I continue the fight for them

I continue the fight for them
Daylen and Grandad at the beach

Wednesday, May 11, 2011

THE SCAMMING OF THE AMERICAN PUBLIC (#1)

Daily I read and investigate a "scam" employed by our federal government to skew data and deceive the public.  This particular blog will be a continual exercise and will most likely re-appear as SCAMMING-Part 2/Part 3/etc.

I will post numerous examples over a period of time on Part 1 before re-addressing the issue in Part 2, and so on.

#1- The AP Poll 

( http://news.yahoo.com/s/ap/20110511/ap_on_re_us/us_ap_poll_obama_boost) has Pres. Obama's approval rating at 60% and 53% say he should be re-elected.  David Axelrod must be turning cartwheels.


Does the AP have no shame?  You then look at the party ID of this particular poll.  46% identify as Democrat or leaning, 29% identify as Republican or leaning, 4% identify as purely independent, and 20% answered "I don't know."  In October of 2010, the AP poll had a split of 43% Democrat and 40% Republican.


What we have in this poll is a 17% point margin in favor of the Democrats.  Notice that in Gallup’s polling, party ID remains pretty stable. In roughly 40 polls since mid-2009, Democrats and Republicans have both ranged in the 40s with leaners. During that time, the split has never been larger than 7 percentage points.


The AP sample contains twice as many Democrats as Republicans (35% Democrat vs. 18% Republican).  While party identification is, itself, a variable to be determined by polling, the normal sample usually contains an approximately equal number of Democrats and Republicans.  Lately, Republicans have actually been slightly in the plurality in most statistically valid samples.

This AP poll on party affiliation (May 2011) -- with its 2:1 Democratic edge -- is nonsense.


PARTY IDENTIFICATION IN AP POLLS

Democrat     Republican
October, 2010 32% 25%
November, 2010 29% 24%
January, 2011 28% 25%
March, 2011 29% 20%
May, 2011 35% 18%
 

In my experience of evaluating polling results for accuracy, I've found Gallup and Rasmussen to be the most honest and accurate.  I've come to this conclusion in that their polling most accurately described the results of elections, etc.  Many Democrats say that Rasmussen tends to favor Republicans in his polling, but my response is that whenever his polling trended towards Republicans, the results confirmed his poll as accurate.
_______________________________________________________
_______________________________________________________

3 comments:

  1. Not sure you can really say Rasmussen's biased claims are justified by election results:

    http://fivethirtyeight.blogs.nytimes.com/2010/11/04/rasmussen-polls-were-biased-and-inaccurate-quinnipiac-surveyusa-performed-strongly/

    As far as the AP poll goes, they use random sampling, just like everybody else. I'm no poll expert but the difference in party ID distribution probably has less to do with the sample and more to do with how people in the sample are asked to identify themselves.

    In any case, as your own table hints at, the party ID distribution probably didn't change in a statistically significant way from their previous poll in March. What this means is that even if their sample is biased, it hasn't shifted meaningfully. Thus, I'm not sure claims of biased sampling can be used to explain away the recent rise in support for Obama.

    While I'm no defender of the Fed and some of the criticisms are justified, some of the things you describe do make sense to me, at least on the surface. Like putting in a cheaper substitute: if prices go up on the steak, won't people be more likely to buy the hamburger instead? So, doesn't it make sense to measure CPI that way? Just b/c a measurement choice makes the inflation rate lower doesn't automatically mean it's corrupt or bad choice, right?

    ReplyDelete
  2. To Anonymous re: Rasmussen

    Thanks for the response. Anything that intellectually challenges my premise is well received and inspirational.

    According to the table that is located within the NYTimes article, it leads an observer towards two interesting observations. The table represents 10 firms accuracy in predicting the success in various Senate and Gubernatorial elections in 2010. I find it interesting that House elections were not included in this analysis.....

    "In December 2009, a full 11 months before Election Day, a Democratic strategist concluded that if the Rasmussen Reports Generic Congressional Ballot data was accurate, Republicans would gain 62 seats in the House during the 2010 elections. Other polls at the time suggested the Democrats would retain a comfortable majority. The Republicans gained 63 seats in the 2010 elections."

    (NYT article) individual U.S. House races are generally quite difficult to poll, and the typical poll issued by companies other than SurveyUSA had missed the margin between the candidates by an average of 7.3 points.

    I wonder how the accuracy data for all 10 firms would have changed with the House results included?

    Also, Rasmussen was evaluated off a total of 105 polls. The other nine firms ranged from 14-45 polls. The lesser the number of polls equates to less reliability in their results.

    (NYT article) "Other polling firms that joined Rasmussen toward the bottom of the chart were Marist College, whose polls also had a notable Republican bias, and CNN/Opinion Research, whose polls missed by almost 5 points on average. Their scores are less statistically meaningful than that for Rasmussen Reports, however, because they had only released surveys in 14 and 17 races, respectively, as compared to Rasmussen’s 105 polls."

    (NYT article) "YouGov, which conducts its surveys through Internet panels, also performed fairly well, missing the eventual margin by 3.5 points on average — although it confined its polling to a handful of swing races, in which polling is generally easier because of high levels of voter engagement."

    I'm not saying that the other firms cherry-picked the races they would poll, but it is obvious that Rasmussen's 105 polls reduces the chances of pre-determining results.

    Rasmussen is known for being accurate more often than not.

    (NYT article) "Rasmussen’s polls — after a poor debut in 2000 in which they picked the wrong winner in 7 key states in that year’s Presidential race — nevertheless had performed quite strongly in in 2004 and 2006. And they were about average in 2008. But their polls were poor this year."

    "Pat Caddell and Doug Schoen, pollsters for Presidents Jimmy Carter and Bill Clinton, say that Scott has an “unchallenged record for both integrity and accuracy.” The Wall Street Journal  proclaims that "Mr. Rasmussen is today's leading insurgent pollster" and "a key player in the contact sport of politics."

    I will admit that Pat and Doug have become friends with Scott Rasmussen. But they are accomplished pollsters in their own right and have been long time Democrats.

    ReplyDelete
  3. To Anonymous re: your statement-

    (Anonymous)"While I'm no defender of the Fed and some of the criticisms are justified, some of the things you describe do make sense to me, at least on the surface. Like putting in a cheaper substitute: if prices go up on the steak, won't people be more likely to buy the hamburger instead? So, doesn't it make sense to measure CPI that way? Just b/c a measurement choice makes the inflation rate lower doesn't automatically mean it's corrupt or bad choice, right?"

    The Steak vs Hamburger analogy may have been a poor choice on my part, but a larger issue is still unresolved.

    If steak makes a dramatic increase, the Fed's will look for a much cheaper substitute. In other words, if the price of hamburger does not significantly help bring down the CPI then they keep going down the list until they get the figures they are looking for. That might be found in hot dogs for all we know.

    IMO, the tools for measurement should remain consistent, month after month after month. I question when any change to the formula always involves finding a cheaper item in a particular category. This equates to selective sampling in order to get the results that you want to achieve.

    IMO, if not corrupt then it is deceitful. They pick and choose in order to achieve the desired result.

    ReplyDelete