As the Presidential Election finally approaches, experts warn that there are good polls and bad polls which can lead to the wrong information about Latino and Asian American voters.
Getting it right is important in a year when registration and early voting is showing record participation among these two groups, said several pollsters and members of Latino and Asian American organizations in a conference call on Friday.
"One fatal flaw of pre-election surveys are too few Latinos in samples, leaving very large margin of errors, sometimes exceeding ten percentage points," said Gabriel Sanchez, a principal at the polling firm Latino Decisions. "Additionally, the wrong Latinos are often surveyed, meaning that they are not representative of the population they are supposed to project."
"Public polling and the exit polls repeatedly get it wrong when it comes to capturing the Asian American electorate, whether by including too few Asian Americans in the sample, failing to conduct interviews in their primary language, or failing to ask about the issues that really matter to Asian Americans," said Taeku Lee, a principal at Asian American Decisions.
Much of what we think we know about Latinos on election day comes from exit polls, which have been notoriously dismissive of minority voters. For instance, an exit poll conducted during the 2014 election in Texas claimed that Republicans were making significant progress with Latinos, but a close look at the data revealed that the data collected in the exit polling was insufficient in supporting the claims being made.
While the methods used to do research on the majority of the population are generally better, gathering data on minorities is more difficult for several reasons.
First, minorities make up a smaller proportion of the electorate and so they are more difficult and expensive to reach.
Latinos and Asian Americans are not always comfortable answering questions by pollsters in English. Reaching them in Spanish, Korean, Tagalog, or some other language becomes expensive for polling firms. Minorities generally live in concentrated geographic regions, making it more likely that any methods which do not specifically seek out minority respondents will miss out on them.
There is still a long way to go, said Gabriel Sanchez, and there is greater difficulty in educating the media and the public about what to expect.
"We should all ask these three questions before running with exit poll results," said Sanchez later in an e-mail with NBC Latino.
"First, what percent of the interviews were conducted in Spanish? Second, do the Latinos in the survey match what we know about Latinos from the US Census and other records? Third, we should look at where these Latinos are from. How many come from majority-Latino precincts?"
Similar questions apply when polling Asian American voters.
Victoria DeFrancesco Soto, an expert on Latino political behavior, recently outlined differences in important characteristics of Latinos in Arizona, North Carolina, and Florida. Hispanics are far from monolithic. There are not only variations regarding their country of origin, such as Cuba, Mexico or some other region, but local nuances regarding how long they have resided in the U.S., which can influence political views.
Latino Decisions will be releasing an election eve poll with a national sample of more than 5,500 Hispanics as well as state-level results in 12 states with large Hispanic populations, including key swing states.