January/February 2003

VOICES
Multilingual Polling
ITS TIME HAS COME

BY SERGIO BENDIXEN

Many pollsters don’t like to hear this, but to get an accurate read of opinion in this country today you must interview in multiple languages.

There are three basic perils in conducting English-only polling among linguistic minority groups:

•Respondents with a limited grasp of English cannot give answers that fully reflect their opinions. In many cases, to avoid the embarrassment of saying “I don’t understand the question,” respondents answer, “I don’t know” or “I’m undecided,” when in fact neither is true.

•With sensitive topics — such as immigration issues, and controversial political and foreign-policy questions — Latino and Asian immigrants sometimes give one answer if asked in English, and a different but more honest answer if asked in their native tongue.

•Worst of all, the respondent may simply refuse to answer any questions, therefore disappearing from the poll entirely. This can lead to underrepresentation of that minority group, and thus a misrepresentation of the group’s opinions.

Consider the opposite outcomes of pre-election polls among Latino voters conducted by the Los Angeles Times and La Opinion, the Spanish-language daily in Los Angeles, during the 1994 political campaign. One of the key issues was Proposition 187, supported by Governor Pete Wilson, which would have denied access to public education and social services to undocumented immigrants. The September 14, 1994, Los Angeles Times poll showed Proposition 187 winning among Latino voters by 52 percent to 42 percent. Another Times poll in October showed a dead heat. But the La Opinion pre-election poll — conducted in early October — indicated a different reality: only 15 percent of Latino voters supported the measure and 69 percent were opposed. (After weeks of such disparity, a Times poll the week before the vote showed 65 percent of Latino voters opposed and 22 percent in favor.) The Times’s exit poll that year found that 23 percent of Latino voters supported Proposition 187 and 77 percent opposed it.

The major methodological difference between the two polls was that La Opinion interviewed more than 50 percent of its Latino voters in Spanish, while the Times, according to Susan Pinkus, who directs the Times poll, interviewed less than 10 percent in Spanish. (Proposition 187 was adopted with 59 percent of the vote, but has since been ruled largely unconstitutional.)

This pattern was repeated in the 1998 election with Proposition 227, which sought to end bilingual education in California. A Los Angeles Times poll in May 1998 showed that 62 percent of Latinos supported the measure, while a La Opinion poll at the same time found only 30 percent Latino support. Exit polls showed that about 37 percent of Latino voters supported 227. (The measure was adopted.)

In 2001 and 2002, my firm conducted two polls in California — not in one or two languages but in twelve languages. The results provide solid evidence that interviewing ethnic and linguistic minorities in their own language yields a depth and richness of opinion that is missing from English-only polls.

One study, involving 1,000 respondents, measured the effects of the September 11 terrorist attacks on California’s ethnic and linguistic minorities. The languages in which we interviewed were Cantonese, Mandarin, Vietnamese, Tagalog, Urdu, Farsi, Dari, Arabic, Korean, Hindi, Spanish, and English. It found that the attacks had significant negative psychological, social, and financial impacts on immigrant groups. Many reported having lost their jobs or making less money, getting depressed more often, feeling insecure about their future, and experiencing heightened discrimination.

Post-9/11 English-language polls didn’t reflect such deep angst among the general population. For example, an August 2002 Knight Ridder poll found that only 9 percent still agreed with the statement “I get depressed more often now than I did before September 11th.” In contrast, 56 percent of Middle Easterners, 50 percent of Hispanics, and 45 percent of Asians agreed with the same statement in our survey. For most Anglo-Americans, financial life returned to normal shortly after the tragedy. The Knight Ridder poll found that only 19 percent of all Americans’ personal finances had been hurt “a lot” by 9/11. Conversely, 37 percent of Hispanics and 36 percent of Asians reported substantial drops in income after 9/11 in our poll.

Why did we conduct these multilingual polls in California? Because according to the 2000 Census, California is the first “majority-minority’’ mainland state. That is, almost 17 million of California’s 33.9 million residents are minorities. But California is only a precursor of sweeping demographic change that is redefining America. According to the census, the Hispanic population of North Carolina increased by 394 percent between 1990 and 2000, and Georgia’s by 300 percent.

Reaching those disparate minority groups, divining their views, and making them feel part of the American tapestry presents profound challenges — and, too often unseen, opportunities — for traditional English-language media and poll-takers alike.

Is it more expensive, more complex, and more time-consuming to interview in multiple languages? Yes. In my experience, multilingual polls cost at least 30 percent more than single-language polls because translators must be hired and because bilingual interviewers are more expensive. The project coordinator must also spend additional time making sure the questionnaires in different languages are compatible.

But Sandy Close, director of New California Media, an organization founded in 1996 that includes more than 400 ethnic media outlets, puts it this way: “I live in a state where 40 percent of the people don’t speak English at home. We journalists are missing the boat by assuming that we know what ‘public opinion’ is. The ancient Greeks referred to people who didn’t have standing in the public forum, the polis, as ‘idiots.’ Here, thousands of years later, we’re relegating to ‘idiot’ status those who aren’t part of the polis” through no fault of their own.

Multilingual polling is an indispensable tool to help minorities enter the polis. “It’s not the ballot box that gives people a sense of belonging,” Close argues. “It’s the sense of having a voice, of this hunger to be visible in the media culture.”

The pollster Rob Schroth, president of Washington, D.C.-based Schroth & Associates, says he has spent ten years “trying to explain to newspaper reporters, other pollsters, and clients that English-only polling is a major methodological failure.”

There is evidence that multilingual polling is becoming a reality. It is important to note that the Los Angeles Times has led the way. The Times has substantially increased its capacity to interview in Spanish over the last decade, and according to Susan Pinkus, it has recently polled in Mandarin, Cantonese, Korean, Tagalog, and Vietnamese as well. The Miami Herald has polled in Spanish for years, and — with the influx of Haitians into South Florida — it now also polls in Creole.

Clearly, the ethnic tide has turned in America. Pollsters and the English-language media alike will simply be emulating King Canute’s futility if they try to hold it back. Instead of drowning in this tide, they should surf it by doing more stories that reflect these immigrant communities, more polls that seek opinions in the respondents’ own language. They can, if they will.

Enjoy this piece? Consider a CJR trial subscription.

Sergio Bendixen is president of Bendixen & Associates, a Miami management and communications consulting group specializing in the Latino population and market in the U.S. and Latin America. He has been an on-air political analyst and commentator for the Spanish International Network, Univision, CNN en Español, and Telemundo.