Why Do People Gather? A Study on Factors Affecting Emotion and Participation in Group Chats

Abstract

Group chat socialization is increasingly central to online activities, yet design strategies to enhance this experience remain underexplored. This study builds on the Stimuli–Organism–Response (SOR) framework to examine how usability, chat rhythm, and user behavior influence emotions and participation in group chats. Using data from 546 users in China, a relevant demographic given the dominance of platforms like WeChat in both social and professional settings, we uncover insights that are particularly applicable to highly connected digital environments. Our analysis shows significant relationships between usability (γ = 0.236, p < 0.001), chat rhythm (γ = 0.172, p < 0.001), user behavior (γ = 0.214, p < 0.001), and emotions, which directly impact participation. Positive emotions (γ = 0.128, p < 0.05) boost participation, while negative emotions (γ = −0.144, p < 0.01), particularly when linked to user behaviors, reduce it. Additionally, we discussed the mediating effects, notably that usability significantly impacts participation through positive emotions, while user behavior exerts a significant influence on participation through negative emotions. This research offers actionable design strategies, such as tailoring sensory inputs to reduce cognitive load and implementing reward systems to motivate participation. Positive feedback mechanisms enhance engagement by leveraging the brain’s reward systems, while optimized error messages can minimize frustration. These insights, which are particularly relevant for China’s active group chat culture, provide a framework to improve platform design and contribute valuable findings to the broader HCI field.

1. Introduction

In the globalized information age, social media platforms have become crucial channels for communication and socialization. A detailed analysis by the Kepios team revealed that, as of October 2023, there are 4.95 billion social media users globally, equivalent to 61.4% of the world’s total population. According to the GWI report, the average user globally spends 2 h and 24 min per day on social media, amounting to approximately 15% of their waking hours. Among the active social platforms facilitating group chats are WhatsApp, YouTube, Discord, and Slack (Kepios). Group chats represent the most intensive form of social interaction within online communities, allowing users to form virtual communities and engage in real-time communication and sharing. A well-designed group chat not only enhances communication experiences but also encourages sustained user participation, thereby enhancing the vitality of the social platform.

Group chat participation focuses on conversational engagement, indicating the extent to which participants are immersed and involved in ongoing social interactions [1]. High levels of participation enhance the perceived quality of social interactions [1,2], whereas low participation levels may lead to negative perceptions of peers and dissatisfaction with conversations. Group chats are prevalent in everyday life and not only have socialization purposes but also serve transactional functions [3]. For instance, group chats serve as effective channels for decision-making, problem-solving, and open debates on specific issues [4]. Group chats offer the advantage of asynchronous communication and maintain a traceable chat log, enabling users to stay informed about ongoing discussions [5], facilitating comfortable group communication among members across different locations and schedules [6] without disrupting their daily routines [7].

Despite the positive effects of group chat socialization on individuals and society, research on the underlying psychological and social dynamics remains relatively limited. The existing research primarily focuses on general trends in social media usage and user behavior patterns, lacking a systematic exploration of the factors influencing group chat participation and insufficiently examining interaction mechanisms, emotional experiences, and their impact on participation specific to the group chat environment.

The components of the SOR model—Stimuli (S), Organism (O), and Response (R)—correspond closely with the key elements of this research: platform environment, user emotions, and user feedback. Furthermore, the SOR theory serves as a comprehensive framework that helps connect the currently fragmented research by integrating platform environment, user experience, and user feedback into a cohesive whole. The SOR theory offers a robust framework for understanding how environmental stimuli affect individual behavioral responses through internal biological processes, making it applicable to analyses of user interactions and emotional responses in group chat environments. A comprehensive understanding of online group chat experience factors, obtained through the SOR mechanism, can aid system builders and designers in developing effective design strategies and offer new insights for the advancement of group chat socialization.

2. Literature Review

2.1. The Stimulus–Organism–Response Theory

The SOR model, derived from environmental psychology, posits that environmental aspects serve as stimuli (S) influencing an individual’s internal state (O), which subsequently drives their behavioral responses (R) [8]. The model suggests that external environmental factors and conditions influence the emotions of the organism, thereby motivating addictive behaviors. It elucidates that pairs of stimuli and external factors reinforce a person’s internal state [9]. The term “organism” refers to the internal state encompassing perception, sensation, and cognitive processes [10]. Previous research has identified these structures as having both positive and negative effects [11,12]. Finally, individuals make a final choice and select their behavioral responses accordingly [8]. In this theoretical framework, stimuli can represent the factors influencing the experience of a social platform, organisms denote the emotional state of the user, and responses signify the reactions of group chat participants to subsequent participation in the group chat. This implies that elements of the experience within the group chat environment stimulate users to generate positive or negative emotions, influencing their participation.

2.2. Interaction Experience in Group Chat

Numerous studies have examined factors influencing the social experience in group chats. For instance, there has been research on online group chats serving business goals, such as Van Dolen et al.’s [13] exploration of the effects of perceived technological attributes (control, enjoyment, reliability, and speed, as well as ease of use) and chat group characteristics (participation, similarity, and acceptance) on customer satisfaction, alongside the moderating role of the advisor’s communication style, from the perspective of structuration theory. Additionally, there have been studies on the necessary characteristics facilitating teamwork in group chats, like Markman’s [14] empirical study summarizing content similarity, negotiation availability, and the need to satisfy burst-like synchronous chats with occasional asynchronous transmissions in work group chats for off-site collaboration. Furthermore, some scholars have focused on the role of chatting styles, such as their effect on mood [15]. However, there is a paucity of research focusing on facilitating a better group chat experience.

This study defines three factors of group chat experience as follows: platform usability, chat rhythm, and user behavior. The definition of a group chat encompasses not only chat rooms but also online communities requiring authentication and comment sections where users can interact with each other on the same post or topic.

2.2.1. Usability

Usability, an established concept in the human–computer interaction (HCI) literature [16], refers to the ease and efficiency with which a system can be used to fulfill tasks [17], as well as the quality of the system that renders it acceptable to users [18]. High usability is crucial for attracting users to a system [19], as usability has been found to correlate positively with increased system usage [20].

Furthermore, usability encompasses the pragmatic aspects of a product or system and is a critical determinant of software success [21]. Good usability enhances users’ creativity, productivity, and overall satisfaction, while poor usability can lead to frustration, as well as wasting time, effort, and resources. Kim et al. [22] expanded the focus on usability to include not only usefulness or ease of use but also the role of emotions in ICT systems [23]. Hassenzahl [24] argues that usability should encompass both “work-related issues” (e.g., validity) and “non-instrumental human needs and related phenomena” (e.g., emotions, feelings, moods, and experiences), reflecting both cognitive and affective concerns.

Usability also leads to more positive emotions over time, as indicated by the peak experience set, demonstrating the intertwined nature of usability and user experience. Validity is essential for fulfilling user goals, with positive outcomes often associated with positive user experiences, such as feeling effective or receiving positive social feedback. Furthermore, good usability is particularly valuable over time as it supports a positive self-image [25]. Different emotional patterns were observed when testing well-designed and poorly designed systems, with the well-designed version resulting in more positive subjective feelings [26].

More specifically, low usability is often associated with complex operational processes, hard-to-navigate interfaces, and unclear feedback mechanisms. These issues can make it difficult for users to complete tasks efficiently, leading to emotional frustration [27]. MA Kuhail et al. [28] found that when users encounter obstacles or experience confusion during their interactions, their emotions quickly turn negative, resulting in a decline in their trust in the platform. Additionally, poor usability has a significant impact on users’ long-term engagement. After facing subpar usability experiences, users tend to abandon the platform in favor of alternatives that offer better usability [28]. This not only diminishes user satisfaction but also leads to a reduction in the platform’s user retention rate. Therefore, the following hypotheses are proposed.

Based on the aforementioned literature, we propose the following hypotheses:

H1a: 

Usability has a positive influence on positive emotions.

Rationale: High usability reduces cognitive load and enhances user satisfaction, leading to positive emotional responses [29].

H1b: 

Usability has a negative influence on negative emotions.

Rationale: Usability issues can lead to frustration and dissatisfaction, thereby increasing negative emotional responses [21].

H1c: 

Usability has a positive influence on participation.

Rationale: When users find a system easy to use and efficient, they are more likely to engage with it and actively participate [25].

2.2.2. Chat Rhythm

Chat rhythm is a critical component in managing conversational turns and maintaining engagement in group chats. It includes various elements, such as the use of emojis, sentence-final particles, and the overall pace of conversation. These elements collectively influence the flow and emotional tone of online interactions.

(1)

Emojis

Emojis serve as digital gestures that convey emotional nuances and compensate for the lack of non-verbal cues in text-based communication. Gawne and McCulloch (2019) [30] argue that emojis function similarly to facial expressions and body language in face-to-face communication, helping to clarify the sender’s emotional state and intention. For instance, the use of a smiling emoji can enhance the perception of a positive tone in a message, while a sad emoji can convey sympathy or concern.

(2)

Sentence-final Particles

Sentence-final particles (SFPs), also known as modal particles, are used to convey the speaker’s attitude and emotional tone in a sentence. In Chinese, particles like “吧” (ba), “啊” (a), and “呢” (ne) are used to soften statements, show politeness, or indicate a question. Chang (1994) [31] and Lee-Wong (1998) [32] highlight that these particles help to moderate the intensity of the message and provide subtle cues about the speaker’s intention and emotional state.

(3)

Conversational Pace

The overall pace of conversation, including the timing of responses and the length of messages, also contributes to the chat rhythm. Tannen (2007) [33] notes that the rhythm and repetition in conversation can enhance coherence and engagement among participants. A balanced pace, where responses are neither too rapid nor too slow, can create a comfortable and engaging communication environment. For instance, Jones (2013) [34] found that a well-paced conversation encourages active participation and reduces the likelihood of misunderstandings.

Therefore, the definition of group chat preference in this study encompasses emotional expression, rhythm of utterance, and attitude during conversation. These elements serve not only as the means of communication in group chats but also as key factors in creating the group’s atmosphere and communication effects. Group chat rhythm elements transcend superficial communication tools, significantly influencing the emotional state of group members. The choice of intonation, use of emoticons, length and density of sentences, and interactive attitude of users in the group chat all play a significant role in shaping the group chat experience.

In summary, the following hypotheses are proposed regarding the relationship between group chat rhythm, online emotions, and participation.

H2a: 

Chat rhythm has a positive influence on positive emotions.

Rationale: The effective use of chat rhythm elements, such as emojis and sentence-final particles, enhances the emotional tone and user engagement in group chats [30,31,32].

H2b: 

Chat rhythm has a negative influence on negative emotions.

Rationale: The inappropriate use of chat rhythm elements, such as the overuse of emojis or poorly timed responses, can lead to misunderstandings and frustration, increasing negative emotional responses [34].

H2c: 

Chat rhythm has a positive influence on participation.

Rationale: A well-managed chat rhythm fosters a more engaging and interactive environment, encouraging users to actively participate [33,34].

2.2.3. Online User Behavior

Numerous studies have attempted to categorize users’ online behaviors, with each offering different perspectives and emphasizing various aspects of user interaction. A detailed examination of these classifications helps to identify their strengths, limitations, and applicability to our study.

(1)

Benevenuto et al. (2009) [35] classified online activities into categories such as search, scrapbooking, messaging, awards, videos, photos, profiles, friends, communities, and others. This comprehensive categorization captures a wide range of user activities on social networks, providing a holistic view of user engagement. However, this approach may become outdated as new technologies and functions emerge, necessitating constant updates and making it challenging to consistently apply this approach across different contexts.

(2)

Dolan et al. (2015) [36] proposed a categorization focusing on social media engagement behaviors on fan pages, distinguishing between active engagement behaviors (creating, contributing, and destructing) and passive engagement behaviors (consuming, dormancy, and detaching). This categorization emphasizes the user’s level of engagement, providing insights into how different behaviors contribute to overall participation. Its limitation lies in its narrow focus on fan pages, which may not fully represent the diversity of user behaviors across various platforms.

(3)

Yan et al. (2024) [37] divided users’ online behaviors into four categories: shaping behaviors, co-acting behaviors, seeking behaviors, and isolating behaviors (

Table 1. Classification of online user behavior.

Behavior Types	Main Purposes	Function Definition	Function Cases
Shaping behavior	Idealized Performance	For self-image building	Post personal status
	Covering up shortcomings		“beautify” to look better
Co-acting behavior	Mystified performance	For interaction or participatory	Comment
	Controlling social distance		Join super (hot) talk
Seeking behavior	Misinterpreted performance	For finding other users or information	“Shake” to add new friend
	Seeking profit		“Drift” Bottle to match new friend
Isolated behavior	Remedial performance	For controlling the level of exposure of personal information	Visibility settings
	Self-protection		Comments prohibited

). Each type corresponds to different purposes and needs, accommodating new features while addressing users’ varying preferences.

Shaping Behaviors: These involve self-presentation activities, such as creating and curating content to project a desired image. Users engaging in shaping behaviors often aim to influence how others perceive them online.

Co-acting Behaviors: These include interactive activities, such as participating in discussions, collaborating on projects, and engaging with others’ content. Co-acting behaviors foster a sense of community and collaboration.

Seeking Behaviors: These involve activities focused on information acquisition, such as searching for content, reading articles, and watching videos. Users engaging in seeking behaviors prioritize gaining knowledge and staying informed.

Isolating Behaviors: These include actions where users limit their interactions, such as lurking, avoiding participation, or setting privacy controls. Isolating behaviors often reflect a desire for privacy or disengagement from social interactions.

This classification is particularly relevant to our study as it encompasses a broad spectrum of user behaviors, capturing the nuances of how individuals interact with group chat platforms.

Consequently, the following hypotheses are formulated:

H3a: 

Online user behavior has a positive influence on positive emotions.

Rationale: Engaging in positive online behaviors, such as active participation and sharing, can enhance users’ emotional experiences and satisfaction [36].

H3b: 

Online user behavior has a negative influence on negative emotions.

Rationale: Negative online behaviors, such as isolation or negative interactions, can increase negative emotional responses and reduce engagement [37].

H3c: 

Online user behavior has a positive influence on participation.

Rationale: Positive online behaviors encourage active engagement and interaction, increasing users’ willingness to participate in group chats [35].

To enhance the comprehensiveness of our model, we integrated additional constructs into the SOR framework. Under the stimulus component, we included social presence and the perceived relevance of the chat content to capture the social and contextual factors influencing user engagement. For the organism component, we incorporated cognitive factors such as perceived usefulness and mental workload to complement the emotional states. This comprehensive approach allows us to better understand how different types of user behaviors function as stimuli influencing emotional states and participation in group chats.

2.3. Basic Emotional Models and PANAS

In design and emotion studies, Desmet [38] proposed that emotions arise when users interact with design outcomes, reflecting changes in emotion based on reactions to design elements. Users typically experience satisfaction when the design is functional and/or appealing, eliciting feelings of enjoyment or excitement and thereby enhancing the user experience.

Understanding emotions in group chat contexts requires a comprehensive approach that integrates various theories and models of emotion. Here, we discuss key models, focusing on how they relate to the Positive and Negative Affect Schedule (PANAS) and justify our hypotheses.

(1)

Basic Emotions Theory

The Basic Emotions Theory posits that there is a set of fundamental emotions that are universally experienced by humans [39]. These include emotions such as happiness, sadness, anger, fear, surprise, and disgust. This theory provides a foundation for understanding the broad categories of emotional responses that users may experience in group chats. Griffiths (2002) [40] emphasized the importance of recognizing these basic emotions in digital communication, where non-verbal cues are limited.

(2)

Circumplex Model of Affect

Russell’s (2000) [41] Circumplex Model of Affect arranges emotions in a two-dimensional circular space, defined by the axes of valence (pleasant–unpleasant) and arousal (high–low). This model helps to visualize the range of emotional experiences and their intensities. For example, emotions such as excitement and relaxation can be plotted on this circumplex to understand their valence and arousal levels.

(3)

PANAS Model

The Positive and Negative Affect Schedule (PANAS) created by Watson and Tellegen [42] is a widely used tool for measuring positive and negative affect. The PANAS model categorizes emotions into two distinct dimensions: positive affect (PA) and negative affect (NA). PA reflects the extent to which a person feels enthusiastic, active, and alert, while NA reflects a dimension of distress and unpleasurable engagement.

The PANAS model is particularly relevant for studying emotional triggers in group chat contexts because it allows for a detailed analysis of both positive and negative emotional experiences. By categorizing emotions into PE and NE, we can systematically explore how different stimuli in group chats, such as user behaviors, platform usability, and chat content, influence users’ emotional states.

In the SOR framework, stimuli in group chats (e.g., user behaviors, chat content, social presence) trigger emotional responses (organism), which then influence participation and behaviors (response).

(4)

Hypotheses

Based on the PANAS model and the integration with the SOR framework, we propose the following hypotheses:

H4: 

Positive emotions have a positive influence on participation.

H5: 

Negative emotions have a negative influence on participation.

2.4. Participation

Participation is defined as “user-initiated action” [43] leading to valuable “co-creation” [44]. Scholars such as Hollebeek [45] regard emotional engagement as a multidimensional concept encompassing behavioral, cognitive, and affective aspects. Engagement entails an individual’s active involvement in interacting with media. The rise of social media platforms has transformed customers from passive content observers to active participants who are now co-producers [46] and creators of content generated through online interactions and behaviors [36]. Platforms like Facebook enable organizations to directly observe public responses to their messages through metrics such as likes, shares, and favorites [47]. These metrics indicate varying levels of engagement, as liking, sharing, and commenting entail different degrees of interaction and may affect users cognitively, emotionally, and behaviorally [48,49].

Participation can manifest in positive or negative forms. Negative participation occurs when members benefit from online groups but refrain from community activities [50]. In contrast, active participation involves high levels of engagement, including activities such as creating and disseminating information and providing emotional support to others [51]. Various metrics can be used to gauge the passive or active intensity of social media engagement [52]. For instance, liking (described as an “emotional response”) and commenting (“positive and public consideration”) are considered positive social media behaviors, while reading content and clicking are considered negative engagement behaviors [53].

Engagement metrics are essential for measuring the level of user interaction within group chats. Common metrics include:

• Functional Interactivity: The extent to which users can interact with the system and each other [54].
• Self-Efficacy: The belief in one’s ability to succeed in specific situations or accomplish a task [55].

Based on the SOR framework, within the context of group chat socialization, this study integrates participation with the level of participation to assess users’ willingness to engage. The results are categorized into four levels: remaining engaged, maintaining interest in engagement, active engagement, and retention within the group.

The SOR framework offers a flexible theoretical lens, providing a comprehensive perspective of how environmental factors influence an individual’s internal state and subsequent responses. It presents the environmental factors contributing to the group chat experience in this study, encompassing both the internal states and subsequent behaviors of participants, and categorizes various factors to elucidate the progression from phenomenon to essence.

In summary, this study is structured in three parts based on the SOR (Stimulus–Organism–Response) framework. The initial stimulus component encompasses three critical facets of group chat experience: usability, group chat rhythm, and users’ online behavior. The organism component includes positive and negative emotions. The reaction component pertains to the level of participation in group chat. This study explores the relationships among these variables within the context of online group chats and their influence on members’ emotions and engagement behaviors. Figure 1 illustrates the research framework.

3. Methodology

3.1. Sample and Data Collection

The questionnaire link was generated on https://www.wj.qq.com, a professional online survey platform, and was widely disseminated through social channels such as WeChat and other media platforms. The questionnaire aimed to assess our research model and comprised 3 demographic questions, 1 inquiry about social software usage duration, and 46 self-reported items. The social software usage questions ensured that all respondents had prior experience with online socialization. The questionnaire encompassed six variables: usability, chat rhythm, online user behavior, positive emotion, negative emotion, and participation.

The survey was conducted online and extensively shared via social media platforms like WeChat. Upon obtaining informed consent, participants were screened with initial questions about their level of comfort with social platforms, experience with group chat socialization, and familiarity with platforms such as WeChat, Weibo, and QQ. Only participants answering “yes” proceeded with the survey. Eligible participants were then prompted to recollect their group chat socialization experiences and respond accordingly.

The survey was conducted from 6 December 2023 to 16 December 2023. All respondents were Chinese residents with experience maintaining ongoing group chats. Of the 1365 individuals who viewed the questionnaire, 546 valid samples were collected, resulting in a response rate of 40%. The sample comprised 282 male (51.65%) and 264 female (48.35%) respondents. Among them, 10.44% were under 18 years old, 51.47% were aged 18–35, 31.5% were aged 36–50, and 6.59% were over 50 years old. The majority of respondents had attained at least a high school education (89.56%), and most reported using social software for over 1 h daily (91.21%).

Table 2. Demographics of the respondents.

Items	Categories	Number	Percent (%)	Cumulative Percent (%)
Gender	1	282	51.65	51.65
	2	264	48.35	100
Age	1	57	10.44	10.44
	2	281	51.47	61.9
	3	172	31.5	93.41
	4	36	6.59	100
Edu	1	31	5.68	5.68
	2	201	36.81	42.49
	3	151	27.66	70.15
	4	141	25.82	95.97
	5	22	4.03	100
Time-consuming	1	48	8.79	8.79
	2	353	64.65	73.44
	3	86	15.75	89.19
	4	59	10.81	100
Total		N = 546	100	100

displays the frequency distribution of respondents’ demographics.

3.2. Instruments and Measures

This study utilized six constructs: usability, chat rhythm, online user behavior, positive emotions, negative emotions, and participation. The usability (US) scale (α = 0.855) was adapted from three items of the Usability Metric for User Experience (UMUX), assessing difficulty of use, difficulty of learning, and integration validity, providing fundamental information about the usability experience of an online platform. The group chat rhythm (CR) scale was adapted from Song [56], comprising six items evaluating online chat content and rhythm, including the use of emoticons, the use of particles, preference for editing long or short text, the expression of opinions, and engagement in conversations.

The User Online Behavioral Scale, adapted from Yan et al. [36], encompassed thirteen questions across four sub-dimensions. Shaping Behavioral (SB) (3 items) focused on self-presentation-related online behaviors; Co-acting Behavioral (CB) (4 items) assessed interactive co-acting-related online behaviors; Seeking Behavioral (SEB) (3 items) measured seeking-related online behaviors; and Isolating Behavior (IB) (3 items) examined isolation-related online behaviors. Positive and negative emotions were measured using the classic PANAS scale by Watson et al. [57], comprising 10 positive emotion items (PA subscale) and 10 negative emotion items (NA subscale). The Participation (PA) Scale, adopted from Taylor and Todd [58], included four levels: staying and participating, staying interested in participating, actively participating, and not leaving the group.

Each variable was assessed on a 5-point scale, where 1 indicated strong disagreement and 5 indicated strong agreement. In the questionnaire, the items were presented in a randomized order to minimize response bias.

3.3. Common Method Variance

Given that reliance on a single source of information, such as a self-report scale, can introduce spurious relationships between variables, we took several procedural steps to mitigate the potential for common-method bias.

Firstly, measurement items were carefully sourced from a variety of established scales and studies. This approach not only diversifies the data but also reduces the likelihood of bias associated with using a single method. The diversity in data sources was intentionally designed to capture a broad range of constructs and to ensure that the measures used were not overly aligned with one specific data collection method.

Additionally, pretests and pilot studies were conducted with representative samples from the target population to refine the questionnaire. These pretests were critical for identifying any ambiguities or leading questions that could result in biased responses. The pilot studies allowed us to test the clarity of the items and ensure that the survey could be comprehended by respondents as intended. Any feedback received from these initial tests was used to revise the survey instrument, which helped to minimize potential response bias.

Further procedural remedies included the randomization of questions and response options throughout the survey. Randomization helps to reduce the impact of order effects and mitigates any tendencies respondents may have to answer questions in a predictable manner. This step ensures that responses are reflective of genuine user opinions and experiences, rather than influenced by the survey structure.

To empirically assess the presence of common-method bias, Harman’s one-factor test [59] was conducted. This test involves subjecting all measured variables to an exploratory factor analysis. If a single factor explains a significant portion of the variance, this indicates the presence of common-method bias. In our study, the analysis revealed that the largest factor accounted for only 22% of the total variance, well below the commonly accepted threshold of 50% [60]. Therefore, the data did not show a significant common-method bias.

Moreover, we applied additional statistical controls to further ensure robustness. For instance, the marker variable technique was considered as a complementary check. This involves including a theoretically unrelated variable to ensure that the measurement system does not confound the results. These statistical measures, combined with the procedural steps taken, confirm that the observed relationships between variables are not significantly influenced by the systematic variance introduced by the measurement method. As a result, there is no substantial evidence of common-method bias in this study.

4. Analysis and Results

4.1. Measurement Model

A confirmatory factor analysis (CFA) was conducted using AMOS 23.0 to assess the validity and reliability of all study variables. Fit indices, including normed chi-square, goodness-of-fit index (GFI), comparative fit index (CFI), root mean squared residual (RMR), and root mean squared error of approximation (RMSEA), were evaluated.

The results, as shown in

Table 3. Confirmatory factor analysis results.

Construct	Items	Loading	α	AVE	CR	Means
Usability			0.855	0.663	0.855	3.632
US1	Easy to use	0.845
US2	Most people can easily start	0.837
US3	Was integrated to make it easy to use	0.841
Chat Rhythm			0.863	0.512	0.863	3.33
CR1	Like to use emoticons	0.721
CR2	Like to use modal particles	0.75
CR3	Like to use large paragraphs to make a point	0.747
CR4	Like to use multiple messages to make a point	0.765
CR5	Like to make a point	0.765
CR6	Like to pick up on what others say	0.789
User Behavior			0.892	0.527	0.892	3.092
SB1	Like to show off my good side	0.638
SB2	Deletes or withdraws content that harms my image	0.676
SB3	Put certain labels on myself in order to be popular	0.744
CB1	Enjoys actively participating in chats on popular topics	0.77
CB2	Will actively interact with others	0.768
CB3	Would be happy to add new friends	0.759
CB4	Will actively share links	0.779
SEB1	Willing to accept content recommended by the system	0.715
SEB2	Interact with others only when necessary	0.699
SEB3	What I want most is information that is useful to me	0.715
IB1	Set the visible content according to the group members	0.747
IB2	Prefer to join a specific group before interacting	0.779
IB3	Setting limits on certain people	0.783
Positive Emotion			0.916	0.522	0.916	3.084
PE1	Interested	0.751
PE2	Excited	0.741
PE3	Powerful	0.717
PE4	Enthusiastic	0.731
PE5	Proud	0.737
PE6	Inspired	0.73
PE7	Determined	0.751
PE8	Attentive	0.72
PE9	Active	0.754
PE10	Alert	0.751
Negative Emotion			0.924	0.548	0.924	3.727
NE1	Distressed	0.733
NE2	Upset	0.749
NE3	Guilty	0.749
NE4	Scared	0.753
NE5	Hostile	0.756
NE6	Irritable	0.766
NE7	Ashamed	0.768
NE8	Nervous	0.773
NE9	Jittery	0.776
NE10	Afraid	0.779
Participation			0.893	0.677	0.893	2.764
PA1	Willing to participate in group interactions	0.84
PA2	Willing to participate in interactions, whether the group chat is interesting or not	0.871
PA3	Willing to actively participate in group chats, whether anyone interacts with me or not	0.844
PA4	Will not leave the group	0.778

, indicate an acceptable fit with the data: χ² = 677.323, df = 614, p < 0.05; normed χ² = 1.102; GFI = 0.937; CFI = 0.994; RMR = 0.031; RMSEA = 0.014.

All factor loadings were statistically significant at p < 0.01 and exceeded the 0.60 threshold, indicating adequate factor loadings [61]. The average variance extracted (AVE) for the constructs exceeded 0.50, and the composite reliability (CR) ranged from 0.855 to 0.924, surpassing the 0.70 threshold, indicating satisfactory convergent validity [61,62]. Reliability was assessed using Cronbach’s alpha, with values ranging from 0.855 to 0.924, indicating satisfactory internal consistency for all scales [63].

Discriminant validity was examined by comparing the squared correlations between constructs and their respective AVE values. As shown in

Table 4. Results of structural model.

PATH			Non-Std. Coef.	S.E.	C.R.	p	Std. Estimate
CR	<---	US	0.158	0.036	4.329	***	0.229
CR	<---	UB	0.122	0.052	2.331	0.02	0.123
PE	<---	US	0.182	0.039	4.702	***	0.236
PE	<---	CR	0.193	0.054	3.607	***	0.172
PE	<---	UB	0.238	0.055	4.311	***	0.214
NE	<---	US	−0.059	0.044	−1.347	0.178	−0.069
NE	<---	CR	−0.039	0.061	−0.631	0.528	−0.031
NE	<---	UB	−0.315	0.065	−4.824	***	−0.255
PA	<---	CR	0.217	0.062	3.527	***	0.172
PA	<---	UB	0.173	0.066	2.631	0.009	0.139
PA	<---	PE	0.144	0.056	2.576	0.01	0.128
PA	<---	NE	−0.145	0.047	−3.124	0.002	−0.144
PA	<---	US	0.108	0.044	2.427	0.015	0.124

Note: *** p < 0.001 indicates a highly significant effects.

, all squared correlations were lower than the corresponding AVEs, indicating acceptable discriminant validity [63]. Overall, the measurement model demonstrated adequate validity and reliability, supporting the robustness of the study’s constructs.

4.2. Structural Model

Structural equation modeling (SEM) was conducted using AMOS 23.0 to test the hypotheses. The fit indices for the structural model were satisfactory (χ² = 686.074, df = 615, normed χ² = 1.116), with GFI (0.936) and CFI (0.993) values exceeding the 0.90 threshold, and RMR (0.041) and RMSEA (0.015) values below the 0.08 threshold, indicating an acceptable fit (Hair et al., 2014) [59].

The results revealed significant relationships between the nine paths. Figure 2 illustrates that all three stimuli had positive and significant effects on positive emotions, with usability (US) having the most significant effect (γ = 0.236, p < 0.001), followed by UB (γ = 0.214, p < 0.001), and CR (γ = 0.172, p < 0.001). Therefore, hypotheses H1a, H2a, and H3a are supported.

Among the three stimuli, only UB had a significant negative effect on negative emotions (γ = −0.255, p < 0.001), while US (γ = −0.069, p = 0.178) and CR (γ = −0.031, p = 0.528) had statistically insignificant effects on negative emotions. Therefore, hypotheses H1b and H2b were rejected, while hypothesis H3c was supported.

Furthermore, all three stimuli had positive and significant effects on participation (PA), with CR having the most significant effect (γ = 0.172, p < 0.001), followed by UB (γ = 0.139, p < 0.01), and PE (γ = 0.128, p < 0.05) and US (γ = 0.124, p < 0.05)). Negative emotions (NE) had a significant negative effect on PA. (γ = −0.144, p < 0.01). Therefore, hypotheses H1c, H2c, H3c, H4, and H5 were supported.

4.3. Mediation Analysis

We tested the mediating role of emotion between the three stimuli and participation, as detailed in

Table 5. Specific indirect effects.

PATH	EFFECTS	Lower	Upper	p	Conclusion
Indirect Effects
US_PE_PA	0.026	0.007	0.057	0.004	Partial mediation
CR_PE_PA	0.028	0.008	0.063	0.003	Partial mediation
UB_PE_PA	0.034	0.009	0.076	0.004	Partial mediation
US_NE_PA	0.009	−0.005	0.028	0.172	Not significant
CR_NE_PA	0.006	−0.015	0.035	0.539	Not significant
UB_NE_PA	0.046	0.017	0.087	0.004	Partial mediation

. The Bias-Corrected Bootstrap program was used to test for mediating effects. Repeated random sampling was utilized to draw 2000 Bootstrap samples from the original data (n = number of samples) to generate a 1-approximate sampling distribution with a confidence interval (CI) set at 95%.

First, we discussed the mediating effect of positive emotions. In the path “Usability—Positive Emotion—Participation”, (Effect = 0.026, 95% CI [0.007, 0.057]), p = 0.004 < 0.05, indicating that positive emotion partially mediates the relationship between usability and Participation. In the path of “Chat Rhythm—Positive Emotion—Participation” (Effect = 0.028, 95% CI [0.008, 0.063]), p = 0.003 < 0.05, indicating that positive emotion partially mediates the relationship between chat rhythm and participation. In the path of “Online User Behavior—Positive Emotion—Participation” (Effect = 0.034, 95% CI [0.009, 0.076]), p = 0.004 < 0.05, positive emotion partially mediates the relationship between online user behavior and participation.

Second, we discussed the mediating effect of negative emotion. In the “Usability—Negative Emotions—Participation” pathway, (Effect = 0.009, 95% CI [−0.005, 0.028]), p = 0.172 > 0.05, showing that the effect of negative emotions on usability and participation is not statistically significant. In the path of “Chat Rhythm—Negative Emotions—Participation”, (Effect = 0.006, 95% CI [−0.015, 0.035]), p = 0.539 > 0.05, indicating that there is no statistically significant effect of negative emotions on chat rhythm and participation. In the path of “Online User Behavior—Negative Emotions—Participation” (Effect = 0.046, 95% CI [−0.017, 0.087]), p = 0.004 < 0.05, negative emotions partially mediate the relationship between online user behavior and participation.

5. Discussion and Implications

The study delves into the influence of experiential factors on users’ emotions and participation within group chat platforms, leveraging the SOR (Stimulus–Organization–Response) theoretical framework. Through an online survey, three key experiential elements and two emotions were identified that shape participation in the context of group chats. This analysis offers valuable insights for platform builders and designers, aiding in the development of targeted participation strategies.

5.1. Direct Feedback Is the Key to Maintaining Positive Emotions

Positive emotions are significantly influenced by all three stimuli, with usability exerting the greatest impact. Usability is often considered a direct aspect because it is experienced first. When users perceive a system as user-friendly, they develop a sense of efficacy and control, fostering confidence and eliciting positive emotions [26]. Moreover, a highly usable environment diminishes cognitive load [29], allowing users to focus more on content and social interactions, rather than grappling with complex interfaces. While user behavior and group chat rhythm also contribute to positive emotions, their influence may be more closely tied to the content and dynamics of group chats. Nonetheless, they still play a crucial role in enhancing the overall user experience and fostering positive emotional responses. By understanding these dynamics, platform developers can prioritize usability enhancements to create more engaging and satisfying group chat environments.

In addition, chat rhythm can also improve participation by promoting positive emotions, highlighting the importance of fulfilling users’ chatting preferences to boost participation [64], and also shows the important role of timeliness and directness in promoting positive emotions.

5.2. Flexible User Behavior Is Beneficial for Reducing Negative Emotions

Among the three stimuli, only user behavior exhibits a significant negative effect on negative emotions. This highlights the crucial role of online user behavior in shaping users’ emotional experiences on the platform. User behavior directly reflects users’ expectations and needs for the platform. When these expectations are not met, users may experience negative emotions such as frustration or dissatisfaction [65]. Technical issues, such as application bugs or crashes, can have an immediate and tangible impact on the user experience, leading to negative emotional responses. In contrast, users tend to be more adaptable to variations in usability and chat rhythm. These factors allow for greater flexibility and subjective interpretation, making it less likely for deviations to trigger negative emotions unless they fall outside the user’s adaptive range [66].

Chat rhythm, although having both positive and negative effects, may not significantly impact negative emotions due to the cancelation of these effects. For example, a fast chat rhythm may evoke excitement but also stress, while a slow rhythm may induce relaxation but also boredom. This complexity may mitigate the direct impact of chat rhythm on negative affect compared to user behavior. Usability, considered a fundamental expectation, typically does not elicit strong negative emotions unless it reaches extreme levels of dysfunctionality. Users may have varying emotional responses to the same experience, with some being more sensitive to technical issues and others more tolerant of changes in chat rhythm or usability.

5.3. Controlling Chat Rhythm and Reducing Negative Emotions Are Essential for Participation

Chat rhythm emerges as the most influential factor impacting participation intention among the stimuli examined in the study. There are several reasons for this finding. Firstly, the rhythm of a group chat directly shapes users’ communication experiences. An optimal pace fosters effective communication and interaction, while extremes in pace—either too fast or too slow—can lead to user discomfort and disengagement. Secondly, the immediacy of the chat rhythm makes it a sensitive indicator of user participation. In real-time group chats, users receive instant feedback on their contributions, influencing their sense of engagement and willingness to participate. Thirdly, the chat rhythm reflects the dynamic nature of group chat interactions. Smooth and well-paced interactions encourage active participation, highlighting the pivotal role of rhythm in driving engagement [3]. Platforms can enhance design elements related to rhythm control, including the use of emoticons, tone-of-voice cues, editing prompts, and moderation mechanisms. By optimizing chat rhythm, platforms can create a comfortable and engaging environment that encourages active participation in group chat interactions [67].

Negative emotions exert a stronger inhibitory effect on participation intention compared to positive emotions. When users experience negative emotions, they may reduce their platform usage and engagement or discontinue usage altogether. Even infrequent occurrences of negative emotions can significantly impact engagement. An asymmetry exists between positive and negative emotions [68]. While positive emotions gradually enhance participation, negative emotions can immediately diminish it. This suggests that negative emotions, although less frequent, can have a substantial negative impact on user behavior. Moreover, a potential “threshold effect” may exist, where further increases in positive emotions beyond a certain level do not significantly boost willingness to participate. Conversely, certain negative emotions may exert a strong immediate influence on participation intention.

Additionally, user behavior affects participation through emotion. Operational behavioral dissatisfaction can directly trigger negative user emotions such as frustration and disappointment. For platform functionality to satisfy operational behavior, specific user performance expectations must be met. If these expectations are not met, users may experience inconvenience and annoyance, resulting in negative emotions. Functional problems often act as “sticking points” in user experience, eliciting strong negative reactions.

5.4. Design Strategies for Group Chat Platforms Based on SOR Theory

The findings demonstrate that a well-functioning stimulus can positively impact users and elicit favorable responses, while a poor stimulus can evoke a more significant negative response. Hence, optimizing the group chat experience and stimuli is crucial to enhance the positive effects of positive emotions. Emotions serve as an intermediary stage influenced by both positive and negative experiences, further shaping responses. Thus, strengthening guidance is vital to ensure positive emotions have an impact while minimizing the amplification of negative emotions. Responses, as the final step, offer essential feedback on stimuli and emotions, guiding subsequent SOR processes. It is essential to facilitate responses that acknowledge user participation and continuously foster positive feedback.

5.4.1. Optimizing Stimulus-Detailed Adjustment Settings

Firstly, to optimize platform functionality, it is essential to consider customization based on user behavior types. Enhancing the platform’s practicality and diversifying its functions ensures that users can fulfil various behavioral needs through different operations. Since certain user behaviors can have a more significant impact on negative emotions, it is important to design the functional layout in a way that can account for these behavior types, preventing any functional imbalances. Adjusting sensory input based on user behavior types is a crucial aspect to consider. Sensory input optimization can significantly influence the learning and adaptation processes by providing stimuli that align more closely with individual needs. Research by Zhu et al. [69] suggests that tailored stimuli can reduce cognitive load and enhance the ease of interaction, ultimately improving user engagement and emotional satisfaction.

Secondly, given the positive correlation between usability and positive emotions, platforms should prioritize simplifying user operations, ensuring they are intuitive and easy to understand, in order to enhance the overall user experience. Integrating software and features with low learning costs can help minimize resistance to use. Giabbanelli et al. [70] emphasize that systems designed with usability in mind, particularly those that reduce operational complexity and enhance feedback through clear sensory cues, can significantly increase user engagement and satisfaction. Usability is fundamental; thus, improving the fluidity of use and reducing operational complexity should be prioritized to facilitate quick adaptation for new users. Customized visual and auditory cues not only enhance emotional responses but also further encourage user participation, reducing frustration. This type of sensory adaptation helps users learn and adapt to new features more quickly, contributing to a smoother and more positive user experience.

Additionally, regulating the rhythm of group chats is crucial, as this significantly affects engagement. Offering various emotional expression methods, including non-verbal cues and new interaction styles, is essential. Schreuder et al. [71] found that consistency among multiple sensory stimuli can enhance positive emotional, cognitive, and behavioral responses. Designing message flow control mechanisms, such as unread message markers and options for undisturbed message types, can prevent information overload. Flexibly adjusting the rhythm of group chats according to user preferences can enhance positive emotions and engagement.

Overall, controlling information flow in group chats through personalized indicators and interaction methods can address the diverse needs of users. By incorporating various sensory cues, platforms can effectively manage users’ cognitive and emotional loads. Diverse interaction methods allow users to engage at their own pace, optimizing emotional responses and participation efficiency.

5.4.2. Internal Organism Processing–Emotional Guidance

To evoke positive emotions in users, it is essential to integrate elements into the design that simultaneously influence cognitive and emotional processing, as supported by psychological and neuroscientific research. Emotions play a crucial role in cognitive processing and decision-making, not only enhancing user experience but also influencing user behavior to some extent [72]. Introducing engaging elements such as emoticons and customizable stickers onto the platform, and offering diverse ways to express emotions, helps create a more positive and interactive environment [73]

Additionally, incorporating feedback mechanisms, such as likes, badges, or achievement systems, can further amplify positive emotions. These mechanisms provide positive reinforcement, activating the brain’s reward system and promoting user behavior, which prolongs user engagement [70]. Simplifying the user interface and interaction flow not only reduces cognitive load but also minimizes the potential for negative emotions, such as frustration, by making the system more intuitive and easier to navigate.

To mitigate negative emotions, platforms can utilize user behavior data to predict the scenarios likely to cause frustration or confusion. For instance, error messages can be optimized through a more supportive and user-friendly design to reduce emotional frustration during operations [74]. By considering both emotional and cognitive factors in user interactions, design strategies can significantly improve overall user satisfaction and engagement, ensuring that negative emotions are minimized during the user experience.

5.4.3. Facilitating Response-Stimulating Participation

Since all three stimuli influence participation intention through positive emotions, platforms should implement positive feedback mechanisms such as liking, commenting, and sharing functions. Research has shown that feedback mechanisms that provide users with social validation can significantly enhance their motivation to participate and contribute [75]. Design incentive mechanisms like point systems and achievement unlocking to motivate users to engage further and contribute more content, as gamification elements have been found to increase user engagement by providing clear goals and rewards [76].

Establishing reward systems and fostering community identity are also crucial for encouraging active participation. Social identity theory suggests that when users feel a sense of belonging to a community, their participation rates increase as they become more invested in the group’s success and culture [77]. Furthermore, implementing user feedback mechanisms, allowing users to provide suggestions or report issues, not only increases participation but also fosters a sense of belonging and ownership among users [78]. This approach encourages users to feel valued, which can lead to stronger community engagement and a positive overall experience for users on the platform.

5.5. Application Discussion in HCI Field

Our study investigates the factors affecting emotion and participation in group chats, contributing to the understanding of how usability, chat rhythm, and user behaviors influence user’s participation. However, to provide a more comprehensive exploration of the theoretical and practical implications, we need to situate our findings within the broader context of recent Human–Computer Interaction (HCI) studies, particularly those that have enhanced group chat experiences through the use of chatbots and conversational agents.

Our findings align with and extend the existing HCI research on group chat interactions. For instance, Kim et al. (2020) [79] explored how chatbots can facilitate group chat discussions by providing a structure and managing conversation flow. Our study complements this by highlighting the importance of chat rhythm and user behaviors in maintaining engagement and positive emotional experiences. While chatbots can effectively manage discussions, our findings suggest that human factors, such as how users interact and the rhythm of their conversations, also play a crucial role in influencing group dynamics.

Additionally, Do et al. (2022) [80] examined the communication strategies of conversational agents in group chats, emphasizing the role of these agents in enhancing user engagement and reducing communication barriers. Our study extends this work by showing how usability and the perceived relevance of the chat content can similarly impact user engagement. By integrating these factors into the design of conversational agents, developers can create more effective tools that not only facilitate communication but also enhance user satisfaction and participation intentions.

From a practical perspective, our study provides several actionable recommendations for improving group chat experiences. These recommendations could be particularly valuable when considered alongside recent advancements in chatbot and conversational agent technologies.

• Enhancing Usability: Ensuring that group chat platforms are user-friendly and efficient can significantly improve user satisfaction and engagement. This aligns with the findings of Kim et al. (2020) [79], who demonstrated that chatbots can assist in managing conversation flow. By combining a high level of usability with chatbot functionalities, platforms can offer a seamless and enjoyable user experience.
• Optimizing Chat Rhythm: Our study underscores the importance of chat rhythm in maintaining engagement. This can be integrated with the strategies proposed by Do et al. (2022) [80], where conversational agents help to maintain an optimal pace of interaction. Developers can design agents that adapt to the natural rhythm of user interactions, providing timely responses and facilitating smooth transitions between conversation topics.
• Promoting Positive User Behaviors: Encouraging positive user behaviors, such as active participation and supportive interactions, can enhance emotional experiences and increase participation intentions. Chatbots and conversational agents can be programmed to recognize and promote these behaviors, creating a more positive and collaborative group chat environment.
• Incorporating Social Presence: Enhancing the feeling of social presence in group chats can improve user engagement. Chatbots can be designed to simulate a social presence by using natural language processing and personalized responses, making users feel more connected and engaged with the conversation.

6. Conclusions

6.1. Conclusion of This Study

In the context of group chats becoming a daily online activity, enhancing the group chat experience has emerged as a critical issue. From a design perspective, it is essential not only to provide users with context-appropriate services from a tool-based standpoint but also to explore the relationship between user psychology and online behavior to promote a higher level of participation.

This paper constructs a theoretical model based on the Stimuli–Organism–Response (SOR) framework. The model identifies usability, chat rhythm, and user behavior as the stimuli in group chat platforms; positive and negative emotions as the organism; and participation level as the response. The relationships among these elements are explored in detail.

The findings suggest that fostering positive emotions requires a concerted effort across various aspects, with system usability being the most crucial factor. Conversely, negative emotions primarily arise when certain online behaviors cannot be carried out, necessitating more targeted design interventions. Additionally, the significant impact of chat rhythm on participation levels highlights the importance of in-group guidance, such as agent bots and context-aware design nudges.

Overall, given the large number of users and dynamic needs in group chat scenarios, it is indispensable to consider a design framework and specific strategies that account for behavior classification and emotional guidance. This approach is vital for improving overall user experience and communication efficiency.

Additionally, this study offers valuable insights into users’ emotions and behaviors regarding group chat engagement. Our study contributes to the understanding of factors influencing emotion and participation in group chats, providing valuable insights for both researchers and practitioners in HCI. By situating our findings within the context of recent advancements in chatbot technologies, we offer a comprehensive framework for enhancing group chat experiences. Our recommendations, when combined with chatbot functionalities, can lead to more engaging, satisfying, and effective group chat interactions.

6.2. Limitation and Future Research

There are several limitations to consider. Firstly, the study relied on existing scales rather than open-ended questions, potentially overlooking important factors. Future research could explore additional predictors of participation in group chats using more diverse methods.

Moreover, the study was conducted in China, focusing on users of WeChat group chats, limiting its generalizability to global contexts. Future research should investigate group chat socialization factors across different cultural settings and include a more diverse sample of participants.

Lastly, future research should explore contextual factors in group chats that may vary by user type and consider population-specific variables to inform the development of targeted marketing strategies for e-commerce platforms. Overall, addressing these limitations will contribute to a deeper understanding of group chat dynamics and inform more effective design and marketing strategies.

Our study opens several avenues for future research. One potential direction is to explore the integration of our findings with chatbot technologies, investigating how chatbots can be designed to enhance usability, optimize chat rhythm, and promote positive user behaviors. Additionally, future studies could examine the long-term effects of these factors on user engagement and emotional experiences in group chats, providing deeper insights into how to sustain user participation over time.