County That Most Closely Mirrors State Results?

[Verily]

I'd say that Pike and Scioto counties are probably the best predictors for Ohio.

What about Stark County? Doesn't it have a bellwether reputation?

For New Jersey, I'd actually say Burlington more than Bergen. Bergen was for a while more Republican than the state as a whole (voting for Franks over Corzine in 2000, for example), and it is heavily influenced by which way the New York metro is swinging.

[Verily]

I'd say that Pike and Scioto counties are probably the best predictors for Ohio.

What about Stark County? Doesn't it have a bellwether reputation?

For New Jersey, I'd actually say Burlington more than Bergen. Bergen was for a while more Republican than the state as a whole (voting for Franks over Corzine in 2000, for example), and it is heavily influenced by which way the New York metro is swinging.

But don't think that Bergen's trend from Republican to Democrat on par with how the state is going have a huge influence?  I think the states numbers are greatly affected by Bergen because it is incredibly populated and can swing either way.

Bergen is large, but it isn't overwhelmingly large. We're not talking about Cook County, IL or LA County, CA here. Middlesex, Ocean, Burlington, Essex, etc. are also all very large; really only Sussex, Warren, Hunterdon, Cape May and Salem are "small" counties (and certainly not by national standards!).

Bergen is also not the only county in New Jersey which is marginal and can swing both ways, nor does it tend to swing substantially more than other counties. In fact, many counties swing much more than it and therefore have a larger effect. New Jersey's pro-Bush swing in 2004 was more or less entirely concentrated in Ocean County and more broadly in South Jersey, for example. (This is probably because Bush advertised almost exclusively in the Philly media market to cover PA, too, but that's beside the point.)

[Verily]

Probably a bay area suburb type like Contra Costa.

But there is probably no county that can capture 'California as a whole'.

Sacramento County comes pretty close, I think, though politically it is more marginal than California as a whole.

[Verily]

Okay, here's a brief numeric analysis for New Jersey.

1996
State: 17.9
1. Atlantic: 17.9 (0.0)
2. Burlington: 16.9 (-1.0)
3. Gloucester: 19.7 (+1.8)
4. Passaic: 21.1 (+3.2)
5. Bergen: 13.8 (-4.1)

2000
State: 15.84
1. Burlington: 15.35 (-0.49)
2. Gloucester: 17.50 (+1.66)
3. Bergen: 13.62 (-2.22)
4. Passaic: 18.70 (+2.86)
5. Atlantic: 18.97 (+3.13)

2004
State: 6.68
1. Cumberland: 6.60 (-0.08)
2. Burlington: 6.96 (+0.31)
3. Atlantic: 5.90 (-0.78)
4. Gloucester: 5.32 (-1.36)
5. Bergen: 4.45 (-2.23)


This non-rigorous analysis indicates that Bergen is not the bellwether county, having been less accurate than Burlington in predicting New Jersey's partisanship in all of the last three Presidential elections. Gloucester also consistently outperformed Bergen while always underperforming Burlington. Additionally, Atlantic averages out to being a better indicator than Bergen though it performed worse than Bergen in 2000.

Bergen has also consistently been 2-4 points more Republican than the state as a whole.

[Verily]

Desnethé—Missinippi—Churchill River went from Conservative in 2004 to Liberal in 2006, but there was some accusations of fraud/something bad going on about the victory (both parties gained % in 2006)

The fraud accusations were made by the Conservative candidate, Jeremy Harrison, and found to have no merit at all. After he only just barely won running for the Sask Party in the 2007 election in a conservative area of his former federal riding, I find it unlikely that there really was any fraud. Much more likely that higher First Nations turnout did him in.