The purpose of this blog is to get readers to think about the complex (or perhaps simple) issues I write about.

The primary topics will revolve around politics and society as a whole, but a mixture of sports opinion may be thrown in from time to time.

Thursday, May 31, 2012

Campaign 2012 Primer: How To Read Polling Data

Tomorrow, June will be on our doorstep while we all ponder why Spring time passes so quickly. The beginning June means were one month closer (now five-ish months-160 days to be exact) to the general election. For political beings and polemics like myself, its when the race for the Oval Office really starts to approach a the need to switch into second gear.

Since June and July produce little in the political world on election years, the only tasty bits we receive on a weekly basis is new polling data. Since I firmly believe that the 2012 campaigns and election are the most important in about three generations, I wanted to do a little "How To" writing on the campaign. In particular I want to talk about the polling that really starts to kick off just after Memorial Day-after the challenging Party has had time to Holiday from its hotly contested (and testy) primary fights, and the incumbent officially begins "campaigning."

All three 24 hour news channels (Fox News, MSNBC, and CNN) will bring their viewers polling data as soon as it is released, and the very inside the beltway, political chattering class chomp on it for 72-96 hours before the newest poll comes out.

Now all sorts of pollsters poll all sorts of topics, and in a political sense, this means national polling, state polling, polling on specific issues, or "feelings." Now of course, every polling company usually has some sort of bias, and I've found the most reliable polls to stick as closely to the center (< 250 people of one political affiliation more than the other) but that isn't a rule, by any stretch of the imagination. I will conclude by highlighting pollsters, their polls, and their biases in a bit, but for now I want to really get into the advice you can use that allows you to adequately read any poll.

The Basics: Almost every poll you'll here about between now and when you head to the ballot box is conducted via telephone. Polling companies analyze voter registration and voter behavior data in order to come up with the people they call. Calls are almost always pre-recorded and use a touch-tone prompt (Hit 1 if "Yes" etc.) to gather the information they are seeking. Many times, pollsters take a broad sampling (Call 1200 people) but report a small sample (100-150) that they have determined by use of special formulas, that represent the larger sample size, and that's what gets reported. This is also how the "Margin of Error" is determined. (Most quality polls have a MoE of 5% or below, anything more than 5% and the results can be scrutinized.)

National Polls: As most of you likely know, national polls are intended to gauge how the general (registered voting) public might vote in the general election if it were held today. Notice the language I chose: All national polling uses registered voters (more on this later) so therefore the question must intrinsically be, "If the election were held today, who would you be voting for?" Adhering to the basics just discussed, the sample and methodology is the same just on a larger scale that spans the country. 

Since the country on the whole cannot really be accurately represented so many months out from an election, the results of a "National Poll" really never paint the clearest picture of how each candidate might do in November. This is primarily due to the quick-fix, results-oriented nature of all political polling. To get to a reported sample size of 100 people, that's just two people from each state-not exactly representative.

Brass Tacks: The inside information, or "help" I can give you when looking at National polling data, is don't read too much into the initial "Who would you vote for today." In the end, the only people to get excited about these numbers are the campaign that came out looking the best, and even they know the number is meaningless. What national polls really will tell you is how the country is feeling on all of the "kitchen table" issues. While the sampling size is the same (obviously) a massive chunk of voters don't really make up their minds until the final weeks of the campaign, but where they're consistent is how they are feeling about certain things (whether the country is on the "right track" or whether or not they think the economy is really getting better). Everyone has a thought on these types of things, which is why the polling data for those questions is important to look at.

Tip: Most national polls can be found online in full-text form and they are extremely thorough. If you have five minutes you can easily scroll through the entirety of the poll-that is, you can read exactly what questions were asked and how the respondents answered.

State Polls: On the whole, state polling data is far more reliable insofar as getting accurate responses, and it is also much more revealing about how the electorate is feeling. New NBC/Marist state polling data was released today, from Iowa, Colorado, and Nevada. The polling data shows that among those states, the President and Mitt Romney are virtually tied. Without analyzing the data, and its political implications, the practicality of a state poll is really unmatched. Why? The Electoral College, that's why.

Regardless of whatever might appear on a given ballot ("down ballot") the Electoral College is a system that puts a premium on the electorate of the 50 States. Therefore, a national poll doesn't really tell us anything because the 2012 presidential election is not a popularity contest-though as we get closer to election, national polls will start to reflect the composite feeling of all the state polling data.

Brass Tacks: If you're planning on being an informed "Politico" this election season, state polls are red meat to the political chattering class. For instance, the three polls I just linked to show that this race is even closer than many of "us" thought, and that Midwestern battleground states will likely hold the key to the White House. State polls also ask similar "kitchen table" questions, and typically the results of those questions are also a bit more pronounced.

Tip: You will not have to look long and hard to find state polling data, especially if you watch any of the 24 hour cable news channels for more than 30 minutes a day. If you aren't interested in that combative resource, realclearpolitics.com does a great job providing ALL polling data and even does the task of averaging polls out.

How to Read the Polling Data: If you have a high school diploma or even a GED, you should be able to decipher the important bits of any poll. Aside from the "big number" which in cases of the presidential race is obvious, the next important thing is to find the margin of error. If President Obama is leading Mitt Romney 50-48 in a poll that has a margin of error of 4%, it means that Romney could be leading 52-46, or that Obama could be leading 54-44. Any time a "lead" is just a point or two higher than the MoE or is actually equal to or less than that number, the poll result suggests a "Statistical Tie" which means the race is dead even.

Just about all polling done between now and October 1, will be of registered voters: The reason for this is simple, their contact information is public record, and polling companies can assume that if they were responsible enough to register to vote, there is a chance they will actually vote. In fact, about 95% of all polling done between now and election day will be of registered voters.

Democrats typically have anywhere from a 2-8 point 'advantage' in registered voter polling: Despite the notion that polling partnerships like NBC/Marist or ABC/USA Today are seeking complete "fairness" most media outlets cannot stray too far from the inherent liberal bias. That is not to say that the polling data they present isn't accurate, what I mean is that most of these types of polls show a 200-600 "registered Democrat" majority in their sample sizes. The reason I mention bias is because it's terribly hard to get those who are "Registered Independent" to participate fully, or to provide decisive answers to the polling questions, which makes it more difficult to really get equal amounts of registered Democrats and Republicans. So when all is said and done, if Democrats are only slightly ahead of Republicans in any one of these polls they really may not be ahead at all, because...

Republicans are at a distinct advantage in polling done of "likely voters": Ironically enough, it's polling data throughout time that has shown this. The reason Democrats do better in registered voter polling is because it's largely a poll of people who may not be all that in tune with what is going on-or to put it more frankly, registered Democrat votes are much more likely to stay home on election day. Except in rare cases where the Republican Party's elected leaders have done a bad enough job to warrant such abstention, Republicans in general, are much more likely to vote - hence why they hold a polling "majority" among "likely voters." For instance, a month out from the 2010 Primary Elections, Democrats had a solid two or three weeks of "comebacks" in the polls but in the month-to-the-day start of Gallup's month-long likely voter model, looked to have squashed the comeback. October 5th's Gallup Weekly Tracking poll of the mid-term races had the GOP holding onto a slim 3 point edge among registered voters well, within the MoE. The day also brought the first results of the likely voter model, which proved the notion true, the GOP held a staggering 13-point advantage among that demographic.

If You Still Can't Figure Out Where the Race is Going to Go: All you should have to do is look at the tone and the narratives coming out of the major media outlets. Let's hypothetically say it's October 21st, about three weeks away from the election and polls look like they look on May 31st. ABC, CBS, and especially NBC will likely be covering every possible story that calls into question Mitt Romney's legitimacy to become President, while likely not covering Obama at all, or providing puff pieces on what his administration has done for four years.

To the contrary, if polls on October 21st show Obama up 3-5 points, expect news coverage to actually be less critical to Romney, but rather of complete ignorance to his campaign and its message. If Obama holds this lead by that point, nightly news broadcasts will spend about a third of their time covering the campaign, and will try to focus on hard news stories that don't make Obama look bad.

Now if Romney is actually up by 3-5 points, first off I will be giddy as all hell, but you can expect to see some of the most biased and slanted news coverage you have ever witnessed. Outside of Chuck Todd (who I will highlight in a moment), MSNBC will be in full-on Romney bashing mode, and should the 3-5 point lead go into the pivotal "two weeks left" mark, look for them to start covering his religion, if they haven't already done so.

Where To Go To Get All This Stuff: As I mentioned previously, 24 hour news channels (pick one) will all be broadcasting these polling results, and most will cover all of the salient "kitchen table question" data as well. Obviously the contextual narrative to follow the reporting of the data will be skewed in various ways, but 24 hour news is the place to go if you want to know what the numbers "mean."

Twitter. Or should I say, #Twitter. If you have not signed up for Twitter because you "don't see the point in telling people about your latest bowel movement," you likely already have a Facebook account and are showing me photographs of your grotesque "rescue" Shi-Tzu mix with a face only a mother could love. Twitter is not about you, in all reality. It's about what is going on in the world.

For instance, I am not a Twitter "presence" I have something in the neighborhood of 200 followers, 80 of which are "bots" for pornography, but Twitter is where I go to get my news, whether it be breaking, or a few hours old. I follow 592 people currently and I would venture to guess that 350 of them are either professional members of the media, or they are political news makers themselves.

You don't have to post anything to Twitter if you don't want, that's the beauty. And, if you're like me, you are white, and thus you inherently like seeing/knowing things way before any of your friends do, THAT IS EXACTLY WHAT TWITTER IS. I will have friends email me articles at 4:15 in the afternoon that I read when I was stuffing my face full of Fruit Loops.

More to the point: All of the major polling outlets are on Twitter. That's right, do you want to know what the people in California are currently thinking about a ballot proposition to mandate every county have an organic ginger farm? @pppolls will tweet that information out before they officially publish it. Who's the smarty pants, now?

Brass Tacks: Join Twitter, you imbecile. 

Polls, Pollsters and The Best Commentators: I will just run through this briefly in list:description format to keep it quick and easy (in no order).

RealClearPolitics: RealClearPolitics.com is a website that is sort of a prettier version of Drudge but without the inherent RW bias. It's keynote feature is its aggregation of all polling data on a given topic or in a given state. The "Real Clear Politics Average" is roundly cited by all news outlets.

Rassmussen Reports: A polling company run by Scott Rasmussen, who has a contract with Fox News, which is where you will see his numbers as they come in. His data has slowly become less reliable, but if you want to know how a Republican might be feeling, these are the numbers to consider.

Progressive Policy Polling: As the name suggests, this firm has a more Democratic bias to its sampling, and after some shoddy results in 2008 and 2010, it has cleaned up its methodology and has been giving very reliable numbers as of late. MSNBC's go-to poll for obvious reasons, but Fox News has no problem citing their numbers as well.

NBC/WallStreetJournal: "Old Reliable" this poll is usually fairly accurate but in some instances can seem to be an outlier. A combination of the most liberal news network and the most "conservative" of mainstream print newspapers, sampling size is always good, although report selection seems a bit odd from time to time. Still, this is a great poll, but the only bad thing about is if polling results don't come in early enough to start the day's news cycle to "make news" all day, NBC will quarantine the data for its nightly news broadcast, which then MSNBC completely gets to play with while CNN and Fox Have to wait until their live shows.

NBC/Marist: This is NBC's long-form version of its polling data. Using a University's polling apparatus, it gives NBC an outlet to conduct larger surveys of a bigger sample size. Same rules apply to the release of the data as the previous poll, but much more accurate data itself.

ABC/USA Today: Another media poll, similar to NBC/WSJ, but typically with a bit of a larger registered Democrat roll. Results are rarely ever favorable to the Republican candidate, making this a prime example of why it's important to read onto the kitchen table data.

NYTimes/Blank: It's the New York Times, while its name is synonymous with "What we all are supposed to think the top news is" it possesses smaller but albeit similar biases to its newspaper, but nothing that is afoul of getting legitimate polling data. But one can say this, if Democrats are struggling in a New York Times poll, they are probably struggling worse than some of the other polls might suggest.

Any University: Several Universities (mostly on the East Coast) have separate polling apparatus/institutions on campus that are constantly polling the American people. Mostly provide national data, but do offer state polling data as the election nears, and will team up with media $ources to get bigger sample sizes and earlier state polls conducted. University polling is always pretty reliable.

Gallup: Perhaps the most comprehensive polling apparatus in the world. They track everything, and they track it on a daily, weekly, monthly, and yearly basis. Somewhat recently since they have increased the volume of polls they conduct, some of their methodology has changed to get data for "cheaper" costs, but what I have found is the criticism is coming from one side and one side only. Democrats have a love-hate with Gallup more than they do with any other poll or polling firm. When it's on their side, no Democratic campaign strategist has a problem with it, and will gladly put it in a press release, but when it's against them, their methodology is questioned and on and on. Republicans generally fear Gallup because for the longest time it was considered just slightly left leaning, but recently, thanks to the President's approval rating being sub 50, they've warmed a bit. Routinely cited by all news outlets.

Top Media Personalities for Poll Analysis and Interviews...

Chuck Todd: Simply the best. There's absolutely no doubt who he's voting for every election day (his wife is a Democratic operative) but he has really become one of the great objective political reporters out there. His show, The Daily Rundown is a must for anyone who is remotely political, and its after his show that MSNBC takes on the character that keeps the Media Research Center in business.

Anderson Cooper: Yes, the grey fox is on this list because he too can objectively present and analyze data and he's not afraid to challenge any guests on his show regardless of their political role. If Fox News toned it down one more notch and hired him to replace Sheppard Smith, he'd get more views than Shep gets now.

Bret Baier: Host of The Special Report weeknights on Fox, he is Fox News' closest thing to Chuck Todd. While I wish his entire hour was like the last 20 minutes of the show (which would make it just like Todd's) he is able to keep himself framed as FNC's hard news political host. The only problem for Bret, is that he works for Fox News, and so literally no one from the Obama Administration, or campaign will consent to a live interview on his show, despite him being a fair host.

There are several others but this is getting long.

I hope you learned something and have some more tools in your tool belt as you navigate often times conflicting poll results as they come your way this summer and fall.

No comments:

Post a Comment