|
|
|
September 20, 2003 CLARK LEAPS INTO THE LEAD....Daily Kos has the latest national poll results taken after Wesley Clark announced his candidacy:
Kos is a little perplexed about where Clark's support is coming from, but otherwise seemingly unimpressed. "So what does this tell us?" he asks. "Not much, I don't think." Man, I couldn't disagree more. We news junkies sometimes forget just how little most people are following this stuff, and the fact that Clark has jumped into first place so quickly says something important. After all, Clark is no Eisenhower, and probably the only thing most people know about him is that he was one of the endless stream of ex-officers who did talking head stints on the cable networks during the war. This strikes me as a very impressive start for Clark. I fully realize there are a ton of practical considerations still in front of him — fundraising, endorsements, organization, etc. — but he obviously brings something to the race that appeals to a lot of people. POSTSCRIPT: In case you missed it below, let me recommend again Tung Yin's account of his lunch with Clark, which really shows the best of what a blog can do. It's not so much that Clark is more candid in such a setting — although there's probably a bit of that — but that Yin's account doesn't have to be homogenized into a "story." The lunch group just asked Clark questions and got some answers, and Yin passes it along with whatever personal observations cross his mind. Of course, I also had to smile at the thought of a tableful of law professors grilling Clark on the details of Bush v. Gore.... Posted by KEVIN DRUM at September 20, 2003 01:19 PM | TrackBackComments
I tend to lurk here, without making comments. But I can't let the above ignorant drivel go unnoted. I have always found David Ehrenstein's comments to be quite informative. In any case, even if David had never said anything worthwhile in his life, comments like those above are drastically inappropriate, and reflect far more on the writer than on the subject. David, I hope trolls like this don't drive you away. And Kevin, I know that you don't moderate these comment boards, but this seems above-and beyond. (Although I admit that I tend to give up on most threads as they get very long, so this may be more common than I realize.) Any chance of deleting this? Posted by: Jesse at September 20, 2003 01:42 PMWith Edwards and Graham currently polling below the level of Al Sharpton, they are effectively toast. If one of them were to drop out and turn over his resources to the Clark camp then the General would be in a very strong position. Posted by: Ross at September 20, 2003 01:43 PMSo. Clark v. Dean and may the best man win looks more and more likely. Fine with me. The interview was impressive, esp. about social security since his huge question mark for me is domestic issues. Posted by: Katherine at September 20, 2003 01:50 PM*shrug*. I tend to agree with Kos. It's one poll, right after the most hyped and best covered nomination in the last few years. A dead man would get a good bounce there. It's not bad news, certainly (and it could have been). Good news, however, is sustained Clark support over the first month or so. Clark needs sustained support. Right now he has support (which is good. As I noted, he could have gotten virtually no support at all after his announcement). Now we wait to see if it sustains. Also, as with all national polls at this point, support and name recognition are often the same thing. At this point, I'd put more faith (at least in predicting nationwide Democratic trends) in results from the early primary states. If you're not a usual primary voter, you're probably not paying much attention at all to the field...and even among primary voters, the people paying the most attention are the ones with early primaries. I want a little more depth in the poll, and some good numbers in early states before I'm ready to call Clark's numbers solid. Right now, it's just as likely it's a media-induced bounce, and it's certainly not based on Clark's campaign performance (he hasn't campaigned long enough for that!). Posted by: Morat at September 20, 2003 01:57 PMThanks for the University of Iowa reference. It read as well as you said. What was that stuff at the beginning of this comment section? thelrd in TEXAS Posted by: Larry Davis at September 20, 2003 02:04 PMI have to go along with Jesse here. There is a big difference between someone picking a partisan fight and someone simply spraying graffiti on the wall. Much as most trolls are annoying they still serve a purpose by cluing us into the freeper argument/talking point Du Jour and allowing us the chance to sharpen our responses. But graffiti is just someone's way of trying to ruin the neighborhood, period. We've learned at our business that if you leave it up it only encourages more, and if it's cleaned off right away they get discouraged and leave you alone. Please Kevin, break out the acetone. [Go Clark!] Posted by: Thumb at September 20, 2003 02:05 PMDid the question ask about Wesley Clark, or General Wesley Clark? People might go for a general, without really knowing who he is. Posted by: Jon H at September 20, 2003 02:11 PMI don't patrol the comments much, but I do delete obvious idiocy when I come across it. It just took a few minutes before I first checked the comments on this thread, that's all. The offending posts are gone. Posted by: Kevin Drum at September 20, 2003 02:12 PMI'm very bothered that Clark hasn't thought about almost any domestic issues (by his own admission). How can we presume to take a candidacy seriously that's so woefully unprepared? It looks like he decided to run because people told him he'd win - not because he has a platform, or even a preference, about anything other than Iraq. Geez, people, this is a Potemkin candidate so far. He's moderate, and I like that, but there's no _there_ there. Sure, he's a quick study, but it smacks of desperation and opportunism to vault him onto the top tier, as if it doesn't matter that he's NEVER EVEN THOUGHT about domestic issues. But wait. Maybe it doesn't matter as long as we've got a popular pretty face to put up front, someone who's been encouraged by a group of kingmakers behind the scenes. After all, it's worked very well for the Republicans so far. Posted by: pundit at September 20, 2003 02:16 PMThanks for the kind words Jesse, but I don't think Kevin's remark reflect ignorance so much as they do the dominant mode of this culture -- and not just in politics. "New" is always supposed to be "better." Therefore since we've all been talking about Dean for the past few months the arrival of Clark as a "new name" produces novelty. There's a ton a an a half of hype out there claiming he's more "electable" than Dean. Maybe so, but we're not anywhere near that starting line yet. Clark appears to be very popular with the media. This reflects the fact that he's basically a Republican who votes for the ar.In fact he continued to support it after it started. Unlike Dean who was against it from the start. Inmy darkest moments I suspect that Clark may be a Trojan Horse for the DLC. Time will tell. Meanwhile he has yet to convince me he's worthy of my vote. Not being named "Bush" is not enough. He may ery well be a Bush in Dean's clothing. Posted by: David Ehrenstein at September 20, 2003 02:20 PMKevin, I think Kos is right on this one. If you read through the
comments for his post, you'd find particulars such as the actual MoE and
number of respondents on which this is based. It means nothing except
that Clark is in the tier of competitive candidates, which shouldn't be a
surprise to anyone. Lieberman topped national polls with better numbers
until a couple of weeks ago: did anyone, apart from himself, seriously
think of him as the frontrunner? For me, not being named Bush really _is_ enough. I don't think Clark is "basically a Republican." We heard that talk in 2000, and I don't think anyone's falling for it anymore. Clark will have plenty of chances to fall on his face. But if he doesn't, I find it hard to imagine that Dean will be a more competitive nominee than Clark. As far as I'm concerned, the only thing that could give Dean the presidency would be the emergence of very strong opposition to Bush among the general public. That opposition may emerge, but if it does, then any reasonable candidate will be able to exploit it. The Democrats should be looking for a candidate who can win even if Bush isn't Public Enemy No. 1 in November of 2004. Posted by: JakeV at September 20, 2003 02:35 PMMara: Yeah, but Lieberman had tons of name recognition from 2000. Clark doesn't. Anyway, I'm not fully convinced or anything, but I do think it's an impressive start. And I'm willing to bet that he'll build a pretty good team pretty quickly. After all, he seems to have the Clintons behind him, even if they haven't fessed up to this publicly. Posted by: Kevin Drum at September 20, 2003 02:35 PMI agree with mara. I posted in Kos's thread thinking ... Wow, good news for Clark! Then I went back and saw the sample size and MOE and wondered. It could be significant. I just wish they had a smaller MOE. Posted by: bink at September 20, 2003 02:35 PMI think this poll reflects more the ambivilance of a lot of Democrats then any strong support for Clark the candidate. A large chunk of his initial support came from the Undecided category after all (and, strangely enough, from Dick Gephardt). Regardless of whether it reflects strong support for Clark or not, it is significant because it will influence the fair-weather voters, the ones who will vote for whoever is (perceived to be) on top. Clark still has to prove that he can be a good campaigner, which is my #1 qualifier for selecting a candidate right now. I do not buy into the belief that a general will automatically trump an AWOL lieutenant. Shiny metals are not bulletproof, despite what people may think. But if Clark can prove to be a good campaigner AND he can persuade the Dean forces that they won't be dismissed as "children" (an attitude I have seen more than once from some of Clark's supporters) than I will support him enthusiastically. Until then, I'm a Dean man. Posted by: Chris Andersen at September 20, 2003 02:50 PMI agree that Lieberman benefitted from higher name recognition. This doesn't make the current poll any more meaningfull. As I said above, a reasonable cautious interpretation of the result is that Clark debuts in the 'competitive' tier of candidates. I think the difference between you and me is that you think this is impressive (perhaps because unexpected?), whereas I think it is not a big surprise. (One can argue that it's impressive given a relatively unknown candidate. One can also argue that this is a further argument for the poll being a fluke.) Overall, I think Clark supporters should definitely be pleased, but not jump with joy quite yet. Posted by: mara at September 20, 2003 03:33 PMClark has been posited as a Democratic Colin Powell, someone who everyone thinks is strong and generally (no pun intended) competent, and agrees with them on all the issues. Clark has also gotten great press lately, much of it positing that he is the only Democrat with a chance against Bush, although somewhat less so once he actually announced he was running. I think that Clark will have to work very hard to keep the momentum alive, especially as he is forced to take positions in public and be seen along side the other 9 candidates. That sort of scrutiny tends to weaken, rather than strengthen, a candidate. Posted by: CADem at September 20, 2003 03:38 PMI am really intriuged by Clark. On foreign policy he pretty much has my vote whenever he wants it. With his vision of muscular multilateral interventionism he's certainly not "pretty much a republican". Hell he sounds a lot like FDR on that front to me. The problem is domestic policy. He's said all the right things that should put democratic minds to rest (for progressive taxation, no cutting of SS benefits, pro AA, pro choice etc.) but I don't see a vision for molding all these positions into a theme that will move the nation forward. I hope he articulates one soon. In a sense Clark is the polar oppossite of Edwards. Well articulated foreign policy and a nebulous domestic policy. Whereas Edwards has some of the best formed ideas on domestic policy but seems inexperienced on the foreign policy front. Perhaps this suggests a winning ticket? Either way anyone who can defeat Clark will be all the better for it. And any one who defeats Edwards better steal his ideas. Posted by: WillieStyle at September 20, 2003 04:02 PMwhatever the reason, I am on record as not expecting this kind of poll presence. So, don't hire me to predict the future. But, we'll see how it is in a few weeks. Posted by: Atrios at September 20, 2003 04:29 PMLast night on the The McLaughlin Show, McLaughlin predicted that Howard Dean would get the Democratic nomination and I think he is probably right. Clark’s okay and I like him but the war on terror has change and the polls are now dipping on the war in Iraq. Like I said before, I think the election will be about diplomacy and not war making abilities that Americans will be interested in for the future Prez. Dean’s main message is that he will go back to the UN and that is the one thing Americans all seem to want right now. AND also Howard recently said, “The most important piece of foreign policy is judgment. The other four fellas who supported the (Iraq war) resolution all now say they were misled by the White House.” (via Josh Marshall) I wonder if Blair will do this too now that his poll numbers and his party are doing very poorly? As far as Clark goes, the dealbreaker is that he thought it would be a good idea to pick a fight with the Russians during the Kosovo adventure. You know, the same guys that everybody from Truman on down went out of their way to avoid fighting directly? Because of their large nuclear arsenal? Maybe there's a time when fighting a nuclear armed power is really the least worst option. (Like when the only other alternative is watching Al-Queda get their own collection of nukes from said nuclear armed power. Or... well, that's all I can think of.) But not over Kosovo, for crying out loud! I'll vote for Clark when Hell freezes over. Posted by: Ken at September 20, 2003 04:39 PMEdwards' ideas are "compelling?" "Show them where the free market betrays the work ethic, and they'll vote for the candidate of the work ethic against the candidate of the free market." The above bit of economic illiteracy on Saletan's part hardly does anything to convince me of that claim. Maybe Edwards' views really are wonderful, and Saletan has just made a hash of presenting them; I don't know. As for the following position: "no cutting of SS benefits" This may set older voters' minds at rest, but for the young it ought to be a very disturbing notion. As the dependency ratio continues to fall, social security payments are either going to have to be cut - whether by reducing payments, or indirectly, by raising the age of eligibility - or young workers are going to have to forego ever more of their income to support an ever increasing number of retirees. Even if all of Bush's tax cuts were reversed, the shortfall in social security would still exist - the only difference would be that the point of crisis would be pushed back a few years, and heaven help all Americans if the immigration restrictionists get their way, or the female fertility rate continues to fall towards Italian-style levels. Posted by: Abiola Lapite at September 20, 2003 04:57 PMYou're right Ken. I mean we can pretty much assume that, say, George W. Bush would have done a much better job as Supreme Allied Commander of NATO than Clark did. Doncha' think? I imagine that most people who say that they aren't voting for Clark because of the Pristina airport incident were definitely not going to vote for him anyway and are just looking for an anecdote to thrash him with. Posted by: JakeV at September 20, 2003 05:01 PMKevin, You left out a key detail: the margin of error of the poll. Kos did too, but you can find it buried in the MSNBC story: 3%. This means that the support of Clark, Dean and Lieberman cannot be distinguished by this poll. All we reall know is that Clark seems to have vaulted into the top three, not necessarily the lead. Posted by: Michael at September 20, 2003 05:34 PMA vast majority of the problems between Dean supporters and Clark supporters fall into the following categories... 1. Overzealous supporters that don't care what 2. Trolls who want more than anything else in the world to get these two guys destroyed by fighting each other, so that another candidate can go forward (whether that person is Democrat, Republican or Green) These two men have gone on the record as respecting each other. Clark has stated that he wants to run his campaign on a positive note and raise the debate in this country from the morase of hate speech in has devolved into. Dean has told his people to stop the attacks on Clark. The sooner we stop this rabid nonsense and quit allowing the Trolls to steer our conversation, the better the party will be as a whole. Are their differences between the two men? Yes, there are, but lets debate these subjects in a tone and a manner that befits people that have graduated from the days of a grade school play ground. Posted by: DanD at September 20, 2003 05:45 PM"All we reall know is that Clark seems to have vaulted into the top three, not necessarily the lead." Very astute. There's nothing quite like a spurious air of accuracy to mislead. Posted by: Abiola Lapite at September 20, 2003 05:50 PMActually no, the margin of error for this SUBSECTION of the poll was plus or minus 6%, even greater. "Newsweek Poll conducted by Princeton Survey Research Associates. Sept. 18-19, 2003. N=377 registered Democrats and independents who lean Democratic nationwide. MoE ± 6." Posted by: mikel at September 20, 2003 05:51 PMMichael, the MoE of this poll is +/-6. Everybody may feel free to draw from it the conclusion that pleases them best. Posted by: mara at September 20, 2003 05:51 PM"Actually no, the margin of error for this SUBSECTION of the poll was plus or minus 6%, even greater." If that is so, it would make the results above well-nigh useless. The most one could say for certain is that Clark has more traction than candidates like Graham, Braun and Kucinich - which is hardly surprising. Posted by: Abiola Lapite at September 20, 2003 05:53 PMBiola, My I point (and I presume Saletan's) isn't to revel in economic minutae. The point is that Edwards has taken Democratic positions and crafted a theme that transcends mere knee-jerk populism. As to that specific point: As for SS, let's leave that for another thread. Forget I brought it up. Posted by: WillieStyle at September 20, 2003 05:57 PMI am one Dean supporter who welcomes Clark to the nomination race. The reason I hooked onto Dean in the first place and the reason I stay with him now is that I believe he can beat George W. Bush AND govern effectively. In order to beat Dean, Clark will have to demonstrate that he can also do both. If he does and wins the nomination, that's fine with me. I am a single-issue voter: Bush must go. Posted by: James E. Powell at September 20, 2003 06:04 PMDean supporters delude themselves about Dean's electablility. In head-to-head matchups with GWB, in poll after poll, Dean fares worse than other democrats. Clark, in the latest Newsweek poll, fares the best. Posted by: obruni at September 20, 2003 06:22 PMAbiola: there's a third option: raising taxes. Currently, the income the social security payroll tax applies only to the first $88,000 or so in income. Clark would lift that cap. It's the same solution Paul Krugman favors and it's a sensible one, since it neatly solves the revenue shortfall, doesn't reduce benefits to those people who can least afford to have them reduced, and reduces one of the most regressive aspects of the current system. Posted by: phil at September 20, 2003 06:42 PMobruni: Dean supporters delude themselves about Dean's electablility. In head-to-head matchups with GWB, in poll after poll, Dean fares worse than other democrats. Clark, in the latest Newsweek poll, fares the best. Reason to be cautious about Dean, yes. Reason to reject him completely? No. Considering the huge changes in polling numbers in just the last 6 months (when Dean was an asterisk, and Bush was invincible), a person would be completely foolish to use this week's polls to determine who to vote for in the primaries. Hell, by that logic, we ought to draft Hillary. The Nov. 2004 election is going to be a referendum on our current president. Any of the leading Democratic nominees can beat him, if they run a good campaign. Posted by: YT at September 20, 2003 07:04 PMInteresting as well in that (based on similar data and MOE) Kos has already announced the collapse of all pre-Clark non-Dean campaigns, and has insisted it's too late for Clark to enter. Posted by: RonK, Seattle at September 20, 2003 07:09 PMKen, regarding As far as Clark goes, the dealbreaker is that he thought it would be a good idea to pick a fight with the Russians during the Kosovo adventure. I had read about this incident in Clark's book and then saw it pulled out of context and used against him by George Will and had to think "now, how could anyone really believe that this was a Cuban Missile Crisis incident?" Well, apparently a lot of people do. I think it's a lot of hype. Please read the very sensible analysis here. Yeah, it's by Clarkophile. It's also accurate, sensible, detailed, and I think fair (well, they're a bit rough on Jackson; maybe they're Irish or just averse to massacres or something) -- in other words, it's something that neither the George W. Will Right nor the Coulterpunch Left seem to want to bother with. I have a lot of questions about Clark (for example, at what point did he start questioning the "Noble Mission" of Vietnam) but this "he's so hot-headed he wanted to start WWIII" is as dumb as Lieberman trying to paint Dean as an anti-Semite. Posted by: billj at September 20, 2003 07:14 PMBillj- If you honestly believe that that trash which you linked to "accurate, sensible, detailed, and I think fair" I think your judgement on everything else has to be suspect. It is a tendatious smear job which contains a number of outright fabrications and reads like a Sinn Fein press release. Is that crap actually anything to do with the Clark campaign? If it is then I have lost all respect for the man. Posted by: Ross at September 20, 2003 07:43 PM"I am not a free market fundamentalist. I believe markets can fail." And who is a "fundamentalist?" I am hardly unacquainted with the notion of externalities. The point is that Saletan is simply talking nonsense; he hasn't identified any market failure, for the simple reason that there isn't one to identify. "There's a third option: raising taxes. Currently, the income the social security payroll tax applies only to the first $88,000 or so in income. Clark would lift that cap." Yes, I failed to mention that, but it is an option - in the short term at least; as long as birthrates remain below replacement level, and immigration fails to occur in sufficient quantities to keep the population structure stable, you're going to have to seek recourse in raising the cap again and again to stave off the mathematically inevitable. As for how politically viable this is, I'll leave to the experts to decide. Posted by: Abiola Lapite at September 20, 2003 07:46 PMOver and over, on many blogs and comments, I keep seeing comments to the effect that "I don't know much about Clark." What are they waiting for? The TV ad campaign? They are using the internet now; why aren't they using it to do their homework like the kids do? Look him up. Posted by: aw at September 20, 2003 07:48 PMOne thing I failed to address is the following: "For instance I think taxing dividend income more favourably than labour income is a very bad thing." While it may strike one as unfair, there are very good reasons why such disparities exist: one extremely important reason is that labor is relatively immobile, while capital isn't. Taxing dividends at a higher rate won't necessarily bring in much by way of income, as there are plenty of ways one can shelter one's investment gains from tax liability; a huge portion of Wall Street's business consists of doing exactly this. Share buybacks, derivative transactions and numerous other financial techniques exist to ensure that investment income escapes the clutches of the IRS. The second aspect of this issue is that while most of us have to work within the confines of our national borders, there are plenty of places in the world one can easily send investment capital, and in many of them, like Switzerland, tax avoidance is perfectly legal. A wealthy investor could choose to put his money in higher return investments abroad, park the receipts in Liechtenstein or the Andorres, and the American government would be none the wiser for it. Workers who have their taxes deducted at source are much easier to milk, a reality even populists like Edwards have to deal with when (if) they get into office. In short, when we come away from the "fundamentalist" fantasy-land of ideal tax policy, we are led to the conclusion that higher dividend taxes don't necessarily do anything to punish "idle" wealth, as those who tend to actually be hit by such taxes are smaller investors saving for their 401(K) retirement plans, who can't afford fancy tax lawyers and offshore shelters with numbered accounts. Posted by: Abiola Lapite at September 20, 2003 08:08 PM"Aw", I suspect that what some people are waiting for is a couple of things: 1. A clearer articulation of Clark's positions on the issues of the day. 2. A clearer understanding of how Clark will handle himself "under fire" in the campaign. Both of these things will become more clear in the next few months. Posted by: PaulB at September 20, 2003 08:28 PMAbiola, But my main objection to the inequitable treatment of dividend income isn't so much that it reduces federal revenus. Rather, it scews the market by providing screwy incentives. Cuts in the capital gains tax in the late 90s led to rampant speculation and the same is occurring today. The net effect is to spurr investment without spurring demand. Recent productivity gains not withstanding, that is simply not tenable in the long term. P.S. I thought you all should see this: Writes one Democrat with national political experience: "I have read the accounts of the Clark interviews and my reaction is despair and anger. Why did my party's best operatives think it would be a good idea to subject their neophyte candidate to the country's savviest reporters for over an hour? Why have my party's elders rallied around a candidate who is so shockingly uninformed about core issues and his own positions? I am not a Dean supporter — but I am angry that our party's leaders have anointed an alternative to him who seems even more ignorant and unprepared — and that this supposed 'anti-war' candidate turns out to have been in favor of both the war resolution and Richard Nixon!! And let's not even talk about the Clintons. Today I am embarrassed to be a Democrat." Good luck. That's from ABC News Posted by: Mike K at September 20, 2003 08:44 PMLast night on the The McLaughlin Show, McLaughlin predicted that Howard Dean would get the Democratic nomination and I think he is probably right. Leaving aside the lameness of Appeal to Authority in a conversation of this kind, how does McLaughlin become The Authority, anyway? He is precisely as fallible a predictor on this issue as any other TV pundit. Maybe more so, given his political views. Posted by: Julia Grey at September 20, 2003 08:55 PMClark's support of middle-class tax cuts helps his electability. Supporting a raise in the cap on social security tax does not. He needs to develop a coherent domestic policy. National security and foreign policy is where Clark shines. Perhaps any Democrat can beat Bush on the economy, but unless the Democratic candidate can convince voters that he (she?) will protect us from terrorism, Bush will win. Posted by: obruni at September 20, 2003 09:03 PMRe: that blind quote from the "embarrassed Democrat." I don't pay attention to supposedly important people "with national political experience" who resort to disseminating their views via well-financed and very visible press outlets while being unwilling to publicly identify themselves. It's cowardly. I try not to take advice from or align myself with cowards. In fact, I'd be tempted to go in exactly the opposite direction simply because of the methodology of this attempt at "persuasion." This is assuming that the quote isn't just made up by the journalist -- always a danger with blind quotes. Posted by: Julia Grey at September 20, 2003 09:04 PMA few words about MOE, since they've been discussed. The usual MOE reported by a poll is simply 1/sqrt(N), which for N=377 would give 5.15%. Interesting that they reported it as 6%. Anyway, here is some info that you may or may not know: 1) The MOE is supposed to represent a 95% probability that the 'true' result (obtained by surveying everybody) is within the MOE of the poll result. This doesn't mean every value from the +MOE to -MOE is equally likely. There is about a 2/3 probability that the 'true' result is within half the MOE of the poll result. 2) The 1/sqrt(N) formula used by the polls for the MOE is significantly wrong for probabilities significantly far from 1/2. The right formula is 2*sqrt( p*(1-p) )/sqrt(N), where p is fraction found in the poll. This reduces to 1/sqrt(N) for p=1/2, and it's surprisingly close for p from .3 to .7, but it begins to differ significantly outside of that range. For Clark's p=0.14, the MOE should be 0.7/sqrt(377) = 3.6%. For Dean's p=0.12, the MOE should be 3.3%. 3) The MOE for Clark's 14% and the MOE for Dean's 12% combine to give more uncertainty, or a larger MOE, to the 2% difference between them. A rule of thumb that should satisfy any statistician (at least after you buy them a beer to make them relaxed and candid) is that the difference MOE is the square root of the sum of the squares: sqrt( 3.6^2 + 3.3^2 ) = 4.9%. 4) The MOE reported assumes that only statistical error exists (i.e. that the answers were obtained by a perfectly random set of people) and does not account for other systematic effects. These effects aren't reported because there is no great way to estimate them, and their size varies widely from poll to poll. But they always act to increase the inaccuracy. 5) if we accept the statistical MOE, i.e. we assume these poll takers
did a bang up job, we find the 2% difference between Dean and Clark to
have a 5% MOE. While that sounds meaningless, it does mean Clark has a
78% chance of really being ahead of Dean (in the 'true' poll that
reaches everyone). Repeat after me: National polls in a system that picks its nominee under a system of state caucuses and primaries mean nothing. It was seemingly just six or eight weeks ago that Joe "I really and a Demorcat! Honest!" Lieberman was leading in the national polls even as he began his sink to the bottom in the primary and caucus states. What we have here is a man that a lot of people recognize from CNN and other networks who just got a ton of publicity leading up to his entrance into the race and a poll that coincides with that blitz. What does it mean? Like I said, repeat after me: National polls in a system that picks its nominee under a system of state caucuses and primaries mean nothing. Posted by: The Ox at September 21, 2003 01:00 AMJust to give a more detailed table to support Ben's analysis above, I find it useful to refer to the following table: http://www.ocs.mq.edu.au/statshare/units/stat823/MarginErrorTable.pdf And just to give you an intuitive sense of why it is that MOE goes down when the percentages go down. Suppose you had a poll of, say, 1,000, with the roughly 3% MOE at 50%. Would it make sense for the MOE for a candidate polling at 5% to be 3%? Of course not -- because at that range the size of the MOE is over half again as much as the number of people who expressed support for him, whereas for a candidate polling at 50%, 3% is only a small fraction of 50%, which would make it a much easier error to come about. Posted by: frankly0 at September 21, 2003 01:02 AMBen Vollmayr-Lee, Many thanks for your rundown of MOEs. A question, where does the 78% probability of Clark being ahead of Dean come from? And can you estimate the probability that the overall ranking, top to bottom, is in the correct order in this poll? Posted by: Michael at September 21, 2003 01:04 AMLike I said, repeat after me: National polls in a system that picks its nominee under a system of state caucuses and primaries mean nothing. What ever do you mean here? National polls mean nothing ONLY IF they change over time in response to early caucuses and primaries. If the national numbers stay basically the same, then sooner or later, in the fullness of the primary elections, those national numbers will turn into substantial victories for the candidate at top. What early primaries can do is to create a bandwagon effect that can change both later primaries AND the national numbers. Yet sometimes too the earliest primaries, namely Iowa and NH, are NOT decisive, and DON'T establish a dominant player or eliminate serious competition. One little noted consequence of Clark's entry into the race is that he adds still another major candidate, in a field already crowded with major candidates, into which voters might sink their votes. The problem this creates for the presumed front-runner Dean is that it makes it less likely that he will win, if he does win, in the early primaries by such a substantial plurality that he will be regarded as the inevitable nominee. What if Dean wins in both Iowa and NH, but in each case barely breaks, or doesn't break, 30%? Who is going to proclaim him the inevitable nominee with numbers like those? And without the presumption of near inevitability, why imagine that he will create the required bandwagon effect? I see a long hard haul ahead of us, unless something dramatic happens.
The poll is based on 377 people! A 6% margin of error. It is totally meaningless. Posted by: Paleo at September 21, 2003 06:26 AMThe fact that a relatively unknown retired general can leap ahead of the rest of the field on day one tends to show just how soft the support is for all of them, including Howard Dean. The Democrats desperately need someone besides Dean to win the nomination since the same Newsweek poll shows him losing to Bush in a head-to-head poll by 14 points - the worst showing of all the major candidates. Clark has something in common with porn star/California gubenatorial candidate Mary Carey so I don't know if he's going to be the Democrats white knight either. Posted by: Randal Robinson at September 21, 2003 06:32 AMI find it interesting that trolls w/ false e-mails like Ross only can
trot out the "start WWIII" hooey as their anti-Clark argument, and then
simply badmouth an actual analysis w/o any facts. 'It is totally meaningless.' Well, that's just factually incorrect. Statistical analysis can give a little meaning, slight shifts in probability estimates, with surprisingly small samples. Just because it doesn't give iron-clad results doesn't make it meaningless. 'A question, where does the 78% probability of Clark being ahead of Dean come from?' Suppose Dean and Clark had tied in the poll. Then your best estimate for what would happen in the 'true' poll would be a 50% chance of Clark beating Dean. Since Clark did slightly better than that, our best estimate of his chance goes up. If the poll were truly meaningless, as some would have it, our best estimate of Clark's chance of beating Dean would only go up a negligible amount over 50%. But we can actually calculate this to see whether it's true. The basic idea, which I can't really go in to, is the central limit theorem, which is the analysis tool behind the MOE, for example. The CLT shows us that a 2% difference with a 5% MOE means chance that Clark is really ahead of Dean in the 'true' poll is 0.5 + 0.5*erf( 2%/(5% * sqrt 2 ) ) = 0.78, where 'erf' is the error function. Now I'm going to rant a little. I really don't like the way poll results are reported. First of all, as noted above, they report a single MOE, but this number is wrong, sometimes quite wrong, for small and large percentages. Second, most people interpret the MOE incorrectly, saying that if the MOE is larger than the difference between two results, we learn nothing. It's true that this situation leaves the null hypothesis in good standing, but that's different than learning nothing. With correct analysis, probability estimates still shift (put it another way: when the MOE is larger than the difference the poll doesn't prove anything, but to demand 'proof' as the only kind of information to take from statistical analysis is to throw away a lot). Basically, many people seem to read MOE as being BOUNDS on the uncertainty, and furthermore that all values between the bounds are equally likely. The phrase 'margin of error' supports this incorrect view (both parts of it are incorrect). It would be much better if 1/2 the MOE were reported as the uncertainty -- so, sqrt( p*(1-p)/N) -- along with a few changes in conventional wisdom: this smaller value represents the TYPICAL error, not a bound. The current MOE neither represents a typical error (it's too big for that) or a bound (it's too small for that). There is roughly a 2/3 chance of being within the +- of the typical error (+-MOE/2) and a 95% chance of being within twice this (+-MOE), and a 99.7% chance of being with three times this (+- 1.5*MOE). That's a few more bits of information to try to pack into CW, but it's not impossible. We got people to wear seat belts, after all. Posted by: Ben Vollmayr-Lee at September 21, 2003 07:14 AMOops. Got on the rant and forgot to answer Michael's second
question. The probability of the entire ranking being 'correct' (see
'true' result definition above) is quite small. There are many possible
swaps of candidates between this poll and the 'true' poll that have a
reasonable chance of occuring. So the odds that none of them happen is
quite small. I think it's great for Dean that Clark is in. Dean's people should stop bitching and find a way to beat Clark. With all the good will they've got, they should have knocked out several rivals by now. Frist off Dean needs to fix his tax message. Posted by: Eric M at September 21, 2003 09:17 AMB.V.L., It's totally meaningless, even putting aside the margin of error, because we don't have a national primary. How Clark polls in Iowa, New Hampshire, and South Carolina is 100 times more important. Posted by: Paleo at September 21, 2003 09:31 AMIt's totally meaningless, even putting aside the margin of error, because we don't have a national primary. How Clark polls in Iowa, New Hampshire, and South Carolina is 100 times more important. This is just more confusion. What the national numbers (together with natural, reasonable assumptions), imply is that, if all state primaries were held today, most likely Clark would win. As I pointed out earlier, it is of course the case that NOT all state primaries are being held at the same time, so that the national numbers now, and even at the time of the first primaries, will be subject to considerable change depending on the perceptions engendered by those primaries. Typically, the winner in the early primaries get a boost from that win. Whether that happens this time around, and to what extent it does, depends on the nature of the win, and, in large measure, how it gets spun in the media. If Dean were to win in both Iowa and NH, his camp would be spinnning it for all its worth as God's sign that he was anointed one. On the other hand, if his win is only in the 30% range, say, his opponents will be spinning those wins as unimpressive at 30%, and deriving in no small part from the relatively liberal base of voters in the Iowa and NH primaries (compared to other parts of the nation), and, in NH, as a geographical bias for a candidate from a neighboring state. How much of boost he'd really get when the spin is all done is anybody's guess. One might also make the reasonable point that the numbers in the early primaries are more useful because they give one a better idea of how a candidate is perceived by the voters after they get to know him better. It's doubtful though that this effect outweighs the considerable bias implicit in choosing Iowa and NH as the first cases. It's a little hard to listen with a straight face to people who find ways to dismiss as "meaningless" exactly the polls that puts their guy in a bad light, and to embrace those polls that show him a good one. It has a way of impressing KoolAid teetotalers as irrational. Posted by: frankly0 at September 21, 2003 11:14 AMMattB- Are you illiterate or just a liar? Where did I accuse Clark of
starting WWIII? The answer is nowhere. I condemned the fact that his
supporters were peddling outrageous smears about General Mike Jackson in
order to defend him. Indeed if you look at the second post in the
thread you will see that I was initially positive towards Clark. Slightly off-topic. I, too, use a false e-mail address (it's an old
one I used to have). The reason for that is that spammers use bots to
pick up e-mail address from forums like this. By keeping my new e-mail
address out of forums like this and by taking a few other elementary
precautions, I've successfully managed to stay free of spammers for
months now. Post a comment
|
|
|
|
|
|||
|
Powered by Movable Type 2.63 |
||||||
|
|||||||