Idea: A (deviously) effective online political campaign, stealth-style

by Alan Cohen

Perhaps the political campaigns have beat me to this, but here’s a somewhat slimy idea for a stealth way to market a political party online. It’s kind of an online equivalent of push polling, those fake, campaign-driven polls that ask questions like, “Do you support Obama’s job-crushing policy of raising taxes on working families, or Romney’s pro-growth policies for promoting a healthy economy?” Except that push polling is blatant, and many people hang up the phone. A well-conducted online equivalent, run by a SuperPAC with a neutral name, could be highly effective.

This can be viewed as a primer in how to lie with statistics. On the other hand, I think that most voters are ill-informed, and that they would vote quite differently if they understood the parties’ true agendas and records. Paradoxically, then, designing a misleading survey could actually produce better-informed voters – see my explanation at the end.

Imagine you’re browsing your favorite web sites, and you come across an ad with lots of flashing red and blue and a donkey and elephant. “Are you a Democrat or Republican? Find out here with our FREE survey…” Personally, I never click on these things, but I assume someone does, which is why they continue to annoy me on my favorite websites. So imagine you click on this.

You’d be taken to an intro page, which would explain that the survey will assign you a score between zero and 100, zero being completely Republican and 100 being completely Democrat. A disclaimer would state that the survey is not affiliated with any political party or candidate, and, though thoroughly researched, is solely for entertainment purposes. You then click on another button to start the survey. One at a time, you work your way through questions with a Likert scale. Here is a sample of what you might find:

1) How strongly do you agree with the following statement: “Women should have the legal right to have an abortion.”

Strongly agree – agree – neutral – disagree – strongly disagree

2) How strongly do you agree with the following statement: “The US constitution prohibits the government from restricting citizens’ possession of firearms.”

Strongly agree – agree – neutral – disagree – strongly disagree

At the beginning, the questions would be designed to separate people out on hot-button issues, giving a sense of fairness and making them think the survey was accurate, as in Questions 1 and 2. However, subtle choice of questions and wording would rig the final outcome so that most people (regardless of their true preference) would arrive at the same outcome.

Here are some examples of questions that would favor Democrats:

14) How strongly do you agree with the following statement: “Health insurance companies should not be allowed to refuse coverage to people with pre-existing conditions.”

Strongly agree – agree – neutral – disagree – strongly disagree

15) How strongly do you agree with the following statement: “Currently, middle-class people pay more than their fair share in taxes, and rich people do not pay enough.”

Strongly agree – agree – neutral – disagree – strongly disagree

These last two questions are carefully worded to be questions that most people of both parties will agree with, and also where agreement will favor Democrats. By choosing lots of neutral questions (like 1 and 2) and lots of Democratic questions (like 14 and 15) but very few Republican questions, the survey can have an air of authenticity, but still be quite rigged to produce a certain outcome, at least for most people.

Another way to think about this is that each question has a weight, implicit or explicit. If each question counts for the same amount, but I ask one question on abortion and 10 questions of health care, I am weighting someone’s opinion on health care 10 times more than their opinion on abortion. One can get around this by asking more abortion questions, or up-weighting the abortion question (letting it count for 10 times as much). But in the end, it is always a subjective exercise in defining what issues are most important in separating people out. A more accurate survey would let people assign their own weights to each topic, regardless of the number of questions. But we want to manipulate things (heehee!) so we won’t do that.

After completing the survey, participants would get their score: “Congratulations! You’re 68% Democrat. Here’s the breakdown of your question answers, where the parties stand on these questions, and how your score is calculated.” Below, you would find things like:

14) How strongly do you agree with the following statement: “Health insurance companies should not be allowed to refuse coverage to people with pre-existing conditions.” You strongly agreed with this statement, giving you two points toward being Democrat.

Explanation: In 2010, 98% of Democrats in congress voted for a bill that prohibited insurance companies for refusing coverage based on pre-existing conditions, and 100% of Republicans voted against it. The democratic platform says “blah blah blah,” and the Republican platform does not mention this issue.

15) How strongly do you agree with the following statement: “Currently, middle-class people pay more than their fair share in taxes, and rich people do not pay enough.” You agreed with this statement, giving you one point toward being Democrat.

Explanation: Over the past 10 years, there have been 127 votes on bills that would have increased the proportion of taxes paid by the middle class and decreased the proportion paid by the rich (top 1%). 93% of the time, Democrats voted against these measures and 96% of the time Republicans voted for them. On the 318 votes on bills with the opposite effect, Democrats were for them 95% of the time and Republicans against them 98% of the time.

From these examples, it should be obvious how a completely factual, honest accounting can be manipulated to make Democrats seem good. Imagine that question 15 were, instead, simply “Middle-class people pay too much in taxes” and if the statistics were then about votes to raise taxes (regardless of income) – it would clearly favor the Republicans.

Sometimes the candidates and their surrogates get into trouble with fact-checking organizations for making inaccurate or grandiose claims. I think this is stupid on their part. If they choose their phrasing and statistics well, it is possible to be completely accurate with uncontroversial statistics AND say something that will sound good for you and bad for the other guy.

Up to this point, this must all sound quite cynical – a primer on how to manipulate public opinion without outright lies. But I firmly believe that the Democrats should use this technique to inform voters about what the real choices are. Recent surveys show that most Republicans (not just most voters) support most of the provisions in Obamacare, as long as the surveys don’t mention that they are provisions in Obamacare. Likewise, most voters believe that the rich do not pay enough in taxes, that corporations have too much power, and so forth. It’s just difficult to sort out what each party is for and what the effects of a vote are amid all the claims and counter-claims.

In my mind then, a Democrat-biased survey such as I propose here would actually be a way to educate (some of) the public about what is at stake and to get them motivated to vote Democratic. Of course, the Republican counter-argument would be that, if we asked questions like, “Is the government too involved in people’s lives?” and “Do you think too much regulation is burdensome and hurts the economy?” we might find that most voters are actually Republican at heart. My counter to that would be that programs like Social Security and Medicare are beloved even by most Republicans who claim to hate government in their lives (“Keep your government hands off my Medicare!”) and that, sometimes honestly and sometimes dishonestly, the Republican party has systematically convinced voters to vote against their own interest.