Geo Products Survey

• Geo Product commissioned a third party research entity to survey a random sample of individuals coded both by Geo Product and by E-tech [sic] in order to validate the accuracy of each company’s appended codes.


• 85,000 individuals were sent the survey, of whom over 664 completed responses were received.


• The survey contained a total of 44 distinct questions and approximately 5-10 minutes were required for completion.


• The survey questionnaire did not reveal any company information and was offered in the language of choice of the individual respondent.


• The survey was administered online using a reputable data collection system and techniques.


• The results were tabulated and illustrate several prominent advantages in favor of Geo Product.


• Key findings are summarized in this document.


• Details can be made available on a case-by-case basis on request.

• "Over" 664 (how much so?) responses of 85,000 is pretty suspicious. That is only .78%, even when they didn't have to buy anything.


• How many of the 85,000 were unknown ethnicity before the survey? Were the 85,000 already predicted by Geo Product (e.g. the Watts neighborhood predicted to be African-American, the Cuban neighborhood in Miami predicted to be Cuban)? Were any of the respondents from non-ethnic neighborhoods, where Geo Product can't find them?


• How clean was the test file? Did it have a full First and Last Name for every record? Did it have a Zip+4 on every record? These are all necessary for the E-Tech software to assign an accurate ethnicity.


• Did they use the current version of E-Tech for this test?


• Was the data provided seeded with subscription data or other non-predictive means and only used when Geo Product got it right?

• Geo Products blurs terminology. Ethnic Technology uses “Hispanic”; Geo Products uses “Latin American/Hispanic”. If a group of Brazilians or Guyanese responded “yes” to Geo Products questionnaire for “Latin American/Hispanic”, Geo Products will claim that Ethnic Technologies misidentified these individuals when in fact, Ethnic Technologies identified them correctly: Neither Brazilians nor Guyanese are typically Hispanic. The terminology creates non valid comparisons in Geo Products attempts to validate results.


• Geo Products uses individuals whose origins are African (ethnicities such as Egyptian, Tunisian, or Moroccan) to blur results. While they are from the continent of Africa, they are not who most Americans consider to be African-American. Ethnic Technologies considers these ethnicities to be Middle Eastern. Although Ethnic Technologies can identify these individuals, because they are not included in the umbrella African American, to more accurately capture the view of ethnicity in the market place, Geo Products is not comparing its African results to the same ethnicities as those identified by Ethnic Technologies. Again, the terminology creates non valid comparisons in Geo Products attempts to validate results.


• Furthermore, there is no such thing as an "origin of Arabic". There is a language with that name and there are countries (like Egypt) that are said to be Arabic countries, most of whose people are Arabic speaking. Then, however, Geo Products would count these individuals twice, once as "from Arabic" and once as "from Africa".


• Caribbean is a location, not an ethnicity. At this time, Ethnic Technologies does not offer a “Caribbean” distinction, not as an ethnicity (since there are multiple ethnicities from the Caribbean, with disparate languages, which Ethnic Technologies identifies separately) nor as a Group Code (since these ethnicities are grouped differently). Geo Products may be finding individuals whose origin is the Caribbean, but they do not have Caribbean data from Ethnic Technologies against which they can compare their results.


• Ethnic Technologies identifies “ethnicity” rather than “origin”. In E-Tech, a person named Mei Lling is coded as Chinese, despite the fact that she may have moved to the United States from France or Ethiopia. It is more precise to identify an individual the way he or she sees him or herself, rather than by the location from whence he or she came. This again demonstrates the discrepancy in the terminology used in the “validation”.

• When measuring language and acculturation, E-tech’s [sic] scoring method is ambiguous and inconclusive.


• For example, acculturation is defined as Spanish dominant or English dominant whereas language is only part of the blend of factors that are commonly accepted as a indicative of acculturation.


• Geo Products method is much more robust and is consistent with how marketing professionals measure acculturation.


• Geo Products offers acculturation measures for both Asian and Hispanic individuals and households.


• Geo Products coding of language out-performs E-tech’s [sic] by a wide margin.


• For example, when measuring Hispanics who are bi-lingual English preferred, Geo Products correctly attributed that value as HL2 81% of respondents whereas E-tech’s [sic] accuracy rate on this measure was 20%.