Gallup Poll Asks Voters The Wrong Question About Divided Government (again) and Arrives at a Wrong Conclusion About Voter Attitudes Toward Divided Government (again)

Graphic Credit: Gallup

This is becoming an election year tradition at The Dividist Papers. Every year Gallup publishes this poll and every election cycle the Dividist explains why there is far less than meets the eye in any conclusion based on this poll. The latest iteration is the absurd headline/conclusion in the 2020 edition of this Gallup Poll - "New High Favors One-Party Control of U.S. Federal Government":

"STORY HIGHLIGHTS

41% of Americans favor unified control of federal government

23% want divided control

52% of Republicans, 43% of Democrats favor one-party control"

"WASHINGTON, D.C. -- A new high of 41% of U.S. adults say it is better to have a president and Congress from the same political party. Twenty-three percent would rather have one party control the presidency and the other control Congress, while 32% say it makes no difference to them."

Here is the problem. Since there is no realistic possibility of the GOP establishing unified one-party control of the government this cycle, the only realistic choice in 2020 is between divided government and unified Democratic one-party control of the government.

The Dividist is sympathetic to their plight. Gallup's intent is to use the same poll question year after year in order to divine trends over many election cycles. To do that, they obviously must ask the identical question every time. The problem is that the question they've asked every year since 2002 is deeply flawed as it has absolutely nothing to do with the actual choice facing American voters when they go to the polls.

As explained in the last election cycle (paraphrased):

Quantifying "A decisive quantum of voters in the middle.."

This happens periodically. The Dividist has been beating the divided government horse on this blog for a dozen years. We think we know all the arguments for and against. Then someone comes along and with a pithy turn of phrase crystallizes in a few words what the Dividist has spilled gallons of ink (megabits of pixels?) trying to explain.

Jay Cost challenges the PoliSci "alignment" theory of American politics in his National Review article"In Praise of Divided Government":

"I would posit another framework to understand politics over the past 40 years, one in which the two parties are basically evenly matched, strong ideologues dominate the bases of both sides, and a decisive quantum of voters in the middle is up for grabs. This process has yielded a general pattern that seems to repeat: One party surges to control the government, but this is short-lived; the opposition quickly gains a foothold; and divided government persists until the opposition finally takes total control, repeating the cycle."

Love that phrase "... a decisive quantum of voters in the middle". Succinct and to the point. This is how the Dividist tried to explain the same concept in one of his first posts in 2006 [Edited to provide some semblance of clarity]:

"Think of it this way. An election is a scale. Pile the large mass of partisan Democrats on one side and the large mass of partisan Republicans on the other, roughly balancing the collective polarized "Partisan Dead Weight" (PDW) that can be relied on to always gets on one side or the other. Then there are some smaller, more mobile weights, that call themselves Independents... an "invisible hand" that votes for divided government ...  [INSERT MANY HUNDREDS OF WORDS] ...  What if this "invisible hand' that prefers divided government becomes visible? What if the "collective unconscious" that prefers divided government, starts making that decision consciously? It could change elections in the same way 3rd parties do, by siphoning partisan support but without the spoiler effect. The beauty of this idea is that this party needs no candidates, no leaders, no platform, no conventions, really none of the trappings of a political party. Dividists are voting by objective, not by platform and not out of party loyalty."

The Dividist got to the point eventually, and subsequently managed to be a bit less wordy in his "About The Dividist" explainer:

"On this blog we advocate a specific voting heuristic that can be implemented by a relatively small percentage of the electorate, perhaps as little as 5%. The target readership are voters who are willing to cast their ballot based on a rational evidence-based voting strategy that will result in better federal government. 

The strategy requires that the voter be capable of casting their vote without consideration of party loyalty or political ideology. The voter must even be willing and able to vote for candidates they dislike, based on accomplishing a greater goal of more fiscal responsibility, stronger oversight, less spending, more deliberately considered, carefully crafted legislation and better overall governance. In short, the voters must be willing to vote for divided government. This blog is for those potential voters."

To be fair, the Dividst was and is describing an aspirational goal, while Mr. Cost is positing a historical political hypothesis. The important question for both of us is whether a small sliver of the electorate who are truly independent (i.e. not self-described "Independents" who actually vote one party as reliably as partisans) are, or could be, a swing vote that determines elections when using divided government as an organizing principle.
Does this "decisive quantum of voters" exist? How big is it? Can we quantify the dividist "quantum" vote? And what does it mean for the 2018 midterms?

Stunning Gallup Poll Finds Voter Preference For Divided Government Slipping vs. Council Of Elrond

Gallup runs this poll every election cycle and every time the Dividist must explain why there is far less than meets the eye in any conclusion based on this poll. The headline conclusion in the 2016 edition of this Gallup Poll is "In U.S., Preference for Divided Government Lowest in 15 Years":

"STORY HIGHLIGHTS

• 20% of Americans say they want divided government

• Fewer Republicans want divided government now than four years ago

• Fewer Democrats want the same party to control Congress than four years ago

WASHINGTON, D.C. -- One in five Americans believe it is best for the president to be from one political party and for Congress to be controlled by another, the lowest level of public support for divided government in Gallup's 15-year trend. The remainder are evenly divided between those who favor one party controlling both the presidency and Congress (36%) and those saying it makes no difference how political power is allocated (36%)."

I am sympathetic to their plight. Gallup's clear intent is to use the same poll question year after year in order to divine trends over many election cycles. To do that, they must ask the identical question every time. The problem is that the question they've asked every year since 2002 is deeply flawed as it has absolutely nothing to do with the actual choice American voters make when they go to the polls.

Gallup asks the wrong question about divided government (yet again), and gets the wrong answer (yet again).

Gallup released the results of their annual governance poll with this headline:

In U.S., No Preference for Divided vs. One-Party Government

Thirty percent prefer one-party government, 28% divided government

by Jeffrey M. Jones

"Americans lack consensus on whether it is better to have one party holding the presidency and the majority in Congress, or better to have control of each branch of government split between the two major political parties. Currently, 30% say it is better to have a one-party government, 28% say a divided government is better, and the highest percentage, 37%, say it makes no difference."

Over the years, this Gallup poll consistently shows little or no preference for divided government, yet the result is at variance with the way Americans vote. As the article goes on to say:

"Divided government has been the norm in U.S. politics for most of the last 45 years, with one party controlling both houses of Congress and the presidency for only 12 of those years -- 1977-1980, 1993-1994, 2003-2006 and 2009-2010, as well as part of 2001. As such, Americans may simply be used to divided government and do not see it as better or worse than the alternative. But Americans have also seemingly rejected one-party government in the midterm elections that took place in 1994, 2006, and 2010, when a single party controlled Congress and the presidency and the public was dissatisfied with the way things were going in the country."

The problem is in the way Gallup asks the question. Presumably, to make the poll comparable year over year Gallup must ask the same question in the same way, so they are stuck with this wording. Unfortunately the way they ask the divided vs. unified question every year obscures the actual voting preferences of Americans and creates this paradoxical result. 

This a pet peeve for the Dividist. I've written about the problem with this survey before and therefore have no qualms about plagiarizing myself with the same explanation to the same problem question in the same problem poll.  To whit: