Interface Dis/Similarities: Investigating Characteristics Influencing Perceived Differences Between GUIs

Paper Title

Interface Dis/Similarities: Investigating Characteristics Influencing Perceived Differences Between GUIs

Publication Info

• Topic area: Human-Computer Interaction (HCI), focusing on user perception of graphical user interfaces (GUIs).
• Keywords: GUI similarity, interface perception, analogical reasoning, user experience, interface design, visual complexity, clustering, modularity, reading flow, feature recognition.

Background and Problem

• Problem / challenge: Existing computational methods for assessing GUI similarity are disconnected from user perception, and designers lack tools to anticipate how interface changes will be perceived. Prior work does not identify which characteristics users rely on to judge similarity or difference between GUIs.
• Significance: Understanding GUI dis/similarities is critical for improving onboarding, redesign, migration, and cross-application knowledge transfer. It helps designers create interfaces that align with user expectations and facilitate analogical reasoning.
• Motivation and related work: Prior research has explored computational metrics for GUI similarity and translation systems for interface adaptation but has not addressed perceptual characteristics influencing user judgments. This paper builds on HCI studies of mental models, visual complexity, and analogical reasoning to close this gap.

Solution

• Proposed approach: A mixed-methods study identifying and quantifying the interface characteristics that influence perceived dis/similarities between GUIs, using card sorting, pairwise comparisons, and scaling methods.
• Novelty:
  1. Identification of seven intrinsic and three contextual characteristics that shape user perception of GUI dis/similarities.
  2. Quantification of the relative importance of these characteristics using Just-Objectionable-Differences (JOD) scaling.
  3. Integration of perceptual findings into a framework for cross-interface knowledge transfer and design implications.
• Procedure and key techniques:
  1. Experiment 1: Open card sorting task to identify intrinsic characteristics of GUIs.
  2. Experiment 2: Online pairwise comparison survey to assess the influence of contextual characteristics.
  3. Experiment 3: Quantification of the relative impact of intrinsic and contextual characteristics using JOD scaling.

Results

• Concrete findings:
  • Seven intrinsic characteristics identified: ColorScheme, Modularity, Clustering, ReadingFlow, Spatialization, FeatureRecognition, VisualComplexity.
  • Three contextual characteristics identified: OperatingSystem, ApplicationContext, Language.
  • Structural characteristics (Clustering, Modularity, ReadingFlow, FeatureRecognition) exert the most significant influence on perceived dis/similarities, with JOD scores above 3.0.
  • Contextual characteristics (OS, Language, ApplicationContext) have a lesser impact, with JOD scores below 2.4.
• Advantage over baselines: Provides a perceptually grounded framework for understanding GUI dis/similarities, addressing gaps in computational similarity metrics and offering actionable insights for design.
• Experiments / evaluation:
  • Experiment 1: Card sorting with 22 participants analyzing 30 productivity software GUIs.
  • Experiment 2: Pairwise comparisons with 60 participants evaluating contextual characteristics across 55 pairs of interfaces.
  • Experiment 3: Pairwise comparisons with 85 participants assessing intrinsic and contextual characteristics across 153 pairs of interfaces.
  • JOD scaling used to quantify perceived differences, with thresholds for statistical significance.
• Limitations and future work:
  • Static screen captures used; dynamic interactions were not studied.
  • Expertise effects were not fully stratified; future work could explore how experience modulates perception.
  • Visual complexity excluded from Experiment 3 due to overshadowing effects; future research could model its impact during interaction.

Summary

This paper identifies and quantifies the characteristics that influence users' perception of dis/similarities between GUIs, combining card sorting, pairwise comparisons, and JOD scaling. Structural characteristics such as Clustering, Modularity, and ReadingFlow dominate user judgments, while contextual factors like OS and Language have a lesser impact. These findings clarify how users form analogical mappings between interfaces, supporting cross-interface knowledge transfer and informing design practices. Future work could extend these insights to dynamic interfaces and generative user interface systems.