The background: There is a special election in Pennsylvania for the vacant House seat in District 18. Clinton lost the district by 20, Obama by 17. This is not a seat we really should have a prayer in. SO these results are shocking:
| Likely voters with leaners | “Surge” Model | High Turnout | Low Turnout |
| Rick Saccone | 49% | 48% | 50% |
| Conor Lamb | 46% | 44% | 45% |
| Other | 1% | 2% | 1% |
| (VOL) Undecided | 4% | 5% | 4% |
| (n) | (320) | (379) | (320) |
Monmouth has noted the recent results in special elections and as a result tried to predict the turnout in this district based on those results. But the different turnout models don’t seem to matter much: under all of them this is a pretty close race.
Trump’s approval rating in 51-47, a bad number in a district he won by 20.
The Cook report has noted a divergence between recent generic ballot polling and polling coming from individual House races. In general, the generic ballot suggests a closer race than district polls are finding. In this regard it is worth remembering that the Generic Ballot is in some ways a crude tool.
As Cook noted:
Private polling from both parties shows some pretty grisly numbers in individual states and districts. Democratic candidate recruiting in House races is going spectacularly well—in fact, some worry it’s gone too well, with too many contested primaries. Fundraising for Democratic challengers is going very well (and you should ignore Democratic National Committee and Republican National Committee fundraising numbers, which are some of the most irrelevant midterm indicators in all of American politics). All of these signs support the very clear historic pattern of midterm-election losses for the party in power.
The Monmouth polls suggests a possible wave of significant proportions.