Design Factors and Correlation Perception in Scatterplots.

Humans systematically under-estimate the relationship (correlation) between two variables when looking at scatterplots.

This project is investigating how to re-design scatterplots, such that people perceive correlations more accurately. This is important to help support communication of information, and in particular scientific results, and improve decision-making from data. The research asks:

- If we systematically manipulate aspects of data presentation, such as the relative size or contrast of data points, can we improve the accuracy with which people judge correlation?
- Do changes in data presentation affect decision-making?

The project will develop a framework for the rapid, inexpensive, and well powered study of data visualisations through crowdsourcing. Best practice in open science will be used throughout.

This work informs the design of data visualisations through empirical study. It also addresses how visual features can affect perception, and more specifically how we understand relatedness and the graphical display of it.

Relevant research areas: Human-Computer Interaction, Graphics and Visualisation