Page 10 - Data Transparency White Paper_FINAL_Neat
P. 10
Addressing Biases in Multicultural & Inclusive Identity Data
10
As seen in the table, White Non-Hispanic coverage is strong (89 percent), lower for Hispanic, Hispanic
Spanish-Dominant, African-American and Asian-American segments. Hispanic Bilingual and English-Dominant
coverage is the lowest (23 percent).
Clearly the complement of these numbers is the size of the “missing” multicultural data in providers’ datasets.
However, this is only half the problem in multicultural identity data. Accurate classification of the multicultural
records that are present in the providers’ datasets is the next concern.
Nielsen Validation Studies Averaged and Aggregated Results:
Accuracy of Multicultural Consumers in Third-Party Datasets
Nielsen*
Segment Accuracy 100%
(Average across 90%
studies and providers)
80%
White Non-Hispanic 89% 70%
60%
Hispanic 74%
50%
• Spanish-Dominant 36% 40%
30%
• Bilingual and English-Dominant 46%
20%
African-American 67% 10%
BILINGUAL
WHITE
SPANISH
ASIAN
AFRICAN
Asian-American 71% NON-HISPANIC HISPANIC DOMINANT AND ENGLISH AMERICAN AMERICAN
DOMINANT
*Source: Nielsen third-party data validation studies from 2011-18.
Nielsen datasets used to measure coverage and accuracy rates vary by study.
As seen in the table, Hispanic consumers were only accurately classified as Hispanic by third-party data
providers 74 percent of the time. We see lower accuracy rates for Asian-American consumers, followed by
African-American consumers. Hispanic Spanish-Dominant accuracy rate is the lowest (36 percent).
Clearly multicultural consumer data can be significantly inaccurate, leading to biases and ineffective and
inefficient marketing spend against the wrong target consumers.