Does a Picture Paint a Thousand Words? Using Visual and Textual Channels to Understand Attitudes and Beliefs

Paper Title

Does a Picture Paint a Thousand Words? Using Visual and Textual Channels to Understand Attitudes and Beliefs

Publication Info

• Topic area: Understanding user attitudes and beliefs through visual and textual elicitation methods.
• Keywords: Human-Computer Interaction, visual elicitation, textual elicitation, attitudes, beliefs, free-form drawing, thematic analysis, survey design, visualization, user studies.

Background and Problem

• Problem / challenge: Existing elicitation methods either prioritize analyzability (e.g., Likert scales) or expressiveness (e.g., open-ended text), but fail to balance these needs. Structured methods often oversimplify complex attitudes, while expressive methods are difficult to analyze systematically. Additionally, prior visual elicitation techniques rely on predefined scales, limiting understanding of how users naturally represent their attitudes and beliefs.
• Significance: Understanding how users mentally represent attitudes and beliefs can inform the design of elicitation methods that are both expressive and analyzable, improving data collection in HCI and related fields.
• Motivation and related work: Prior work has explored structured and visual elicitation methods, such as sliders, emoji-based surveys, and interactive diagrams, but these rely on researcher-defined constructs. Cognitive psychology and visualization research suggest potential benefits of free-form drawing and diagrams for capturing user perspectives, yet these approaches remain underexplored for eliciting attitudes and beliefs.

Solution

• Proposed approach: A two-round qualitative study using free-form drawing and textual descriptions to explore how users visually and textually represent their attitudes and beliefs.
• Novelty:
  1. Identification of basic visual elements (e.g., pictorials, facial expressions, directional symbols) and their combinations in user-generated drawings.
  2. Comparison of visual and textual representations to uncover gaps and complementarities.
  3. Design of speculative visual elicitation techniques inspired by participant strategies, including adjustable facial expressions, multi-slider interfaces, and pictorial diagrams.
  4. Exploration of how attitudes and beliefs differ in visual representation across topics and cultural contexts.
• Procedure and key techniques:
  • Conducted a two-round study with 41 US-based participants using tablet devices.
  • Participants expressed attitudes and beliefs on four topics (e.g., vaccination, generative AI) through free-form drawing and textual descriptions.
  • Thematic analysis was applied to identify visual and textual patterns, with inter-rater reliability assessed and codebooks developed.
  • Findings informed the design of speculative visual elicitation prototypes.

Results

• Concrete findings:
  • Identified five basic visual elements (pictorials, facial expressions, posture, text, symbols) and six composition strategies (e.g., stories, diagrams, equations).
  • Textual responses were more effective for expressing nuanced attitudes (e.g., hedging, conditional beliefs), while visual responses excelled at conveying core attitudes and causal relationships.
  • Participants preferred textual responses in 84.1% of cases, citing their ability to articulate complexity.
• Advantage over baselines: The study highlights the potential of free-form drawing to complement textual methods by capturing visual mental constructs, offering richer insights than structured scales or predefined visual tools.
• Experiments / evaluation:
  • Two rounds of qualitative studies with 41 participants (N1=20, N2=21).
  • Topics included vaccination, education spending, generative AI, and bicycle usage.
  • Visual and textual responses were analyzed using thematic coding, with inter-rater reliability scores of 0.79 (round 1) and 0.44 (round 2).
• Limitations and future work:
  • Participants’ drawing skills and graphical literacy influenced the clarity of visual responses.
  • Cultural specificity of findings (US-based participants) limits generalizability.
  • Future work should validate proposed elicitation techniques, explore cross-cultural differences, and address task-order effects.

Summary

This paper investigates how users visually and textually represent attitudes and beliefs through a qualitative study involving free-form drawing and textual descriptions. The findings reveal key visual elements and composition strategies, as well as the complementary strengths of visual and textual channels. Inspired by these insights, the authors propose speculative visual elicitation techniques, such as adjustable facial expressions, multi-slider interfaces, and pictorial diagrams, to balance expressiveness and analyzability. While the study highlights the potential of visual elicitation, future work is needed to validate these techniques and explore their applicability across diverse cultural contexts.