Words Can Hurt: Examining the Association Between the Prevalence of Latino Threat Narrative and Anti-Hispanic Hate Crime

Background and purpose: Although hate crime (i.e. bias motivated crime) incidences have been decreasing overall in recent years, anti-Hispanic hate crime (AHHC) incidences have increased significantly. The link between AHHC and anti-immigrant rhetoric in the media has been suggested in previous studies. One manifestation of the rhetoric is Latino threat narrative (LTN). As defined by Chavez, LTN labels Latinos as a group unwilling or incapable of integrating into the majority society in the U.S., and as an invading force destroying the American status quo. When widely spread in news media, the threat narrative could justify bias and lead to violence against Latinos. In this paper, we hypothesized that LTN was positively associated with AHHC. To obtain the empirical evidence of the association, we used media sources to quantify the prevalence of LTN from 1997 to 2011, and analyzed Uniform Crime Reporting Program (UCR) data to identify trends in AHHC during the same time period. 

Methods: To quantify the prevalence of LTN, we first extracted the newspaper articles (n = 13,039) published between 1997 and 2011 by applying the search term “illegal w/1 immigrants” on the LexisNexis newspaper database. We then identified 37 terms (e.g. threat) that could detect LTN in the subject of illegal immigrants, and applied text mining techniques to quantify the prevalence of LTN. To detect the trend of AHHC, we measured the prevalence of AHHC by using UCR Data from 1997 to 2011 (n= 7650). Using both data sources, we conducted Mann-Kendall trend analysis to model the observed change of the prevalence of LTN and AHHR. We then examined the association between the two time series by using cross correlation analysis.  

Results: We found that there was a significant increase of LTN from 1997 to 2011 (τ = .48, p < .05) with a spike observed from 2003 to 2007 (τ = 1.00, p < .05). We also found a significant increase of AHHC from 1997 to 2011 (τ = .51, p < .05), with a spike from 2003 to 2007 (τ = 1.00, p < .05). By using cross correlation analysis, we discovered that the prevalence of LTN and AHHC evolved concurrently (r = .93, p< .05).

Conclusions and implications: This study highlights the possible connection between macro level processes, negative portrayals and bias in the media, and micro level processes, AHHC. While our study does not rule out the possible confounding factors, such as the increased migration flow, policy changes, or economic fluctuation, it does suggest a statistically positive association between the prevalence of LTN and AHHC. The relationship between the threat narrative spread in the media and individuals’ bias motivated behaviors suggests a fertile area for prevention efforts. Advocating for the use of person-centered language in media outlets may have important impacts on individuals’ attitudes and behaviors.