This open access book demonstrates how human biases affect the process of visual data analysis, a subject which has typically been left to researchers in cognitive and perceptual psychology and the social sciences. Human biases affect the way that people interpret and experience the world and how they operate within it and make decisions. These can include cognitive biases such as confirmation or anchoring bias, perceptual biases including visual or auditory illusions, and implicit biases such as racial or gender bias that are often borne of harmful cultural norms and stereotypes. In the context of visual data analysis, this book explores (1) what these biases are, (2) how to characterize them, and (3) how to mitigate them through designing digital interventions. This book synthesizes years of work on detecting and mitigating biases in visual data analysis and project directions for the next decade of research and practice. It represents an accessible entry point to understanding the prevalence of biases in computing before taking readers on a deeper dive into empirical studies on the efficacy of various bias mitigation interventions. It will synthesize years of research into a digestible portal to technical work on visual data analysis. Data scientists and citizens alike can benefit from this book by reflecting on their own unique privileges and susceptibility to biases and scrutinizing how digital interventions, sometimes as simple as adding one extra step to verify the decision by checking "yes," might be integrated or enacted in their own personal and professional decision making settings.
This is an open access book. It demonstrates how human biases affect the process of visual data analysis, a subject which has typically been left to researchers in cognitive and perceptual psychology and the social sciences. Human biases affect the way that people interpret and experience the world and how they operate within it and make decisions. These can include cognitive biases such as confirmation or anchoring bias, perceptual biases including visual or auditory illusions, and implicit biases such as racial or gender bias that are often borne of harmful cultural norms and stereotypes. In the context of visual data analysis, this book explores (1) what these biases are, (2) how to characterize them, and (3) how to mitigate them through designing digital interventions. This book synthesizes years of work on detecting and mitigating biases in visual data analysis and project directions for the next decade of research and practice. It represents an accessible entry point to understanding the prevalence of biases in computing before taking readers on a deeper dive into empirical studies on the efficacy of various bias mitigation interventions. It will synthesize years of research into a digestible portal to technical work on visual data analysis. Data scientists and citizens alike can benefit from this book by reflecting on their own unique privileges and susceptibility to biases and scrutinizing how digital interventions, sometimes as simple as adding one extra step to verify the decision by checking "yes," might be integrated or enacted in their own personal and professional decision making settings.
In addition, this book:
- Demonstrates the impact and implications of human biases on visual analytics
- Explains how readers can identify and mitigate biases in their own data analysis
- Explores the implications of unconscious biases in data and by users for analysis and decision making
