Peer effect in educational decision-making – a field experiment in Hungary
Last registered on June 11, 2018


Trial Information
General Information
Peer effect in educational decision-making – a field experiment in Hungary
Initial registration date
June 04, 2018
Last updated
June 11, 2018 4:03 AM EDT

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Research Center for Educational and Network Studies, Hungarian Academy of Sciences, Centre for Social Sciences; TÁRKI Social Research Institute, Budapest
Other Primary Investigator(s)
Additional Trial Information
On going
Start date
End date
Secondary IDs
We are conducting a large-scale randomized field experiment in Hungarian schools that randomly assign pupils to desk-mates within classrooms. We investigate how desk-mates’ parental background and baseline educational preference affect pupils’ own choice for the most demanding, college-bound secondary track called grammar school.
External Link(s)
Registration Citation
Keller, Tamas. 2018. "Peer effect in educational decision-making – a field experiment in Hungary." AEA RCT Registry. June 11.
Experimental Details
We have conducted a large-scale randomized field experiment that randomly allocate pupils to desk-mates. Our primary treatment is sitting next to a high-status desk-mate. Our research is restricted in grade 6, 7 and 8 since educational track choice at the secondary level is generally performed in grade 8 and outcome data on anticipated choices (6th and 7th) and made choices (8th grade) have been asked only in these grades.

Our primary treatment is the parental background of pupil i’s desk-mate j, j≠i, in classroom c and grade g in school s, defined as a dummy variable which =1 if either the mother or the father of i’s desk-mate j has completed the high-school final examination (érettségi), and 0 otherwise.

As a secondary treatment, we will use desk-mate’ baseline educational intention measured in 6th grade.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Track choice
Primary Outcomes (explanation)
Track choice (TCH) is assessed by the question #13 in pupil’s endline questionnaire whether pupils are planning to apply (grade 6 & 7) or have applied (grade 8) to the grammar school track. In grade 6 and 7 the question was as follows: “After finishing the primary school, are you planning to apply to the secondary grammar school track?” The answer categories were the following: “Yes”, “No”, “I don’t know”. Category “Yes” will be coded as 1, the two other categories will be coded as 0. In grade 8 the survey question was the following: “Please indicate whether or not you applied to grammar school in February 2018! If you applied to several high schools were any of these grammar schools?” The answer categories were the following: “Yes”, “No”, “I don’t remember”. Category “Yes” will be coded as 1, the two other categories will be coded as 0. The 8 graders educational choices obtained from the endline survey will be verified by their real application behavior provided by HEA. The category 1 will be used if pupil i has applied at least to one secondary grammar school, and 0 will be used otherwise. If endline educational track choices are missing in the survey, or if the survey data and HEA data disagree, the HEA data will be used.
Secondary Outcomes
Secondary Outcomes (end points)
Perceived likelihood of admission to secondary grammar school
Admission to grammar school
Secondary Outcomes (explanation)
Perceived likelihood of admission to grammar school is (PLA) assessed by the question #14 in pupil’s endline background questionnaire “Regardless of whether you did or did not apply to grammar school, do you think you would/will be accepted?” Pupils indicated their answers on an 11-grade scale from 0 (Definitely will not accept me) to 10 (Definitely will accept me).

Only for the 8 graders, we will use actual admission to grammar school (Admit) as a further outcome variable, which equals 1 if pupil i has been admitted to the grammar school track and 0 otherwise. This variable will be obtained from HEA in July 2018.
Experimental Design
Experimental Design
We execute a field experiment in 149 6th, 7th, and 8th grade classrooms

Outcomes are collected through pupil surveys between February and April 2018. Outcomes data comes from our survey and will be obtained from the Hungarian Educational Authority (HEA). Survey data will first become available to the research team in June 2018 (at the date of this submission, data did not arrive). Data on 8th graders real application decisions and actual admission will be obtained from HEA in July 2018.

Baseline variables are driven by the National Assessment of Basic Competencies (NABC) or are collected via teacher reports. Baseline variables include parental educational level, pupils’ educational intentions (for 7th and 8th graders) and baseline math and reading competency scores (for 7th and 8th graders). These variables will be obtained from the NABC. NABC measures the entire 6th-grade school cohort math and reading comprehension test with a 4×45 minutes long school-based test, which is accompanied by a take-home pupil questionnaire. Linking the data to NABC is feasible with a personalized ID. Merging data to NABC will be performed by the HEA in July 2018 for the current 7th and 8th graders. Linking current 6th graders NABC data will be performed in HEA in March 2019, since these data were collected in May 2018 and will be published in next spring. For the current 6th graders we will use only parental educational level as baseline variable.

Pupils’ social background is defined as the parents’ highest educational level and it is measured once pupils completed the 6th grade. Parental education is asked separately about the mother and about the father. Baseline educational intention (“What is the highest educational level you would like to achieve”, question #14 in the NABC background questionnaire) and NABC competency scores in math and reading comprehension will be available for 7th and 8th graders in our sample and were measured in grade 6.
Experimental Design Details
Not available
Randomization Method
Within each classroom, we assigned students to free-standing two-person desks via unconstrained random partitioning. Randomization was based on Spring 2017 class rosters, using a computer. To account for entries and exits from the Spring class roster, we stipulated a replacement algorithm and defined the "intended seating chart" as the randomized seating chart after the application of the replacement algorithm.
Randomization Unit
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
Approximately 1,700 pupils across 90 classrooms of 36 schools
Sample size: planned number of observations
Approximately 1,700 pupils
Sample size (or number of clusters) by treatment arms
This is a networked trial for peer effects. We analyze the outcomes of approximately 1,700 students. Each student is both subject in the analysis and serves as the peer for his or her desk mate.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
University of Wisconsin-Madison
IRB Approval Date
IRB Approval Number
IRB Name
Hungarian Academy of Sciences
IRB Approval Date
IRB Approval Number