September 10, 2012
A new RAND Corporation poll has begun tracking American public opinion about the 2012 presidential election.
The poll is conducted each day among members of a set panel of Americans, with results updated each evening. The poll, which began in early July, tracks opinions about which candidate the public favors and which candidate is expected to win the November election.
The poll is conducted among members of the RAND American Life Panel, an Internet-based panel of 3,500 Americans organized for the purpose of conducting online surveys. Since the survey repeatedly polls the same people, researchers say the survey should lead to more stable results as compared to surveys that randomly sample new respondents each time.
"The changes in opinion we see are true changes in people's opinions and not the result of random fluctuations in who gets asked the questions," said Arie Kapteyn, the poll's leader and a senior economist at RAND.
Since the poll began on July 11, it has generally shown President Barack Obama and challenger Mitt Romney neck and neck. Over this period, the difference between both candidates never was large enough to be statistically significant. At the same time, respondents generally find it considerably more likely that Obama will win the election than Romney.
The poll results and an explanation of the methods used are available at this link: https://alpdata.rand.org/index.php?page=election2012 .
A unique feature of the RAND American Life Panel poll is that participants are asked how likely they are to vote for a particular candidate and how confident they are that one candidate will win the election. They express that likelihood as a percentage from 0 to 100 percent. Researchers say the approach acknowledges that respondents may not be completely sure who they will vote for or whether they will even vote.
In fact, the poll takes into account both the stated likelihood that a respondent will vote and the likelihood that he or she will vote for Obama, Romney or someone else. Respondents who are likely to vote for Romney report a higher likelihood that they will actually vote in the November election than Obama supporters. Currently Romney supporters express about a 3 percent higher chance that they will actually vote than the Obama supporters.
Thus the statistically indistinguishable share of the votes that Obama and Romney are expected to get is really the result of two different facts: on average more respondents say they expect to vote for Obama, but the Romney supporters think they are more likely to actually go to the polling station.
The questions asked by the poll are: What is the percent chance that you will vote in the Presidential election? What is the percent chance that you will vote for Obama, Romney, someone else? What is the percent chance that Obama, Romney, someone else will win?
The questions are asked daily of about 500 members of the RAND American Life Panel. The results are weighted to produce a nationally representative sample and the daily updates report averaged weekly results.
The margin of sampling error for the poll is generally about 2 percent with a 95 percent level of confidence, but it may vary somewhat over time and by the different outcomes presented at the website. Every graph on the website shows the exact level of confidence so that the reader can see immediately whether differences are statistically significant or not.
Created in January 2006, the RAND American Life Panel is an Internet panel of people aged 18 and over. Participants use either their own computer to log on to the Internet or they are provided a small laptop or a Web TV by RAND if they did not have their own Internet access.
While most Internet-based surveys use convenience samples, the RAND American Life Panel is based on probability samples of the population and hence includes respondents who would otherwise not participate in Internet surveys. Panelists are recruited from all 50 states and the District of Columbia and represent a wide cross section of the American public.
The American Life Panel has fielded more than 275 surveys in such diverse areas as financial decision making, the effect of political events on self-reported well-being, inflation expectations, retirement preferences, health decision making and Social Security knowledge. A total of 60 researchers have conducted surveys using the panel and more than 300 researchers worldwide have registered to use data from the panel's surveys.
The RAND American Life Panel and the Multimode Interviewing Capability (the software that powers the panel) are proprietary capabilities of the RAND Corporation. The tools were developed by RAND with its own funds and with the support of numerous clients and grantors who have commissioned social science and economics research and analysis at RAND. Use of the RAND American Life Panel and the Multimode Interviewing Capability is available to outside researchers for a fee. The use of the RAND American Life Panel and the Multimode Interviewing Capability for the continuous presidential election poll was initiated and financed by RAND to demonstrate the distinctive capabilities of the tools to a broader audience.