Clinton Up 10 Points, Or One?

By Jim Ellis

June 28, 2016 — Two new national media polls were released over the weekend, and even though they were conducted over the same sampling period their conclusions are quite different.

The Washington Post-ABC News poll (June 20-23; 1,001 adults — undisclosed number of registered voters, 650 certain voters) finds Hillary Clinton leading Donald Trump and Gary Johnson, 47-37-7 percent. But, the NBC/Wall Street Journal survey (June 19-23; 1,000 registered voters) finds only a one-point Clinton edge, 39-38-10 percent, over Trump and Johnson, respectively.

In both cases, Clinton’s lead is stronger without Johnson included. The fact that the Johnson-Weld Libertarian Party ticket will be on the ballot in all 50 states makes the third party inclusion more accurate. Without Johnson, WaPo/ABC finds a 51-39 percent Clinton spread; NBC/WSJ sees a 46-41 percent margin.

Green Party candidate Jill Stein was also added to the mutli-candidate question in both surveys. She scored three percent on the WaPo/ABC study, and six percent from NBC/WSJ. It is unlikely, however, that she will gain 50-state ballot standing. Therefore, her national poll position is largely irrelevant at this time.

The Post-ABC’s usual data firm, Abt-SRBl of New York City, also conducted this poll. A decided liberal skew has been evidenced, here, and well beyond the ballot test for at least two fundamental reasons:

First, the Obama job approval result greatly exceeds any other known public data. According to WaPo/ABC, the president’s ratio is 56:41 percent positive to negative. Though he has performed better recently across the board, no pollster has found such a favorable Obama split (as has WaPo/ABC) in, literally, years. Currently, in comparison, the prolific Gallup organization finds Obama’s job approval ratio to be 50:45 percent on their rolling trend from a respondent pool exceeding 1,500 snapshot answers.

Rasmussen Reports, also a daily presidential job approval pollster, with a three-day rolling sample group of 1,500 respondents (500 per day), finds a virtual dead-even favorable to unfavorable split, 50:49 percent.

Second, the partisan division skews Democratic. According to the self-described party identification question, 36 percent of the respondents classify themselves as Democrats, 24 percent as Republican, and 33 percent Independent. Looking at the Gallup partisan national track, sampled continually from 1988 through 2015, 29 percent identify as Democratic Party members, 26 percent as Republican, and a whopping 42 percent, Independent. This particular sampling universe is approximately seven points high on the Democratic scale, however, and two points low on the Republican side. Therefore, the net skew could conceivably be as much as nine percentage points.

This brings us to the NBC/Wall Street Journal poll, which appears to be much closer in constructing an accurate sample. The pollsters here screened only for registered voters, which presents a better understanding of the electorate as opposed to the populous.

Second, the management is bipartisan. Working for NBC and the WSJ on this particular project were Republican pollster Bill McInturff from Public Opinion Strategies, and Democratic surveyor Fred Yang, of the Garin Hart Yang Research Group. This suggests that the sample was better screened since the nuances of both party’s voter groups were considered in the founding equation that produced the relevant sample.

The disparity in these two media polls conducted in the same time frame illustrates how widely results can differ from simple differences in polling methodology. It also tells us how quickly the national political picture can change.

The key to determining political status is tracking many surveys over a long period in order to detect trends. The aggregate summation of the WaPo/ABC and NBC/WSJ surveys tell us that this campaign is far from ending, and that we can continue to expect wide swings all the way to Election Day.

Leave a Reply

Your email address will not be published. Required fields are marked *