Welcome back to The Plebian's Guide to Polls.  In last week's issue, "Why We Poll", we discussed how polling provides a stable estimation for public opinion.  Today's issue, "The Four Polls", will address key distinctions among the types of polls that are conducted.

There's more...

----------------------------------------

The Four Polls

In our discussion of polling, it's important that we understand the different reasons why someone might want to conduct a poll.  Not all polls are measures of political support.  I divide the field of opinion polling into four categories, which we'll discuss below.  The categories are:

• Political Polling - polls conducted to measure support for a political candidate
• Policy Polling - polls conducted to gague public opinion on political conditions or policies
• Parse Polling - polls conducted to ascertain how to craft a message with more public appeal
• Push Polling - polls conducted to explicitly change public opinion

Although these categories share many commonalities and some overlap, I want to distinguish them, each in turn.  There are differences in how these polls are constructed and conducted, and more importantly for us, there are very significant differences in how these polls should be interpreted.  I will be providing one real-world example for each type of poll, to help illustrate these differences.

Political Polling

Political polling is the category with which we are most familiar.  As our example, let's take the Oct. 31st, 2004 FOX News poll from Ohio.  The specific question polled and the results are as follows:

"If the election for president of the United States were held today, would you vote for: [ROTATE] 1. Republican George W. Bush 2. Democrat John Kerry 3. (Not sure/Other)
4. (Would not vote)"

Bush: 50%
Kerry: 47%
Other: 3%
No Vote: -

Political polling is the most well-known form of polling for a very simple reason: it also carries the highest stakes.  Careers are made, fortunes won and lost, on the outcomes of national elections.  For this reason, in political polling, predictive accuracy is paramount.  Candidates need to have a clear picture of where they stand with the electorate.  Good polling not only presages the electoral outcome, it allows a candidate to target their message toward groups where they can have the largest impact.

This FOX News poll from Ohio is an excellent example of the need for accurate political polling.  The final margin in Ohio, 2004, was Bush: 50.8%, Kerry: 48.7%.  Had the polling been less accurate in this, a very competitive state, the candidates might not have contested it so hotly.  Polling provides an excellent measure of where a candidate's resources are best spent.  It is critically important, in end-of-campaign budgeting, to know which states remain competitive and which are likely out of reach.  Although I fully support the Fifty State Strategy (which appears to have done exceptionally well in 2008), broadcasting money without targetting is a good way to lose any election, especially in the closing days.  Political polling is the key to effective targetting, and accuracy is critical in ensuring that money goes to the most competitive targets instead of being squandered on longshots.

Accuracy in political polling is also important because polling drives the media narrative.  Even though random variation (noise) is inherent in all polls, the media has a perverse tendency to read two days of upswing or downswing as "momentum".  Accurate polling helps keep the media narrative on real trends, instead of leading reporters on a wild outlier chase.

Policy Polling

For our second example, let's take a survey question from the aptly named Public Policy Polling (PPP).  The following was part of a phone poll of North Carolinans, taken on Feb. 12, 2007:

"Which would you prefer: the current health insurance system, where most people get their health insurance from their employers, but some people have no insurance; OR would you prefer a universal health insurance program, where everyone is covered under a program like Medicare that is run by the government and financed by taxpayers?"

Current system: 37%
Universal health care: 51%
Don't know: 12%

Policy polling goes hand-in-hand with political polling, but where political polling deals with electoral races, policy polling deals with... well... policy.  Or more generally, it deals with opinions about current events and government programs.  Gallup's consumer confidence tracking poll (as opposed to the Consumer Confidence Index, published by The Conference Board) is another type of policy poll.

Policy polls help politicians understand what issues are important to their constituents, and what solutions their constituents favor.  Policy polls also help to set the media narrative on issues.  Polls like the one by PPP are instrumental in bringing attitudes about the health care system into the public discourse.  Ideologues on the left and the right may argue about the advantages of universal vs. free-market health care, but at the end of the day, a smart politician knows not to go completely against the will of his or her constituents.

Policy polls help to keep the government on track, focusing on the issues that matter to everyday people.  The government and the media both use them to highlight the issues that are most important to common citizens.

Parse Polling

Parse polling is intimately related to policy polling, but with a twist.  Where policy polls focus on communicating popular opinion, parse polls focus on the most popular ways to communicate an opinion.  Parse polling is about crafting a message, not crafting policy.

Unfortunately, I'm going to have to go back on my word here.  I promised to provide real-world examples of each type of poll that I'd discuss, but I can't seem to get my hands on a poll report for a parse poll.  In fact, I've never seen one - perhaps unsurprising given the negative publicity often associated with using this type of poll.

So to illustrate the difference between policy polling and parse polling, let me give you a very short tale of two presidents.

Bill Clinton got in a lot of heat for his marriage.  No, not to Hillary; to Dick Morris, the Clinton pollster.  The following is an excerpt from the PBS series Frontline (Jan. 16, 2001):

GEORGE STEPHANOPOULOS: Over the course of the first nine months of 1995, no single person had more power over the president, and therefore over the government, than Dick Morris. No question about it.

CHRIS BURY: The favorite West Wing joke was that the president had developed a split personality. Daytime Clinton worked with his regular staff, but nighttime Clinton belonged to Dick Morris. The team feared the president now relied on polling data more than principles. It got so bad that even Clinton's vacation plans were determined by Dick Morris's polls.

LEON PANETTA: My view was, "You're president of the United States. You've got to make some very important decisions on some very difficult issues." And I mean, I remember at one point basically saying, "Look, you know, Abraham Lincoln did not have to have a pollster in this office to decide what's right and wrong. And you don't need a pollster, either."

ROBERT REICH: The assumption that everything had to be poll-tested, that you couldn't have a good idea for what the president ought to do or where this country ought to go unless some poll indicated that it would be particularly popular with the public- that alone I thought a little bit galling.

Bill Clinton used policy polling to an heretofore unexplored level of excess.  But Bill Clinton always stuck to the policy polling side of the divide - looking for the most politically expedient moves to appeal to the American people and bolster his public support.

George W. Bush, on the other hand, took the Clintonesque notion of polling every move and turned it on its head.  Bush and his pollster, Jan van Lohuizen, didn't use policy polling to decide what to do.  Where Clinton had been a pragmatist, Bush was a consummate ideologue.  George W. Bush knew what policies he wanted to implement.  He just didn't know how to sell them to the public.

In "The Perils of the Permanent Campaign", Time's Joe Klein spells out how the Bush White House used polls not to gague the public's opinion about matters of policy, but to gague what language would make the public most receptive to (potentially unpopular) policies George W. Bush wanted to implement.

The difference between policy polling and parse polling, then, is this: policy polls are conducted to determine how the public feels about a particular issue (or policy); parse polls are conducted to determine how best to spin an issue (or policy) to make it well-received by the public.

Push Polling

Perhaps the most famous example of push polling - George W. Bush's phone campaign against John McCain in South Carolina during the 2000 Republican primaries - wasn't actually push-polling at all.  Viewed in isolation, however, the following question (taken from that very campaign) is indistinguishable from a push-poll question:

"John McCain calls the campaign finance system corrupt, but as chairman of the Senate Commerce Committee, he raises money and travels on the private jets of corporations with legislative proposals before his committee. In view of this, are you much more likely to vote for him, somewhat more likely to vote for him, somewhat more likely to vote against him or much more likely to vote against him?"

You'll notice that I didn't provide results on that question.  This is because push polls, despite the name, are polls in form only.  A true push poll is short, with a handful of questions lasting only a few minutes.  A true push-poll is also broadcast - sent to tens of thousands of potential voters - rather than targetted to a select group of poll participants.  The responses to the poll are never analyzed.  The entire purpose of a push poll is to spread a negative message about a candidate or issue, under the guise of polling (which most people will assume to be fundamentally fair and unbiased).  Push polls deserve to be mentioned with the other types of polls because they do take the same form as other polls, but push polls are propaganda, plan and simple.

If the above question is such a good example of a push poll question, though, why do I say that the phone campaign against McCain was not a push poll?

Despite the media saturation it received, it turns out that George W. Bush's polling against McCain in South Carolina was parse polling, not push polling.  The above question was part of a 20-minute phone survey, given to 300 people across the state.  The Bush campaign was, in fact, using these calls John McCain has a black to determine how effective certain accusations would be against John McCain.

To be fair, though, the most infamous question of the phone campaign against John McCain may well have been push-polled... though the calls were fielded anonymously, and have never conclusively been traced back to the Bush campaign.  That most incendiary question was:

"Would you be more likely or less likely to vote for John McCain for president if you knew he had fathered an illegitimate black child?"

----------------------------------------

Well, that's it for this week's installment of The Plebian's Guide to Polls.  The series will be on hiatus next week - there's an election Tuesday, in case you hadn't heard, and I'll be driving over 1300 miles to cast my vote (2600+ round trip).  Overdetermined.net will have plenty of election coverage in the meantime, however, so please make sure to stop by and check us out!