Tuesday, July 29, 2008

Polls: Dumb and Dumber

Yesterday, this blog was again proven right (as if there were ever any doubt) about how MEANINGLESS polls are, and how STUPID the media is in the way they report them as "news".  That is apart from the fact that they ear EVIL things, which we (the public) should sabotage at every opportunity by LYING (or at least not cooperating).
In case you missed it, or properly ignored it, the Gallup "tracking poll" of registered voters" yesterday had Barack "world" Obama ahead of John "Hit Man" (leftist designated "hit man" when they want to target a conservative, or conservatives in general) McCain by a margin of 8 percentage points.  Meanwhile, ,a Gallup/USA Today poll of "likely voters" (the kind of poll you get from EVERYONE at election time, since a poll of merely "registered voters" is KNOWN to be an inaccurate predictor of actual voting results) gave MCCAIN a 4 percentage point LEAD over Obaba.  This poll was, predictably, NOT publicized much--for example, on AOL--even though the previous USA Today poll giving Obama a six percentage point lead was played up big on AOL. 
Now it is known that polls of all people, polls of registered voters, and polls of "likely" voters vary substantially in their results.  Since we do not elect people by polls, it is VOTERS who are the only ones that matter.  And it is only VOTES that count.  Logically, however, IF you are going to play up polls (and be a stupid idiot), it is polls of LIKELY VOTERS that are the most meaningful.  That would mean that it is ONLY the USA Today Poll giving McCain the LEAD that MATTERS.  EXCEPT, all polls are meaningless at this stage.  How can you even determine a "likely voter" more than 90 days before the election?
Yet, the media (ALL the media--as I have said repeatedly, Fox News and Sean Hannity are among the worst) reports poll numbers (but only the selected poll or polls they WANT to report) as if the numbers mean something.  They don't, and they shouldn't.
Is a poll of likely voters REALLY supposed to differ by 12 percentage points from a poll of registered voters?  Nope.  Even if it does, that merely highlights how MEANINGLESS the polls are, since it is the ACTUAL VOTERS that matter.  And the media will STILL say Obama is "leading" in polls, even while saying that the poll is within the margin of error.  In other words, the media insists of treating the poll as an exact measurement of the ENTRIE POPULATION, when it is not.  This is not mere stupidity,  It is a LIE.  The media--all of the media--LIES to you every single day.  A poll is a STATISTICAL PROJECTION of the results of a SMALL sample (generally around 1000 people) projected onto a LARGE population (say 100 MILLION people).  By chance alone, you could have an error of 100% (chance of throwing 100 heads or 100 tails in a row)  You have an EXPECTED error of plus or minus 4% or so (a LANDSLIDE in recent elections).  That means that EXPECTED chance variations in a poll having Obama "ahead" by 3 points could mean that MCCAIN is actually "ahead" in the larger population.  Polls are NOT "news".  It is a LIE to suggest that they represent the "opinion" of the larger population of 100 million plus.  They are--this is NOT a matter of opinion; I can assure you as a person having a B.S. in physics--merely an ESTIMATE of the opinion of 100 million plus people based on a small SAMPLE of people.  in other words, polls are only an EXACT reflection of the opinion of the 1000 person sample, and that assumes the people in the sample DO NOT LIE (as I suggest they do, and as I would).  The ESTIMATE has a built in, expected error (usually 4% or so).
But the ESTIMATE is a statistical estimate based on the sample being REPRESENTATIVE.  In other words, if you flip a coin 10000 times, you EXPECT that it will come up 500 heads and 500 tails.  However, it is UNLIKELY that it will come up exactly 500 heads and 500 hundred tails.  Chance alone will cause the number to vary 3 or 4% either way.  That is ALL that is meant by the "margin of error".  Remember, however, that it is POSSIBLE for a coin flipped 1000 times to come up heads 1000 times--just unlikely.  Still, chance alone is why polls are an ESTIMATE, at best, and NOT a "measurement" of opinion.  "Issue" polls are even worse, because they depend on HOW THE QUESTION IS ASKED.  There is NO "scientific" basis for the validity of "issue polls", since there is every indication that the sample is ALWAYS invalid because the way the question is asked affects the reults.  To even "report" "issue poll" results is an OUTRAGEOUS LIE.  The results are even more meaningless than straight voter polls ("Who do you plan to cast your vote for?").
Even in "simple" polls of how people expect to vote, the mathematics of statistics does NOT tell you that the sample is valid  There are numerous ways the sample may be unreliable.  As I said, people in the sample could LIE.  The sample may be CHOSEN in a way (as the TELPHONE poll that put Alf Landon ahead of FDR) that the people in the sample are not representative of the people who actually vote (or of the general population).  The people WILLING to participate in the poll may NOT be representative of the overall population (if you follow my advice to SABOTAGE polls, it will not be).  The POLLSTERS (people asking the questions) may not be honest.  They may either send UNCONSCIOUS signals as to the answer they want, or they may DELIBERATELY skew the results.  That is because polls have become an ever more obvious means of PUSHING one candidate or another as a "winner" on whose bandwagon you should get.  Logically, of course, your vote should NOT be affected by a poll.  But the way the media reports polls suggests that polls should MEAN SOMETHING to you.  They SHOULDN'T, as a matter of fact.  The mere suggestion that they should is an EVIL distortion of the election process, and of the political process (like on "approval" or "issue" polls).
See why I say that polls are meaningless?  They are little more than an "educated" guess, and NOT "news". 
Beyond all of the above, polls are an attempt to gauge opinion at a SINGLE MOMENT OF TIME.  The citizen wo you ask may have a DIFFERETN opinion 24 hours later.  MANY people (unlike this blog) may not think much about an overall philosophy, and their opinions may shift with the wind (and every sensory input).  Thus, polls this far out from an election are especially meaningless--of NO significance at all.
Yet, this is now the MAIN way the media reports elections  And some wonder why I hold the media in nothing but contempt!!!!
Then we come to Real Clear Politics and the Fox News habit of reporting their AVERAGE of all polls, as if it adds validity to that number.  It does not.
Gain, this is NOT a matter of opinion.  It is a "scientific" FACT that you cannot add any validity to ERRONEOUS polls by averaging the results.  I can illustrate this to you with a thought experiment. Say there are two polling organizations who MAKE UP poll results (dont' even ask the questions).  One shows McCain ahead by 10% and the other shows Obama ahead by 10%.  Neither has any validity  However, if you AVERAGE them out, you will REDUCE THE MAGNITUDE OF ERROR in any one of the FAKE polls, simply because of the averaging process.  If Obama is really ahead by 10%, then the other poll is in error by 20 percentage points.  But if you AVERAGE the two polls, the AVERAGE is only in error by 10 percentage points.  However, the average is not ANY more valid than the two FAKE polls.  You just reduce the possible "error" of any one of the polls.  However, if ONE of the polls is valid, and the other is not, then the average makes it IMPOSSIBLE to have the correct result. 
In short, averaging adds NOTING to the validity of the individual polls averaged, and the average is no more likely to be CORRECT than any individual poll.  In fact, the average is almost guaranteed to be INVALID, since it is likely to include invalid polls in the average.  To cite a Real Clear Politics "average" as meaningful is another one of those OUTRAGEOUS LIES, with no mathematical basis for such a conclusion.  For example, what SENSE does it make to "average" polls of "likely voters" and "registered voters".  They are not even picking their samples the same way!!!!  That is true across the board.  If ONE poll's sample is invalid, the "average" is NOT VALID.  In short, an "average" (by definition) is NEVERF as bad as the WORST individual poll and almost NEVER as good as the BEST individual poll.
I can't tell you how serious I am about this  We need, desperately, to ELIMINATE POLLS.  And we, the people, CAN do it by sabotaging them.
P.S.  Do I know whether McCain or Obama is ahead?  Nope.  But I have a good idea.  Right now, at a meaningless time, I think they are about 50-50.  That is NOT based on polls, but on the fact that--at a Presidential level--we have basically been a 50-50 country for some time  That dos not mean the ELECTION will end that way, simply because the way we are NOT a 50-50 country is if one of the candidates is viewed as DANGEROUS or hopeless (George McGovern, Walter Mondale. Barry Goldwater, etc.).  I have confidence that EITHER McCain or Obama (more likely, just because of his lack of experience and objective qualifications) could lose 60-40.  But it will take the people who do not live and breathe politics ALL shifting one way or another because they view on candidate or the other as DANGEROUS.  In that case it does not matter whether people "like" the "safer" candidate.  NO ONE "liked" Richard Nixon in 1972  Nevertheless, they voted for him because he was a KNOWN quantity, while McGovern was a far left extremist.  Hence a 60-40 vote.  In this election, Obama is MORE extremist than who the Democrats nomminated n 1972.  If he is PERCEIVED that way in the end, he is going to lose big.   However, McCain is McCain, and I am tempted to say that if he is PERCEIVED that way he is going to lose big.  I know.  One of them has to win.  Is it not sad?

No comments: