dimits-ts / college_analysis

A statistical study about US college admissions, featuring a full report in LaTeX.

A Statistical Study on Test Scores in US College Admissions

A statistical study on US college admissions, featuring a full report in LaTeX.

Abstract

US College Admissions have and continue to be a subject of great debate among scholars and analysts. Such educational institutions have an interest in selecting the most qualified applicants using limited data, while the applicants themselves often protest admission requirements, especially those deemed discriminatory in nature. In this study we investigate how admission test scores can be explained by various candidate traits and prior performance. We discover a relationship between the candidate's gender and previous program and their overall test scores. We also identify a positive correlation between test scores, which we attribute to a confounding variable which we name "Competence".

Data Analysis

The analysis has been implemented in the R programming language. We use it to generate graphs and figures directly, as well as dynamic LaTeX tables used in the final report.

Our analysis primarily uses traditional statistical techniques such as t-tests, ANOVA, ANCOVA and OLS regression models. We use these tools to test our hypotheses and rule out alternative explanations.

The report

The report is written entirely in LaTeX and exclusively uses dynamically generated figures from our R script. It should be understandable to readers with no prior experience in Statistics.