letters

Language barrier

Thank you for the article on Hispanic-American consumers [PROMO July]. Your readers may not know, however, that many of these consumers' views and opinions are either never counted or misunderstood.

Pollsters may randomly select a sample, but that is useless if the interviewers cannot fully communicate with a significant cross section of that sample. What happens when the interviewer only speaks English and the person on the other line only speaks Spanish? In most cases the call is terminated, labeled LB (language barrier) and is put back in queue to be reassigned to a native speaker. The problem with this data gathering method is that callers will not answer the phone the second time around thereby further reducing the Latino-Hispanic sample size for the poll. Hence the weakness of referring to such polls as indicative of a “national representative sample.”

What if the poll is being conducted in Southern California (or Miami, Southern Texas, New York City, etc)? At least 30% of the sample may get flagged as a non-response due to language barrier. How valid can the results be if the survey misses a third of the pool?

[Researcher Peter Coy has said], “It's important for researchers to ask questions in a neutral way and reach a representative cross-section of the public.” Coy alludes to the scientific responsibility of designing an unbiased survey and counting, by representation, almost everybody. In the multicultural U.S., what is neutral? In order to speak with “the representative cross-section of the public,” national surveys and polls would need to be conducted by a completely dual-speaking/bi-lingual interviewing pool of people.

Having interviewing staff that is fully-fluent in reading, writing and speaking Spanish and English makes it possible to switch back and forth, instantly, between Spanish and English, at the first intercept or call. This decreases the chance of miscommunication that occurs when forcing respondents to speak English when that is not a dominant language.

You get a much higher response rate when the respondent can converse in the language of choice. Interviewers aren't the only ones who need to be bilingual. Supervisors, trainers and managers need to be bilingual. How else could Spanish interviews be monitored, as are English interviews?

For example, an English speaking supervisor would not be able to catch when the Cuban interviewer interpreted “mueble” to mean “furniture,” when the Mexican respondent used “mueble” meaning “vehicle.” (Similar situation to bad meaning good in ‘teen speech’). Standard quality control measures need to extend to Spanish and Spanglish interviews as well.

It makes sense for bilingual interviewing to be a standard practice for polling and surveying today's USA? Unfortunately, of the thousands of telephone call centers in the U.S., only a handful have this capability.

“Quien bien atiende bien aprende, si ademas de oir entiende.” (“Who pays attention, learns well, if besides hearing, understands.”)
Suzanne Irizarry de López
Eastern Research Services

A reader makes a plea for more conscientious consumer research in more complex ethnic markets