Wednesday, September 19, 2012

Erin Burnett: Liar on The Liar Network (CNN/Media Poll Lies)

Erin Burnett is a piece of work, and I do NOT mean that in a good way. Bouth competence and teh truth are not in her.

Ms. Burnett yesterday:  "Does his mean Rmney is DONE--his campaign is over. No, not quite (clear disappointment here).  In 1980 Jimmy Carter led Ronald Reagan by 1 percentate pont in Septermber, and Reagan won by 2."  erce

Ms. Burnett, it may surprise you to learn that Ronald Reagan efeated Jimmy Careter in a LANDSLIDE (although not a landslide of the magnitude Reagan would defeat Walter Mondale in 2984). Reagan won, in 1980, by about the SAME margn that Barack Obama defeated John McCain.  And the MEDIA was still calling the 1980 race a "toss-up" based on the POLLS, the WEEKEND before the electin (not just in september). 

"How do I LIE to thee; let me count the ways."  (again, with apologies to Elizabeth Barrett Browning).

It is a LIE--and media peole KNOW it is a LIE--to say that either Carter was "leading Reagan by 1 percentage point, or that Obama is NOW "lleading" Romney by 1 percentage pont (latest Gallup poll).  Forget about all of the many problems with polls.  Even taking polls at their face vaule, and as meaning someting, the EXPECTGED MARGIN OF EROR in these polls is 3-4%, and the plls are often not even measuring the same people.  It matters a LOT whether a poll is lookng at "likely voters" or "registered voters"--much less lthe truly disgraceful "all peole".  IF Galluyp is right, the "race' between Romney and Obama is now a dEAD HEAT: well within the EXPECTED margin of error. 

But does it not "mean something" if he AVERAGE of polls shows an Obama lead.  Sob.  NO.  This is a STATISTICAL LIE.  An 'average" of polls adds NOTHING to the validity of any individual poll.  In fact, if the polls differ much, that mrely shows gthat POLL S ARE UNRELIABLE.  Nope.  This is NOT a matter of opinion. It is a statistical TRUTH.  Yo CANNOT "bootstrap" polls by 'averaging" all of the polls.  Sure, you can reduce teh margin of error, usually by only a little, by ADDING TO THE SAMPLE SIZEW.  But atht is NOT whaqt you do when yuou "average" olls.  You aonly add to the sample sie IF yu use the SAME POLLING METHODS AT THE SAME TIME.  This "averaging" of polls "averages" polls using DIFFERENT polling methods (different samples, often), and usually the polls are taken at DIFFERENT TIMES.  "Averaging" helps you NOT AT ALL. Oh, "averaging" does do one thing because of the nature of 'averaging" (not becaue any "accuracy" is added).  By definition, "averaging means that the "avrerage" is AL:WAYS SOMEWHEERE BETWEEN THE EXTREMES (unless all of the plls are the same, in which case ou don't need "averaging') . Taht means that an AVERAGE (of anything) ALWAYS will reduce the possible MAGNITUDE of error. Look, for example, at this ossible set of percentages: 30%, 40$ and 50%.  The AVERAGE is 40%, but statistically that is nO morfe likely to be right than the plls that found 30% or 50%.  Hoever, notice that IF you say that you "agree" with 30%, and use that poill, you may be a full 20% OFF (maximum error--the eifference between 30% and 50%).  However if yu choose to go with the AVERAGE, you can "only" be 10% in error, because that is the MAXIMUM error if ANY of the polls are right.  And, if there is truly this spread, NONE of th epolls is probably right.  Of course, what I said is not quite true.  What if the actual percentate, in my example, is 20%?  Then the average can be wrong more than 10%, but still CAnNOT be wrong as much as if you go with either extgreme.  Thus, all "averaging" does is reduce your MAXIMUM possible error.  It tells you NOTHING about the correct number. 

Erin Burnett's LIES are not over.  Not only did she INGORFE the "margin of error" for AGENDA purposes, but she went to the "battlegound" state polls (namely ONE on Virginia) to try to say that Rmney is really pretty much falling WAY behind.  I have previusly informed you of the LIE in this stupditiy. True, Al Gore showed that you CAN lose the electin, even though you win the popular vote.  But Al Gore also showed that is ONLY true if the popularer vote is CLOSE (within half a percent or so)  Thus, if either Romney or Obama wins the popular vote by as much as a full percentatge point, that person will (99% or mroe certain) WIN TH EELECTIn . It does not matter that the electin is a "state-bystate electin, when the NATINAL PESULT is almost SURE to be the SAME as the state-by-state result UNLESS the electin is within about HALF OF A PERCENT in the popular vote.

Erin Burnett is a LIAR.  With a straight face, she quoted a Washington Post Poll showing Obama ahead by 8 percentage pints in Virginia, while Obama is TIED (that one percentage point) with Romney in the natinal oll.  Nope.  That is IMPOSSIBLE.  One of the plls has to be WRONG, and Ms. Burnett MUST be a LIAR (not to pont that out).  It si INCONSISTENT for Obama to be TIED natinallyl, and 8% ahead in Virginia.  Isn't it POSSIBLE? Sure, but it is also POSSIBLE for ALL of tghese  polls to have an ERROR of about 100%.  Oh. The ODDS are long. But there is SOME chance, by cance alonge, that you will pick the ONLY 1000 people in Virginia who support President Obama for yuouir poll, wile EVERYONE else supportes Romhey . True, the odds are aoubt the same as the universe ending TOMORFROW, but it is OSSIBLE.  In the real world, whre polls CAn easily get a sample wrong by 5, or even 10, percent, Obama is NOT "leading" Romney in Virginia by 8 percent. That is espeically true wehn Bruneett said taht Viirginia was pretty tyupical of what is happening to Romney in a NUMBER of the battleground states.  Nope.  NOT POSSIBLE for the NATIONAL POLL to stay the same (actually moving SLIGHTLY in favor of Romney from the preivous poll_, while there are these dRAMATIC moves in "battlegournd states. 

Message to CNn and Erin Burnett (as well as all media) You should FORGET polls alltogether, and reort REAL NEWS.  But sicne yo are BAD PEOLE, I don't expect that.  Failing that, at least don't be this DISHONEST about plls.  There is SOMETHING WRONG if the "battlegound statates" show a supposed DRAMAATIC move in the plls, while the natinal polls remain the SDAME.  Cannot be.  IT is a LIE to report that with a straight fgace. 

P.S No proofreading or spell checkng (bad eyesight). 

No comments: