Page 9 - Data Transparency White Paper_FINAL_Neat

Addressing Biases in Multicultural & Inclusive Identity Data
9

A total of 10 providers were evaluated in the set of validation studies.

Key Finding: Low Coverage and Inaccurate Multicultural Data
Nielsen revealed two bias issues.
1. Underrepresentation: Identity data is missing for a large number of homes Nielsen knows to be
multicultural. This means the providers’ classification of ethnicity/race and language preference
did not match many of those diverse households. Those households are simply “not seen” and
underrepresented in the data. In the table below Coverage is reported as the average percent
of multicultural Nielsen homes found in the third-party data; the complement of the percent
underrepresented.

2. Misclassification: The ethnicity/race assigned to the household record is incorrect. Accuracy measures
the correct classification of ethnicity and race. The households or records are included in providers’
databases, but they are misclassified.

Nielsen Validation Studies Averaged and Aggregated Results:
Coverage of Multicultural Consumers in Third-Party Datasets

Nielsen*
Segment Coverage 100%
(Average across 90%
studies and providers)
80%
White Non-Hispanic 89% 70%
60%
Hispanic 73%
50%
• Spanish-Dominant 81% 40%
30%
• Bilingual and English-Dominant 23%
20%
African-American 67% 10%

BILINGUAL
AFRICAN
ASIAN
SPANISH
WHITE
Asian-American 44% 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.