Questions Of Validity In Hispanic Survey Research
"Proven" survey methods for general market studies may yield invalid research data when used on the Latino population. This article describes survey research methodologies and questionaire design that will more effectively obtain the Latino opinion.
By now everyone knows about the growth of the U.S.'s Hispanic population, and many companies now include Latinos in their research plans. The research industry has responded eagerly with an unprecedented number of companies offering Hispanic research services. This is a positive development.
When I started suggesting including Latinos in general market research in the mid 80's my clients thought I was crazy; and I probably was, because there were then very limited options for appropriately fielding a Latino survey. Things have changed! Or... have they?
Today almost every big player (and many smaller ones) claims to have the ability to conduct Hispanic market research. The problem is that research companies are surveying Latinos using the same "proven" process that have been established to be successful and appropriate for general market studies. There is ample evidence, however, that conducting research with Latinos using this "proven" approach yields invalid research data.
To understand how Latinos respond to surveys we need to appreciate the cultural differences between Hispanics and non-Hispanics. A lot can be said regarding the fact that the "Hispanic community" is really a U.S. marketing invention that was fueld by our industry's eagerness to classify individuals who did not fit well into our standard classification categories. After all, Hispanics come from as many as 20 different countries of origin and include individuals of every race.
However, there is no denying that most Latinos share a common language as well as certain cultural characteristics and values that unite them as a group. In fact, many Hispanics in this country have come to identify themselves with their "Hispanic community." These commonalities set Latinos apart as a market segment and differentiate them from non-Hispanics. We need to look closer at the Hispanic/non-Hispanic differences and why they affect data collection methods and data validity.
Many Hispanics are new immigrants who were born outside of the U.S. and are not as familiar with opinion polls and survey research. Opinion research is such a big part of American society that we can assume respondents know about surveys and polls; but in most of Latin America consumers are not as exposed to marketing research as we are in the U.S.
As a result, many Latinos approach survey questions as if they are an academic exam or a government form because this is their only frame of reference. In that mindset, the Latino respondent struggles to come up with the correct answers to the survey questions. Logically, giving the wrong answer always has negative ramifications when completing tests or government forms. In researching Latinos, especially unacculturated new immigrants, great care must be taken in explaining the research process.
Not being familiar with surveys also has other consequences. Again, in our industry's American mentality we take much for granted. For example, we assume that people are familiar with the concept of a number scale. When asked to rate something on a scale of 1 to 10, many new Latino immigrants will select either 1 or 10 because they do not know that they are allowed to use numbers in between.
Education also plays an important role in the respondent's ability to complete a complicated survey instrument, and U.S. Latinos as a group have a lower educational level than non-Hispanics. Another factor that creates confusion in Spanish language surveys is the language used in the Spanish translation. Very often researchers purposely instruct translators to translate a survey verbatim because to avoid interpretation errors it has to be exactly the same as the English version. The problem is that this always yields a very awkward and confusing question narrative that often results in more significant data errors.
This is an abstract, the full article is in pdf form here. This article was originally published in the July 2008 issue of MRA’s Alert! Magazine.