 |
| Published on redlineproject.org September 20, 2012 |
Technology and changes in sampling are giving polls a more accurate reading on the upcoming elections, DePaul University Political Science Professor Molly Andolina said Wednesday.
Because polling institutions will try to predict the outcome of the 2012 presidential election, it is important that they get an accurate prediction of who will get out and vote, Andolina said.
"There's a real art in trying to get the mood of the electorate to make sure that you include enough or not too many likely voters," she said.
Andolina spoke at a series of weekly lectures by faculty on the 2012 presidential election. The events are hosted by DePaul's Department of Political Science.
Institutions use probability sampling to collect a fair sample of the voters scientifically. This means that everyone has an equal chance of being selected to take the poll but should not select the poll themselves, Andolina said.
"We want to be confident that our sample of the population is an accurate representation of the population at large," she said.
The most common way to do this is through random digit dialing (RDD) calls. This is also the most effective way to use probability sampling because 98 percent of the population has a phone.
"It use to be that most campaigns didn't include cell phones," she said. "Gallup surveyors could see the trajectory of where cell phones were going and stated to include them in their samples."
Research shows 25 percent of people in the U.S. only have a cell phone. That percentage is disproportionally young people, majority male, and majority are minorities. Additionally, 18 percent of the populations have both a landline and a cell phone, however primarily use the cell phone.
Beginning with the 2008 campaign, the Pew Research Institute started doing tests to compare the results of landline-only surveys with cell phone-only surveys. Pew found that including cell phone numbers is the RDD surveys gives a more accurate sample. This is now a common practice among pollsters.
Aside from using the probability sample to determine who will show up to the polls, institutions consider barriers placed by our political system that will prevent some voting on election day.
“The most important thing to know about our political system is that there are some systemic elements that prevent some people from voting,” Andolina said.
Voting is a two step process because you must first register and then vote explained Andolina. Research shows that young people and people move around more are less familiar with the process.
"There is disturbing information about how misinformed young voters are," she explained. "They need to be registered 30 days before the election." Institutions also factor last minute voters into the outcome predictions. This is typically known as allocating the undecided.
Based on the results from the 2008 presidential campaign 30 percent of voters were undecided before the conventions, while 20 percent were undecided following the convention. Ten percent of voters in the 2008 elections were last minute voters.
Voter participation has increased each presidential election. The 2008 presidential election had the highest percentage of voter turnout at 56.9 percent.
Andolina explained it is important to understand the past numbers in order for polling institutions can make an accurate prediction of the turnout for an upcoming election. According to Andolina, websites like pollster.com is good to reference because it shows trends over time.
Despite an institutions attempts to provide an accurate outcome of the election every poll will have a certain level of sampling error also known as margin of error.
Simply put, this is an estimate of likely difference between estimate generated by numbers and the true numbers in the population. The smaller the sample size the larger the margin of error.
"Sampling error in a really close race could really mean a switch of leads," Andolina added. "But the bottom line at the end of the day is who won?"
No comments:
Post a Comment