Visualization According to Statisticians: An Interview Study on the Role of Visualization for Inferential Statistics

Area of Science:

• Statistics
• Data Visualization
• Human-Computer Interaction

Background:

• Data visualization is integral to statistical analysis.
• Understanding statisticians' use of visuals can inform best practices for sensemaking.
• Few studies explore how statisticians mentally model and visually represent statistical inference.

Purpose of the Study:

• To investigate statisticians' utilization of data visualization in their analytical processes.
• To explore statisticians' mental models of inferential statistical methods.
• To gather statisticians' design recommendations for visualizing statistical inferences.

Main Methods:

• Conducted interviews with 18 professional statisticians (average 19.7 years experience).
• Elicited participant-generated visual designs for statistical inference.
• Analyzed interview transcripts using thematic analysis and open coding.

Main Results:

• Statisticians employ visualization across all analytical phases, not solely for reporting.
• Mental models of inferential statistics are predominantly visual.
• Many statisticians prefer nuanced representations over dichotomous (e.g., significant/non-significant) outcomes.

Conclusions:

• Visualization is a critical tool for statisticians throughout the entire analytical workflow.
• Visually-based mental models highlight opportunities for improved statistical representations.
• Multi-faceted visual displays incorporating effect sizes and uncertainty can enhance statistical inference communication.