[ad_1]
Labor Day is over, which in an election year means that Americans are about to sink into the polls, leading the mood of voters who will decide which candidate to support in 435 congressional districts and dozens of elections to the US Senate and Senate on November 6th.
The mid-term elections also mark the first major test of the accuracy of pre-election polls since 2016, when polls were canceled to show that Donald Trump was lagging in the last days and that some forecasters were predicting a victory. more than 90%. .
According to a 2017 report by the American Association for Public Opinion Research, national surveys have been quite accurate in assessing Clinton's advantage at 2.1 percentage points in the national popular vote.
[The 2016 national polls are looking less wrong after final election tallies]
The recent unexpected results of the Democratic primaries in New York, Florida and Massachusetts have revived questions about the accuracy of polls, although they also highlight the truth about non-residential polls: Few quality polls are done in races not very competitive. election observers with a blind spot.
What lessons can election observers draw from the 2016 elections and past mid-term elections to be wise consumers of this year's elections? Elections and polls contacted by the Washington Post this week have highlighted several unique challenges facing surveys this fall and proposed strategies to make the most of this year's pre-election surveys – as well as the types of results to be tested. pull salt (or a whole salt shaker).
1. The most direct poll for Congress can also be one of the least accurate
The main competition in November is to determine which party will win the majority of the 435 congressional districts to control the House, and that 60 to 70 districts should be competitive. While national polls provide highly predictive indicators such as presidential job approval and generic voting – that voters would like a Republican or an unnamed Democrat to prevail – these are measures indirect by nature. Surveys of voting preferences in congressional districts are comparatively rare in comparison, making the situation of these competitions more blurred than during the presidential years.
In addition, the FiveThirtyEight data journalism website found that the polling error in congressional district surveys was 6.2 percentage points on the margin, higher than 5.4 points for election surveys at the state level and at four points for national surveys.
"Pre-election polls are like many other consumer products – you get what you pay for," wrote Michael Traugott, a professor of political science and communication at the University of Michigan. "Most of the published polls are sponsored by news agencies, and local newspapers and TV stations are under severe economic pressure."
Just keep in mind that national and local polls have strengths and weaknesses; they expect larger errors in local polls, while national polls are less direct indicators of what is happening in the country's competitive districts.
2. Read beyond who is conducting the last survey
Survey readers tend to cling to the numbers, unaware that they are smoother than they might appear. While accuracy is an unrealistic expectation, experts have encouraged more focus on what polls about the general mood of voters say.
"Read the mid-term polls to find out the state of mind and public concerns, but do not read them thinking you can guess who will win if you consider a tight election," wrote Courtney Kennedy, Director of Research, Pew Research Center. . "Polls work, but they are not designed to provide the kind of precision that can call a 50-50 election."
Republican investigator Whit Ayres recommended focusing on President Trump's approval rate rather than the "generic" polls often cited by congressional support.
"The approval of the presidential election, which is more stable than the generic ballot, is a better indicator in a mid-term election that will be a referendum on the president," Ayres wrote in an e-mail posted Thursday. "With one exception, when the approval of the president's post was 45% or less," writes Ayres, "his party lost enough seats to control the House in 2018". of the house this year.
Most national polls of registered voters show that Trump is below this limit, about 41% nationally. In 2014, President Barack Obama's low approval ratings also weighed on Democratic candidates in the US Senate, with no candidate exceeding Obama's approval rating by more than 9 points in exit polls.
[Exit Polls: Democrats could not escape Obama’s unpopularity]
3. Determining "likely voters" may be more difficult for polls this year
Before Labor Day, most surveys focus on the results among registered voters, but as election day draws closer, survey results will focus on a subgroup of voters deemed most likely to vote. Andrew Baumann, of the firm of Democratic firm Global Strategy Group, believes that identifying potential voters in mid-term elections can be more difficult when less than half of eligible voters have cast ballots.
"Usually, pollsters have looked at previous similar elections and assume that participation rates will be similar," said Baumann, but he noted that Democrat enthusiasm demonstrated during the recent governors' runs in Virginia and New Jersey clear advantage of participation. It is essential to find a way to effectively balance voter turnout and the current enthusiasm for voting, Baumann said.
[Why the ‘likely voter’ is the holy grail of polling]
Republican pollster Kristin Soltis Anderson of Echelon Insights offered similar advice. "If an investigator relies primarily on the history of past votes to judge who is a" likely voter ", he must ensure that he includes a means of" likely new voters "- those who are unusually excited or the midterms – to clear a path in the investigation.
One of the ways in which election observers can monitor the impact of voter turnout is to measure the excitement of voting. While polls consistently show Democrats are certain to vote, Republicans are unlikely to have the same voter turnout as in 2014 and 2010.
4. Polls are "snapshots" and the margin of error is different in pre-election polls
Most readers are aware of the margin of sampling error commonly reported in surveys, but it is often misinterpreted in pre-election surveys.
As Baumann explained, "if a poll has a margin of error of plus or minus four points and the horse race is 48% for the Democrat and 43% for the Republican", the margin of error applies to the estimation of the support of each candidate. "This means that there is a 95% chance that the Democrat will have between 52 and 44% and that 95% of Republicans have between 47 and 39%."
In practical terms, this means that the candidate's advantage must be between 1.5 and 2 times the margin of error for one to be 95% certain that a candidate actually has an advance. This does not mean that a candidate does not really have a lead if the result is within the sampling error range, but the survey is not precise enough to detect it with a level of accuracy. high confidence.
In simple terms: if a poll indicates that a candidate "leads" to a low margin, it is unlikely that this lead would be statistically significant.
Kennedy of Pew Research also cautioned against over reading the reported election forecasts with decimal point accuracy. This "implies a false sense of precision as to the extent to which these forecasts are likely to correspond to the actual result," he said. Baumann, of the Democratic firm Global Strategy Group, warned polls against publishing results with decimals for the same reason.
A third warning from the experts cited is that polls conducted weeks before polling day may not reflect the late fluctuations among undecided voters, which could lead to seem less accurate when the results are published. in a series of battlefield states.
[How America decided, at the last moment, to elect Donald Trump]
"The first thing for observers is that polls are a snapshot of a race, not a direct prediction of polling day results," Doug Schwartz, director of Quinnipiac University Poll, said Thursday. "That's why pollsters are different from forecasters like 538, who are actually looking to predict the outcome of the election based on current and historical data."
Emily Guskin contributed to this report.
[ad_2]
Source link
