It’s not easy to prove racism — this study does

What’s in a name? Ev­ery­thing, es­pe­cially if your moniker has an African-Amer­i­can ring to it, writes Justin Wolfers

Bangkok Post - - ROUNDUP -

Ateam of economists has un­cov­ered per­sua­sive ev­i­dence that lo­cal gov­ern­ment of­fi­cials through­out the United States are less re­spon­sive to AfricanAmer­i­cans than they are to whites. The re­searchers sent roughly 20,000 emails to lo­cal gov­ern­ment em­ploy­ees in nearly ev­ery county. The emails posed com­mon­place ques­tions, such as, “Could you please tell me what your open­ing hours are?”

The emails were iden­ti­cal ex­cept that half ap­peared to come from a DeShawn Jack­son or a Ty­rone Wash­ing­ton, names that have been shown to be as­so­ci­ated with African-Amer­i­cans. The other half used names that have been shown to be as­so­ci­ated with whites: Greg Walsh and Jake Mueller. The email sent to each lo­cal of­fice­holder was de­ter­mined by chance.

Most in­quiries yielded a timely and po­lite re­sponse. But emails with black­sound­ing names were 13% more likely to go unan­swered than those with white­sound­ing names. This dif­fer­ence, which ap­peared in all re­gions of the coun­try, was large enough that it was sta­tis­ti­cally un­likely to have been a mat­ter of mere chance.

These trou­bling re­sults were doc­u­mented in the pa­per, “Racial Dis­crim­i­na­tion in Lo­cal Pub­lic Ser­vices: A Field Ex­per­i­ment in the US”, by Cor­rado Gi­uli­etti of the Univer­sity of Southamp­ton in Bri­tain, Mirco Tonin of the Free Univer­sity of BozenBolzano in Italy and Michael Vlas­sopou­los, also of the Univer­sity of Southamp­ton. The study is to be pub­lished in the Jour­nal of the Euro­pean Eco­nomic As­so­ci­a­tion.

The find­ings ap­peared to be a strik­ing in­di­ca­tion of racial dis­crim­i­na­tion in mun­dane in­ter­ac­tions. The ten­dency to ig­nore emails sent by African-Amer­i­cans was pro­nounced in sher­iffs’ of­fices, but it was also ev­i­dent in school dis­tricts and li­braries.

In a clever twist, the au­thors an­a­lysed whether the replies were po­lite, count­ing re­sponses that in­cluded ei­ther the sender’s name or words like “hi”, “Mr”, “dear”, “good” (which cap­tures “good morn­ing,” “good af­ter­noon” and “have a good day”) or “thank” (which cap­tures both “thanks” and “thank you”). By this mea­sure, those with ap­par­ently African-Amer­i­can names re­ceived 8% fewer po­lite re­sponses than those with white names.

While many stud­ies have found dif­fer­ences in treat­ment for African-Amer­i­cans and whites in em­ploy­ment, hous­ing and the crim­i­nal jus­tice sys­tem, it hasn’t al­ways been clear whether these dif­fer­ences re­flect dis­crim­i­na­tion or other fac­tors.

The usual dif­fi­culty is that it’s im­pos­si­ble to find, say, job seek­ers who are ab­so­lutely iden­ti­cal in ev­ery re­spect ex­cept race. As a re­sult, it is dif­fi­cult to con­clude whether a white job seeker suc­ceeded — and a black one didn’t — be­cause of dis­crim­i­na­tion. While sta­tis­ti­cal tech­niques can ad­just for some of these fac­tors — ed­u­ca­tion, ge­og­ra­phy and the like — no anal­y­sis can ac­count for all of them.

But the new re­search al­lows for a clearer con­clu­sion: It ap­pears to have doc­u­mented straight­for­ward dis­crim­i­na­tion.

As a real-world ex­per­i­ment, it built on ear­lier “au­dit ex­per­i­ments”, as they are known in so­cial sci­ence. Per­haps the most fa­mous is a study by Mar­i­anne Ber­trand of the Univer­sity of Chicago and Send­hil Mul­lainathan of Har­vard. In that ear­lier ex­per­i­ment, Ber­trand and Mul­lainathan sent fic­ti­tious re­sumes to em­ploy­ers, find­ing that peo­ple with white-sound­ing names were more likely to re­ceive a pos­i­tive re­sponse than those with black-sound­ing names.

The new find­ings pro­vide fur­ther in­di­ca­tion of the many ways in which dis­crim­i­na­tion shapes the lives of African-Amer­i­cans. What’s more, when it’s harder to get your neigh­bour­hood li­brar­ian to re­spond to a sim­ple email about open­ing hours, it’s not much of a leap to imag­ine other in­ter­ac­tions — deal­ing with a com­puter help desk, the front of­fice at a school or just the dry cleaner — that go less smoothly.

Economists tend to group ex­pla­na­tions of dis­crim­i­na­tory be­hav­iour into two buck­ets: Taste-based and sta­tis­ti­cal. If a li­brar­ian chooses not to re­spond be­cause a per­son is black, that’s taste-based dis­crim­i­na­tion. There’s a sim­pler la­bel: Racism.

Sta­tis­ti­cal dis­crim­i­na­tion, on the other hand, oc­curs when a li­brar­ian uses a per­son’s name or race as a marker for other char­ac­ter­is­tics. Per­haps an African-Amer­i­can-sound­ing name sig­nals that a per­son is more likely to be poor. The li­brar­ian hap­pens to be bi­ased against poor peo­ple. In this case, race is be­ing used as a statis­tic for in­fer­ring poverty, and the per­cep­tion of poverty causes the dis­crim­i­na­tory be­hav­iour.

But two pieces of sug­ges­tive ev­i­dence in this study point to the prob­lem here as be­ing straight­for­ward, taste­based dis­crim­i­na­tion.

First, the au­thors re­peated the ex­er­cise — send­ing an ad­di­tional 20,000 emails to the same re­cip­i­ents — although this time with a twist. They added a sig­na­ture line, iden­ti­fy­ing the sender as a real es­tate agent. This ex­tra in­for­ma­tion made the sender’s name less rel­e­vant for in­fer­ring in­come or so­cioe­co­nomic sta­tus. If sta­tis­ti­cal dis­crim­i­na­tion had driven be­hav­iour in the first round, this ex­tra in­for­ma­tion should have led to less dis­crim­i­na­tion in the fol­low-up. It did not.

Sec­ond, the pat­tern of ev­i­dence was con­sis­tent with taste-based dis­crim­i­na­tion. While the re­searchers didn’t de­ter­mine the race of the peo­ple who re­sponded to their emails, they did have data on the racial break­down of the mu­nic­i­pal work­forces. The racial gap in email re­sponse rates was greater in coun­ties where the pro­por­tion of whites was higher.

Taste-based dis­crim­i­na­tion — ba­si­cally, racism — isn’t nec­es­sar­ily the re­sult of con­scious thought. In an email, Prof Tonin, one of the study’s au­thors, said that it’s pos­si­ble “this be­hav­iour is due to some sort of un­con­scious bias” and, there­fore, that “mak­ing peo­ple aware of the prob­lem may con­trib­ute to the so­lu­tion”.

The study may be help­ful in re­fin­ing our un­der­stand­ing of racial dis­crim­i­na­tion in the United States. It oc­curs not only in the labour mar­ket and the crim­i­nal jus­tice sys­tem, but also in count­less small fric­tions ev­ery day.

The cul­prit might not be a hate-spew­ing white na­tion­al­ist but rather a li­brar­ian or a school ad­min­is­tra­tor or a county clerk, un­aware that she’s help­ing some clients more than others.

This dif­fer­ence was large enough that it was sta­tis­ti­cally un­likely to have been a mat­ter of mere chance.


Pro­test­ers from Stu­dents For a Demo­cratic So­ci­ety and Black Lives Mat­ter demon­strate on the Univer­sity of Utah cam­pus last month. A re­search study sug­gests African-Amer­i­cans are still af­fected by racial dis­crim­i­na­tion.

Newspapers in English

Newspapers from Thailand

© PressReader. All rights reserved.