Development and validation of a robot social presence measurement dimension scale

Na Chen, Xiaoyu Liu, Yanan Zhai, Xueyan HuScientific 2023 · DOI

TL;DR

This paper introduces and validates a new 5-dimensional scale to measure "robot social presence," which is the human perception of a robot's ability to exhibit human-like social behaviors and build relationships. This is crucial for designing more effective and engaging human-robot interactions, moving beyond just functional capabilities.

What Problem Does This Paper Address?

As robots become more integrated into our lives, particularly in roles requiring social interaction (e.g., elder care, customer service), understanding how humans perceive their "socialness" is critical. While "social presence" is a well-studied concept in human-human and human-computer interaction (especially in virtual environments), there's been a significant gap in defining and measuring it specifically for physical robots. Existing research often focuses on a robot's intelligence or functional capabilities, neglecting the nuanced social attributes that influence human acceptance and interaction effectiveness. This paper aims to fill that gap by developing a robust, validated scale for measuring robot social presence from a social interaction perspective.

Methods

The researchers developed a theoretical model of robot social presence based on existing theories of interpersonal and network social presence. This model was iteratively refined through a multi-stage process:

  1. Initial Model Development: Based on literature review, an initial 6-dimensional model was proposed.
  2. Expert Evaluation & User Interviews: Three experts and five users evaluated the initial model and scale items. This led to the removal of the "emotional understanding and expression" dimension due to practical limitations in current robot capabilities and human perception, resulting in a 5-dimensional modified model.
  3. Pilot Questionnaire Survey (Study 3): A 19-item questionnaire based on the modified model was administered to 172 participants. Exploratory Factor Analysis (EFA) was used to assess construct validity and refine the scale.
  4. Final Questionnaire Design & Validation (Study 3): After further refinement based on EFA and validation factor analysis, a 17-item, 5-dimensional scale was finalized. This final scale was administered to 494 participants to confirm its reliability and validity.

The validation process involved various statistical analyses, including Cronbach's alpha for reliability, Kaiser-Meyer-Olkin (KMO) and Bartlett's sphericity tests for suitability of factor analysis, and model fit indices (e.g., χ²/df, RMSEA, CFI, TLI) to assess the overall fit of the proposed model.

Datasets

The study utilized data from two questionnaire surveys:

  • Pilot Survey: 172 valid questionnaires were collected.
  • Final Survey: 494 valid questionnaires were collected.

These datasets were used for the validation of the developed robot social presence scale. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Robot Social Presence is Distinct: The study highlights that robot social presence isn't just about mimicking human behavior but about the human perception of that behavior in a social context. It's a subjective experience.
  • Emotional Expression is Less Critical (Currently): A key insight was the removal of "emotional understanding and expression" as a primary dimension. Experts and users indicated that current robots' limited emotional capabilities and human acceptance levels make this dimension less impactful for measuring social presence compared to other factors. This suggests that for current robotic technology, functional and interactive aspects are more salient for perceived socialness.
  • A Multi-Dimensional Construct: Social presence isn't a single trait. The research confirms it's composed of distinct, measurable dimensions: presence, attention allocation, interactive expression and information understanding, perceived emotional interdependence, and interaction behaviour perception.
  • Practical Tool for Design: The developed 17-item scale provides a concrete, validated tool for researchers and developers to assess and improve the social aspects of robots, guiding design choices for better human-robot collaboration.

Key Results

The study successfully developed and validated a 5-dimensional, 17-item scale for measuring robot social presence.

  • Model Fit: The final model demonstrated good reliability, high inter-factor correlation, high model fit (e.g., RMSEA = 0.048, CFI = 0.966), high internal consistency of dimensions, and high convergent validity.
  • Reliability: Cronbach's alpha values for the scale and its dimensions were consistently high (e.g., overall alpha = 0.900), indicating excellent internal consistency.
  • Dimensions: The final scale measures:
  • Presence: The perception of being with the robot.
  • Attention Allocation: How attention is distributed and perceived between human and robot.
  • Interactive Expression and Information Understanding: The perceived clarity and mutual understanding in communication.
  • Perceived Emotional Interdependence: The degree to which emotions and attitudes influence each other.
  • Interaction Behaviour Perception: The perceived influence of each other's behaviors.

Limitations and Future Work

  • Sample Demographics: The authors acknowledge the need for a larger and more diverse sample in future research, including a wider range of age groups and educational levels.
  • Cross-Cultural Validation: The study notes that cultural differences, particularly between Asian and Western perspectives on technology, might influence the interpretation of the scale. Future work should include cross-cultural validation.
  • Measuring Emotional Interdependence: The authors suggest that further research might require specific experimental sessions to accurately measure the effect of emotional dependence on robot social presence in human-computer interaction.
  • Refining Interaction Dimensions: The paper mentions that the "interactive expression and information comprehension" dimension could be further refined with more targeted questions in future studies.

Summary updated

Abstract

Robots that can exhibit human-like behaviour, build social relationships with humans, and carry out social interaction that can be considered to have a social presence. Measuring the social presence of robots can provide an important basis for optimizing the design of robots' human-like behaviour, but until now, there has been no research and discussion on the presence of robots from the perspective of social interaction. Based on related studies, such as the theory of the presence of interpersonal interaction society, the mechanism of network social presence measurement and human-computer interaction, this study constructs a theoretical model of robot social presence, and develops corresponding measurement scales with five dimensions, namely, the presence, attention distribution, interactive expression and information understanding, perceived emotional interdependence, and interaction behaviour perception. The 5-dimensional robot social presence model was finalized, and a 17-question questionnaire scale was developed. The validation of the model and the development of the scale includes one expert assessment (involving three experts), one user interview (involving five interviewees), and two questionnaire surveys (involving 172 valid questionnaires and 494 valid questionnaires respectively). The final model shows good reliability of the measurement questionnaire, high inter-factor correlation, high model fit, high internal consistency of the dimensions, high reliability of the scale, and high convergent validity of all dimensions. This study provides a theoretical basis for the study of robot social presence and human-robot collaboration and provides a reference measurement tool for social robot-related development and research.
BibTeX
@inproceedings{chen2023development,
  author = {Na Chen and Xiaoyu Liu and Yanan Zhai and Xueyan Hu},
  title = {{Development and validation of a robot social presence measurement dimension scale}},
  booktitle = {Scientific},
  year = {2023},
  doi = {10.1038/s41598-023-28817-4},
}
Figure