Friday, November 10, 2006

Why Kathleen Falk REALLY Lost

There's been a lot of speculation, much of it wrong, about why Kathleen Falk lost the Attorney General's race to JB Van Hollen. As Daniel Patrick Moynihan said, "Everyone is entitled to his own opinion, but not his own facts." So join me to look at the facts in this race, and learn, to the extent possible, the truth about why this was the only statewide race we lost in Wisconsin this election cycle.

This was a Democratic wave election, and every statewide Democrat seemed to benefit except for Kathleen Falk. She significantly underperformed Jim Doyle, and that's what cost her the race. Doyle received 7.3% more votes than Falk did. Had she been able to perform just one percentage point better she would have won the election. Here are the four main theories I've seen given to explain why Falk wasn't able to pick up those votes:

1. The "Blame Madison First" crowd, which includes a significant faction within the Democratic Party, claims that Falk lost because liberal Dane County (Madison) voters were mad she beat Lautenschlager in the primary, and withheld their votes to get even. This article in the Capital Times makes a slightly broader version of this claim: "Although Falk tried to reach out to Lautenschlager's supporters after the primary, Tuesday's results indicated that many of them did not come back to her."

2. A favorite among progressive bloggers is the belief that Falk lost because she didn't have a strong and consistent message, while Van Hollen did. To quote Folkbum, "The campaigns that won here in Wisconsin on Tuesday were the ones with the clear, unambiguous, crystalized messages... JB Van Hollen ran on a single clear message ... I want to fight crime! I'll get the terrorists! I'll clear out the illegal immigrants!" In Effect made a similar point.

3. The far left favors the "Doyle is so corrupt that voters wanted an AG who will keep an eye on him" theory. See this post from The Nate Report for example. Perhaps this will give away my own ideological leanings, but my belief that Falk wouldn't be able to effectively police Doyle was one of the biggest reasons why I supported Lautenschlager in the primary.

4. Kathleen Falk and her supporters complain that Wisconsin Manufacturers and Commerce simply bought the race with several million dollars of attack ads. This is a rather ironic claim since, as I showed here and here, the only reason Falk was even in this race was because she bought her primary victory over Peg Lautenschlager.

So let's look at the data and sort out these somewhat conflicting claims. You can click on any of the maps to get larger versions.

This map shows Falk's underperformance relative to Doyle. As you can see, there's a strong and fairly clear pattern, and this pattern is the clue we need to solve the mystery. But what can we determine from this?

It's certainly clear from the map above that Falk didn't lose because of Dane County, which doesn't stand out as being significantly worse than the rest of the state. In fact, Falk underperformed similarly in both Dane County and Milwaukee County despite losing the primary in Dane County by 20% and winning it in Milwaukee County by 13%. And the pair of maps below shows that bad feelings from the primary race had little to do with the general election. Falk's underperformance is on the left, and the primary election results are on the right. Very different patterns, as you can see.

So we can safely lay to rest the idea that Falk's loss had anything much to do with her nasty primary battle against Peg Lautenschlager. The Democratic base was indeed mad at Falk, Doyle, and the Dem Party establishment for deposing Lautenschlager in the primary, but they still turned out and voted for Falk in the general election.

The next thing to check is if Falk's underperformance compared to Doyle matches with her overall electoral performance. Maybe some voters just didn't vote for the downticket races. But the map below shows it's more complicated that that.

Once again, the map on the left shows Falk's underperformance compared to Doyle, but this time the map on the right shows the AG general election results. They're closer, but still not really a match. I'm not going to bore you with the details, but I also compared Falk's underperformance to the results of the Governor's and the '04 Presidential race, and neither matched very well.

Which finally brings me to the meat of the issue. The thing that correlated most closely with the pattern of Falk's underperformance is Wisconsin TV media markets. This final (I promise!) map again shows Falk's underperformance, but also includes the media markets:

The markets are:
1. Duluth/Superior
2. Wausau/Rhinelander
3. Green Bay/Appleton
4. Minneapolis/St. Paul
5. La Crosse/Eau Claire
6. Madison
7. Milwaukee

Look at, #4 on the map, the Minneapolis/St. Paul market. Three out of the four counties where Falk actually overperformed Doyle are in this market. What's unusual about this market is that Wisconsin campaigns rarely advertise here because it's very expensive and reaches few Wisconsin voters. So it provides us a "pristine" model of what happens when voters don't see TV ads. The fact that Falk did so well there shows that in the absence of TV ads she was quite capable of matching Doyle almost vote for vote. Notice that this also tends to weigh strongly against the theory that voters supported Van Hollen because they wanted an AG to keep an eye on Doyle.

But voters in all the other media markets were inundated with attack ads against Falk from both Van Hollen and Wisconsin Manufacturers and Commerce (WMC). Negative ads are designed to suppress votes, and that's exactly what they did.

But there's still one more piece of the puzzle that I need your help to solve. Check out the La Crosse/Eau Claire market, #5 on the map. Falk's performance wasn't great there, but it was still significantly better than in any of the other remaining markets. Had Falk done this well across the state she would have won. My guess is that either Van Hollen or WMC didn't advertise in that market. If you live in that area please leave a comment and let me know what ads you saw and on which stations.

But what about the message theory? I just don't have enough information to make a good judgment on this one, although it's clearly secondary to the attack ads. But my gut feeling is that Van Hollen gained some additional advantage from his strong and consistent message, and this contributed to his significant overperformance compared to GOP Gubernatorial candidate Mark Green.

There are some anomalies in the maps that I can't explain, such as Shawano and Sauk Counties. If you have any idea what was different about these counties, or any of the others that didn't fit the general pattern, please let me know in the comments.

The results of this election don't bode well for Kathleen Falk's political future. Rightly or wrongly she'll get the blame for costing us the AG's race in what should have been a Democratic sweep. But perhaps even more important is that WMC has shown they don't like Falk, and will go to the mat to keep her out of statewide office. To have any chance of success at running for AG or Governor in the future Falk would probably have to make a deal with WMC, which would mean selling out on her progressive history and values.

The involvement of WMC in this race is one of the reasons I've always felt, and still do, that Peg Lautenschlager would have been a stronger candidate than Falk in the general election. I may be wrong, but I think WMC would have had a much harder time demonizing Lautenschlager, if they had chosen to get involved at all. But this is just speculation.

I've shown pretty conclusively that Falk's loss wasn't due to her divisive primary battle or to voters wanting an AG to keep an eye on Doyle. Turns out the insiders are probably pretty much correct when they say that WMC bought the election. Which seems almost like cosmic justice given that Falk used exactly the same technique to win the primary.

Live by the sword, die by the sword...


Anonymous said...

Fascinating analysis! Thank you for doing this. You make a convincing case in debunking various reasons why Falk didn't get elected and also that media markets corelate with election results. However, I can't quite swollow the leap you make to giving WMC the credit for putting JB over the top.

Give the voters some credit! Perhaps they evaluated both candidates and found JB to be more qualified. Both Lautenschlager and JB hammered Falk for not having ever prosecuted a single criminal case, "not one." I think Falk just never made the case for why she was qualified to fill Lautenschlager's shoes. JB on the other hand with his experience as a DA and a US Attorney did.

Can't the reason simply be that Falk was a poor candidate? Otherwise, this theory would suggest that WMC could buy any election! For Falk to lose in a year with a strong wind at Democrats' backs, I think she has no one to blame but herself.

proletariat said...


I found it interesting too.

I think you make a good case that those under votes were not disgruntled lefties, but on the center right. But..

You more or less blow off the various arguments with the Falk / election map, but the effect of under votes is as clear as day. In the center of the state where undervotes are strongest, Van Hollen wins by the pinkest of margins.

Your media explanation is a shot in the dark at best. While it shows the lack of a Falk undervote, it changes nothing. At the end of the day Falk and Doyle still won those counties.

As far as the media, that may explain the under vote. With only information from the national media one would vote strictly along party lines irrespective of candidate. In other words true but irrelevant.

Do we know WMC advertised throughout the state? In which areas of the state were they the strongest? I think you are giving way too much power to WMC. Yes, they are a terrorist organization but they didn't make or break the election.

What I do think is that there were enough voters who voted for Doyle, but who in the end selected Van Hollen. Was it the media, WMC, check on Doyle, or as one (female) progressive stated the bitch factor? I see nothing in your analysis that tries to separate these factors.

I still hold that corruption, a national theme, was a factor in the end. This corruption vote however was not disgruntled progressives, but those on the center.

Ben Masel said...

The Shawano anomaly: I suspect anti-Doyle more than pro-Falk, and tied to the casino issue. Unfavorable geography means the Menominee casino spills less economic benefit to their neighbors than other tribal gambling, so lingering tensions take longer to fade, and the perpetual compact imbroglio damages Doyle more.

Anonymous said...

Paul Soglin has his own take on Falk's loss:

(See his post titled "Why Falk Lost")

Soglin suggests the following, the logic of which is a little murky to me:

"I don't have the time to do the research, but the next step would be to take a few Wisconsin counties, liberal and conservative, and see if the Falk vote dropoff (from Lautenschlager) was greater in liberal or conservative counties. If it was greater in the liberal counties, the resentment against Falk on the left was the reason. If it was greater in the conservative counties, then it was the moderates and swing voters that cost Falk the election."

Russell Wallace said...

Thanks for the comments! Some thoughts:

Anonymous, money does buy elections. It's certainly easier when the opposing candidate is weak, as was likely the case here, but enough of a money advantage can swing almost any election. It appears that WMC outspent both candidates put together, and that's a hell of a lot of attack ads hitting in the last two weeks before the election.

Proletariat, I don't understand how you could look at the correlation between Falk's undervote and media markets, and say my explanation "is a shot in the dark at best." TV ads are the single largest determinant of the outcome of statewide and national elections in the US. We may not like it, but that's the way it is.

You're correct that in the absence of any information about candidates people tend to vote straight party tickets, but you're making the additional assumption that voters who don't see TV ads are uninformed, and this isn't necessarily the case.

Keep in mind that media market essentially means TV market. The Wisconsin voters in the Minneapolis/St. Paul market do have access to Wisconsin media through newspapers, radio, and the Internet. They just can't see Wisconsin TV stations. And because their opinions aren't shaped by TV (money), they may in some ways have a more balanced and fair (not fair and balanced!) view of candidates than the rest of the state.

Then again, I'm constantly amazed by just how ignorant and uninformed most voters are, regardless of where they live...

proletariat said...


What we don't know is where WMC ads were played. My guess is the Falk ads probably did more damage than WMC itself. Why in quadrant 4 does Falk have over, under, and neutral votes. You'd think that if it was solely WMC and TV, they'd be one color.

What was most striking to me was the center part of the state where the most "under votes" took place Van Hollen won by the pinkest margins. Many of these Doyle won. The nastier WMC ads were directed at Doyle, how does that fit into your analysis.

With all due respect it seems you went into your analysis with an idea beforehand. I just don't think it holds up. How would those under votes be explained by WMC. I'd be curious where Falk and that Van Hollen is not tough on crime ad played. That would certainly be a more plausible explanation.

Anonymous said...

Hey Fraley why don't you put your name to things instead of hiding behind the anon label.

Without the WMC $ how many people would have even known she "never prosecuted a crim case"???

Aaron said...

proletariat: Why in quadrant 4 does Falk have over, under, and neutral votes. You'd think that if it was solely WMC and TV, they'd be one color.

Go back and read the analysis again -- market 4 likely had no or very few ads, because it's the Twin Cities market.

proletariat said...


That's the point, if it was solely the tv market you'd assume there would be no under performance of Doyle in St Croix. St. Croix has the same market as the other counties why did so many voters who went Doyle say no or not vote for Falk.

Looking at the maps again, they are interesting, it seems pretty clear that Waukesha and Washington counties gave Van Hollen the victory. Falk's most populous county of similar strength is Dane.

I could see a WMC media blitz having a role in those two counties. But, in no way does quadrant 4 and the Minn media market even come into play here.

Anonymous said...

Another factor?
Falk trailed Doyle's vote totals in Dane Co. by more than her margin of loss statewide. When you look ward by ward in Dane, it's clear that some old animosities toward Falk in Dane outside Madison was in play. Rural and suburban voters were never as high on Falk as their city counterparts.
Also, Doyle's unfavorables in Dane were high, but stem cells/abortion etc. trumped them. Falk didn't have a similar trump card.

Russell Wallace said...

Proletariat, my point was that because there was no TV advertising in the Minneapolis/St. Paul market, it lets us see what effect TV ads had in the rest of the state. And the net effect was to significantly depress the vote for Falk.

St. Croix is a red herring. It's a conservative county in the middle of a pretty Democratic area. That Falk performed somewhat differently there than in the surrounding counties isn't terribly surprising.

Keep in mind that three out of the four counties where Falk outperformed Doyle are in the Minneapolis/St. Paul media market. The odds of that occurring randomly are astronomically small.

Washington County is the most conservative county in the state, and Waukesha is probably the second. You would naturally expect Van Hollen to do very well there. But keep in mind that the maps show percentage margins, not the number of votes, so you have to be very careful when you try to interpret them in terms of which counties provided the margin of victory.

You're correct that I went into this analysis with a set of preconceived notions, but they probably weren't what you think. My politics are pretty left-leaning, and I basically agreed with you prior to making the maps. But my background is science and engineering, and I've learned that when reality conflicts with ideology it's usually wise to bet on reality.

I would have liked nothing more than to show that Falk was a terrible candidate who shot herself in the foot by running a nasty and divisive primary. But the data just doesn't support that.

Anonymous (the most recent one) I addressed your point in my analysis, so you might want to look at it again. The county by county data doesn't support the idea that Falk's underperformance resulted from primary grudges. They could have been a factor, but only a relatively small one.

Aaron said...

proletariat, the point is that since ads didn't run in market 4, there's no reason to expect to see unified results among those counties. If the ads were the deciding factor, we would expect markets that got the ads to look different -- that is, less varied among counties -- than the market that didn't get the ads.

proletariat said...


Here is what it comes to for me, you see over votes in a quadrant than has no Wisconsin media, where there are major undervotes through the rest of the state. You correlate a uniform pattern of undervotes with a WMC ad blitz.

Here's the problem, neither WMC nor Falk advertised through the whole state. Do you have evidence otherwise. If it was WMC I would expect to see less uniformity, larger under votes in some media markets, less in others.

I think what the maps show are interesting but its kind of like saying there are no pine trees in quadrant 4, so it was the pine trees that cost the election.

In all honesty, I am suspicious of a grand narrative such as WMC. It may have made a difference in a few counties to give the election to Van Hollen, but the whole state thesis is a bit too much.

Your whole thing was patterns, but are they really needed to tilt the election. Certainly corruption, primary anger, bitch factor, WMC, Falk's ads while not strong enough to alter any pattern may have been enough to tilt the election to Van Hollen.

Your conclusion I agree with. Wisconsin needs strong campaign finance legislation which bans most ads from August to December. Legislation that will also limit the enormous amounts of PAC money into Falk, Van Hollen, Doylie, and Green's campaigns.

proletariat said...


You got me thinking with the under votes. I looked at my ward in Madison (32) and there were only two under votes for Falk. So, I began to wonder why the rest of Madison saw the ads but not Ward 32. We must have seen the Wisconsin no ads because we said no to both refs by over 90%. That got me thinking whatever pattern is out there it probably does not include us, we are the outliner as they say. Eisman got almost as money votes as Mark Green.

Looking at Madison however Falk lost 11,000 of the Doyle votes, and Van Hollen got 14,000 more votes than Green. There is something about being in Madison that made me wonder how 11,000 of us could be swayed by a WMC ad.

I then looked at 2002 in which Peg gained 12,000 more votes than Doyle did. 112 k to 98 k. In 1998 Doyle got 10,000 less votes than Garvey.

Or Milwaukee in which in 2002 Doyle got 2000 votes Garvey, and in 2006 Falk got 5000 votes under Doyle. In 1998 Doyle got 30,000 more votes than Garvey.

Russell Wallace said...

Proletariat, if I find out that the patterns of advertising by WMC or the campaigns weren't as I believe then I'll change my analysis accordingly. But until then I stand by what I wrote.

I think that Falk's undervotes came mainly from moderate to right swing voters, and if that's the case I wouldn't expect to see many undervotes in a ward like yours, which voted for Doyle more than ten to one over Green. In Dane County the undervotes would probably tend to be concentrated in the outlying areas, not near downtown. I haven't mapped the data, but it sure would be interesting to do so...

Anyway, you're trying to cherry-pick small bits of data to prove me wrong. There are always local factors that can override the larger trends in particular wards or even counties. But we need to remember that this was a statewide race, and it was the macro factors that decided it.

The whole point of my analysis is to understand the factors that determine electoral success so that I can better support liberal/progressive candidates in future elections. That means trying to find the actual truth, not what I want the truth to be. I certainly don't claim to be perfect, but I try hard to be unbiased.

Lon Noel said...

As for Falks overperformance in far western Wisconsin, many of these voters are transplanted liberal DFL escapees from Wellstoneland, so their heads aren't exactly screwed on right to begin with, they didn't get the barrage of truth that Falk is in NO WAY SHAPE OR FORM QUALIFIED TO BE A.G. Falk as A.G. would drive more business out of Wisconsin than Doyle and the DNR combined. The sad thing is, despite getting beat, she holds hope that 9000 plus votes are going to flip (maybe she knows people at the State Democratic Party Elections Board who can help her after the canvass is done, so they can change the vote totals in her favor) and what's worse, she'll probably run for Governor in 4 years if Doyle hangs it up, that is, if anybody's left in the state to vote for them!!!!

proletariat said...


No, that's not what I was trying to do. One thing Dane Co shows is that a 10,000 over / under vote in the norm (2006, 2002, 1998). Interestingly, Dane Co had twice the undervotes as Milwauukee.

If you want to put all your eggs in the WMC basket, go right ahead. I would also assume the WMC media blitz was larger this year than in 2002 and 1998. An historical analysis would show that to be false. This was a tame year for WMC, a more truthful scenario would be the Repub blaming them if Falk had actually won.

Anonymous said...

"The involvement of WMC in this race is one of the reasons I've always felt, and still do, that Peg Lautenschlager would have been a stronger candidate than Falk in the general election. I may be wrong, but I think WMC would have had a much harder time demonizing Lautenschlager, if they had chosen to get involved at all. But this is just speculation"

WHAT A JOKE! Peg demonized herself when she drove her state owned car into a ditch under the influence! Peg had absolutely no chance in the general election if us Dems had been dumb enough to put her there.

FOM said...

Then again, I'm constantly amazed by just how ignorant and uninformed most voters are, regardless of where they live...

Well gee, I wish we could all be as smart as you.

Interesting analysis and while you can discern some patterns here and there, correlation doesn't necessarily prove cause. I think it was a combination of factors, including a leftist(or maybe anti-corruption leftists) revolt, the strength of JB as a candidate, lesser evil votes for Doyle, the WMC, etc. etc. I do think you're attributing a bit too much power to the WMC or else how come they couldn't do more to stem the Democrat wave?

Also, while we don't know what kind of candidate Peg would have made, we do know what kind of candidate Falk is.

Anonymous said...

Percentages and maps are not as informative as raw numbers, notwithstanding a nice analysis.

For instance, in Dane County, Doyle got 11000 more votes than Falk as many have noted. More telling, however, of that 11000, nearly 3000 voters did not vote for either AG candidate (2500 did not vote and 500 wrote in.) There's your progressive backlash, I'd posit.