1 Introduction

Live streaming is a recent socio-commerce and a new online marketing tool (Xie et al., 2022) to broaden revenue and consumption opportunities, such as travel live streaming (Xu et al., 2021). Marketers believe that Live-streaming is an effective channel that can prompt customers to make purchasing decisions quickly (Wang et al., 2022a, 2022b). Prompted by the benefits of live-streaming and its novelty, researchers investigate the live-streaming e-commerce phenomena (Ye & Ching, 2022), such as the influencing factors of consumer purchase intention and behavior (Ma et al., 2022). Many research findings have been found concerning consumer behavior in live streaming e-commerce, for instance, the roles of emotional and information support (Mao et al., 2022), linguistic patterns of gender (Yang & Wang, 2022), trust and uncertainty reduction (Lu & Chen, 2021), value-based marketing (Xie et al., 2022), perceived persuasiveness (Gao et al., 2021), interactivity and customer engagement.

Theoretical coverage is one of the essential components of the existing literature on live-streaming e-commerce. To explore the flow experience state of consumers in live-streaming e-commerce, Sun et al. (2021) use the affordance theory. Afforestation “do not rely on the environment as heroes, nor on individuals as heroes”, but rather on the interaction between individuals and environmental cognition,” according to Parchoma (2014: 361). By altering how a person interacts with their surroundings, marketers hope to increase the potential for customer actions to achieve an instant concrete effect (Bygstad et al., 2016: 87). Sun et al. (2021) consider the IT affordance perspective, including visibility, meta-voicing, and guidance shopping. Visibility refers to the visible form of product information to facilitate decision-making (i.e., the product service that meets their interests and needs). Affordance theory strongly emphasizes psychology (Rahman & Sciara, 2022). The live-streaming e-commerce setting frequently focuses on the socio-technical environment (Li et al., 2021).

However, there is a general paucity of knowledge regarding the factors that influence customers’ holistic experiences when fully engaged (specifically, flow experience; Csikszentmihalyi, 1975), as well as how flow experience affects compulsive purchasing during live streaming shopping. Thus, the purpose of this study is to:

apply flow theory to conceptualize and validate consumer behaviors in live-streaming e-commerce, which examine the flow state drivers and post-flow state mediators as significant ability to shape compulsive buying.

According to Muller et al. (2005), compulsive buying manifests as a “regular obsession with buying or impulses to buy that are seen as irresistible, obtrusive, and senseless,” and it can be a consumer’s primary means of relieving stress, frustration, and disappointment (Phau & Woo, 2008). Csikszentmihalyi (1996: 29) defines flow as a “condition in which people are so engrossed in a task that nothing else seems to matter; experiencing it is so pleasurable that people will do it even at enormous cost for doing it.” Flow experiences can occur depending on the many contextual circumstances; for example, e-sport supporters are more likely to focus on the match when identifying the athlete’s unique abilities and surprising game plans (Chou & Ting, 2003). The basic tenets of flow theory report enjoyment, cognitive engrossment, and a loss of sense of location and time as flow characteristics (Kim & Kim, 2020).

The structure of the theoretical concepts addressed by the research objective is depicted in Fig. 1.

Fig. 1
figure 1

Structure of theoretical concepts addressed by the research objective

Full size image

2 Literature review

Live-streaming shopping and e-commerce research is of increasing interest, and additional studies can provide a more thorough explanation of the rationale underlying consumer behavior. One of the cognitive problems is understanding the concept of “flow.” When a person is engaged in an activity, they are fully immersed in a state of energized attention, involvement, and delight, known as “flow” (Buzady, 2017: 205). The “flow” concept’s originator is Mihaly Csikszentmihalyi (1975, 1991, 2003), from whom this study draws its direct inspiration. The original, reliable source for the conceptual framework creation is used in this section to review the idea of flow theory. This study focuses on consumers’ emotions while live-streaming or engaging in online shopping, noting that “things flow freely, like being carried away by a river” (Buzady, 2017: 206).

2.1 Literature analysis

To some extent, the annual distribution of literature can reflect the research status of this field at a specific stage. An important indicator of scientific output is also the total number of papers published each year. CiteSpace is used to extract annual data from WOS data that analyzes the overall research trend in the field of e-commerce, and the number of time chart documents is obtained, as shown in Fig. 2.

Fig. 2
figure 2

Distribution map of annual total literature in the e-Commerce live streaming field

Full size image

According to the trend of the number of publications, the development of E-commerce Research can be divided into three stages: (1) Stage of initial; (2) Stage of slow; (3) Stage of development. 2008–2009 was the initial research phase, and several papers were published or unpublished every year. This was a slow development from 2010 to 2014. The number of articles published every year starts to increase, and sometimes it drops. The number of published papers increased from six in 2011 to 11 in 2014. The year 2015 is a stage of rapid development of the publication of articles. The number of articles published each year shows an obvious rapid growth trend, and the number of articles published is a rapid growth trend, with the number of published papers decreasing from 11 in 2014 to 10 in 2021.

2.2 Institutional analysis

We pulled organization data from WOS and ranked us first in terms of number by citation counts is Simon Fraser Univ, with citation counts in 3. The second item is Univ Mississippi of Cluster, with citation counts in 2. The third item is Tsinghua Univ, with citation counts in 2, as shown in Fig. 3.

Fig. 3
figure 3

Top 10 institutions in the field of e-Commerce live streaming

Full size image

The 4th item is Huazhong Univ Sci & Technol, with citation counts in 2. The 5th item is Jinan Univ, with citation counts in 1. The 6th item is I Shou Univ, with citation counts in 1. The 7th item is LinkedIn Corp, with citation counts in 1. The 8th item is Chinese Univ Hong Kong, with citation counts in 1. The 9th item is Univ Minnesota, with citation counts in 1. The 10th item is Georgia Inst Technol, with citation counts in 1.

2.3 Key scholars analysis

In this field of research, leading scholars are ahead of the curve. The author's emphasis on the research field and the measure of the author's influence can be reflected by the author's reference frequency (Fig. 4).

Fig. 4
figure 4

Analysis of key scholars in the field of e-Commerce live streaming

Full size image

Previous analysis believed that this is a faster pace of growth since 2013. Through the analysis of a large number of authors, the researchers can find the latest representative authors in the field of e-commerce. The top ranked item by centrality is JIANGCHUAN LIU, with centrality of 11. The second item is FENG WANG, with centrality of 7. The third item is HAO YIN, with centrality of 6. The 4th item is GEYONG MIN, with centrality of 6. The 5th item is XIAOQUN YUAN, with centrality of 6. The 6th item is JINHONG LIU, with centrality of 6. The 7th item is QING FANG, with centrality of 6. The 8th item is YI DING, with centrality of 6. The 9th item is QIONG LIU, with centrality of 6. The 10th item is YUAN LIU, with centrality of 5.

2.4 Research hotspots analysis

Through the analysis of recent hot spots and frontiers, researchers can understand the latest research trends and determine the recent research questions. We select the data from 2008 to 2022 as the source data for hot and cutting-edge analysis based on the final development stage. In CiteSpace, set “TimeSlicing = 2008–2022”, “Node type = keywords” and “TOP = 10” to obtain the visual knowledge map of common literatures (Fig. 5). In the figure, NoteClusters with means that the data is too sparse to select meaningful features: adaptive resource management for p2p live streaming systems; adaptive resource management for p2p live streaming systems; optimized selection of streaming servers with goods for delivered live streaming; optimized selection of streaming servers with goods for delivered live streaming; cloud-assisted live streaming for crowdsourced multimedia content and cloud-assisted live streaming for crowdsourced multimedia content; proxy caching for peer-to-peer live streaming; proxy caching for peer-to-peer live streaming. The highlight articles in each block are nodes, as shown in Fig. 5.

Fig. 5
figure 5

Keyword clustering graph of e-Commerce live streaming

Full size image

Keyword co-occurrence is another method to detect and study hot spots. CiteSpace is used to generate co-occurrence matrix, build co-occurrence network, and identify research hotspots in this field. We can find that the research object in the field of e-commerce live streaming are live streaming, video, p2p live streaming, cloud computing, peer to peer, facebook live, social networking site, engagement, user motivation, channel streaming quality.

3 Research hypothesis

The central idea of Csikszentmihalyi’s (1975) “optimal experience theory” is “flow experience,” which is understood as a state of complete engrossment or strong focus in activities (Bao & Huang, 2018). Humans require a flow state to function, and according to Csikszentmihalyi (2022), they keep pushing themselves to achieve one to advance personally. People can take advantage of possibilities for discovery, push themselves to reach higher levels of performance, or experience a creative emotion that allows them to enter a different reality or previously unimaginable realms of awareness (p. 89). The optimal or joyful and engrossed state of the flow experience might change in the next moment or occasion, typically marked by anxiety and boredom, meaning that flow experiences are dynamic (ibid, pp. 90–91). Thus, maintaining the flow is a challenging task for live-streaming anchors and hosts. For this purpose, the question of whether the flow experience state is a valuable construct still has to be answered. If so, the proposed conceptual framework addressed by the research goal in Fig. 1 should be supported by empirical validation.

According to Csikszentmihalyi (2022), flow-induced activities, also known as flow activities, should not divert customers but rather provide an incentive for them to concentrate and create a pleasurable condition for flow to arise. They can also cause total concentration and a sense of time passing quickly, as well as a lack of self-consciousness due to the merging of action and awareness, which is a state of immersion and absorption (Csikszentmihalyi, 1990a). As a result, flow activities allow people to channel their psychic energy commitment, which is more closely linked to their intrinsic motivation.

Numerous situations correspond to the flow experience. For instance, students whose abilities match the obstacles provided by the learning environment will pay close attention to what they are learning (Rathunde & Csikzentmihalyi, 2005), leading to flow to arise. According to Csikszentmihalyi (1990b), a pleasant moment is crucial for the emergence of a flow experience because it creates the right conditions for being swept along like a river current. The flow-like experience differs from ordinary consciousness by combining action and awareness, or “the mind falls into the activity as if actor and action have become one” (Csikszentmihalyi, 1990b: 127). Csikszentmihalyi (2020: 104) depicts two pathological social states—anomie and alienation, which have similar meanings to anxiety and boredom, respectively—that make it difficult to feel the flow. People become uncomfortable when there are no rules in an anomie setting, as with live-streaming e-commerce. Accordingly, the hosts or anchors of live streaming services should work to influence consumer trust. Without feeling the possibility of accomplishing a goal, as emphasized in Kristine Marin Kawamura’s interview with Csikszentmihalyi (2014), the flow will not arise. Thus, trust implies a decrease in ambiguity (Lu & Chen, 2021). Gaining consumer trust reduces anxiety (Hong et al., 2022) and creates an environment conducive to experiencing flow, such as immersion and a state of focus (Csikszentmihalyi, 1990a, 1990b).

The rationale put forward by Csikszentmihalyi (1975, 2022) and the considerations above lead this study to assume trust as a prerequisite for flow to occur rather than assuming flow experience to alter trust (Bao & Huang, 2018; Jamshidi et al., 2008). According to Sztompka (2006), trust is a significant study subject in sociology and political science; consumer behavioral research is no exception (Murphy et al., 2022; Sztompka, 2006).

The second barrier to people experiencing flow is boredom. Csikszentmihalyi (2022) uses the social pathology concept of “alienation,” which is “in many ways the opposite of anomie,” to illustrate how boredom occurs when the social system forces people to act in ways that are counter to their goals. Overly rigidity inevitably results in boredom (p. 104). Therefore, because customers will not put their psychic energy into what is desirable owing to a lack of enjoyment (Csikszentmihalyi, 2022: 104), concentration and flow states will not occur (Atombo et al., 2017). Whether they are enjoyment (a mild but affective mood) or emotion (very intense and focused on a specific item), they are conditions for flow experience (Russell & Griffiths, 2008).

The concepts of trust and enjoyment can also be viewed as cognitive and affective evaluations interconnected through the lens of the cognitive appraisal theory. In addition, experiences that provoke strong emotions are typically handled mentally (Brockman et al., 2017; Lazarus, 1991). Particularly, cognitive evaluation is connected to a person’s perception of the live-streaming retail deliverable (e.g., represented by consumer trust). Comparatively, customer enjoyment is represented by affective evaluation (cf. Kim et al., 2021).

As a result, the following theories are put forth.

H1

The flow state drivers will have a positive effect on consumer trust.

H2

The flow state drivers will have a positive effect on consumer enjoyment.

H3

Consumer trust will have a positive effect on the flow experience.

H4

Consumer enjoyment will have a positive effect on the flow experience.

H1 and H2 are not only an essential insights into the intellectual base of consumer behaviors, which aim to explain the factors that lead the consumers to form the subjective belief that they will fulfill their obligations (Patil et al., 2020) and enjoyment, but also a condition to make flow easy to arise. Similarly to H1, Csikszentmihalyi (1991) stresses the role of the specific nature of experiences needed for order or attention in everyday life, such as free-floating artistic experiences (Moholy-Nagy, 1947). Thus, the drivers must motivate the consumers to direct the psychic energy to satisfy their needs and to keep consciousness tuned towards (Csikszentmihalyi, 2000) the live-streaming shopping as noted in H3 and H4.

A balance of autonomy and competence is just one of many characteristics that might lead to a positive, trusting, and flow-like experience (Reer et al., 2022). The flow theory is used in this investigation. The idea of flow theory emphasizes activity-personal inducement (Csikszentmihalyi, 2022) and incorporates ideas of cognitive evaluation theory, affordance (Parchoma, 2014; Sun et al., 2021), and environmental psychology (Uhrich & Koenigstorfder, 2009). For instance, Uhrich and Koenigstorfder (2009) conceptualize how emotions (such as pleasure, arousal, and dominance) and behavioral reactions (such as short-term or long-term behavioral reactions) can be initiated by both the environmental stimuli that can be caused by the organizer, spectators, and game action, as well as personality. Mehrabian and Russell (1974) proposed the environmental psychology behavioral model for sports events (individual pre-dispositions, active versus passive spectators, frequent versus infrequent visitors, VIP versus ordinary spectators). By using cognitive evaluation and coping, the cognitive appraisal theory (Otterbring et al., 2021) assumes that a person has an equal and bidirectional relationship with their environment.

Both activity and personal levels, thus, become the two drivers that can significantly influence trust formation and enjoyment and also, enact as antecedents to flow experiences.

Environmental psychology with activity level as the theme. This is a place that focuses on trading between humans and their environment (Tam & Milfont, 2020). As noted in Csikszentmihalyi and Bennett (1971), when the action (e.g., the live-streaming shopping) resonates with the environment, enacted, for instance, by supportive environment (e.g., prompt feedback to the actors), flow experience can more readily arise. This study considers two activity-related constructs: the value perceptions of the products and live-streaming participation (represented by utilitarian value and social value) and social influence in the live-streaming shopping session. Utilitarian value is a direct reason for the task at hand—attending to the live-streaming e-commerce for specific products or services. Goods not only provide practical value, but also help people gain social status; They are the material symbols of a person, and they are also the people they want to be (Xu, 2008). Social influence is a structure that explains the role of live streaming platforms. Defined as the level of support that consumers perceive as important in their lives and others should have for their participation (Baishya & Saamalia, 2020; Patil et al., 2020) in the live-streaming shopping. Examples of social influence include, for instance, consumer-anchor interaction and consumer-consumer interaction (Ma et al., 2022). Social influence represents a two-way communication flow, which is a significant theme of social interaction theory that aims to explain the interactive relationships among people (Liu, 2003; Varey, 2008) to enhance communication quality and reduce the uncertainty of consumption in live-streaming commerce (Yang et al., 2022). It implies that social influence shapes the trust level of the consumers, as noted in H1.

At the personal level, this study considers personal innovativeness and compulsive buying tendency (Kukar-Kinney et al., 2009). Innovativeness is a personal trait that tends to try new things and has a more remarkable ability to take risks (Wu & Yu, 2022). Researchers acknowledge this trait as personal innovativeness associated with creativity and innovative work behavior (Maqbool et al., 2019). Personal Innovation refers to a person's belief that they are actively inclined to use new technologies, products, or services (Cheng, 2014). Thus, personal innovativeness is, by nature, a risk-taking propensity (Lee et al., 2007) and implies a fit of skill challenge that is a condition for a flow-state to arise (Csikszentmihalyi, 1975).

Compared to other consumer behaviors, compulsive buyers tend to fantasize and be anxious that the promotion may not be unavailable on other occasions, thus, motivating compulsive buying (Scherhorn et al., 1990). Based on 112 Israeli individuals, Shoham and Brencic (2003) support their hypothesis that compulsive buying tends to affect unplanned events positively. Although compulsive buying generally has a somewhat negative tone in that the “In the environment of physical stores, compulsive buyers feel guilty and regret frequent purchases (299), it may not be so in most cases. Instead, compulsive buyers, according to Faber and O’Guinn (1992), desire to experience positive, stimulating feelings while buying. That is, as illustrated in compulsive buyers purchase impulsively to help uplift their feeling. In addition, greater product variety or utilitarian values can give compulsive buyers a way to achieve more positive feelings (McAlister & Pessemier, 1982). As a result of an enjoyable experience that is associated with a positive, uplifting feeling, then, according to Csikszentmihalyi (2022), a flow state can arise.

Correspondingly, a further hypothesis, H5, is assumed:

H5

The flow state drivers will have a positive effect on the flow experience.

Figure 6 presents activity-trait nature of flow state drivers as inferred from Csikszentmihalyi (2022) and the above discussions.

Fig. 6
figure 6

The flow state drivers

Full size image

Although flow experience is crucial, purchasing behavior may not directly relate to it. According to Csikszentmihalyi (2000), the flow experience state is a mental or emotional condition that keeps people focused on their activities (products or services). The flow experience state is a prerequisite for further influencing a psycho-cognitive state that is thought to be more likely than a simple flow state to motivate customers to make purchases. The two constructs that this study takes into account for the psycho-cognitive states that are more buy action-oriented are consumer addiction and loyalty. The incentive sensitization hypothesis of addiction can be used to understand the consumer addiction phenomenon (Robinson & Berridge, 2008). An addiction to live-streaming e-commerce can be interpreted as incentive sensitization, which results in a bias of attentional processing toward live-streaming e-commerce stimuli, as deduced from drug-associated addiction. Addicts repeatedly expose themselves to live-streaming e-commerce events, honing a psychological mechanism to induce purchase.

Depending on the situation, an enhanced incentive salience may appear in behavior through implicit (as unconscious desire) or explicit (as conscious seeking) mechanisms (Robinson & Berridge, 2008: 3137). According to the consumer behavior vocabulary, compulsive buying is the implicit (unconscious desire) process, while loyalty-associative is the explicit mechanism. Compulsive purchasing behavior (CBB) is defined by a robust and uncontrollable drive to buy things and mounting stress that can only be eased by purchasing. In addition to having an uncontrollable urge, compulsive conduct is commonly understood to have “repetitive and seemingly deliberate” activities that are “done according to particular rules or in a stereotyped pattern”. Loyalty and forced purchase are positively correlated in this sense. Similarly, Lim et al. (2020) distinguished between two types of compulsive purchasing behavior. Compulsive and obsessive–compulsive. Impulsive buying occurs when a consumer wants to compulsively buy something (Fenton-O’Creevy et al., 2018).

Consequently, the following theories are put forth.

H6

Consumer loyalty will predict consumer addition to live-streaming e-commerce.

H7

Consumer addiction will predict compulsive buying.

H8

Consumer loyalty will predict compulsive buying.

Arranging H1 to H8 into the proposed conceptual model yields Fig. 7.

Fig. 7
figure 7

The conceptual model with hypotheses

Full size image

4 Method

4.1 Procedure and sample

Before the main study, a pre-test and a pilot study were conducted to see if the research tool and data collection strategy were appropriate for the study, if the measurements accurately reflect the research tool has intended meaning and purpose and if they were used to survey a small subset of consumers who had experience with live-streaming e-commerce.

The data was gathered using a self-administered questionnaire, which gave participants the freedom to complete the survey on their own. The questionnaires were disseminated via social media and during live streaming events with the anchor’s consent. The questionnaire was pre-tested in June 2022 using a sample (n = 5) of professors with experience in e-commerce to ensure that the questions were accurately answered and that respondents could understand their meanings. There were some updated questions. Later, in July 2022, a pilot survey of a second independent sample (n = 30) was conducted to check readability and comprehension, the time it took to complete the questionnaire, and the fundamental validity and reliability tests. The reliability, represented by Cronbach’s alpha, is over 0.85.

4.2 Measurement instrument

In order to measure the different variables included in this study and ensure that these items fit the context of live streaming e-commerce, a questionnaire was compiled based on relevant literature and carefully revised. Positive responses such as loyalty are essential but it still lacks the revenue growing power of firms intending to push the performance to a new height (Ye & Tan, 2022a, 2022b). The following presents the definition of the construct and the measurement items:

Compulsive buying tendency: A person with a tendency for impulse buying often experienced as irresistible, intrusive, or senseless (Muller et al., 2005). The construct and the measurement items include CBT1, CBT2, CBT3, CBT4.

Innovation: Personal innovation refers to individuals who are willing to try new things. The construct and the measurement items include I1, I2, I3.

Social Value: The benefits received in the aspect of the social domain (Ciccarino et al., 2022).The construct and the measurement items include SV1, SV2, SV3.

Utilitarian Value: The utility aspect of consumers’ evaluations of the consumption experience (Ryu et al., 2010).The construct and the measurement items include UV1, UV2, UV3.

Social impact: Social impact refers to a change in a person's behavior, attitude, and beliefs caused by another person or group (influencing factor) (Kelman, 2001: 11).The construct and the measurement items include SI1, SI2, SI3.

Trust: What is built on faith is trust. That is, the behavior of service providers is based on the interests of consumers (Martinez & Rodriguez del Bosque, 2013; Utz et al., 2022).The construct and the measurement items include T1, T2, T3, T4.

Enjoyment: Enjoyment is a pleasant state from the services (Lee & Park, 2022).The construct and the measurement items include E1, E2, E3.

Flow Experience: According to Csikszentmihalyi (1975) and Schiefele & Csikszentmihalyi, 1995), flow is an experience that manifests the merging of action and awareness, potency (alert, active, etc.), cognitive efficiency (concentration, etc.).The construct and the measurement items include F1, F2, F3, F4, F5.

Loyalty: Loyalty is defined as consumers’ commitment to the repurchase or attitude or behaviors of recommending the products or services to others (Gao & Huang, 2021).The construct and the measurement items include L1, L2, L3, L4, L5.

Addiction: Consumer addiction adapts some relevant constructs, such as Internet addiction, which refers to “a person who has lost control of their internet usage” and has been excessively surfing the internet in a “state” (Israeli et al., 2019: 151). The construct and the measurement items include A1, A2, A3.

Compulsive Buying (Fenton-O’Creevy et al., 2018), The construct and the measurement items include CB1, CB2, CB3.

5 Analysis and results

The quantitative survey generated 517 valid responses, of which 42.7% of men and 57.3% of women participated. The respondents were split around 50–50% between the 18–25 and older age group. The majority (51.5%) have at least a bachelor’s degree. The bulk, at 36.2%, are employees of the company, followed by students at 25.3%, government agencies at 18.2%, and independent contractors at 11.6% (Table 1).

Table 1 Demographics and live-streaming e-Commerce experience profiles
Full size table

The study uses structural equation modeling (SEM), and multi-perceptron artificial neural networks (NN) for data analysis. SEM is appropriate for model validation or theory testing (Hair et al., 1998). ANN simulation is appropriate when the study concentrates on theory formulation and prediction rather than thorough theory confirmation.

5.1 Measurement robustness

The research constructs’ reliability and validity were assessed initially. To assess the instrument’s convergent validity, loadings had to be greater than 0.7 and average variance (AVE) had to be greater than 0.5. (Churchill, 1979; Carmines & Zeller, 1979; Fornell & Larcker, 1981). The constructs were internally consistent since their composite reliabilities were higher than 0.7. The correlations between the constructs that were less than the square root of the AVE support the discriminant validity. Tables 2, 3 and 4 presents the reliability, convergent, and discriminant assessments.

Table 2 Reliability and convergent validity assessments
Full size table
Table 3 Discriminant validity for antecedents to flow structure
Full size table
Table 4 Discriminant validity for flow to consequence structure
Full size table

5.2 Neural networking simulation

This section results from an artificial neural network simulation as a pre-structural equation modeling (SEM) analytics guide. SEM is appropriate for traditional linear statistical analysis, but neural networks get over this restriction. The ANN has one input, one output, and one hidden layer, as illustrated in Table 4. One hidden layer is adequate for continuous functions, but multiple hidden layers are advised for discontinuous functions. Each layer consists of neurons linked together through feed-forward and backward propagation through a changeable synaptic weight. As indicated in Table 5, the input of each neuron is multiplied by its synaptic weight, and then this signal is converted into an output value, which is then used as the "identity" function. The calculated error propagates backwards through the network and is learned through repeated supervision. And used to modify all synaptic weights to reduce estimation errors.

Table 5 Neural network simulation
Full size table

The ANN structure and the predictive regression outcome are given in Fig. 8. There is a good match between the measured and the predicted in the regression plots in Fig. 8. Trust tops the list as the most critical predictor of flow experience state, followed by compulsive tendency, social value, enjoyment, utilitarian value, social influence, and personal innovativeness, in sequence.

Fig. 8
figure 8

Multilayered perceptron neural network structure and simulation results for flow experience

Full size image

5.3 Structural model and hypothesis testing

The hypotheses and constructs’ relationships were analyzed based on examining standardized paths. The model accounted for 70 per cent of the variation in the flow state and 61 per cent of the variation in compulsive buying. The results indicate that trust (β = 0.20, t = 3.769, Sig. 0.000), compulsive buying tendency (β = 0.21, t = 5.881, Sig. 0.000), social value (β = 0.19, t = 4.484, Sig. 0.000), utilitarian value (β = 0.14, t = 3.467, Sig. 0.001), and enjoyment (β = 0.20, t = 3.432, Sig. 0.001) were statistically significant in explaining the flow experience. Thus, H3–H5 were supported with the following exceptions: personal innovativeness (β = 0.08, t = 1.639, Sig. 0.102), and social influence (β = 0.01, t = 0.277, Sig. 0.782, not shown the line, as insignificant) are not significant predictors of flow experience.

In addition, H1 is supported from the predictive strengths of the following constructs (see Fig. 9): compulsive buying tendency (β = 0.15, t = 0.491, Sig. 0.000), innovativeness (β = 0.26, t = 7.147, Sig. 0.000), social value (β = 0.30, t = 8.601, Sig. 0.000), utilitarian value (β = 0.23, t = 7.405, Sig. 0.000), and social influence (β = 0.10, t = 3.537, Sig. 0.000). For the support of H2, the evidences of the path coefficients that predict the strength of the antecedent factors’ predictive capability of enjoyment are: compulsive buying tendency (β = 0.09, t = 2.523, Sig.0.012), personal innovativeness (β = 0.22, t = 5.106, Sig. 0.000), utilitarian value (β = 0.28, t = 7.234, Sig. 0.000), social value (β = 0.16, t = 0.156, Sig. 0.000). H1 is supported except for social value, with β = 0.03, t = 0.651, Sig. 0.515. Trust is an additional predictor of enjoyment, with β = 0.20, t = 3.865, Sig. 0.000.

Fig. 9
figure 9

SEM—flow model

Full size image

The incremental and absolute fit indexes of the SEM in Fig. 9 fit the threshold requirement (Hair et al., 1998): \({\chi }^{2}\)/df = 0.078, p = 0.78, NFI = 1.000, RFI = 0.999, IFI = 1.000, TLI = 1.008, CFI = 1.000, RMSEA = 0.000. The full version of Fig. 9 with the observation and measurement items is shown in Fig. 10, and the fit indexes also satisfy the model fit requirement (Hair et al., 1998).

Fig. 10
figure 10

SEM—flow model with measured items

Full size image

Hypotheses H6 to H8 are supported by the SEM result shown in Fig. 11. The fit statistics are: \({\chi }^{2}\)/df = 0.4676, p = 0.00, NFI = 0.937, RFI = 0.927, IFI = 0.953, TLI = 0.946, CFI = 0.953, RMSEA = 0.072. Evidences of the path coefficients in support of H6 to H8 are: addiction → loyalty (β = 0.21, t = 7.404, Sig. 0.000), addiction → loyalty (β = 0.39, t = 11.691, Sig. 0.000), and loyalty → compulsive buying (β = 0.24, t = 8.319, Sig. 0.000), respectively. The fit statistics are: \({\chi }^{2}\)/df = 0.739, p = 0.390, NFI = 1.000, RFI = 0.996, IFI = 1.000, TLI = 1.001, CFI = 1.000, RMSEA = 0.000.

Fig. 11
figure 11

SEM—compulsive buying

Full size image

By including the observation items, the SEM, presented in Fig. 6, also complies with the fit index statistics (Hair et al., 1998): Statistics: \({\chi }^{2}\)/df = 3.79, p = 0.00, NFI = 0.952, RFI = 0.944, IFI = 0.964, TLI = 0.958, CFI = 0.964, RMSEA = 0.07.

Figures 12 and 13 present the density plots of the relationships between the constructs of the study (Fig. 14).

Fig. 12
figure 12

SEM—flow model with measured items

Full size image
Fig. 13
figure 13

Predicting compulsive buying

Full size image
Fig. 14
figure 14

Predicting flow experience

Full size image

5.4 Comparative analysis

The statistical tools to analyze consumer views across various demographic and usage profiles include correlation analysis, ANOVA, and t-tests, with results given in Table 6. The female consumers had higher positive perceptions of social influences. High school students have more favorable perceptions of the constructs when compared to older consumers. The perceptions of the years of e-commerce experience follow a parabolic upward-downward trend, peaking at “1–2 years.” The involvement in weekly live-streamed shopping and the monthly disposable income correlate favorably with the constructs. The “Mushroom Street” platform has the most outstanding consumer ratings across all constructs. Compared to consumers who have had unfavorable experiences, individuals who have never had a bad experience with live-streaming e-commerce offer more positive perceptions. Figure 15 provides the density plots of the statistically significant ones.

Table 6 Comparative analysis
Full size table
Fig. 15
figure 15

Some statistically significant differences

Full size image

6 Discussion

This study aims to bridge a knowledge gap on what influences the flow experience state (immersion and concentration state) and how flow drives speedy purchasing or compulsive buying. Since there is a growing desire for materialism (Xu, 2008) and increased product availability and accessibility through both offline and online retail channels, compulsive shopping is emerging consumer behavior. Still, it is understudied (Tarka & Kukar-Kinney, 2022). Ridway et al. (2008) and Mason et al. (2022) note that compulsive buying might not entirely imply a loss of compulse control over obsessive shopping-related thoughts and feelings. Their findings align with this study as evidenced by the loyalty and addiction base of compulsive buying, implying both planned and unplanned characteristics and rational versus impulsiveness in purchasing.

Specifically, flow theory provides the central concept in this study. With flow as the focus, it is shown that trust and enjoyment are the most critical flow-driver mediators, whereas loyalty and addiction are the post-flow mediators. The SEM-validated model reveals a structure that depicts antecedents and consequences of flow experience with embedded mediators. Practitioners often acknowledge the practical value of simplicity in model-based management (Ashton, 2007; Denning, 2021; Schwaninger, 2010).

Trust and enjoyment that this study proposes, as the antecedents to flow state inferred from the flow-theory concepts discussed in Csikszentmihalyi’s publications (), are antidotes to anxiety and boredom, respectively. Both anxiety and boredom prevent flow experiences from arising (Csikszentmihalyi, 1975, 1991, 1996). The rationale is as follows: Trust shifts consumers from a sense of disequilibrium (anxiety, Stephan & Stephan, 1985; Turner, 1988) to equilibrium. While the extant literature highlights a direct relationship between anxiety (e.g., panic over COVID-19) and compulsive buying (Omar et al., 2021), this study presents otherwise in live-streaming e-commerce, which leverages the power of trust (an antidote to anxiety). In addition, roles of social values and influence share the parasocial interaction in live-streaming e-commerce in Rungruangjit (2022). Enjoyment is shown to shape trust and steer customers to buying, sharing the finding of Hanaysha (2022), Kim et al. (2007), Musarra et al. (2022). Both trust and enjoyment exhibit, to some extent, dependable cognitive evaluation of the purchasing experience (Chun et al., 2017).

In the live-streaming e-commerce context, the empirical result of this research supports the flow activities characterized by social value, utilitarian value, and social influence. Flow activities are the sources of flow experience (Leung, 2020).

According to the comparative t-test, women see social influence as much more positive than men. The results are consistent with that of Venkatesh et al. (2000). They found that women are more likely to be sensitive to the opinions of others and, as a result, value social influence more when developing bonds of trust and pleasure. While other studies have noted that older people are more likely to place more salience on social impacts, such as Rhodes (1983) and Venkatesh et al. (2003), the findings of this study are inconsequential; the tendency rises to peak at age group “26–35,” but go down from there. The “Mushroom” platform draws the most favorable perceptions. Usage profiles deliver loyalty attributes: The higher the frequency of weekly live-streaming shopping participation, the more favorable the perceptions of the constructs, with correlations between 0.26 and 0.35; monthly disposable income is another significant factor, with correlations ranging between 0.11 and 0.19, sig. at 0.01 (2-tailed). Consumers who have had no negative experience with the live-streaming show more favorable perceptions as well, with correlation coefficients between − 0.13 and − 0.26.

6.1 Theoretical contributions

The validated SEM structure shares the pattern of the SOR model (Mehrabian & Russell, 1974), capturing the stimuli (both personal and flow activity levels), the organism (trust, enjoyment, and flow experience), and responses (represented by loyalty, addiction, and compulsive buying).

By linking to social influence and values, trust facilitates the interactions between the live-streaming anchor and the consumers, leading to a parasocial relationship (Huang et al., 2022). As an antidote to anxiety and uncertainty, the trust-based factor in shaping flow experience will find applicable theories of uncertainty reduction and signaling (Lu & Chen, 2021). Uncertainty refers to the degree to which future environmental conditions cannot be precisely forecast for various reasons (Pfeffer & Salancik, 1978). When ambiguity lessens, trust develops (Melewar et al., 2017). The fact that trust has both cognitive (e.g., positive assessment of the quality of goods, Ozdemir et al., 2020) and affective components (e.g., displaying genuine care and concern, Ozdemir et al., 2020) leads to trust as a significant predictor of both consumer addiction and loyalty and, in turn, induces compulsive buying, in which signaling theory has a robust validity (Dirks & Ferrin, 2002).

This study demonstrates that flow activities that induce flow experience should deliver social and utilitarian values and be able to foster social influence. Social influence represents a two-way communication flow, which is a significant theme of social interaction theory that aims to explain the interactive relationships among people (Liu, 2003; Varey, 2008) to enhance communication quality and reduce the uncertainty of consumption in live-streaming commerce (Yang et al., 2022). It implies that social influence shapes the trust level of consumers. The activity level is the subject of environmental psychology, a field concerned with transactions between humans and their environments (Tam & Milfont, 2020). As noted in Csikszentmihalyi and Bennett (1971), when the action (e.g., the live-streaming shopping) resonates with the environment, enacted, for instance, by supportive environment (e.g., prompt feedback to the actors), flow experience can more readily arise.

The positive relationship between addiction and loyalty, which this study identified, also shares the finding of Le (2020) and Mrad et al. (2020). Together, they contribute to compulsive buying. Thus, compulsive buying has both uncontrollable (urging) and rational, loyal elements. The incentive sensitization hypothesis of addiction can be used to understand the consumer addiction phenomenon (Robinson & Berridge, 2008). An addiction to live-streaming e-commerce can be interpreted as incentive sensitization, which results in a bias of attentional processing toward live-streaming e-commerce stimuli, as deduced from drug-associated addiction. Addicts repeatedly expose themselves to live-streaming e-commerce events, honing a psychological mechanism to induce purchase. Depending on the situation, an enhanced incentive salience may appear in behavior through implicit (as unconscious desire) or explicit (as conscious seeking) mechanisms (Robinson & Berridge, 2008: 3137), which provides the logic supporting loyalty and addiction as predictors of compulsive buying.

6.2 Practical implications

Consumer behavior should be guided by notions of social capital, social exchange, and trust (Irun et al., 2020; Schilke & Cook, 2015). The social context is an essential stimulant in a socio-commercial environment like live streaming e-commerce. This study gives several examples, such as the capacity of perceived social values to increase consumer trust predictably; and the social influence on consumers to elicit affective emotions like enjoyment through interactions and support from others throughout the decision-making process and in the environment of live-streamed shopping.

7 Conclusion

This study draws on the flow theory (Csikszentmihalyi, 1975, 1990a, 1990b, 1991, 2003) as its foundation to propose the flow-state drivers and mediators, post-flow state mediators, to examine compulsive buying. The study assumes that the flow experience state, which has a sense of concentration, deep involvement, and a sense of being in control (Kazancoglu & Demir, 2021), is highly desirable for anchors or hosts of live-streaming shopping who would like the consumers to be engrossed in the temporal and action cycle of the event by feeling motivated in moments of pleasure. However, the study also assumes that it has the potential to trigger an urge for compulsive purchases.

The quantitative method employs artificial neural network (ANN) and structural equation modeling (SEM). ANN is particularly suitable for non-linear relationships of constructs. Theoretical validation shares a stimulus-organism-response (SOR) structure. Activities that promote flow should be able to deliver social and utilitarian values and foster social influence. The stimulus also exhibits a strong personal influence, as evidenced by personal innovativeness and compulsive buying tendency. The important organism factors are trust and enjoyment, which function as counterbalances to anxiety and boredom, respectively. According to Csikszentmihalyi (1991, 1996, 2003), anxiety and boredom are the barriers that prevent the establishment of a flow experience. Post-flow state mediators, which include loyalty and addiction, can significantly leverage the relationship between flow and compulsive buying.