Scribd
Upload
15 views7 pages

Female-Sounding Names and Callbacks Analysis

The document discusses the impact of having a female-sounding name on callback rates in various industries, revealing that individuals with such names are 2.4 percentage points less likely to receive callbacks in manufacturing and related sectors. It asserts that the study's randomized design supports the conclusion that the observed effects are unbiased estimates of the average treatment effect. Additionally, while there appears to be no overall discrimination based on perceived gender, significant industry-specific differences are noted, indicating varying levels of bias across sectors.

Uploaded by

Krish Goyal
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
15 views7 pages

Female-Sounding Names and Callbacks Analysis

The document discusses the impact of having a female-sounding name on callback rates in various industries, revealing that individuals with such names are 2.4 percentage points less likely to receive callbacks in manufacturing and related sectors. It asserts that the study's randomized design supports the conclusion that the observed effects are unbiased estimates of the average treatment effect. Additionally, while there appears to be no overall discrimination based on perceived gender, significant industry-specific differences are noted, indicating varying levels of bias across sectors.

Uploaded by

Krish Goyal
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

ECO372 - Winter 2024 - Tutorial Week 6

Sean Elliott

University of Toronto

February 16, 2024

1/7
Solution to AA1 Stage 3 Q1(a)

Refer to Column 3 of Table 2. Interpret the coefficient for “Female


Sounding” in this regression, commenting on its economic and sta-
tistical significance.

2/7
Solution to AA1 Stage 3 Q1(a)
Here we are looking for the marginal effect of “female sounding”
on the callback rate. The model here looks something like,

CallBacki = β0 + β1 FemaleSoundingi + β2 IndustryTypei +


β3 FemaleSoundingi × IndustryTypei + · · · + ui

Thus, the marginal effect is given by

∂CallBacki
= β1 + β3 IndustryTypei
∂FemaleSoundingi
implying we need to interpret β1 as the effect when
IndustryType = 0.

Recall, that IndustryTypei = 0 refers to jobs in the manufacturing,


wholesale trade, construction, or automotive industry (MWCA).
3/7
Solution to AA1 Stage 3 Q1(a)1
▶ Interpretation: On average, those with female-sounding
names are 2.4 pp less likely to recieve a callback in the
MWCA industries, relative to those with male-sounding
names, holding all else constant.
▶ Practical significance: In Table 1 we can see that the mean
callback rate for females in the MWCA industries is around
25% – thus 2.4 pp represents ∼10% difference which is
relatively large.
▶ Statistical significance: There are many ways to go about
this. I think the easiest is to just divide the coefficient
(-0.024) by the standard error (0.007) which gives you a t-stat
of t = −3.43 which is larger than 1.96 =⇒ statistically
significant at the 5% level.
1
Any answer that clearly conveys the same ideas I have written here will be awarded full marks – this is simply
one example of how to answer the question.
4/7
Solution to AA1 Stage 3 Q1(b)

Refer to Column 1 of Table 2, where the coefficient on “Female


Sounding” is near zero, measured with precision. For this population,
do you think that coefficient provides an unbiased estimate of the
average treatment effect (ATE) for having a female vs male sounding
name? Use the study design and evidence from the data to support
your assessment.

The study uses a randomized design implying that the SDO (as
measured by the coeffient on “Female Sounding”) will be
equivalent to the ATE.

5/7
Solution to AA1 Stage 3 Q1(b)

We can confirm that the SDO=ATE by comparing the effect


across Specifications 1 & 2.

When adding control variables, we observe a negligible difference in


the two effects despite the fact that some of these controls are
relevant for callback rates. This implies that there is no correlation
between “Female Sounding” and the attributes of the applicants
being controlled for.

6/7
Solution to AA1 Stage 3 Q1(c)

Given the evidence shown in the supplement, do you conclude that


firms in this population do not discriminate call backs by perceived
gender? Explain why or why not using evidence from the data.

If we look at the SDO in Column 1 (which we argued is equivalent


to the ATE), there does not appear to be any discrimination on
the basis of perceived gender.

However, this is not the case once we allow for effect hetergeneity
by industry. In fact, we see that in the MWCA industries those
with female-sounding names are 2.4 pp less likely to get a callback
whereas this effect is a positive 0.7 pp for the retail, finance,
transportation, public utility, and service industries.

7/7

You might also like