Count me in: the impact of ‘other’ on research respondents

Editor’s note: Ryan Sullivan is quality control manager at iModerate, a Denver research firm.

151902402Last year’s elections saw a number of historic firsts – the first Mormon to run for president on a major party ticket, the first openly LGBT and Buddhist members of the Senate and the first Hindu to serve in the U.S. Congress. Add these members to the mix of Latinos, American Indians, Muslims, Jews, atheists and other minority members already serving in Congress and you have a poignant representation of the increasing diversity of the United States. Far too often, however, market research fails to acknowledge this reality. In the quest to find insights that will let our clients reach the greatest populations, we often let the less visible minorities fall by the wayside.

When being asked to identify their ethnicity, Caucasians, African-Americans and Latinos almost invariably find their groups listed while smaller groups are forced to relegate themselves to an “other” category. From a statistical standpoint, this makes sense. More than 80 percent of the U.S. population identifies as one of those three dominant ethnic groups and you’re more likely to reach the base sizes necessary to recognize a statistically significant trend. However, consider how the term “other” can be interpreted by somebody whose ethnicity is central to their identity. They’re essentially being told that their ethnicity is so inconsequential that it doesn’t even merit being listed. They will only be viewed in the aggregate, lumped together with other individuals whose only similarity is being a member of a less-represented ethnicity. Not only is this frustrating for the respondent but it could potentially even engender their ill will and affect the quality or veracity of their responses.

Fortunately, this invisibility and frustration can be easily avoided by creating more inclusive answer choice lists. Consider expanding the available options to include Asian, Pacific Islander or Native Hawaiian, and American Indian. Similarly contemplate adding a multiracial option or allowing respondents to select all of the races with which they identify. Because a growing percentage of the population identifies with more than one race, only allowing them to identify one aspect of their racial identity forces them to misrepresent themselves. By being more inclusive, you allow respondents to represent themselves as accurately as possible and improve the quality of your data.

Questions asking for a respondent’s religion similarly emphasize the largest groups and exclude the smallest. These questions most often revolve around the Judeo-Christian tradition, offering respondents a list of different denominations such as, Other Christian, Jewish and Other. As with ethnicity, the United States is becoming increasingly religiously diverse and we should consider adding Muslim, Buddhist, Hindu, and atheist/agnostic to the commonly listed answer choices.

If you’re asking your respondents to answer a question about their relationship status, consider including “In a civil union” and “In a domestic partnership” as potential options. Because the laws governing same-sex marriage and these statuses vary from state to state, and even between different cities within a state, you run the risk of leaving this categorization open to personal interpretation. One person in a domestic partnership might label himself or herself as married due to the level of commitment in their relationship while another might mark himself or herself as single because gay marriage isn’t recognized where they live. Recent research estimates that anywhere between 3 percent and 10 percent of the U.S. population is LGBT and that means that your data is likely to include a few of these individuals. By including these options, you’re allowing people to more accurately identify themselves while leaving yourself the option to group these categories however you see fit.

The beauty of this more inclusive approach is that it doesn’t have any drawbacks. The time commitment is minimal – literally only adding a handful of minutes to programming and processing times. Depending on the sample size, greater inclusiveness could yield some interesting, unexpected and statistically significant results. If not, you still have the freedom to group the data for all of these individuals under a generic “other” umbrella on the back end. Accounting for diversity allows people to more accurately identify themselves, improves the quality of your data and validates minority groups that are so often overlooked.

This entry was posted in Consumer Research, Data Collection/Field Services, Demographics, Ethnic/Multicultural Research, Market Research Best Practices. Bookmark the permalink.

2 Responses to Count me in: the impact of ‘other’ on research respondents

  1. Chris Lewis says:

    Why pigeon-hole people by’choices’ at all rather than allow them to fill in the information on their own to specific questions? Closed-end questions are legalistic and strip people of their individuality and freedom of identification in lieu of freedom of choice which under such circumstances is quintessentially enslavement of personality if not person. How much better can one learn the consumer if not through freedom of interpretation? People and culture are not relegated to 7 continents but by the cultures and communities of billions of people.

  2. Cleo Parker says:

    I spent part of 2010 working for the Census, which reminds me of the many ethnic categories offered and NOT offered on the standard census questionnaire. Many Asian ethnicities are called out, while those from the Middle East and Latin America are not so lucky. There must be some interesting stories behind how those options were selected. We heard from a number of people whose ethnicity was much more common in our area than some of those listed, and they were not pleased.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>