Personalized Emails from Course Instructors and First-Year Student Performance and Persistence: A Randomized Field Experiment

Students assigned to the treatment group receive three short, personalized emails from their course instructors during the first teaching block. The emails are sent at key moments in the course: at the start of the course, after the midterm, and after the final exam. Students in the control group receive standard course communication.

2026-08-31

2027-12-31

Course performance in introductory economics and econometrics courses; first-year academic progress and persistence.

Course performance will be measured using administrative records on midterm grades, final exam grades, and course grades. First-year academic progress and persistence will be measured using administrative records on credits earned in the first year, binding study advice outcomes, early dropout before 1 February, and progression to the second year.

Students’ perceptions of instructor involvement, clarity, and support; use of available student support channels.

Perceptions of instructor involvement, clarity, and support will be measured using regular course evaluation surveys, including additional items where available. Use of support channels will be measured using available administrative or course-level records, such as contact with study advisers or office-hour participation, where available.

This study is a blocked randomized field experiment among approximately 2,800 first-year students across two waves at an economics school within a broad-based university in the Netherlands. The experiment takes place in introductory economics and econometrics courses and covers both Dutch-language and international tracks. Students are randomly assigned within their study programme to either a treatment group receiving three short, personalized instructor emails or a control group receiving standard course communication. Outcomes are measured using administrative education records and course evaluation data.

Not available

Randomization done by computer, blocked by study programme/track.

Individual student.

No

Approximately 2,800 individual students across two waves.

Approximately 2,800 first-year students across two waves.

Approximately 1,400 students assigned to treatment and 1,400 students assigned to control across two waves.

The study uses a blocked randomized design with randomization within the 2 × 2 programme/track blocks in each two waves: Economics and Econometrics, each split into Dutch-language and international tracks. We assume a two-sided test with a 5 percent significance level, 80 percent power, individual-level randomization, no clustering, and equal treatment-control allocation. The calculation assumes 8 blocks across the two waves, an average block size of 350 students, and approximately 2,800 students in total, with about 1,400 students assigned to treatment and 1,400 to control. We use a conservative specification with no additional level-1 covariates and no assumed variance explained by blocks or covariates. Under these assumptions, the minimum detectable effect size for continuous main outcomes is approximately 0.11 standard deviations. For course grades measured on a 1–10 scale, this corresponds to about 0.16 grade points, assuming a standard deviation of 1.5 grade points. For binary persistence outcomes, the minimum detectable effect depends on the baseline rate. With this planned sample size, the study can detect effects of approximately 4.6 percentage points for an outcome with a 25 percent baseline rate, 5.3 percentage points for a 50 percent baseline rate, and 4.6 percentage points for a 75 percent baseline rate. The actual MDE may be slightly smaller when including block fixed effects and baseline covariates in the analysis.