Wednesday, October 17, 2012

binders full of women

While numerous platitudes and promises of questionable merit were thrown around last night at the presidential debate, the most popular aspect of the debate on the internetz right now is Romney's "women in binders" comment.  When asked about women's equality issues in the workforce, Romney touted his record in Massachusetts where his cabinet and senior staff had more women in positions than any other state in America. He commented that this was accomplished by looking though "whole binders full of women."

And these "binders full of women?" Well, here is what the (best of the) memes have to say about that:









Sunday, October 7, 2012

"unskewing" the data

im still stuck on talking about the current Presidential election. i am actually debating starting a solely political blog. i wish wonkette wasn't already taken. that's a good one. well, whatever. let me know if you think i should create a politics blog. maybe I will call it Merica. like, because i am "Meredith" and this is "America." good thing i didn't go into advertising, eh?

OKAY! SO last week i kept coming across a report that when you use "unskewed" data, Republican presidential candidate Mitt Romney is actually leading Barack Obama in the polls.

The story goes like this: an October 1st ABC News/Washington Post poll showed Obama leading Romney on average 49% to 46% amongst those most likely to vote.


To understand the meaningfulness of polls, you need to understand sampling. To "sample" is to select a segment of the population that can be generalized to the larger population from which you are sampling. So, if you want to understand attitudes of college students at a large state university, you need to sample a population that would represent those views. One way to do this would be to administer the attitudes questionnaire to several large general education courses; this way you're likely to get a variety of majors, races, ethnicities and both men and women. The best kinds of samples are those that are selected randomly. in the described scenario, then, a researcher would compile a list of gen ed classes, and then use a random numbers table to generate the specific classes that would receive the survey. we can never achieve perfect randomness, but we can try. A "skewed" sample, or nonrandom sample, on the other hand, would be if you gave the attitudes survey to an upper division philosophy section--those beardy weirdos don't represent the rest o' us! (JK i love philosophers).

okay, so when the poll came out showing that Obama was leading Romney, Virginia blogger Dean Chambers cried foul play regarding the sample. he claimed the ABC News/Washington Post poll oversampled Democrats, and this is why the poll shows that Obama is leading. Chambers then went on to create Unskewedpolls.com where he "unskews" the data. From what I can tell he reruns the numbers from the original polls, like ABC's, so that quotas of Republicans and Democrats are sampled to match Rasmussen's "real numbers" on party identification (Rassmussen's numbers show that of likely voters, 37.6% are Republicans, 33.3% are Democrats, and 29.2% are Independents).

THUS, he negates the use of a "random" sample for a quota sample based on party identification, to come to his conclusion---which is that Romney is in fact leading in the polls.




anyone who studies polling knows that quota samples are inherently flawed--especially when those quotas themselves come from POLLS! Furthermore, the numbers from Rassmussen are contested. This is precisely why you don't do quota sampling based on a non-static indicator, like party ID. and second, to poll on political election outcomes using a sample based on party ID (essentially the dependent variable) is absurd. party ID is directly correlated with vote choice. if you're interested in who people are going to vote for, you don't pad your sample with equal parts Democrats and Republicans! variation in the dependent variable is a mainstay of good research.  In an interview with TalkingPointsMemo.com, Public Policy Polling's Tom Jensen said this:
“We don’t have any preconceived notions about the party breakdown of a poll before we conduct it. The only things we make any adjustments for are gender, race, and age." 
thus, at PPP, they ensure that a representative sample of Americans based on gender, race, and age are sampled.

PPP's criticism notwithstanding, conservatives will still flock to Unskewedpolls.com because it confirms their biases toward a Republican win in November. but this kind of worldview confirming behavior is not unique to conservatives. Just ask Al Gore why Obama lost the debate last Wednesday. and unfortunately this kind of worldview confirming behavior is more common and overt this election season...which is really troubling.

Andrew Sullivan, a conservative British author and political commentator, wrote a blurb about this phenomenon in a piece for The Daily Beast, called Who's Delusional? Essentially, he writes that Democrats and Republicans have such different experiences, today, that their realities are fundamentally different. And thus, their interpretation of the "facts" will not coalesce. Ben Smith at Buzzfeed expanded on this by suggesting this divergence of realities is much more amplified this election season because in the United States now have such stark divisions in our media, that our "realities" are reinforced by an "authority" (the media). And i would take it one step further and extend this media reinforcement to include venues where likeminded people come together via online groups and forums to agree on perceived realities. in sum, falsehoods can become facts when enough people confirm them, and more than ever today, there are places to go for the affirmation.

a different interpretation of the facts and subsequent confirmation of the misinformation, is being seen with the recent jobs numbers, which show unemployment to be down, and at its lowest since 2003. Republicans are claiming that the Bureau of Labor Statistics is lying, while others suggest Democrats were lying when they filled out the form, because they see the world through rose colored glasses, as long as Obama is in office.

it all makes me think of Stephen Colbert and TRUTHINESS, which describes truth as "What you feel." merely hilarious in 2005 when it made its debut, today its all too "true."


The Colbert ReportMon - Thurs 11:30pm / 10:30c
The Word - Truthiness
www.colbertnation.com
Colbert Report Full EpisodesPolitical Humor & Satire BlogVideo Archive