Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
United States

Journal pudge's Journal: Even More Lies 13

Journal by pudge

Most of the time when the media reports a poll result, it quotes them as hard figures. An MSNBC/Reuters poll is reported on MSNBC as Dean in front with 28%, Gephardt in second with 23%, Kerry in third with 17%, and Edwards in fourth with 14%.

The problem is that the poll is +/- 4.5%, a nine-point swing. This means that Edwards could be as 9.5-18.5%; Kerry 12.5-21.5%; Gephardt 18.5-27.5%; and Dean 23.5-32.5%. Gephardt could be in the lead, and there is no rational basis for concluding otherwise.

The +/- 4.5% means "we are not saying Gephardt is at 23%, we are saying he is between 18.5% and 27.5%." That is what it means. Edwards and Gephardt could be even. Kerry could be ahead of Dean. Not tomorrow, right now, according to this poll. This poll says there is a three-way tie for the lead right now, not that Dean is in the lead by himself.

While I am on the subject of polls, some of you may not be aware, but who is allowed to participate in Presidential Debates is essentially decided by a private bipartisan commission that works to keep out third parties. One of the ways they do this is by saying you have to achieve a certain percentage in certain polls to be considered a viable candidate that is allowed to be in the debates.

This is such an atrocious thing on many levels, first and foremost because our choices are being limited arbitrarily by private people who have political interest in doing so. But the criteria they use is astonishingly questionable. We have seen many polls be flat-out wrong over the last several years, such that to rely on them as a basis for the slection of our leaders is patently ridiculous and borderline criminal.

But take it to the next step: who is to say these polls are not being falsified? We have put a lot more emphasis on verifying voting data, which is a good thing, but we are using unofficial private polling data to actually determine whom we get to vote for, and there is not the slightest bit of public oversight of the data. If it is possible that state officials are committing fraud in actual elections to hurt Bush or Gore, why do we not question whether it is possible that private people are committing fraud in the polls to hurt Nader or Perot?

This discussion was created by pudge (3605) for no Foes, but now has been archived. No new comments can be posted.

Even More Lies More

Even More Lies

Comments Filter:
  • The worst I've seen so far is +/-6% margin of error in a recent poll shown on Fox. I couldn't believe it. Technically, it put Edwards in fourth place (at 9%) within striking distance of Dean (at 22%). I didn't know whether to laugh or cry. Why even bother publishing the poll at that point? It doesn't say a damn thing. If I tried to use data like that to make decisions at work, they'd fire me in a second, and they'd be justified.

    What gets me even more worked up is that they rarely tell you the sample

  • Do polls report their +/- range at the 50% or 95% confidence level? I suspect 95% but honestly don't know. Scientists use 95% but ...

    Pudge, it doesn't mean there's a 3-way tie for the lead. That is a statement that is of course unknowable with certainty. What we can say is that, based on the numbers given, "it is n% likely that the top three candidates are within e percentage points of each other," where e is small enough to meet anyone's definition of a dead heat. I don't know offhand how to make that ca

    • An excellent point! Personally, I'd love to see the confidence intervals used, the deviation estimates, etc. I wouldn't have nearly as much of a problem if the data was presented more realistically like you did above ("There is a 90% likelihood that Dean is in the lead") rather than the talking heads saying "Dean is 2 points ahead of Kerry" and nodding sagely into the camera.

      I won't even start on the "two polls makes a trend" stories. =)

    • Professional polling organizations use a 95% confidence interval [janda.org] for the reported "margin of error" [notrain-nogain.org].

      So, it is 97.3% likely that Dean leads Gephardt in the Iowa poll Pudge reported.

      • 97.3% likely Dean is in the lead? If your math is right, then MSNBC is being far too cautious when it says [msn.com]:
        Democratic presidential front-runner Howard Dean has widened his lead over Missouri Rep. Richard Gephardt,according to an MSNBC-commissioned poll released Tuesday, although the pair essentially remain deadlocked within the poll's margin of error.
        • Yeah, reporters don't understand what the margin of error means any more than most people.

          However, I think it's reasonable to say that they're very close, because of the volatility of the polls. Public opinion turns on a dime these days, and if someone was two standard deviations ahead last week, they could easily be a standard deviation behind next week.

    • The margin of error is similar in all polls where Bush leads(and in his approval rating). Pre-thanksgiving stunt, Howard Dean might well have been leading Bush overall.

      Still, I'm with pudge(wow, I agree with him completely... interesting). I don't buy poll data, I wish everyone would just stop using it as fact. It's not. It's merely a vague idea. 90-95% CI, come on! Unless the margins are incredibly wide... it just doesn't indicate much of anything. Everything(nov 2004 and primary elections) is stil
    • Jamie, you're absolutely right that the MOST correct thing to say would be to qualify it, such as "there is an n% chance that Dean has a y% lead over Gephardt," or somesuch. But without qualification, there is no leader. Without qualification, they are in a statistical tie. Only when it is broken out in some way, such as how you describe, can you say there is a leader. And they don't do that, so they are lying.
  • The Democrats and the Republicans have the candidate selection mechanism sewn up. The FEC has always been and will always be (until a major change in the entire process) made up of 3 DNC sponsored candidates and 3 RNC sponsored candidates. The agency responsible for fair elections has no incentive to look into fraud against other parties.

    Can you imagine if we abolished the party system? If legislators had to make decisions based on the issue at hand rather than "fall in line" with the party view? If vo

  • An article [cnn.com] on CNN claims that students who were used to conduct the survey to determine if the Peterson trial should be moved have admitted to faking the results in order to make sure that they got good grades. This is very unsettling since in many areas the local univerisity poli sci or statistics proof is the main pollster and uses students to conduct the polls.

  • I've been trying to tell people for months that NOTHING matters in the primary race until Iowa, except for one thing. The only candidate to have any real effect on the primary in the pre-season has been Graham, who dropped out of the race. Going into Iowa, primary candidates fall into two tiers. The first tier is serious candidates to be considered in the running. These are Dean, Kerry, Lieberman, Gephardt and Clark. The second tier is everybody else. In this tier you'll find race warlord Sharpton, fo

    • Well yes, Iowa and New Hampshire are all about positioning yourself for the following races, but people vote and continue to donate based on how well you place there.

      I don't think any of the three bottom candidates are expecting to be elected. All are in it to make points. I wouldn't be surprised to see none of them drop out. Who wins the first two is instructive on who will win the others. For example, if Gephardt loses Iowa big, he may drop out. If Kerry loses New Hampshire big, he may drop out. et

Slashdot Top Deals

Swap read error. You lose your mind.

Close