This repository contains a statistical analysis project completed during my undergraduate studies at the University of Toronto in Winter 2022.
The project investigates how increasing visual distraction affects participant accuracy and reaction time in a visual spot-the-difference task, using classical statistical modeling and hypothesis testing in R.
Participants were shown pairs of images under varying levels of visual distraction and asked to identify differences within a fixed time window. The goal was to quantify how distraction influences:
- Identification accuracy
- Reaction time
- Performance trends across difficulty levels
The analysis focuses on measuring correlation, fitting regression models, conducting ANOVA tests, and validating assumptions through diagnostic plots.
- Language: R
- Visualization: ggplot2
- Statistical techniques:
- Pearson correlation tests
- Linear regression
- ANOVA
- Residual diagnostics and model validation
- InputModeling.R — Data ingestion, preprocessing, and exploratory calculations
- StatsTesting.R — Statistical modeling, hypothesis testing, regression, and ANOVA
- PDF report — Final written analysis with methodology, results, and interpretation
- README.md — Project documentation
- Does increasing visual distraction significantly reduce participant accuracy?
- Is reaction time positively associated with difficulty level?
- Are the observed relationships statistically significant?
- Do linear model assumptions hold for the collected data?
If I were extending this project today, I would explore:
- Increasing the sample size and number of repeated trials
- Randomizing experiment order to reduce learning effects
- Mixed-effects or hierarchical models to account for participant-level variation
- Bootstrapped confidence intervals for regression estimates
- Non-linear relationships between distraction and performance