1 Introduction

China Internet Network Information Center [9] reports about 617 million live streaming audiences, which accounts for over 60% of the total Internet users at the end of 2020, and over 15% of e-commerce sales in China, approximated at over 300 billion dollars in 2021 [27]. A subset or an extension of e-commerce, social commerce (s-commerce), has gained momentum, partly being driven by the platform design in facilitating the online participations between the presenter and, for instance, the live streaming viewers [17]. In today's era, live streaming has gradually occupied a significant place in e-commerce. More and more customers, manufacturers and investors realize this benefit. Though the live streaming concept in electronic commerce is not new, its use as a critical revenues-driver and strategy for social- and e-commerce platforms expansion is only a recent movement, and different countries are gaining the pace of influence differently. In Forbes, Greenwald [12] reported that live streaming e-commerce is another recent strategic catalyst propelling China's highest e-commerce of as a per cent of total retail sales (over 35% in 2019) to even higher, crossing $1 trillion in 2020. The creativity part of the live streaming fad in e-commerce in China, according to Greenwald [12], is about "promoting and selling goods through influencer streams on their own social media channels, most often housed on China's online shopping malls", and as such, sprouting Alibaba's Taobao Live, with the lion's share of live-streaming at the time of reporting in Forbes at around 80%. Although Gen-Z and Millennials are the primary audience group of acceptance, the middle-aged Chinese and seniors are now widely joining the bandwagon [12]. More specifically, nowadays, live streaming in China has transcended the original live streaming events, without much interaction and for video content purposes, to all about shopping, as evidenced in Taobao Live and its rival JD.com. Seeing the trend in acceleration, the e-tail giant JD.com has also ramped up its live-streaming efforts, opening up many bases across China. According to Williams [43], live streaming e-commerce in China is a platform attractor that exploits the live streaming talent resources of the presenters. Based on the previous research on live streaming, the real-time interaction between the audience and the broadcaster in live streaming brings a strong sense of existence to the audience, affects the audience's needs, and then affects the attitude and behavior of the potential audience consumers [13]. On the impact of live delivery on customers' purchase intention: Tong [36] found that the interaction, authenticity, and vividness significantly affect consumers' purchase intention, social presence, and spatial, social presence play an intermediary role. Wongkitrun Grueng and Assarut (2018) found that customers read the barrage while watching the live delivery. These barrages will make them feel the attitudes of other customers and affect customers' perceived value and behavioral intention.

The recent trends of shopping-targeted live streaming e-commerce, now being propelled by leading e-commerce and social commerce platforms in China, provide the frontrunner advantage for research efforts to enhance the theoretical understanding of the phenomena. In addition, as live streaming is a relatively untapped phenomenon in the West [12, 43], the theoretical and practical implications can provide the necessary psychological boost to accelerate investments, as noted, for instance, through the theory of planned behavior [41, 45]

2 Literature Review

2.1 Stimulus–Organism–Response (SOR) Theory

Stimulus–organism–response (SOR) theory is a widely acknowledged parsimonious model capable to validly explain consumer and human behaviors. The SOR theory was motivated by Woodworth [42] and refined and made popular by Mehrabian and Russel [29] and Jacoby [19]. By tracing the SOR theory of consumer behavior to having strong root in environmental psychology, it is also not difficult to find the influences of other researcher scholars with a strong emphasis on environmental psychology thrusting the SOR concept further to the frontiers. The rationality lies in the fact that consumers are constantly in search of cues to help them judge the functionality and utilities of the products and services, and physical environment, or Kotler [23] calls it atmospherics, is one important variable that is capable of judging the capabilities of the service or product providers and the quality of the products and services [40]. Due to multiple environmental cues in play [38], it thus makes stimulus construct as widely flexible that is contingent upon the context of applications. Moreover, the SOR model provides a framework or conceptual platform to influence a presenter of organistic, psychological, perceptual states and experiences [17] of consumers, to further impact on the responses.

In relation to livestreaming, Ming et al. [27] discusses the over-emphasis of the prior research of live streaming commerce in the motivation domains of live streamers, for instance, found in Chen and Lin [8], and ignore other aspects. As a result, Ming et al. [27] study the influences of live streaming’s environmental factors, which include social presence of live streaming platform and viewers, and the streamers. Ming et al. [27] approaches the understanding of the social presence of live streaming platform and viewers from the environmental psychological perspective, and simultaneously take note a possible psychological distance effect between the viewers and streamers.

Due to the interactive features of s-commerce and the recent technological upgrades of many e-commerce platforms, interactivities become common driver for brand community engagement [35]), which promotes social word of mouth, and causes “flow state” or “flow experience” as an important organistic phenomenon to arise [17]. Flow experience is a concept which will be employed by this research, and its original conceptual map would be extended to guide the organistic variables development in the SOR model of this research. Owed to Csikszentmihalyi [10], flow experience of consumers is an organistic state that reflects the absorptive or concentrated states of the consumers in the enjoyment of the stimulating events. Thus, to some extent, it is what the marketers hope to be able to achieve, as the so-called “optimal” experience state of consumers [11]. Though s-commerce and the contemporary e-commerce platforms do have the technological advantages to design and promote optimal experience atmospherics, research studies relating to the use of flow states are scarce [17, 27], notwithstanding a more detailed examination of the flow experience states of consumers. To this end, this research fills the gaps in the extant literature through contributing towards identifying not only the stimulating factors from the SOR framework, but also by looking directly into the original conceptual map of “flow experience” of Csikszentmihalyi [10] in searching for the organistic stimulation.

3 Use of Flow Theory in Organistic Variables Development

Flow theory centralizes on the “flow experience” concept which is defined as the “holistic experience that people feel when they act with total involvement” ([10], p. 36). The significant role of the flow experience phenomenon in livestreaming e-commerce is that, when the consumers are in the flow states, as reflected by the definition in Csikszentmihalyi [10], the consumers would be absorbed into the livestreaming events, and although they may lose some level of self-consciousness [32], they do maintain control of the environment, just like in game-playing context [20]. In other words, flow is “the subjective experience of effortless attention, reduced self-awareness, and enjoyment that typically occurs during optimal task performance” [15]. The depiction shown in Hartmart et al. (2015) relating to flow experience definition is shown in Fig. 1, which sees the opposites of boredom, that is enjoyment, in the play. Accordingly, this research exploits the definitional context of “flow experience” [10] and the various enrichments from other researchers such as Hartmart et al. (2015), to guide the deductive or immersion development of variables that can be used to characterize the factors influencing flow experience of livestreaming viewers, as potential consumers, in e-commerce platform. The outcome is shown in Fig. 2.

Fig. 1
figure 1

(Source: Developed for this Research)

The proposed flow conceptual model

Full size image
Fig. 2
figure 2

The conceptual model for the research

Full size image

By the fact that as anxiety increases, the flow experience state of the consumers is reduced [10] [24], this research extends by posing the question around: What if we have the opposites of anxiety and boredom, as originally advocated in Csikszentmihalyi [10] in the regions without flow states, would the opposites become parts of the factors supporting and enabling flow experience of the consumers or livestreaming viewers? If yes, by means of empirical validation and some other theoretical supports from the extant literature, then, it certainly is contributing to the original flow theory, and is also providing a groundwork for identifying the theory to study the “organism” states of the SOR model in livestreaming e-commerce studies.

As noted in Fig. 1, when consumers form trust with the service providers, the anxiety is reduced for the consumers have gained the confidence, and thus, have reduced their uncertainties and risk perceptions over the products and services offered [37]. In addition, in anxiety, people tend to procrastinate [14], and by deductive inference to the livestreaming e-commerce research here, it is hypothesized that trust should lead to flow experience, which in turn, lead to compulsive consumer behavior.

4 Research Hypotheses

Not just in the first-entry and mass adoption stages [39], but also in today's technological advancement stage, compulsive buying behavior is an often-overlooked variable in interactive e-commerce and s-commerce [27]. Nonetheless, based on the above-mentioned flow experience idea, compulsive purchase behavior appears to be a critically important dependent variable. Since stated in Ajzen [1,2,3], the rationale is that the flow experience state implies non-procrastination of customers' decision making, as otherwise it would be tilted toward rationalized or planned activities [7, 16]. Furthermore, flow has a strong emotional impact, and compulsive buying is a troublesome and uncontrollable buying activity that is also known as compulsive shopping, addicted buying, impulsive-compulsive buying, or spendaholic. Prior to the internet era and today's tremendous technical advancements, compulsive customer behaviors were mostly influenced by product design, retail stimuli, and personal variables [30]. However, in today's e-commerce, characteristics including performance expectancy, effort expectancy, synchrony, social impact, interactivity, system and service quality, and vicarious learning can all influence compulsive purchase behaviors. The role of compulsive buying behavior has been discovered to be more widespread in the internet environment, which accounts for about 30% of all online purchases [26], and has the compulsive and addictive nature of buying [25].

Customers will generate both cognitive and affective perceptions [44] when aroused, according to the SOR idea, and perception of values, such as utilitarian and hedonic value [18], is one of the organic states of customers. The correlational correlations are proven in Kim and Thapa [21], albeit the rationale for linking perceived value to impact the flow experience of tourists are not presented. In Kim and Thapa [21], four categories of perceived values are identified as influencing flow experience: quality, emotional, price, and social. Instead, this study focuses on the directional focus of flow experience, inferring to the characteristics of flow as "a focus of awareness, loss of self-consciousness, responsiveness to clear goals and unambiguous feedback, and a sense of control over the environment" [21]: 374), and perceived values are at least, from the definition, representing the instant feedback of the stimuli translated to organic perceptions, as well as compulsive behaviors.

Given the foregoing discussions, as demonstrated by the use of the flow theory to generate trust and flow experience, and taking customers' perceived values as the focal attention and flow experience characteristics ([21]: p. 374); [10], a conceptual model shown in Fig. 2 is developed.

The conceptual model, in particular, has three broad-based hypothetical structures, namely H1:

H1: Stimulus factors, consisting of performance expectancy, effort expectancy, synchronicity, social influence, interaction, system and service quality, and vicarious learning, significantly predict the organic states of the customers in terms of utilitarian value, hedonic value, social value, perceived value, trust, and flow experience.

In variable-to-variable forms, the following hypotheses provide the details.

H1a1: Performance expectancy is a significant predictor of utilitarian value.

H1a2: Effort expectancy is a significant predictor of utilitarian value.

H1a3: Synchronicity is a significant predictor of utilitarian value.

H1a4: Social influence is a significant predictor of utilitarian value.

H1a5: Interaction is a significant predictor of utilitarian value.

H1a6: System and service quality is a significant predictor of utilitarian value.

H1a7: Vicarious learning is a significant predictor of utilitarian value.

H1b1: Performance expectancy is a significant predictor of hedonic value.

H1b2: Effort expectancy is a significant predictor of hedonic value.

H1b3: Synchronicity is a significant predictor of hedonic value.

H1b4: Social influence is a significant predictor of hedonic value.

H1b5: Interaction is a significant predictor of hedonic value.

H1b6: System and service quality is a significant predictor of hedonic value.

H1b7: Vicarious learning is a significant predictor of hedonic value.

H1c1: Performance expectancy is a significant predictor of social value.

H1c2: Effort expectancy is a significant predictor of social value.

H1c3: Synchronicity is a significant predictor of social value.

H1c4: Social influence is a significant predictor of social value.

H1c5: Interaction is a significant predictor of social value.

H1c6: System and service quality is a significant predictor of social value.

H1c7: Vicarious learning is a significant predictor of social value.

H1d1: Performance expectancy is a significant predictor of perceived value.

H1d2: Effort expectancy is a significant predictor of perceived value.

H1d3: Synchronicity is a significant predictor of perceived value.

H1d4: Social influence is a significant predictor of perceived value.

H1d5: Interaction is a significant predictor of perceived value.

H1d6: System and service quality is a significant predictor of perceived value.

H1d7: Vicarious learning is a significant predictor of perceived value.

H1e1: Performance expectancy is a significant predictor of trust.

H1e2: Effort expectancy is a significant predictor of trust.

H1e3: Synchronicity is a significant predictor of trust.

H1e4: Social influence is a significant predictor of trust.

H1e5: Interaction is a significant predictor of trust.

H1e6: System and service quality is a significant predictor of trust.

H1e7: Vicarious learning is a significant predictor of trust.

H1f1: Performance expectancy is a significant predictor of flow experience.

H1f2: Effort expectancy is a significant predictor of flow experience.

H1f3: Synchronicity is a significant predictor of flow experience.

H1f4: Social influence is a significant predictor of flow experience.

H1f5: Interaction is a significant predictor of flow experience.

H1f6: System and service quality is a significant predictor of flow experience.

H1f7: Vicarious learning is a significant predictor of flow experience.

The perceived value of customers in a live delivery situation influences customer happiness and purchase decisions. Despite the fact that value has received a lot of attention in recent years in the marketing literature, there isn't much research on it in the context of online purchasing, particularly in terms of empirical hypothesis testing [31]. Bridges and Florsheim [6] discovered that the practicability of an online shopping website will cause clients to have a purchase intention based on the experimental condition of an online shopping website. As a result, H2 is considered as the second hypothesis:

H2: The organic states of customers, which include utilitarian value, hedonic value, social value, perceived value, trust, and flow experience, significantly predict the customer responses in terms of compulsive buying behaviors.

As a result, according to Gadosey et al. [14], customer behavioral compulsiveness becomes a critical ultimate variable for live broadcasting in e-commerce. Because live streaming is real-time interactive, as seen in Taobao's markets and current e-commerce platform design [12, 43], it should be targeted. It also fits into Csikszentmihalyi's [10] notion of flow theory, in which consumers show a desire to buy that outweighs self-consciousness and self-control (Wang, Lu, and Wang, 2020).

The following hypotheses are the breakdown versions of H2, namely:

H2a: Utilitarian value is a significant predictor of compulsive buying behavior.

H2b: Hedonic value is a significant predictor of compulsive buying behavior.

H2c: Social value is a significant predictor of compulsive buying behavior.

H2d: Perceived value is a significant predictor of compulsive buying behavior.

H2e: Trust is a significant predictor of compulsive buying behavior.

H2f: Flow experience is a significant predictor of compulsive buying behavior.

5 Research Method

This research supports a positivistic worldview that promotes a methodological approach capable of empirically examining and verifying objectivist facts. Customers who have shopped via live streaming shopping platforms in leading e-commerce live-streaming platforms in China, including Taobao.com, are surveyed using convenience-based sampling. To assess the quality of the measuring items relevant to the constructs, the questionnaire instrument is pilot-tested.

6 Population and Sampling

The live streaming viewers who shop on the e-commerce site are the study's target population. When choosing the samples, it is necessary to ensure that the respondents have seen at least one live streaming session on the e-commerce site. However, because it is difficult to pinpoint specific groups of live streaming users or enthusiasts, sampling in many cases will have to resort to convenience sampling, which is also dependent on snowballing [34].

Given the general statistical z-value formula [4], the sampling size for the questionnaire-based survey is 517:

$$s= \frac{{z}^{2}(p)(1-p)}{{E}^{2}}= \frac{{1.96}^{2}(0.5)(1-0.5)}{{\left(0.05\right)}^{2}}=517$$
(1)

Eventually, 207 men and 310 women participated in the survey, with a total of 517 sample size, which is also in line with an online poll about live streaming purchases, which found that the majority of people who watch live streaming are women.

The 18–25 and 26–33 age groups were the largest, accounting for 243 and 111, or 45.85% and 20.94%, respectively. As for the monthly disposable amount, the most significant number is more than 5,000 Yuan, with 118 people, indicating that the trend of watching live streaming with goods has moved to middle and high-income groups in China. In terms of occupation, private enterprises accounted for the highest proportion of 36.42%.

7 Questionnaire Design

The questionnaire is a measurement tool that allows researchers to inspire respondents to answer by standardizing questions and response categories, as well as employing suitable phrasing, question flow, and appearance to achieve the study objectives [5]. Most crucially, Brown et al. [5] state that a successful questionnaire design is one in which the layout gives immediate readiness for analysis. Nonetheless, questionnaire design presents obstacles and constraints for a research field that is still emerging, such as live streaming e-commerce, where most theoretical breakthroughs are still in their infancy. The length of the questionnaire design, which necessitates the omission of some additional factors, is a key problem and limitation in this study domain [33].

The stimulation factors are centralizing on (1) performance expectancy, (2) effort expectancy, (3) synchronicity, (4) social influence, (5) interaction, (6) system and service quality, and (7) vicarious learning received by the customers. At this point, the ultimate response variable being addressed is compulsive buying behavior. The questionnaire is split into two sections: the first contains the respondents' basic information, and the second has the particular items for 14 measurement variables. The questionnaire items were designed using a Likert five scale, with one representing "strongly disagree" and five representing "extremely agree." The finished questionnaire in this study sought the advice of experts and scholars in the field and was steadily enhanced to construct the initial questionnaire. The questionnaire was collected using Questionnaire Star software in this study. A total of 600 questionnaires were distributed, with 517 of them being found.

8 Results

8.1 Reliability Analysis

The consistency, repeatability, and stability of the test results, as well as whether the test results represent stability and consistency, and the actual features of the subject, are the most important aspects of survey reliability performance. When the same object is measured over and over, the results must always be consistent before they can be considered legitimate.

SPSS 25.0 software is utilized to test the questionnaire's reliability and validity in this paper. Table 1 displays the results. Cronbach's coefficients for all variables are greater than 0.9, indicating that the questionnaire is robustly reliable.

Table 1 The reliability analysis statistics
Full size table

9 Validity Analysis

The consistency between the test score and the traits you want to evaluate is referred to as validity. That is, whether the testing mark can accurately reflect the features that are being measured. Distinct variables have different measurement characteristics, so their measurement findings should be different as well. The overall KMO value of the questionnaire sample is 0.972, and the results of the Bartlett spherical test are significant, indicating that the sample variables are suitable for factor analysis (Table 2).

Table 2 The validity analysis statistics
Full size table

10 ANOVA and T-Test Results

The notable findings in Table 3 are as follows. First of all, within one year, the most recent e-purchase experience has more favorable perceptions than those having long years of experience in e-purchase. The more frequent weekly e-purchase, the more favorable are the perceptions across the stimuli and the organic states, flow experience, and compulsive purchasing, except for customer engagement and fan-becoming. The trend is also evident for those having frequent purchases during the e-live streaming session compared to those on least. Thus, fan-becoming is a weak spot that urgently needs improvement. Freelancers are the only career segment that differentiates significantly from the other career categories; their customer engagements and fan-becoming are also relatively low. Two age groups also show relatively more positive perceptions than other groups: age < 18 and between 26 and 35. Higher levels of customer engagement and fan-becoming are age groups 18–25 and 36–45. Rather than becoming fans following after the live-streaming sessions, the viewers continue to hop around from one platform to another platform, as shown in the positive correlations between the number of e-live streaming platforms used by the customers and all the ten theoretical constructs studied, as evidenced in the last row in Table 3.

Table 3 ANOVA and T-test results
Full size table

10.1 Confirmatory Factor Analysis

The SEM path model is depicted in Fig. 3. While Fig. 3 confirms the theoretical derivations presented in the literature review section, it also highlights a few key aspects that are often overlooked in the existing literature. Table 4—the SEM statistics, with Chi-square = 2991.995, Degrees of freedom = 920, shows both absolute and incremental SEM fits. The SEM fit is assured with values nearly 0.9. NFI is 0.910, RFI is 0.898, IFI is 0.936, TLI is 0.927, and CFI (Comparative Fit Index) is 0.935, which are the incremental model fitting statistics.

Fig. 3
figure 3

The validated structural equation model (SEM)

Full size image
Table 4 The SEM model fit summary statistics
Full size table

Specifically, Fig. 3 shows that performance expectancy, effort expectancy, synchronicity, social influence, interaction, system and service quality, and vicarious learning are the essential variables that positively influence compulsive buying, which is directly a result of a customer experiencing the touchpoint design of the platform system. The finding shows the ability of search popularity to positively influence performance expectancy, effort expectancy, synchronicity, social influence, interaction, system and service quality, and vicarious learning.

This paper has shown the constituent elements of live streaming through empirical analysis. Performance expectancy, effort expectancy, synchronicity, social influence, interaction, system and service quality, and vicarious learning will increase consumers' perceived value and further trigger consumers' loyalty to enhance purchase intention. Customer trust and flow experience can positively regulate the relationship between store live streaming and compulsive buying. Under the regulation of customer trust and flow experience, consumption and purchase intention will change.

11 Conclusion

In the Internet era, technology communication and innovation have entered a faster and more efficient pace. In this context, we can use a visual method to show the results of quantitative analysis, so as to show the research trend and background more clearly. Literature measurement based on knowledge map is a new scientific measurement method. The traditional method of literature analysis is a simple statistical analysis of literature, while the new method is in-depth excavation of literature. It uses deep learning method to recognize entities and their relationships in literature, and displays them as knowledge map. This new method opens up more possibilities for the study of knowledge.

The livestreaming anchor would strengthen the professional capabilities with the widespread availability of big data, marketers, and companies. On the one hand, the store should increase the anchor's training to improve their awareness of live broadcast items so that they can accurately and timely describe product quality, logo, and culture during the live delivery process, thereby increasing customer awareness and faith in the products.

On the other hand, in order to increase the quality of after-sales service, we must follow the state's relevant regulatory rules, closely monitor all links from manufacturing to after-sales, and guarantee that product quality and publicity are consistent. This research demonstrates the utility of a structural understanding of live streaming, which illuminates the values aspect of live streaming, by relying on the deductive approach of research methodology, using the existing literature as the intellectual base, and the structural equation modeling (SEM) method.

This research shows that the value of live streaming with goods is not only proved in reality. The academic community has also done a lot of research on it, which could provide a richer explanation of the statistical structure of the research finding. As noted in the research problem section, identifying a valid theoretical structure to the phenomenon of live streaming e-commerce [28] in China could provide a basis for investment decisions in other countries. As noted in Warsame and Ireri [41] and [45], the investment decision is favorable depending upon the perceived behavioral control, subjective norm perceived to be able to influence the investor and the industry, as a whole, and attitude of the investment. Although the theory of planned behavior provides a theoretical context to explain intentional human behavior in general, it implies the value or utility of theoretical understanding or structure of a researched phenomenon [22].

Based on the research finding and the limitation, further research can also use mixed qualitative in-depth interviews and quantitative surveys to identify the sellers' subtle concerns, specific motivations, and relevant socio-demographic characteristics so that a more holistic picture of livestreaming, like that of Taobao, can be represented. It would also establish the bridge between the perceptual domains of customers and the quantitative data offered by live streaming anchors. Future research can enrich the descriptive and explainable details of the stimulus–organism–response (S–O–R) model of customer behaviors to benefit business development, where the organic states reflect the perceptual details significant to influence compulsive buying behavior.The exploratory factor analysis (EFA) separates it, leaving only a single item. Nevertheless, with many significant differences revealed in ANOVA and T-tests, this study maintains a robust level of generalizable utility.