[Paper Review] Anchored in a Data Storm: How Anchoring Bias Can Affect User Strategy, Confidence, and Decisions in Visual Analytics

Ryan Wesslen, Sashank Santhanam+4 more|arXiv (Cornell University)|Jun 7, 2018

Data Visualization and Analytics3 citations

TL;DR

This study investigates how visual anchors and strategy cues in visual analytics systems influence user decision-making, confidence, and performance in identifying Twitter news misinformation. Using two rounds of controlled lab experiments with 94 participants, it finds that visual anchors significantly affect user confidence and behavior, but have mixed effects on accuracy and speed, highlighting the need for careful training design to avoid unintended bias.

ABSTRACT

Cognitive biases have been shown to lead to faulty decision-making. Recent research has demonstrated that the effect of cognitive biases, anchoring bias in particular, transfers to information visualization and visual analytics. However, it is still unclear how users of visual interfaces can be anchored and the impact of anchoring on user performance and decision-making process. To investigate, we performed two rounds of between-subjects, in-laboratory experiments with 94 participants to analyze the effect of visual anchors and strategy cues in decision-making with a visual analytic system that employs coordinated multiple view design. The decision-making task is identifying misinformation from Twitter news accounts. Participants were randomly assigned one of three treatment groups (including control) in which participant training processes were modified. Our findings reveal that strategy cues and visual anchors (scenario videos) can significantly affect user activity, speed, confidence, and, under certain circumstances, accuracy. We discuss the implications of our experiment results on training users how to use a newly developed visual interface. We call for more careful consideration into how visualization designers and researchers train users to avoid unintentionally anchoring users and thus affecting the end result.

Motivation & Objective

• To investigate how visual anchors and strategy cues impact user behavior, confidence, and decision accuracy in visual analytics for misinformation detection.
• To understand the role of cognitive biases—particularly anchoring bias—in exploratory visual analysis with coordinated multiple views (CMV) systems.
• To evaluate the effects of different training methods (e.g., scenario videos, strategy cues) on user interaction patterns and performance metrics.
• To explore how user interaction logs can reveal underlying strategies and susceptibility to anchoring effects.
• To inform best practices for training users on visual analytics tools to prevent unintentional anchoring and its downstream consequences.

Proposed method

• Conducted two rounds of between-subjects, in-lab experiments with 94 participants using a visual analytics system (Verifi) for detecting misinformation on Twitter.
• Designed three treatment groups: control (no anchor), strategy cue group (guided approach via video), and visual anchor group (video emphasizing specific views as primary).
• Collected and analyzed interaction logs to track user navigation paths, view usage, time-on-task, and data coverage across views.
• Measured user confidence via post-decision self-reports and assessed accuracy against ground truth labels for news accounts.
• Integrated psychological theories of anchoring bias into the experimental design to simulate real-world cognitive influences on visual analysis.
• Used quantitative analysis to compare performance across treatment groups on metrics including accuracy, speed, confidence, and view utilization.

Experimental results

Research questions

• RQ1 How do visual anchors and strategy cues affect user confidence in decisions made during visual analytics tasks?
• RQ2 To what extent do visual anchors influence user interaction strategies and view selection in coordinated multiple view (CMV) systems?
• RQ3 What is the impact of visual anchoring on decision accuracy and task completion time in misinformation detection tasks?
• RQ4 How do individual differences (e.g., experience, cognitive style) moderate susceptibility to anchoring effects in visual analytics?
• RQ5 Can interaction log analysis reveal patterns of anchoring and help identify users who are more or less influenced by visual cues?

Key findings

• Visual anchors and strategy cues significantly increased user confidence in their decisions, even when accuracy was not improved.
• Participants in the visual anchor group spent significantly more time focusing on the highlighted view (e.g., Tweet Panel), indicating strong reliance on the anchor.
• Accuracy was not consistently improved by anchoring; in some cases, it decreased, especially when users ignored other relevant views.
• The strategy cue group showed faster decision-making and better data coverage across views compared to the control and visual anchor groups.
• Some users were unaffected by the treatments, suggesting individual differences such as prior experience or cognitive ability may reduce susceptibility to anchoring.
• Interaction log analysis revealed distinct strategies: anchored users followed a rigid path, while non-anchored users explored more flexibly across views.