Discrimination Flashcards
What is lookism?
Discrimination on the grounds based on someones appearance. ( good looking people likely to be paid more for example, eve. though they are not more able)
Why is that good looking people likely to be paid more for example, even. though they are not more able
As they are more confident, thus seen as more able
Why is this possibly the case?
A lot of women work in HR
What is the paper related to this topic ?
Are Emily and Greg more Employable Than Lakisha and Jamal? A field Experiment on Labour Market Discrimination ( Bertrand and Mullainathan (2004)
What is the Main message of the paper by Bertrand and Mullainathan(2004)?
Field experiment on racial discrimination
Experimenters respond to ‘help wanted’ ads with fictitious CVs - use either White (W) or African American (AA) sounding names,
Cvs are the same.
Main result: W receive 50% more interviews.
Lets look at the Experimental design!, so they take real –CVs ( from job search websites) with names and contact details changed.
-For each ad they send 4. CVS: high( experience)/low quality and W/AA name.
->1300 ads, about 5000 CVs
- Use high frequency W/AA names ( from birth certificate data)
- See who gets an interview or callback( CB)
Interpret these results.
We can see that the white CVs are clearly getting more interviews.
With one exception ( female in sales job) they are all significant the t tests ( difference in p-value).
Interpret this
1) Resume quality manipulation obviously works here; higher quality resumes receive more callbacks.
2) White sounding applicants with more likely to get a call back with higher quality resumes with a significant p value 2.51
3) Whereas AA experience or cv quality doesn’t have much of a difference, the p value is not statistically significant.
In the table they compute what have they done here?
In table 2 they compute the fraction of employers that treat White and AA applicants equally, the fraction of employers that favour White applicants and the fraction of employers that favour AA. Equal treatment occurs when either no applicant gets called back, one white and one AA gets called back or 2 white and 2 AA gets called back. Whites are favoured when either only one White gets called back, two whites and no AA, two whites and one AA get called back. Finally AA are favoured in all other cases.
Now that we know what the table is interpret it?In table 2 they compute the fraction of employers that treat White and AA applicants equally, the fraction of employers that favour White applicants and the fraction of employers that favour AA. Equal treatment occurs when either no applicant gets called back, one white and one AA gets called back or 2 white and 2 AA gets called back. Whites are favoured when either only one White gets called back, two whites and no AA, two whites and one AA get called back. Finally AA are favoured in all other cases.
Equal treatment occurs for about 88 percent of the help wanted ads = EXPECTED as high proportion of ads for which no call backs were recorded. Whites are favoured by nearly 8.4 percent of the employers, with the majority coming from one white applicant. AA on the other hand are favoured only about 3.5% of employers .
the difference between the fraction of employers favouring whites and the fraction of employers favouring AA is statistically very significant ( p - 0.0000).
Table 3 shows means and SD’s of the most relevant resume characteristics for full sample as well as race and quality of surveys. What are some things we can see?
High quality applicants more likely to have an email address and a year in military service in comparison to low quality applicants. They don’t differ in education level.
Based on table 3, we use resume characteristics to classify resumes, instead of subjectivity. so we control for sex dummy, city dummy, six occupation dummies and vector of job requirements as listed in the employment ads. We classified high as those resumes that have above median predicted call back and similarly low those resumes that have below median predicted callback, Interpret results.
As we can see AA do appear to significantly benefit from higher quality resumes under this alternative classification, but benefit less than whites.
What are 3 weaknesses of this experiment?
1) resumes do not report race but instead suggest race through personal names. Some employers may not recognise the racial content in name.
2) not every AA has an African sounding name.
3) getting a callback doesn’t mean you will get the job, we cant validate our findings for hiring rates.
We want to find out here if AA are helped here by living in more affluent neighbourhoods. So columns 1 3 and 4 is a city dummy, 2,4 and 6 is a city dummy interacted with a race dummy. So all these columns reports the results of probit regression of the callback dummy on a specific zip code characteristic and a city dummy.
Interpret results ( NOTE THESE NUMBERS ARE MARGINAL EFFECTS, AS THEY ARE REGRESSIONS)
We find a positive and significant effect of neighborhood quality on the likelihood of a callback.
Applicants liibng in whiter( column 1 ), more educated( column 3 and or higher income (column 5 ) have a higher probability of receiving a callback.
In column 2 4 and 6 where the zip code characteristic with a dummy variable for whether the applicant is AA or not, we find that there is no evidence that AA benefit any more than whites from living in whiter, more educated zip code. ( very insigifncant and small)
In terms of p values how do we know whether they are statiscially significant?
A p-value less than 0.05 is typically considered to be statistically significant. We reject null hyphothesis.
What is the summary meaning of table 7?
If names mainly signal social background ( which could be interpreted in table 6), one might have expected the name gap to be higher for jobs that rely more on soft skills or require more interpersonal interactions, however we found no evidence to this.
Table 8 is a very interesting table, it displays the average callback rate for each first name we used in the sample along with this proxy for social background .
‘Average’ reports represent the average fraction of mothers that have at least completed high school for the set of names listed in that gender-race group.
‘Overall’ reports the average fraction of mothers that have at least completed high school for the full sample of births in that gender race group.
So consistent with the social background interpretation the A names chosen fall below the AA average
Now interpret it. ( also interpret correlation)
1) the difference between AA names with males chosen in study and population is negligble.
2) For white names both males and females are above average.
3) What is interesting is the difference between name heterogeneity in social background, e.g. AA babies called Kenya or Jaml are affiliated with higher mother education than AA babies named Latonya or Leroy, Conversely, White babies named Neil or Carrie have lower social background than those named Emily or Geoffrey.
4) The correlation between call back and mothers education is negative, which we predicted it was positive. The p value not significant so we cannot reject null hyphoesis, thus we cannot say that social background drives the measured race gap.
So to conclude what does table 8 disprove?
