means it could easily filter through to voters. We're not talking about an experienced successful politician with a long history of a close connection with his constituents, either. However, it's hard to know whether it plays out that way without any polling confirmation at all. Presumably we'll get some soon enough.
On the other three: Nevada sure would seem to be connected to questions of turnout and the presidential election. So whatever you think about that is probably what you think about this election. The other two look less likely to me for the Dems.
So what do we have? Best case for Republicans, with a big surge to Romney (or it turns out that Romney-leading polls had it right), it's plausible for them to get as many as 54 seats. For the Democrats, best-case is 57 seats. That's a huge swing! However, the Democratic path to 57 is basically to keep all their leads (at least one of which seems quite tough), win Indiana, and then overcome leads in just three other states. That's few enough, and the leads are small enough, that any one could easily be a consequence of either local factors changing the polls or the polls turning out to have missed something. All three falling that way, plus everything else, seems highly unlikely...it would probably take a real Obama surge to make it happen.