Back to History

Fields Changed

Registration

Field Before After
Trial End Date March 01, 2024 December 31, 2026
Last Published August 10, 2023 01:22 PM May 21, 2026 12:39 PM
Intervention (Public) We will partner with a black owned company to place an employment ad for for an administrative assistant position. The advertisement will include instructions to email a company contact for more details on the position and application instructions. After an applicant is hired, a demographic survey will be sent to everyone who requested more information. Respondents will be randomized to receipt of more job details including either: [VERSION 1] "Thank you for your interest in a position with [COMPANY]." [VERSION 2] "Thank you for your interest in a position with [COMPANY], a Black owned company." We partnered with a Black-owned small business (Moorhead Apparel) to place a national online employment advertisement for a remote temporary administrative assistant position. Interested job seekers were instructed to email [email protected] for more information. Upon receipt of each inquiry email, an automated system randomly assigned the inquirer to one of two treatment conditions (50/50) and sent the corresponding job description. The only difference between conditions is the inclusion or exclusion of the phrase "a Black owned company" in the opening sentence of the job description. To formally apply, inquirers must complete and return an attached employment questionnaire.A pilot wave was conducted under a Concordia College IRB in 2023 (n=778; 389 treatment, 389 control). This registration is updated to reflect a replication wave conducted under a North Dakota State University IRB exempt determination. Both waves are included in the combined analysis.
Intervention End Date March 01, 2024 December 31, 2026
Primary Outcomes (End Points) The outcome variable will be the proportion of people who applied for the position after receiving more information in each treatment, broken down by the race of the prospective applicant. The primary outcome is the binary application decision — whether an inquirer subsequently completed and returned the employment questionnaire — compared across treatment arms for all inquirers regardless of race identification status. The primary test is a two-sided Fisher's exact test at α=0.05.
Primary Outcomes (Explanation) An inquirer is defined as any individual who emailed [email protected] in response to the job advertisement. An applicant is defined as any inquirer who subsequently completed and returned the employment questionnaire. The primary analysis tests whether application rates differ between treatment and control arms across all inquirers.
Experimental Design (Public) We will partner with a black owned company to place an employment ad for for an administrative assistant position. The advertisement will include instructions to email a company contact for more details on the position and application instructions. After an applicant is hired, a demographic survey will be sent to everyone who requested more information. Respondents will be randomized to receipt of more job details including either: [VERSION 1] "Thank you for your interest in a position with [COMPANY]." [VERSION 2] "Thank you for your interest in a position with [COMPANY], a Black owned company." The outcome variable will be the proportion of people who applied for the position after receiving more information in each treatment, broken down by the race of the prospective applicant. Natural field experiment. A real job advertisement for a remote temporary administrative assistant position was posted nationally on Indeed.com on behalf of Moorhead Apparel, a Black-owned small business. Interested job seekers emailed [email protected] to request more information. An automated system randomized each inquirer to treatment or control (50/50) and sent the corresponding job description. The treatment job description included the phrase "a Black owned company" in the opening sentence; the control version was otherwise identical without this phrase. To apply, inquirers completed and returned an attached employment questionnaire. This study pools two data collection waves: a pilot wave conducted at Concordia College (2023, n=778) and a replication wave conducted at North Dakota State University (2025-2026). Pilot results are disclosed in full in the hidden fields of this registration prior to commencement of the NDSU wave.
Randomization Method Subjects will be assigned to one of the two treatments in alternating order of their email request for more information. Automated 50/50 randomization via Google Apps Script using JavaScript's Math.random() function, triggered upon receipt of each inquiry email. Each inquirer is independently randomized.
Planned Number of Observations 500 1,621 total inquirers across both waves (778 from Concordia pilot wave already collected; 843 additional inquirers targeted from NDSU replication wave).
Sample size (or number of clusters) by treatment arms N/A Approximately 811 treatment, 811 control (combined across both waves, 50/50 randomization).
Power calculation: Minimum Detectable Effect Size for Main Outcomes Based on pilot wave overall treatment effect (application rate 53.5% treatment vs. 58.4% control, Cohen's h=0.098). Required n=1,621 total for 80% power at α=0.05 two-sided. Pilot wave contributes 778 observations; NDSU wave targets 843 additional observations.
Intervention (Hidden) Pilot wave results (disclosed prior to NDSU data collection): Overall application rate 53.5% treatment vs. 58.4% control (p=0.194). Among racially identified inquirers: Black applicants 83.7% treatment vs. 94.6% control (p=0.166); White applicants 68.3% treatment vs. 79.2% control (p=0.074). Race identified for 48.5% of pilot inquirers via social media search following Leibbrandt and List (2018). The NDSU replication targets 843 additional inquirers to reach a combined sample of 1,621 (80% power, Cohen's h=0.098, α=0.05).
Secondary Outcomes (End Points) Exploratory analysis of treatment effect heterogeneity by race. Specifically, whether application rates differ between treatment and control arms within racial subgroups (Black or African American; White, non-Hispanic).
Secondary Outcomes (Explanation) Race is determined by searching social media profiles using name and resume information following Leibbrandt and List (2018). Two independent identification passes are conducted. Inquirers classified as Not Found are excluded from the race-stratified analysis. Given approximately 48.5% coverage in the pilot, bounding analysis will be reported as a robustness check.
Back to top

Irbs

Field Before After
IRB Name North Dakota State University
IRB Approval Date April 13, 2026
IRB Approval Number Protocol #IRB0005864
Back to top