The hidden axis: the left-right spectrum has a non-ideology problem

This article presents more results from our November Strength In Numbers/Verasight poll. You can read the first article about this poll here:

Introduction to this essay

Dissatisfaction with American politics and the two major parties is at a modern high. In our new Strength In Numbers/Verasight poll, 54% of voters said the Democratic Party doesn’t get people like them — and 55% said the same about the Republicans. Nearly a quarter said both parties are out of touch, and only a third of voters told us they are happy with their current choice of political parties. Today, most voters say that a major third party is necessary to get the country back on track.

While general anti-party sentiment is growing, Democratic strategists and center-left pundits are engaged in a fierce debate over whether to move the party’s “brand” and policy positions substantially to the right as a way to appeal to more moderate voters. Republicans will soon face similar soul-searching efforts, addressing the question of what the party looks like after Donald Trump is gone.

To address the questions from all of these groups, we need data. It’s obvious that Americans broadly want something new from their political leaders, but “something new” is not a policy platform or ideology. In order to start a successful third party — or to meaningfully change the brand of one of the major ones — we have to know what, exactly, voters actually want from their leaders. We need clean, clear, hard data.

Many analysts try to approximate an answer to these question by first looking the policies that are popular with voters, or by forcing voters to pick one of several pre-determined party platforms. Then, they use that data to infer the policy positions a party should take, and the messaging strategy it should use, in order to win the next election. They use their data, in other words, as a way to back in to an analysis of the thing they actually care about — as proxy for answering the question “What should my party campaign on to win the next election?”

These approaches can be useful, and are often cool from a data (esp. visualization) perspective. However, they suffer from two problems. First, the use of proxy variables (asking voters “do you favor this?” to answer the question “would you vote for a party that campaigns on this?”) introduces many ...