Republicans maintain an edge among voters ahead of next month’s congressional elections, but Democrats are gaining ground, according to a Washington Post-ABC News poll released on Tuesday.

Republicans are favored by 49 percent of likely voters compared to 43 percent for Democrats, said the poll, which was published by the Washington Post.

A Washington Post-ABC News poll last month showed Republicans with 53 percent support versus 40 percent for Democrats, who are fighting to keep their majority in Congress.

The findings were in line with recent polls showing Democrats making slight gains in terms of voter preferences for Congress and voters’ enthusiasm for the parties, although most surveys still show the Democrats trailing Republicans.

The Republicans’ advantage on the question of which party voters say they will support in the November 2 election has been cut in half, the poll showed.

Democrats also gained some ground on the question of which party people trust most to handle big issues such as the economy and health care, the poll said.

Forty-one percent of respondents said they trusted Democrats to do a better job of coping with the main problems the United States faces over the next few years, while 39 percent favored the Republicans.

Fifteen percent said they did not trust either party to do the job.

Nearly three-quarters of voters who identified themselves as Republicans called the 2010 elections more important to the country than previous elections in their lifetime, the newspaper reported.

Voter dissatisfaction with lawmakers in Washington remained high, with Congress getting a 23 percent approval rating in the poll. Congressional Democrats have a disapproval rating of 61 percent, while 67 percent disapprove of Republicans.

One third of voters think Tea Party candidates would change the culture of Washington in a positive way, and about 15 percent said they strongly supported the populist conservative movement, the newspaper said.

The poll of about 1,000 people was conducted between September 30 and October 3. The margin of sampling error was plus or minus 3.5 percentage points.

Copyright © 2010 Reuters Ltd.

### 3 Responses to "Democrats closing gap in midterm election poll"

Comments are closed.

If the sampling error is 3.5 points, and the difference is six points, that’s the same as no difference. This poll is essentially saying they’re neck and neck. When will reporters ever learn to interpret polls correctly?

That is not what sampling error means; this is the most frequent misinterpretation of statistical data drilled into our heads by a lazy press.

The margin of error can be defined for any desired confidence level, but usually a level of 90%, 95% or 99% is chosen (typically 95%). This level is the probability that a margin of error around the reported percentage would include the “true” percentage.

The most common confidence level is 95 percent, and statisticians use some other margin of error only if such margin is disclosed. If the confidence level is not stated you may assume it to be 95%.

The interpretation of the data presented is:

There is a 95 percent confidence that the true level of support is within 3.5 points either side of the level of 49 percent. In other words,there is only one chance in 20 that the Republicans would garner more than 52.5 percent or less than 45.5 percent of the vote. Similarly, there is only one chance in 20 that the Democrats would garner more than 46.5 percent of the vote or less than 39.5 percent.

And the closer you get to the edges of the two ranges the less likely the possibility. Each range is in effect its own bell-shaped curve.

The data above does NOT REFLECT a dead heat.

Your analysis is almost certainly incorrect, although there is an outside possibility it’s true. The sampling error term is not a “mandatory” correction factor, but a statistical limit on the probability of whether any poll has actually measured what it purports. In this case, given that the results are reported in percentages, the sampling error suggests that should the poll be conducted a minimum of 100 times, the value of each approval rating would fall within 3.5 percentage points of its current value. Thus, if 3.5 points was added to the Dem score and 3.5 was subtracted from the Repub score, Dems would take the lead by a point. Alternatively, if 3.5 was subtracted from the Dem score and 3.5 was added to the repub score, the Repub lead would blossom to 13. The range of possible scores would fall in between if the poll results are valid.

Of course the current value could be woefully incorrect in the first place, due to actual statistical error and/or sampling bias, which would be much more important regarding whether pols are “interpreted” correctly.

The following article discusses factors that affect the accuracy of polling:

http://www.pollingreport.com/sampling.htm