|
|||||||||
|
|||||||||
|
 |
Asian Pacific Americans, Polls and the 2004 ElectionIn the barrage of election polls, it behooves us to recall the old saying: There are lies, there are damn lies, and then there are statistics
The 2004 Presidential election is in full swing and it appears that the results will be just as tight and hotly-contested as in 2000. As George W. Bush and John Kerry pound away at convincing voters that he is the best person for the job, we get bombarded by surveys and polls about voter preferences almost on a daily basis. All of these polls claim to be scientifically valid and indeed, are conducted by well-established and well-respected organizations and institutions. So why do they consistently come up with inconsistent findings? For example, as of late September, many polls found that Bush has increased his lead over Kerry into double digits. However, some polls now find that the race is still a statistical dead heat with Bush having only a slight lead of a few percentage points. Still others even claim that Kerry has a very small lead. If each of these polls is scientifically valid, why can’t they all agree on who is leading and by how much? This issue is not just about statistical methodology. With so many political decisions and policies these days being based on (or at least strongly influenced by) public opinion, it’s an important issue that deserves serious attention -- it is more than just an “academic” question. This is especially true for a constituent group such as Asian Pacific Americans, who are notoriously difficult to poll and as a result, policies are often made without our direct input.
THE BASICS OF A PROPER SURVEYLet’s begin with a basic summary of how these polls are supposed to work. The first requirement of having a valid poll is having a large sample size. Small samples tend to have more chances for error or bias. Generally, this is not a problem with the polls we see these days. However, the second major component of a valid poll is more difficult to get completely correct. In order for a poll to be scientifically valid, its sample (the respondents who are asked for their opinions) must resemble the target population as closely as possible.
One famous example of the pitfalls of not having a valid sample comes from polls conducted by Literary Digest magazine in 1936 in which they predicted that Republican challenge Alf Landon would easily defeat President Franklin Roosevelt. As history shows us, Roosevelt crushed Landon with 61% of the vote. The problem was that the sample that Literary Digest used in their poll was drawn from telephone registries and automobile registrations -- samples that included a disproportionate number of wealthy respondents. Therefore, it’s absolutely critical to have a sample that is truly representative of the population you’re trying to measure. With that in mind, how representative and valid are the samples used in the current political polls? Each organization who conducts the sample will claim that its methodology for choosing a sample is scientifically valid but others aren’t quite as sure. Earlier this week, Michael Moore, of Fahrenheit 9/11 and Bowling for Columbine fame, reportedly sent out an email to his supporters. The email was basically a pep talk to counter the increasing skepticism many liberals have about Kerry’s chances of winning the election. As Moore bluntly notes, “The polls are wrong. They are all over the map like diarrhea.” He went on to point out that most polls only include “likely voters” who are normally categorized as those who have consistently voted in past elections. Moore argued that this in effect excludes first-time voters and particularly, young people (who other surveys show are more likely to support Kerry). Further, Moore notes that typical polls don’t include respondents who use their cell phones as their primary phone numbers (referencing a recent article in Newsday, cited below), another group where young people predominate. Finally, Moore points out an article written by a prominent pollster, John Zogby, in which Zogby argues that most polls apparently include a disproportionate number of Republicans [citation below], despite their claims to be nationally representative. If all of Moore’s points are valid, it would go a long way toward explaining why so many polls can have such different findings. It further alerts us to be cautious in interpreting poll results and other statistical data, despite claims that they are valid. As the old saying goes, “There are lies, there are damn lies, and then there are statistics.”
ASIAN AMERICAN POLLSHistorically, trying to poll or survey the APA population has been a struggle at best. These difficulties generally revolve around large proportion of the APA population who are immigrants -- not being fluent in English, residing in densely populated urban areas that make physically locating them tricky, respondents’ suspicions that such surveys are really attempts at government surveillance, etc. Even a large national operation such as the Census Bureau concedes that it may have undercounted the APA population by about two percent in 2000 [citation below].
Despite these reservations about the overall validity of polls, there are some examples of polls and surveys that seem to be done right. In particular, On September 14, New California Media released the results of their “National Poll of Asian Pacific Islanders on the 2004 Election” [citation below], conducted between August 19-29, 2004. This national survey included 1,004 registered voters and was designed to be representative of the approximately 2.9 million APIA “likely voters” in the U.S. Their main finding was that Kerry holds a 43 percent to 36 percent lead over Bush among their sample respondents, but with a large 20 percent still undecided. However, the real value in this particular survey was that it included several additional and detailed measures of the sample respondents. These included (in addition to their ethnicity) their age, educational attainment, residence in a “battleground state,” voting pattern in the 2000 election, opinions on the Iraq War and domestic issues, and experiences of discrimination, to name just a few. In other words, the NCM poll appears to have learned that in order to make a poll truly valid and useful, you need more than just a large sample -- you also need more detailed information on the respondents. For example, among other things, the NCM survey found that Vietnamese and Filipinos are the most likely to vote Republican while Chinese, Asian Indians, and Hmong are much more likely to vote Democratic. Further, the NCM survey notes that young APAs and those with at least a college degree tend to support Kerry while older APAs and those with less education tend to be Bush supporters. At the same time, I would have liked to see even more detail, such as cross-tabulating the respondents’ views on particular issues with which candidate they support, or by cross-tabulating respondents’ ages within each Asian ethnic group with his/her likely choice for President. Another interesting result would have been cross-tabulating those who have recently experienced discrimination with their voting preferences. Nonetheless, in many ways, the NCM survey represents an excellent example of how to conduct a properly representative survey, particularly on a group that has historically been difficult to sample.
WILL IT MATTER IN THE END?
In the end, the question comes down to, how important will the APA vote be in the 2004 election? In other words, do APAs represent a sufficiently powerful voting bloc and constituency group in the same manner as Blacks or Jews? The jury still seems to be out on this question. On the one hand, organizations such as the 80-20 Initiative have worked very hard to organize the APA vote toward one candidate in order to maximize the impact of our collective voices. On the other hand, we cannot escape the fact that a tremendous amount of diversity still exists within the APA population -- ethnically, economically, and politically. As reinforced in the NCM survey, while the APA population (or at least those who are likely to vote in November) tends to lean Democratic, the margin of support is by no means overwhelming. Nonetheless, as the 2000 elections showed us, every vote counts (assuming of course that it is actually counted). With that in mind, APAs have an unprecedented opportunity to influence the outcome of the 2004 election. Flexing our collective muscle begins with lifting a finger to vote on November 2.
Also of Interest at AAV
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
