泛娱乐移动直播作为泛娱乐大背景下的一部分借着极强的互动性和内容的丰富性迅速占领市场，其强大变现能力引起了广泛舆论关注，但目前学术界和行业内关于泛娱乐直播用户付费意愿的研究较为缺乏，并且以往研究数据获取多基于问卷调查的方式，主观性较强。本文通过爬取某国内大型泛娱乐移动直播平台用户的真实数据，在测量变量的数据获取上保证了数据的客观性。通过学习总结前人对网络付费意愿、心流体验及心流体验对付费意愿的影响、移动泛娱乐直播用户付费意愿的相关研究，提出影响泛娱乐移动直播用户“付费意愿”的四个影响潜变量，分别为“网络互动”、“使用深度”、“社会影响”和“心流体验”。本文基于前人已验证得到“心流体验”的中介变量作用，进一步探索“网络互动”、“使用深度”、“社会影响”对于“付费意愿”在泛娱乐移动直播中的直接效用。提出假设和理论的结构方程模型，通过Amos软件实证分析，得到结论：在影响用户“付费意愿”的模型中，“心流体验”起到中介变量作用，用户的“网络互动”和“使用深度”通过作用于“心流体验”间接对“付费意愿”产生影响。而“社会影响”不通过中介变量，而是直接作用于“付费意愿”发挥效应。通过路径分析，我们发现“心流体验”和“社会影响”与“付费意愿”关系最大。 为进一步探索十二个解释变量“评论数”、“点赞数”、“弹幕数”、“人气点击数”、“发红包数”、“新关注用户数”、“分享直播数”、“贡献上榜次数”、“新增好友数”、“观看时长”、“观看直播数”、“进入直播频道次数”与三个被解释变量“充值金币额”、“购买礼物次数”、“消费金币额”的定量关系，本文建立基于逐步回归法的多因变量线性回归模型。由回归模型结果可以看到，并非所有指标都能最终进入因变量的回归模型中，对于三个被解释变量，与“充值金币额”显著相关的指标有“新增好友数”、“观看时长”、“贡献上榜次数”、“弹幕数”、“发红包数”。与“购买礼物次数”显著相关的指标有“观看时长”、“发红包数”、“分享直播数”、“评论数”、“新关注用户数”、“观看直播数”、“弹幕数”。与“消费金币额”显著相关的指标有“红包数”、“观看时长”、“新关注用户数”、“弹幕数”、“贡献上榜次数”、“观看直播数”、“评论数”。其中“观看时长”和“发红包数”对于三个被解释变量均起到相对重要的影响作用。它们分别从属于“心流体验”与“社会影响”的分类，与前文结构方程模型得到的结论相吻合。 因此在实际泛娱乐移动直播运营中，企业应当考虑增强“社会影响”，即社会接受度，树立正面直播形象，引导用户健康直播，将直播“有用化”，增强用户及用户熟人圈的使用信心，从而也能为企业带来更好的效益。在用户体验上，企业应当注重用户的“心流体验”，提高用户沉浸式体验，提升用户使用产品功能时的快感，在优化用户体验的同时，也能带来双赢的效果。基于三个定量回归方程，我们可以通过检测用户行为数据来预报用户付费意愿，有助于平台挖掘潜在用户。另外可以通过检测异常值，进而审查违规和异常消费，维持泛娱乐移动直播平台健康发展。

Pan-entertainment mobile live broadcasting, as a part of pan-entertainment background, quickly occupies the market by virtue of strong interaction and richness of content. Its strong liquidity has attracted wide public attention. However, there is a lack of research on the willingness of users to pay for pan-entertainment live broadcasting in academia and industry at present, and previous research data acquisition is mostly based on questionnaires, which has strong subjectivity. 。 This paper guarantees the objectivity of the data by crawling the real data of the users of a large-scale pan-entertainment mobile live broadcasting platform in China. By studying and summarizing the previous studies on the influence of network payment intention, cardiac fluid experience and cardiac fluid experience on payment intention, and mobile pan-entertainment live broadcasting users' payment intention, four potential variables affecting the "payment intention" of pan-entertainment Mobile Live Broadcasting users are proposed, which are "network interaction", "depth of use", "social impact" and "cardiac fluid experience". Based on the mediating variable function of "heart fluid experience" which has been verified by predecessors, this paper further explores the direct effect of "network interaction", "depth of use", "social impact" on "willingness to pay" in pan-entertainment mobile live broadcasting. The hypothesis and theoretical structural equation model are put forward. Through the empirical analysis of Amos software, it is concluded that in the model that affects users' willingness to pay, the "heart flow experience" plays an intermediary variable role, and the "network interaction" and "depth of use" of users indirectly influence the "willingness to pay" by acting on the "heart flow experience". The "social impact" does not mediate variables, but directly acts on the "willingness to pay" to exert its effect. Through path analysis, we find that "heart flow experience" and "social impact" have the greatest relationship with "willingness to pay". In order to further explore twelve explanatory variables "comment number", "point praise number", "bullet screen number", "popularity click number", "bonus package number", "number of new users concerned", "number of live broadcasts shared", "number of contributions to the list", "number of new friends", "length of viewing", "number of live broadcasts watched", "number of times entering the live channel" and three explanatory variables "recharge value" In this paper, a stepwise regression model of multi-variable linear regression based on stepwise regression method is established for the quantitative relationship between the amount of gold coins, the number of gifts purchased and the amount of gold coins consumed. From the results of the regression model, we can see that not all indicators can finally enter the regression model of dependent variables. For the three explanatory variables, the indicators that are significantly related to the "replenishment gold coin amount" are "new friends", "viewing time", "contribution number of lists", "number of bullets" and "number of bonuses". Significant correlations with the number of gifts purchased were "viewing time", "number of red packets", "number of live broadcasts shared", "number of comments", "number of new users concerned", "number of live broadcasts watched" and "number of bullet curtains". The indicators that are significantly related to the amount of gold coins consumed are "red envelope number", "viewing time", "number of new users concerned", "number of shots", "number of contributions to the list", "number of live broadcasts" and "number of comments". Among them, "viewing time" and "red envelope number" play a relatively important role in the three explanatory variables. They are subordinate to the classification of "heart fluid experience" and "social impact", which are consistent with the conclusions of the previous structural equation model. Therefore, in the actual operation of pan-entertainment mobile live broadcasting, enterprises should consider enhancing the "social impact". That is, social acceptance, establishing a positive image of live broadcasting, guiding users to live broadcasting healthily, making live broadcasting "useful", enhancing users' circle of acquaintances' confidence, thus bringing better benefits to enterprises. In user experience, enterprises should pay attention to the user's "heart fluid experience", improve the user's immersive experience, enhance the user's pleasure in using product functions, while optimizing user experience, it can also bring win-win results. Based on the three quantitative regression equations, we can predict the user's willingness to pay by detecting the user's behavior data, which is helpful for the platform to mine potential users. In addition, the healthy development of pan-entertainment mobile live broadcasting platform can be maintained by detecting abnormal values, and then reviewing irregularities and abnormal consumption.