The Pardu

The Pardu
Watchful eyes and ears feed the brain, thus nourishing the brain cells.

Saturday, March 29, 2014

AP/GfK: Flawed And Conservative Survey Company Reports Only 26% Support ObamaCare


Let's take a look at AP/GfK to provide perspective on the 26% who "support" the ACA. My past experience with AP/GfK was along this line. As we approached the 2012 General Election, I ran across this headline:
"Majority Harbor Prejudice towards blacks."  

Read more after the break below

Now, why do you think such a headline would attract my interest? May I suggest,  if you are a normal person with any interest beyond sitting each day watching six different TV judge shows, or flipping between various mindless reality shows (including various derivatives of talent panel contests), the headline should have piqued your interest.

My askance and curiosity took me a few steps farther.  The following is an except from a piece I wrote after reading the October 2012 headline posted above. The TPI AP/GfK piece is long. Out of respect for your time and possible lack of deep interest in the genesis of the headline, I am posting a Awesome Screenshot Capture & Annotate image below the table box (below). 

TPI (October 2012)
If you never ever review poll data, you have to give a look at the AP/GfK Poll  data released just a few days before the November 6th Elections. The poll's major tickler was, a "Majority harbor prejudice towards blacks".   [See Link above]
Page 33 of the Growth for Knowledge (GfK) poll reads like this:

The TPI Link provided in the table will take you to even more startling and ridiculous survey findings related to perceptions of candidate religion. The 1,071 survey respondents (predominately white and from the South) seem to have been comprised of a potential respondent pool stacked for achievement of desired results. The headline: "Majority harbor prejudice towards blacks," was both indicative of what many should consider flawed survey administration via GfK, and indicative of the dangers of subsequent headlining via media or political operatives.

Now for the Russert, NOW W/Alex Wagner, segment. It is a good story, once you pass through the first 1:50 seconds which include suspect survey findings from the very "suspect" Growth for Knowledge (GfK).

Despite the good ACA segment, I believe it necessary to provide additional perspective on a polling authority that will surely receive wide quoting from the Right and conservative pundits as we move into next week.

But more than two-thirds distrust polls conducted by political parties or candidates and automated telephone surveys. Media polls fall in the middle, with somewhat more distrusting than trusting.  Note: the "News media" category below.
2013-09-04 Which polls do people trust
Nate Silver looked at which polls scored best coming out of the 2012 elections. His review also included look and reporting of poll bias. If you look about midway through the table below you will find AP/GfK; you will quickly notice the AP/GfK poll has a strong bias to the Republican Party. You may also notice GfK uses the very survey methodology shown by Silver as biased towards the Republican Party with moderate margins of error.
The New York Times 538 Blog  
"Which Polls Fared Best (and Worst) in the 2012 Presidential Race"Excerpt
We can also extend the analysis to consider the 90 polling firms that conducted at least one likely voter poll in the final three weeks of the campaign. One should probably not read too much into the results for the individual firms that issued just one or two polls, which is not a sufficient sample size to measure reliability. However, a look at this broader collective group of pollsters, and the techniques they use, may tell us something about which methods are most effective.
You Gov Dot Com
Figure 1 below shows the point estimates (and reported margins of error) for final national polls from different polling organizations for the Obama share of the national major party vote. The final YouGov poll, released on Sunday November 4, based on 36,472 interviews of likely voters between October 31 and November 3, had Obama at 48.5 percent, Romney at 46.5 percent, other candidates at 2.3 percent, and the remaining 2.7 percent of voters undecided. This translates into a 51.1 percent share of the national major party vote for Obama. As of this writing, the Associated Press reports that the Obama share of the national two party vote was 51.0 percent, so the YouGov estimate had an error of less than 0.1 percent. Two other polling organizations also pegged the Obama lead at two percent.
A few words about AP/GfK. As I sought information about AP/GfK, I ran across quite a few survey/poll authorities that employ survey takers and some pay the respondents on a per survey basis.  I suppose I was naive enough to think political polling took place predominately on random sampling by land line or cell phone. My knowledge of online "for pay" survey was at the level of nil; I did not know. I was also not aware that some survey firms actually have their "members' (as they call them), identified by profiles that includes information related to family size, rent or own home, race, gender, religion and social economic groupings.

While visiting one survey company website, I saw a comment complaint from a family respondent who complained about others in the household receiving multiple survey opportunities while stating they appeared cut-out of survey opportunities. 

I understand many companies and political organizations use focus groups and they do so extensively. So, my surprise at for fee survey taker, again, was embarrassingly naive and downright ignorant. A question from the revelation, "How difficult would it be to discern the inclinations of the survey and adjust answers to facilitate more calls or online opportunities?" Could contribute to a busy Pay Pal account since at least one authority stated they affect pay via Pay Pal.

We have knowledge GfK is a for fee polling (gift points - small dollars) authority. I also found GfK has an invitation or coded survey application process, See the image that follows. Please excuse the fuzzy screenshot.

Back to MSNBC and other media that are using the AP/GfK survey results. Commerce is commerce; viewers listeners and readers are objects to acquire revenue. I get all of that. If they media is going to use, what seems a very biased survey authority like GfK, why not include a broadcast or written disclaimer. The perfect example is CNN. 

When CNN reports on the Pew and Quinnipiac survey results you hear this: "Left leaning." why not give the quoted poll source perspective that is readily available via a couple hours of research?

I should out of fairness state, " I have never heard any host on CNN refer to Rasmussen" as "leaning Right."

Polls are critical to the American political process. Americans are not heavy viewers of news and we probably do not read enough for our own good.  Polls provide perspective on public sentiment. Public sentiment is constantly and effectively manipulated by the poli/social Right. Why would anyone believe surveys are not a tool or strategy item for Right-wing operatives?  

If you think I am off-base with this piece consider a few final points.

Why do you think Karl Rove literally freaked out the night of the Obama defeat of Mitt Romney? Rove's survey results and political data organization acumen was flawed, while Obama's via the Narwhal Project was impeccable.  

Why do you think Romney never developed a concession speech? He knew via polling he had the eliciting won!

Why would AP/GfK even publish a pre-2012 poll results that led to headlines: "Majority harbour prejudice towards lacks." Especially, after conducting a poll via online methodology with over almost 70% of survey takers white and from the US South?

Final question. Do you seriously believe the AP/GfK results showing only 26% of 1,012 respondents "support" the ACA?

No comments :

Post a Comment