This repository contains all the R code used to answer the research question: To what extent can students “recover” from mistakes made in earlier stage(s) and still reach an estimate close to their true ability? I did this project for an internship at Cito, from 10.1.2022 to 18.2.2022 as part of an elective for my Research Master's Programme.
This repository contains the following folders:
- Data: Contains both the MST design rules, needed to repeat the simulation, as well as all the data produced by the simulations and the analyses presented in the report;
- Functions: Contains the functions for calculating the reference scores of each subject, as well as a function that classifies students in one of the classes of secondary education
- Proposal: Contains the proposal for the internship sent to the University;
- Report: Contains the markup manuscript for the report of this project;
- scripits_mistakes: Contains the
Rscripts written to calculate the number of mistakes in Modules A and on Day 1; - Simulation 1: Contains the code needed to produce the simulated MST responses, and the code for the analyses presented in the report for the first simulation;
- Simulation 2: Contains the code needed to produce the simulated MST responses, and the code for the analyses presented in the report for the first simulation.
Note: In the first analyses of Simulation 1, Welch's tests are performed to test whether the mean of the ability re-estimates for each level of ability differ significantly from ther true theta. Theoretically, instead of a Welch's test a one-sample t-test should've been perfodmed. However, practially, both analyses yield the same results. A proof of this has been added to the welch-tests.R script.