Regional Negative Bias in Word Embeddings Predicts Racial Animus--but only via Name Frequency
This work highlights a critical methodological flaw in using word embeddings to measure social biases, cautioning researchers about reliance on black-box models for studying human cognition and behavior.
The study found that anti-black bias estimates from word embeddings in social media correlate with racial animus measures, but these correlations are spurious and fully explained by the relative frequency of Black names in the data, leading to misleading results in areas with fewer Black residents.
The word embedding association test (WEAT) is an important method for measuring linguistic biases against social groups such as ethnic minorities in large text corpora. It does so by comparing the semantic relatedness of words prototypical of the groups (e.g., names unique to those groups) and attribute words (e.g., 'pleasant' and 'unpleasant' words). We show that anti-black WEAT estimates from geo-tagged social media data at the level of metropolitan statistical areas strongly correlate with several measures of racial animus--even when controlling for sociodemographic covariates. However, we also show that every one of these correlations is explained by a third variable: the frequency of Black names in the underlying corpora relative to White names. This occurs because word embeddings tend to group positive (negative) words and frequent (rare) words together in the estimated semantic space. As the frequency of Black names on social media is strongly correlated with Black Americans' prevalence in the population, this results in spurious anti-Black WEAT estimates wherever few Black Americans live. This suggests that research using the WEAT to measure bias should consider term frequency, and also demonstrates the potential consequences of using black-box models like word embeddings to study human cognition and behavior.