The Dems versus Fred Thompson - state-level polls by Survey USA
I found an interesting poll of Survey USA, which pits the three main Democratic candidates against Fred Thompson, and measures the outcome state-by-state across 16 states. Pollster.com noted it on May 9, tho the actual data was gathered in mid-April.
Not having followed state-level polls much, I've gotten used to national match-up polls that pretty consistently see the victory margin of Giuliani over the respective Democrats vary 3 or 4 points this way or that way, Hillary doing a little worse, Obama doing a little better, and the margin of the different Democrats over McCain vary in equally modest ways - two, four, five points. I've taken note of state polls in NH and Iowa over time showing McCain, then Romney, and on the other side Edwards, often doing clearly better, and Giuliani and Hillary worse, than in the national polls. But I assumed that outside the much-visited primary states, the national picture would be pretty consistent. So what struck me right off the bat, in this listing, is the huge differences in how the respective Democrats did against Thompson from state to state.
I was also interested in the outcome because Fred Thompson is still relatively unknown. That, I'd speculate, makes him a sort of human equivalent to the hypothetical 'generic' Republican candidate. The proportion of respondents saying they'd vote for him, I am guessing, is close to the number that would vote for the typical kind of Republican candidate they'd conjure up in their mind.
That should make, I'm thinking, the scores of the respective Democrats matched against him more of a neutral reflection of
their personal appeal to cross-over voters. Whereas with a poll that pits, say, Giuliani against the three Democrats, you dont know whether any respondent's preference for the Democrat is inspired by his or her personal appeal, or by their response to
Giuliani, and how they look at the Democrat in comparison to
him.
See a listed overview of the poll's state-by-state results on Pollster.com
Now some caution is probably warranted, since sample size is relatively low: in each state they polled 500 randomly selected people (whereas national polls usually take some 1,000). But that's still a substantive sample, and makes for a total of 16 x 500 = 8,000 respondents.
First thing that strikes one is how well the Democrats do against Thompson. All three win easily against him in all but the most hardcore red states. But this should be taken with a grain of salt: polls in this phase of the campaign are still to a large extent measures of name recognition more than anything else, and Thompson's low name recognition automatically puts him at a disadvantage against any major national Democratic candidate.
It's therefore the comparison between states that interests me most. But with the long list I linked in, I couldnt really map it out in my mind. So I made a little graph. Click the thumbnail below to see it full size (you may have to click on the graph again in the window that opens, if it resizes it to fit your screen). It maps the margin of victory (or defeat) that the poll has the respective Democrats at, vis-a-vis Thompson. Below the thick black line means Thompson wins. The state names (abbreviations) and data are at the bottom.
(The GIF is of crappy quality, but there's a better-quality PNG version
here - dont forget to click on the graph again if it has resized to fit your screen.)
OK, so what do I see? And what do
you see?
What I see is this:
- Edwards outdoes both Hillary and Obama clearly when matched up against Fred, and does so consistently across states. Only in New York, her homestate, does Hillary do better, while Obama, interestingly, outdoes Edwards in California and, almost, Oregon.
- Surprisingly, Obama fares badly in this poll, matching up worse than Hillary in 11 of the 16 states. Thats odd, as in national polls he usually outdoes Hillary in match-ups against Republicans. (I dont know about match-ups against Fred Thompson though.)
I'm thinking that one possible explanation could be something I noted here before. In general election match-ups, Obama doesnt outpoll Hillary and Edwards because he actually gets a greater percentage of people who'd say they'd vote for him; but because the percentage of people who'd vote for the other guy drops, with more voters reverting to the "dont know" category. In the case of Fred Thompson as opponent, who is relatively unknown, this advantage might disappear because the number of dont knows will be high in any match-up.
Mind you, that does mean that this poll does not appear to bear out the perception that Hillary will provoke a specifically guaranteed high anti-vote whoever the Republican in the race is.
What is the pattern from state to state, if you look at their ranking in this graph from "red" (in this context: largest average lead for Thompson) to "blue" (greatest average defeat for Thompson)?
- First of all, about that ranking. It's pretty much in sync with how the popular vote stacked up across states in the 2004 presidential elections results, with three major exceptions. Ohio is much further up towards the blue end than it was in the 2004 election results; and Wisconsin and Virginia are much further up towards the red end. Might be statistical noise, who knows? It's just conspicuous because otherwise the ranking is really almost exactly that of 2004.
- Both Hillary and Edwards "defeat" Thompson by some 30 points or more in New York and Massachusetts (in both of which Obama lags).
- Edwards' lead over Thompson holds up at a pretty consistent 20-25% across a further 8 of the 16 selected states, ranging from Kentucky and Missouri on the edge of the South via midwestern states like OH, WI, MN and IA, to Washington and California on the Pacific coast. Hillary's lead on the other hand drops in these states the more conservative they become, from a comforting 20+ points in California to 10% or less in Missouri, New Mexico, Iowa and Oregon.
- Edwards' lead drops to 10-15% in red states like Texas, Virginia and Kansas, but Hillary actually drops beneath Thompson's number in the latter two. Alabama is the great equaliser, in which all Dems lose in similar fashion.
The Edwards vs Hillary comparison thus seems to suggest a narrative of sorts: they both lose in similar fashion in the worst of states, and both win comparably big in the best of states, but in the purple and all but the most deep-red of red states, Edwards has a clear advantage.
But what about Obama? His numbers appear to vary more randomly.
- One thing stands out: he does comparatively very well on the Pacific coast. He gets his three best match-up scores in California, Oregon and Washington.
- Another observation is that the four states in which he loses to Thompson are all in the South: AL, VA, TX and KY. That's actually all the Southern states in the sample bar Missouri. This is noteworthy because in Texas and Kentucky, Edwards and Hillary each best Thompson, in Kentucky even both with 10+ percent. Is it the Southern racism factor? That's significant if you believe the CW that the Democrats cant win the Presidency without winning any Southern states.
Two notes of relativation on how Edwards outdoes the other Dems almost across the board in this particular set of match-ups. The first is that the data was collected some three weeks after the news of Elizabeth Edward's cancer came out, so there may still have been a sympathy bounce.
The second is that I imagine that, here, Thompson's personal image may play a role after all. It is easy to imagine that Thompson's folksy (if faux-folksy) Southern profile works well in pulling red- and purple-state voters in, when it is contrasted with the "cold", blue state profile of Hillary (or, dare I say, the blackness of Obama). But against Edwards, who shares much of the same profile, he doesnt have any particular advantage there.
That raises the question whether the narrative I mentioned above, about Edwards having an advantage over Hillary that shows up as states turn purple, then red (with the exception of the deepest of reds of Alabama), is indeed a general one or also one that's relative to Fred Thompson as opposite party, in particular.
It is in any case still kind of hard to gauge how the degree in which Edwards does better than Hillary and Obama varies from state to state from this graph, which after all primarily focuses on how they do better than
Thompson. So I created another one.
In the graph above, you'll see the thin grey line that represents how the three Dems do
on average against Fred Thompson. OK, so how much better or worse
than their common average do the three candidates do, state by state? That's in the graph below. The higher their point above the thick black line, the better they do in comparison with the other two; the lower their point below the line, the worse they do than the other two. Click the thumbnail for full size.
(Better-quality PNG version
here - dont forget to click on the graph again if it has resized to fit your screen.)
I found it useful to filter out Obama, and compare only Hillary and Edwards - so here is the same graph but with just the two of them, and their scores set off against the average of just the two of them:
(Better-quality PNG version
here - dont forget to click on the graph again if it has resized to fit your screen.)
The latter one is very clear, and makes for specifying the narrative I suggested above. Edwards does not, as one might expect, have an advantage over Hillary that
increases the "redder" states are. His advantage over Hillary opens up as soon as you leave the bluest states (New York, California) behind - and remains roughly stable at 5-10% all the way until the Dems' collective "defeat" in Alabama closes the bracket. There's just three or four exceptions: Edwards' advantage over Hillary opens up beyond 10% in Kansas, Virginia, to some extent Missouri -- and Iowa.
Iowa's position in that list is significant of course because of its leading role in the primaries. Edwards has put a lot of sustained effort in campaigning in Iowa, from before Hillary or Obama had even started their campaigns, and it seems to be paying off for now - suggesting a tough run for Hillary.
The former graph, though, just increases the puzzle. The four states in which Obama did worst
against Thompson may have been four of the five Southern states in the selection, which is a clear pattern. But the states in which he does worst in comparison
with Hillary and Edwards are a seemingly random bunch: Kentucky, Massachusetts, New York, and Ohio, followed at a distance by Texas and Minnesota.
OK, all of that just from the top of my head.. and mind you, all based on just one set of polls. Would probably be worthwhile to follow this up with other state-by-state match-up polls with Thompson or, if available, a generic Republican, over time.
In the meantime, any of you any thoughts?
Why not?
And which side of the line did that post of mine come down on?
)
