Skip to content

Physical Disability and Labor Market Discrimination: Evidence from a Video Résumé Field Experiment

icpsr_175061

Charles Bellemare, Marion Goussé, Guy Lacroix, Steeve Marchand (2023). American Economic Journal: Applied Economics 2023, 15(4): 452–476

DOI: 10.1257/app.20210633

field value
venue American Economic Journal: Applied Economics 2023, 15(4): 452–476
paper 10.1257/app.20210633
replication data openICPSR 175061
data license CC BY-NC 4.0 — “The data are licensed under a Creative Commons/CC-BY-NC license.”
regression datasets (in this corpus) 3
graduated / eligible / exported 3 / 4 / 27 (policy claims_manual_regression_cap)
routed / gap / residue (at source) 17 / 3 / 0

Claims & tasks

Every claim the paper makes in its main body about a parameter, joined to the benchmark task (dataset · coefficient) that captures it. The effect, s.e. and p are the captured regression's own reported values (the paper's Stata figures, carrying the paper's SE method); the row shows the paper's verbatim quote (page · exhibit), so you can confirm the regression matches what the paper says. The prominence column is the within-study inclusion signal (⭐ headline → primary → secondary). The sensitivity column is the privacy level of the most-sensitive variable in the regression (🔴 high / 🟠 medium / ⚪ low).

# prominence sensitivity claim effect s.e. p task instance (dataset · coef)
C1 ⭐ primary 🔴 high “On the other hand, our estimates suggest that including a video résumé increases the callback rate by around 10 percentage points.” — p. 14
“We also find that the video résumés increase callback rates by 10 percentage points for disabled and nondisabled applicants alike, which suggests that they convey a positive signal to potential employers.” — p. 3
Table 2, col 1 (Video experiment (a))		 +0.104 0.036 0.004 Linear Regression reg_1 · 1.vid
n=2021 · d=3
C2 ⭐ primary 🔴 high “We find that revealing a disability decreases callback rates by 25 percentage points.” — p. 1
“In all specifications, we find that the callback rate decreases by 25 percentage points when the application reveals the disability.” — p. 14
“We find that revealing a disability decreases callback rates by 25 percentage points for applications both with and without videos, even after controlling for accessibility constraints.” — p. 3
Table 2, col 1 (Video experiment (a))		 -0.250 0.038 9.1e-11 Linear Regression reg_1 · 1.disab
n=2021 · d=3
C3 ⭐ primary 🔴 high We find that including a video résumé of a well-spoken applicant significantly increases callbacks by 10 percentage points for persons with and without disabilities, suggesting that discrimination is unaffected by quality signals in our context.
p. 1 · Table 2, col 1 (Video experiment (a))		 -0.030 0.046 0.517 Linear Regression reg_1 · 1.vid#1.disab
n=2021 · d=3
C4 ⭐ primary 🔴 high “On the other hand, our estimates suggest that including a video résumé increases the callback rate by around 10 percentage points.” — p. 14
“We also find that the video résumés increase callback rates by 10 percentage points for disabled and nondisabled applicants alike, which suggests that they convey a positive signal to potential employers.” — p. 3
Table 2, col 2 (Video experiment (b))		 +0.104 0.038 0.006 Linear Regression reg_2 · 1.vid
n=1788 · d=5
C5 ⭐ primary 🔴 high “In all specifications, we find that the callback rate decreases by 25 percentage points when the application reveals the disability.” — p. 14
“We find that revealing a disability decreases callback rates by 25 percentage points for applications both with and without videos, even after controlling for accessibility constraints.” — p. 3
Table 2, col 2 (Video experiment (b))		 -0.256 0.048 1.2e-07 Linear Regression reg_2 · 1.disab
n=1788 · d=5
C6 ⭐ primary 🔴 high We find that including a video résumé of a well-spoken applicant significantly increases callbacks by 10 percentage points for persons with and without disabilities, suggesting that discrimination is unaffected by quality signals in our context.
p. 1 · Table 2, col 2 (Video experiment (b))		 -0.026 0.048 0.585 Linear Regression reg_2 · 1.vid#1.disab
n=1788 · d=5
C7 ⭐ primary 🔴 high “On the other hand, our estimates suggest that including a video résumé increases the callback rate by around 10 percentage points.” — p. 14
“We also find that the video résumés increase callback rates by 10 percentage points for disabled and nondisabled applicants alike, which suggests that they convey a positive signal to potential employers.” — p. 3
Table 2, col 3 (Video experiment (c))		 +0.104 0.036 0.004 Linear Regression reg_3 · 1.vid
n=2021 · d=4
C8 ⭐ primary 🔴 high “We find that revealing a disability decreases callback rates by 25 percentage points.” — p. 1
“In all specifications, we find that the callback rate decreases by 25 percentage points when the application reveals the disability.” — p. 14
“We find that revealing a disability decreases callback rates by 25 percentage points for applications both with and without videos, even after controlling for accessibility constraints.” — p. 3
Table 2, col 3 (Video experiment (c))		 -0.250 0.038 9.2e-11 Linear Regression reg_3 · 1.disab
n=2021 · d=4
C9 ⭐ primary 🔴 high We find that including a video résumé of a well-spoken applicant significantly increases callbacks by 10 percentage points for persons with and without disabilities, suggesting that discrimination is unaffected by quality signals in our context.
p. 1 · Table 2, col 3 (Video experiment (c))		 -0.037 0.049 0.449 Linear Regression reg_3 · 1.vid#1.disab
n=2021 · d=4
C10 primary 🔴 high Finally, our result that wheelchair accessibility plays no role in callback rates may be due to small sample size.
p. 15 · Table 2, col 2 (Video experiment (b))		 +0.052 0.034 0.121 Linear Regression reg_2 · 1.firm_access
n=1788 · d=5
C11 primary 🔴 high We thus present novel insights for this literature by showing that workplace accessibility plays a marginal role in explaining the gap even though a sizable share of firms are inaccessible.
p. 3 · Table 2, col 2 (Video experiment (b))		 -0.001 0.044 0.981 Linear Regression reg_2 · 1.disab#1.firm_access
n=1788 · d=5
C12 primary 🔴 high We test this hypothesis by interacting a dummy variable for late disclosure with the interaction of video and disability dummies (column c) but find no significant effect of revealing the disability later.
p. 14 · Table 2, col 3 (Video experiment (c))		 +0.014 0.033 0.673 Linear Regression reg_3 · 1.vid#1.disab#1.vid_late
n=2021 · d=4

Regression datasets

The 3 regression dataset(s) graduated into the benchmark corpus from this study's replication package — each backs one or more claims above. Expand each for its features, response, (sound, data-independent) public bounds, and reproduction grade.

icpsr_175061_reg_1 — response ~ 1.vid + 1.disab + 1.vid#1.disab

Sensitivity proposal: 🔴 high maximum across columns (advisory; reviewed manually). Reproduction grade: ✅ match (benchmark-eligible).

Features

name description type sensitivity bound lo bound hi estimate s.e.
_cons intercept continuous — 1 1 +0.4496 0.031
1.vid video résumé included in application indicator categorical ⚪ low 0.0 1.0 +0.1043 0.036
1.disab applicant discloses physical disability status categorical 🔴 high 0.0 1.0 -0.2503 0.0384
1.vid#1.disab interaction: video résumé x disability disclosure categorical ⚪ low 0.0 1.0 -0.02964 0.0457

Response

name description type sensitivity bound lo bound hi estimate s.e.
response employer callback/response to job application continuous 🟠 medium 0.0 1.0 — —

Public bounds (data-independent) sourced from: binary employment/labor-market outcome indicator; binary indicator; binary indicator (treatment/arm variable); interaction of two binary indicators.

n = 2,021 samples.

icpsr_175061_reg_2 — response ~ 1.vid + 1.disab + 1.vid#1.disab + 1.firm_access + 1.disab#1.firm_access

Sensitivity proposal: 🔴 high maximum across columns (advisory; reviewed manually). Reproduction grade: ✅ match (benchmark-eligible).

Features

name description type sensitivity bound lo bound hi estimate s.e.
_cons intercept continuous — 1 1 +0.4196 0.0379
1.vid video résumé included in application indicator categorical ⚪ low 0.0 1.0 +0.1041 0.0381
1.disab applicant discloses physical disability status categorical 🔴 high 0.0 1.0 -0.2559 0.0482
1.vid#1.disab interaction: video résumé x disability disclosure categorical ⚪ low 0.0 1.0 -0.02642 0.0484
1.firm_access firm's workplace wheelchair accessibility indicator categorical ⚪ low 0.0 1.0 +0.0524 0.0338
1.disab#1.firm_access interaction: disability disclosure x firm wheelchair access categorical ⚪ low 0.0 1.0 -0.001029 0.0443

Response

name description type sensitivity bound lo bound hi estimate s.e.
response employer callback/response to job application continuous 🟠 medium 0.0 1.0 — —

Public bounds (data-independent) sourced from: binary employment/labor-market outcome indicator; binary indicator; binary indicator (treatment/arm variable); binary indicator; firm-level (non-person) attribute; interaction of two binary indicators.

n = 1,788 samples.

icpsr_175061_reg_3 — response ~ 1.vid + 1.disab + 1.vid#1.disab + 1.vid#1.disab#1.vid_late

Sensitivity proposal: 🔴 high maximum across columns (advisory; reviewed manually). Reproduction grade: ✅ match (benchmark-eligible).

Features

name description type sensitivity bound lo bound hi estimate s.e.
_cons intercept continuous — 1 1 +0.4496 0.031
1.vid video résumé included in application indicator categorical ⚪ low 0.0 1.0 +0.1043 0.036
1.disab applicant discloses physical disability status categorical 🔴 high 0.0 1.0 -0.2503 0.0384
1.vid#1.disab interaction: video résumé x disability disclosure categorical ⚪ low 0.0 1.0 -0.03697 0.0488
1.vid#1.disab#1.vid_late interaction: video x disability x late disclosure timing categorical ⚪ low 0.0 1.0 +0.01408 0.0333

Response

name description type sensitivity bound lo bound hi estimate s.e.
response employer callback/response to job application continuous 🟠 medium 0.0 1.0 — —

Public bounds (data-independent) sourced from: binary employment/labor-market outcome indicator; binary indicator; binary indicator (treatment/arm variable); interaction of two binary indicators; triple interaction of binary indicators.

n = 2,021 samples.