A team of researchers at Harvard University have found wide-spread anti-Latino bias (PDF) among election local election officials nation-wide. Specifically, the researchers found that local election officials were less likely to respond to voting-related questions from Latinos, and the quality of responses to Latinos was significantly lower than for Whites.
(n/t Political Wire)
What did they do?
The research team sent emails to "every local official or election commission responsible for overseeing elections for each county or municipality at which elections are administered in 48 states". The authors did not contact officials in Maine and Alaska for "data constraints". They did not elaborate. They also excluded responses from Minnesota and Virginia because some officials received multiple emails and others discussed the emails with other officials.
This left the researchers with 6825 emails sent to officials in 48 states. Each official received one of two emails. Half of the officials received an email asking about voter ID laws:
I've been hearing a lot about voter ID laws on the news. What do I need to do to vote?
The other half received an email asking about a different voting issue:
I've been wondering about this. Do you have to vote in the primary election to
be allowed to vote in the general elections?
In addition, the researchers varied the names on the email accounts. Half of the election officials received an email with a Latino-sounding name (either Jose Martinez or Luis Rodriguez) while the other half received an email with a White-sounding name (Jake Mueller or Greg Walsh). They essentially had four experimental conditions:
1. Voter ID Email/White author
2. Voter ID Email/Latino author
3. Generic Voting Email/White author
4. Generic Voting Email/Latino author
Each election official was randomly assigned into one of these four conditions.
What did they find?
Response Rate. The team received responses from 4557 local election officials and they found the response rate was different for Latino and White names. Election officials were significantly less like to respond to emails from Latino names than White names.
Responses to Latino names are three-and-a-half to four percentage points less likely than to non-Latino white names.
The authors predicted that this anti-Latino bias would be stronger for the Voter ID Email than the Generic Voting Email, but it wasn't. The authors also expected that anti-Latino bias would be greater in states with strict voter ID laws. Again, this hypothesis was NOT supported. Anti-Latino bias (in the response rate) was no greater in the 11 states with strict voter ID laws than it was in the states without such laws.
Quality of Responses. The research team used blind coders to rate the quality of the responses to the Voter ID emails on three dimensions--absolute accuracy, informativeness, and friendliness. "Blind" means that the coders did not know the name on the email account. I'm happy to discuss that and other steps they took with coding in the comments.
Anti-Latino bias emerged on two of the three dimensions. Latino emailers were less likely to receive accurate and informative emails than White emailers.
A key reason Latino emailers received fewer absolutely accurate responses appears to be that local officials asked Latino emailers for more information, rather than answering their question directly.
Alternative explanations addressed. The researchers addressed some alternative explanations--something other than anti-Latino bias. One possibility they addressed is that officials were unlikely to respond to emails if they didn't think the emailer was actually a constituent. Maybe they lived in a community with a small Latino population? The researchers were able to rule this out.
While Figure 4 shows that bias in response rates increases as the population becomes more non-Latino, there is still statistically signicant bias against Latinos in places where as much as 10% of the population is Latino and officials plausibly interact regularly with Latino constituents.
They also looked at whether this was a problem in small communities, where officials know (or think they know) everyone and ignore emails from people they don't know. But they were able to rule out that possibility as well. Moreover, they note that the Latino names used in the study are actually statistically more common in the US than the White names.
Local officials are significantly less responsive to questions from Latino than White voters. As the authors note, the cost of replying to an email with even a one word answer is "exceptionally low." That such significant bias was found in email communication suggests that such anti-Latino bias exists among government officials who provide other essential services to the community. I suspect that these findings are unsurprising to people of color, but this is a big wake-up call for the rest of us. We have a big problem, a big national problem.