A lot of the VP candidate talk (for both candidates) has focused on how the choice of a running-mate can affect the Electoral College math. The conventional wisdom is that a VP candidate can "carry" his/her homestate. I've long believed (and argued) that this is a myth, and should never be a factor in choosing a running-mate. At the very least, the potential for this "home state advantage" is so small that there are undoubtedly many other factors in the selection that should play a much more important role. For instance: choosing someone who would make a good President, appealing to a certain demographic, strengthening a vulnerable issue area, choosing something with media/campaigning chops, etc.
So I did a bit of a historical study into Vice Presidential candidates. My method was fairly simple. I took a look at a VP candidate's homestate's results and compared them to that state's results in the prior Presidential election and the results in the next Presidential election. (If a candidate ran for VP or the Presidency multiple times, I looked at the results for every election until they were no longer a candidate.)
However, as we sometimes forget after having lived through the 2004 and 2000 elections, not every Presidential election is a nailbiter. So, when looking at each state's election results, I looked at them as a comparison to the national results. For example: Bush Sr. won the 1988 election by an 8% margin nationally. He won Louisiana by 10%, and lost Iowa by 10%. So, Louisiana went 2% more Republican than the nation as a whole, and Iowa went 18% more Democratic than the nation as a whole... I'd call Louisiana R+2, and I'd call Iowa D+18.
Though I'm well aware that national shifts in support aren't cleanly reflected in the relative shifts for each state, I think that looking at state results compared to the national result creates a much more accurate picture than does looking at state results without any kind of frame of reference.
___________________________________________________________________
1960 Democrat - Lyndon Johnson, TX. Traditional wisdom says that, without the benefit of LBJ, the Democrats would've lost Texas (and the election) in 1960. The results don't make a great case for the LBJ effect in Texas: D+4 pre-LBJ, D+2 with LBJ running for VP, D+4 with LBJ running for President, and D+2 post-LBJ. I'd like to point out that the results in 1960 are polluted by two things: the fact that LBJ was running for re-election as Senator from TX at the same time he was running for VP, and the allegations of extensive vote fraud in TX.
1960 Republican - Henry Cabot Lodge, MA. Any homestate advantage in Massachusetts in 1960 would have benefitted JFK, not HCL.
1964 Democrat - Hubert Humphrey, MN. Results: D+1 pre-Humphrey, D+5 with Humphrey running for VP, D+13 with Humphrey running for President, and D+18 post-Humphrey. The only thing I think we can learn from these results is that Minnesota was increasingly trending Democratic during these years.
1964 Republican - William Miller, NY. The results: D+5 pre-Miller, D+15 with Miller, and D+6 post-Miller. If anything, there's a reverse home-state advantage in this case.
1968 Democrat - Edmund Muskie, ME. The results: D+15 pre-Muskie, D+13 with Muskie running for VP, and D+0 post-Muskie. There's no evidence of a VP effect. Other notes: Maine appears to be a fickle state, and/or seems to be trending more Republican during these years.
1968 Republican - Spiro Agnew, MD. The results: D+8 pre-Agnew, D+2 during Agnew's first VP run, R+1 during Agnew's second VP run, and D+4 post-Agnew. For once, we see a logical VP home-state advantage.
1972 Democrat - Sargent Shriver, MD. It's probably not worth doing an analysis here, because the 1972 Republican VP candidate (Agnew) was also from Maryland. And Shriver never had very deep political connections to his home state.
1972 Republican - Spiro Agnew, MD. See analysis above.
1976 Democrat - Walter Mondale, MN. The results: D+18 pre-Mondale, D+11 with Mondale for VP the first time, D+14 with Mondale for VP the second time, D+18 with Mondale running for President, and D+15 post-Mondale. While he arguably gave Minnesota a boost in 1984, when he was running for the Presidency, there's no evidence of a VP effect. In fact, the two years he was running for VP were the two worst years for Democrats in Minnesota over this timespan.
1976 Republican - Bob Dole, KS. The results: R+15 pre-Dole, R+10 with Dole, and R+15 post-Dole. Once again, there's a reverse home-state effect.
1980 Democrat - Walter Mondale, MN. See analysis above.
1980 Republican - George H. W. Bush, TX. This one's complicated, because George Bush Sr. was on a Presidential ticket for 4 elections in a row. The results: D+1 pre-Bush, R+4 with Bush-for-VP, R+9 with Bush-for-VP the second time, R+5 with Bush running for his first Presidential term, R+9 during his second run, and R+13 post-Bush. One thing is certain: Texas was definitely turning more Republican between 1976 and 1996. The individual effect of Bush is clouded by this, as well as by the long period of time between Bush-less elections.
1984 Democrat - Geraldine Ferraro, NY. The results: D+7 pre-Ferraro, D+10 with Ferraro in the ticket, and D+12 post-Ferraro. Like the results for Hubert Humphrey, all we can tell from this is that NY was trending more Democratic during these years.
1984 Republican - George Bush, TX. See analysis above.
1988 Democrat - Lloyd Bentsen, TX. Definitely not worth looking for a Bentsen effect, considering that the Republicans had a fellow Texan at the top of their ticket.
1988 Republican - Dan Quayle, IN. The results: R+6 pre-Quayle, R+12 with Quayle the first time, R+12 with Quayle the second time, and R+14 post-Quayle. Though there's a big jump for the Republicans between 1984 and 1988, it seems less likely to be due to the Quayle effect, and more likely due to Indiana trending more Republican during this time. Similar to Democrats Hubert Humphrey and Geraldine Ferraro.
1992 Democrat - Al Gore, TN. The results: R+9 pre-Gore, R+1 with Gore-for-VP, R+6 with Gore-as-VP, R+4 with Gore running for the Presidency, and R+12 post-Gore. I'm willing to call this a homestate bump. The 1988 election is when Democrats had their best performance in TN, and that's also the year that Gore should've seen his biggest homestate advantage (as he was still a Tennessee Senator, rather than a national figure).
1992 Republican - Dan Quayle, IN. See analysis above.
1996 Democrat - Al Gore, TN. See analysis above.
1996 Republican - Jack Kemp, NY. The results: D+10 pre-Kemp, D+20 with Kemp, and D+24 post-Kemp. Kemp had no effect in this strongly Democratic state. The Democratic trend is part of a larger Democratic shift in NY that's pretty steady between 1976 and 2000.
2000 Democrat - Joe Lieberman, CT. The results: D+10 pre-Lieberman, D+17 with Lieberman, and D+13 post-Lieberman. Holy crap, he had Joementum! (Random aside: I love this poll.)
2000 Republican - Dick Cheney, WY. The results: R+21 pre-Cheney, R+41 during Cheney's first run, R+37 during Cheney's second run, and R+19 projected for 2008. Who knew? The people of Wyoming seem to like Dick more than your average Republican.
2004 Democrat - John Edwards, NC. R+13 pre-Edwards, R+10 during Edwards' VP run, and R+7 projected for 2008. Well, the good news is that North Carolina is slowly trending Democratic; the bad news is that Edwards isn't responsible.
2004 Republican - Dick Cheney, WY. See analysis above.
_____________________________________________________________________________
Of all of the above, only 3 candidates had a demonstrably positive homestate effect: Spiro Agnew, Joe Lieberman, and Dick Cheney. And two of those three involved states that couldn't possibly have any effect on the election anyway. Connecticut is a solidly Democratic state where the Democratic VP candidate got a bounce; Wyoming is a solidly Republican state where the Republican VP candidate got a bounce. Spiro Agnew's Maryland was a swing state at the time. The problem is that the Agnew effect wasn't powerful enough to actually swing Maryland to the Republican column in 1968. And in 1972, the effect was irrelevant because Nixon/Agnew won by a landslide anyway.
Meanwhile, two candidatess demonstrated a reverse homestate effect: William Miller (NY, 1964) and Bob Dole (KS, 1976). In other words, not only were they not able to swing their homestates more towards their party, the evidence is that they actually hurt their party's ticket in their homestates.
One more candidate demonstrated a possible positive homestate effect: Al Gore. Even if we concede that Gore helped turn Tennessee more Democratic in 1992 and 1996, neither of those elections were very close anyway.
The conclusion is that, in the last 48 years, not a single Vice Presidential candidate has helped alter the election results by virtue of the state they happened to come from.
.... So can we finally stop talking about picking a certain running-mate based on what state they live in?