\(~\) \(~\)
bm %>%
group_by(race) %>%
summarise(
Avg_education= mean(education) ,
Avg_jobs= mean(jobs)
) ## # A tibble: 2 x 3
## race Avg_education Avg_jobs
## <chr> <dbl> <dbl>
## 1 b 3.62 3.66
## 2 w 3.62 3.66Yes. Given the average of education and jobs in White-sounding names (w) and Black-sounding names (b) are very similar, and rounded to two decimal points identical, we can infer that education and jobs are balanced across race. If they werenโt balanced between race, we cannot draw the conclusion that the causal effect of racially distinguishable names on callbacks is attributed solely to whether the name is White-sounding or Black-sounding, and not also influenced by resume quality. Furthermore, if education and jobs were not balanced across race, the process of randomization would not have worked in as we would be committing selection bias. Randomly assigning Black-sounding or White-sounding names to resumes would not be representative considering the quality of resumes, whether in the high quality or low quality pool, would differ. Hence, we cannot randomly assign Black-sounding and White-sounding names to resumes in either of those groups (high or low). In other words, the process of randomization would only work if for example within the low quality resumes, we could randomly assign a Black-sounding or White sounding name and that those two are almost identical on both education and job experience levels.
mean(bm$call)## [1] 0.08049281bm %>%
group_by(race) %>%
summarise(
Avg_callback= mean(call)
) ## # A tibble: 2 x 2
## race Avg_callback
## <chr> <dbl>
## 1 b 0.0645
## 2 w 0.0965Average callback rate of resumes with White-sounding name: 0.0965
Average callback rate of resumes with Black-sounding name:: 0.0645
The above results show that there is a causal effect of racially distinguishable names on average callback rates. Individuals with White-sounding names got 3.2 percentage points more callbacks than individuals with Black-sounding names. These results suggest that resumes with White-sounding names on average receive more callbacks than those with Black-sounding names, when the resumes have relatively identical qualities whether in the low quality or high high quality pool. From these results, we can infer that employers advertising jobs in certain newspapers discriminate against resumes with Black-sounding names compared to White-sounding names in Chicago and Boston.
bm %>%
group_by(race,gender) %>%
summarise(
Avg_callback= mean(call)
) ## # A tibble: 4 x 3
## # Groups: race [2]
## race gender Avg_callback
## <chr> <chr> <dbl>
## 1 b f 0.0663
## 2 b m 0.0583
## 3 w f 0.0989
## 4 w m 0.0887Average callback of a Male with a White-sounding name: 0.0887
Average callback of a Male with a Black-sounding name: 0.0583
Average callback of a Female with a White-sounding name: 0.0989
Average callback of a Female with a Black-sounding name: 0.0663
The results above allow us to look at discrimination against racially distinguishable names within combinations of race and sex. On average, a female with a White-sounding name receives 3.26 percentage points more callbacks than a female with a Black-sounding name. Similarly, males with White-sounding names receive 3.04 percentage points more callbacks than males with Black-sounding names. Furthermore, the average callback rates for Females with White-sounding and Black-sounding names is the higher than those of Males, according to this experiment. However, as suggested in the paper, this reverse gender gap is statistically insignificant, especially when compared to the gap in racially distinguishable names that can be seen above.