If reports of Republican panic on Capitol Hill are accurate, a big reason is likely the latest NBC/Wall Street Journal survey, which purports to show the public overwhelmingly disapproves of the GOP's role in the government shutdown.
And on the surface, the news does indeed look grim for the GOP, as explained by NBC's political editor, Mark Murray:
"The Republican Party has been badly damaged in the ongoing government shutdown and debt limit standoff, with a new NBC News/Wall Street Journal poll finding that a majority of Americans blame the GOP for the shutdown, and with the party's popularity declining to its lowest level.
"By a 22-point margin (53 percent to 31 percent), the public blames the Republican Party more for the shutdown than President Barack Obama – a wider margin of blame for the GOP than the party received during the poll during the last shutdown in 1995-96."
And worst of all, from the Republican perspective anyway, is this:
"And one year until next fall’s midterm elections, American voters prefer a Democratic-controlled Congress to a Republican-controlled one by eight percentage points (47 percent to 39 percent), up from the Democrats’ three-point advantage last month (46 percent to 43 percent)."
But there is much less here than at first appears. For one thing, the survey is based on interviews with "adults," not "likely voters," which significantly lessens its usefulness in predicting the shutdown's impact on the 2014 election.
The sample size is only 800 when 1,000 or more is preferable because it usually boosts the confidence level in the results and lowers the margin of error.
But what really detracts from its usefulness is the fact the survey, which was conducted jointly for NBC and the Wall Street Journal by Democratic pollster Peter Hart and GOP pollster Bill McInturff, uses a seriously flawed sample.
As BlueCollarPerspective's Charles Hawkins points out, the sample is heavily over-weighted in two ways: By the presence of government workers and households in which at least one person works for government at some level — 20 percent, which is two-and-a-half-times the rate for the general public — and there too many Democrats and too few Republicans being interviewed.
From a statistical perspective, public opinion surveys absolutely depend upon randomness in sample selection. The most accurate survey is produced when every potential respondent has exactly the same chance of being selected as every other potential respondent.
It's impossible to achieve perfect randomness, but the closer a sample gets to it, the more accurate it will be. Hart and McInturff are reputable pollsters, so it appears they simply drew what amounts to an outlier survey sample.
That makes the survey all but useless as support for strategic decisions by anybody involved in the shutdown debate. But it is perfectly consistent with the conventional wisdom of the traditional media, which is heavily skewed against the Republicans, so expect to hear it cited repeatedly in the next few days.