Table 2. Open in new tab Assessment of...

	Admission process	Hiring process
Panel A: Quasi-random assignment
Guryan et al. (2009)
Leave-one-out mean	0.002*	−0.001	−0.012	−0.005
	(0.001)	(0.001)	(0.027)	(0.028)
$R^{2}$ (within)	0.998	0.998	0.710	0.707
Jochmans (2023)
Test statistic	0.695	−0.048	0.710	1.037
p-value	0.487	0.962	0.478	0.300
Panel B: Quasi-random ordering
Guryan et al. (2009)
Lag (⁠ $t - 1$ ⁠)	0.000	0.002	0.024	0.013
	(0.006)	(0.006)	(0.017)	(0.023)
$R^{2}$ (within)	0.009	0.024	0.002	0.000
Jochmans (2023)
Test statistic	0.915	0.960	1.426	0.967
p-value	0.360	0.337	0.154	0.333
N	26,970	26,970	5,165	5,165

Admission process

Hiring process

(1)

(2)

(3)

(4)

Std. TPA

Std. predicted rating

Std. TPA

Std. predicted rating

Panel A: Quasi-random assignment

Guryan et al. (2009)

Leave-one-out mean

0.002*

−0.001

−0.012

−0.005

(0.001)

(0.027)

(0.028)

R^{2}

(within)

0.998

0.710

0.707

Jochmans (2023)

Test statistic

0.695

−0.048

0.710

1.037

p-value

0.487

0.962

0.478

0.300

Panel B: Quasi-random ordering

Guryan et al. (2009)

Lag (⁠

t - 1

⁠)

0.000

0.002

0.024

0.013

(0.006)

(0.017)

(0.023)

R^{2}

(within)

0.009

0.024

0.002

0.000

Jochmans (2023)

Test statistic

0.915

0.960

1.426

0.967

p-value

0.360

0.337

0.154

0.333

26,970

5,165

	Admission process	Hiring process
Panel A: Quasi-random assignment
Guryan et al. (2009)
Leave-one-out mean	0.002*	−0.001	−0.012	−0.005
	(0.001)	(0.001)	(0.027)	(0.028)
$R^{2}$ (within)	0.998	0.998	0.710	0.707
Jochmans (2023)
Test statistic	0.695	−0.048	0.710	1.037
p-value	0.487	0.962	0.478	0.300
Panel B: Quasi-random ordering
Guryan et al. (2009)
Lag (⁠ $t - 1$ ⁠)	0.000	0.002	0.024	0.013
	(0.006)	(0.006)	(0.017)	(0.023)
$R^{2}$ (within)	0.009	0.024	0.002	0.000
Jochmans (2023)
Test statistic	0.915	0.960	1.426	0.967
p-value	0.360	0.337	0.154	0.333
N	26,970	26,970	5,165	5,165

Table 2.

Open in new tab

Assessment of quasi-random assignment & ordering

	Admission process		Hiring process
	(1)	(2)	(3)	(4)
	Std. TPA	Std. predicted rating	Std. TPA	Std. predicted rating
Panel A: Quasi-random assignment
Guryan et al. (2009)
Leave-one-out mean	0.002*	−0.001	−0.012	−0.005
	(0.001)	(0.001)	(0.027)	(0.028)
$R^{2}$ (within)	0.998	0.998	0.710	0.707
Jochmans (2023)
Test statistic	0.695	−0.048	0.710	1.037
p-value	0.487	0.962	0.478	0.300
Panel B: Quasi-random ordering
Guryan et al. (2009)
Lag (⁠ $t - 1$ ⁠)	0.000	0.002	0.024	0.013
	(0.006)	(0.006)	(0.017)	(0.023)
$R^{2}$ (within)	0.009	0.024	0.002	0.000
Jochmans (2023)
Test statistic	0.915	0.960	1.426	0.967
p-value	0.360	0.337	0.154	0.333
N	26,970	26,970	5,165	5,165

	Admission process		Hiring process
	(1)	(2)	(3)	(4)
	Std. TPA	Std. predicted rating	Std. TPA	Std. predicted rating
Panel A: Quasi-random assignment
Guryan et al. (2009)
Leave-one-out mean	0.002*	−0.001	−0.012	−0.005
	(0.001)	(0.001)	(0.027)	(0.028)
$R^{2}$ (within)	0.998	0.998	0.710	0.707
Jochmans (2023)
Test statistic	0.695	−0.048	0.710	1.037
p-value	0.487	0.962	0.478	0.300
Panel B: Quasi-random ordering
Guryan et al. (2009)
Lag (⁠ $t - 1$ ⁠)	0.000	0.002	0.024	0.013
	(0.006)	(0.006)	(0.017)	(0.023)
$R^{2}$ (within)	0.009	0.024	0.002	0.000
Jochmans (2023)
Test statistic	0.915	0.960	1.426	0.967
p-value	0.360	0.337	0.154	0.333
N	26,970	26,970	5,165	5,165

Notes: TPA, third-party assessment of candidate quality (see Section 3.2 for details). Panel A presents tests for a relationship between an individual’s quality and the leave-one-out mean quality of the other candidates assigned to the same interview sequence. The test proposed by Guryan et al. (2009) controls for the leave-one-out mean quality at the workshop or candidate pool level. This test has limited power in the admission process (Columns 1 and 2) due to limited variation in the size of workshops. Therefore, we additionally provide test statistics and p-values from an alternative bias-corrected test for random peer assignment developed by Jochmans (2023), which does not require variation in the size of randomization units. In Panel B, we test for a relationship between the quality of the current and the previous candidate, conditional on the leave-one-out mean quality at the sequence level. All regressions control for gender and workshop/candidate pool fixed effects. Standard errors are clustered at the workshop/candidate pool level (N = 312/N = 63). ^* p<0.10, ^** p<0.05, ^***p<0.01.

This Feature Is Available To Subscribers Only