在泛娱乐的大背景下，短视频行业异军突起，以其空前的内容丰富度迎合了当代空前的用户需求繁荣，站在了互联网的风口并引发众多关注。2014年至今，短视频行业一跃成为用户量仅次于移动通信业的黑马。在互联网产品中没有流量就等于没有生命。但现阶段，学术界和互联网公司的用户增长部门对于短视频用户使用行为影响因素的研究较为缺乏。并且以往的研究形式以用户访谈和问卷调查为主，主观性强。本文通过某头部互联网公司互动文娱部门对于短视频软件进行全链路埋点所得到的用户行为数据进行分析，在测量数据上保证了数据的客观性。通过研究和总结用户使用意愿，先前对心流体验的研究以及心流体验对使用意愿影响的相关研究，提出影响泛娱乐移动短视频用户使用行为的四个影响潜变量。变量是“ 网络互动”，“使用深度”，“社会影响”和“心流体验”。本文基于先前的研究中验证得出“心流体验”的中介变量作用，并进一步探讨了“ 网络互动”，“使用深度”，“社会影响”对于泛娱乐移动短视频用户使用行为的直接影响，提出了假设和理论结构方程模型。

首先探索12个解释变量“评论数”、“点赞数”、“热评数”、“热门点击数”、“发版数”、“新关注用户数”、“分享短视频数”、“贡献上榜次数”、“新增好友数”、“观看时长”、“观看短视频数”、“进入短视频频道次数”与三个被解释变量“冷启数”、“完播率”、“DAU/MAU”的定量关系，本文建立基于逐步回归的多因变量线性回归模型，回归模型的结果表明，并非所有指标最终都能进入因变量的回归模型，对于所描述的三个被解释变量，与“冷启数”最显著相关的指标是“新增好友数”、“观看时长”， “贡献上榜次数”、“热评数”、“发版数”。与“完播率”显著相关的指标是“观看时长”、“发版数”、“分享短视频数”、“评论数”、“新关注用户数”、“观看短视频数”、“热评数”。与“DAU/MAU”显著相关的指标是“发版数”、“观看时长”、“新关注用户数”、“热评数”、“贡献上榜次数”、“观看短视频数”、“评论数”。其中，“观看时长”和“发版数”在所描述的三个被解释变量中起着相对重要的影响作用。

进一步，通过Amos软件进行的分析表明，在影响用户使用行为的模型中，“心流体验”充当中介，而“ 网络互动”和“使用深度”通过“心流体验”间接对用户使用行为产生影响”。“社会影响”直接作用于“用户使用行为”产生效用，而不是通过中介变量。路径分析表明，“心流体验”和“社会影响”与“用户使用行为”关系最密切。

因此，在娱乐移动短视频中，公司考虑“社会影响”，即增强用户认可度，树立积极的产品形象，引导用户以健康的方式进行使用，将短视频有用化。增强用户及用户熟人圈的使用信心，可以为企业带来更好的利润。从用户体验的角度来看，公司需要关注用户的“心流体验”，以在使用产品功能时增加用户的沉浸感和用户的愉悦感。还可以在优化用户体验的同时实现双赢。基于三个定量回归方程，可以通过检测用户数据来预测用户的使用行为意愿。这有助于平台挖掘潜在用户。此外，通过检测异常值进而审查违规和异常行为，维持泛娱乐性移动短视频平台的健康发展。

In the context of pan-entertainment, the short video industry has sprung up to cater to the unprecedented content needs of contemporary users with its unprecedented content richness, standing on the Internet and attracting a lot of attention. From 2014 to the present, it has become the dark horse after the mobile communication industry. No traffic in Internet products is equal to no life. However, at this stage, the user growth departments of academia and Internet companies are lacking in research on the usage behavior factors of short video users. In addition, the previous research forms were mainly based on user interviews and questionnaire surveys, with strong subjectivity. This article analyzes the user behavior data obtained from the full-link buried point of the short video software through the interactive entertainment department of a head Internet company, which guarantees the objectivity of the data on the measurement data. By researching and summarizing user wishes, previous research on traffic experience and related research on the impact of traffic experience on usage intention, four latent variables affecting the usage behavior of pan-entertainment mobile short video users are proposed. The variables are "internet interaction", "depth of use", "social impact" and "flow experience". Based on the verification of previous studies, this paper concludes the role of intermediary variables for "flow experience", and further discusses the direct impact of "network interaction", "depth of use", and "social impact" on the usage behavior of pan-entertainment mobile short video users. Hypothesis and theoretical structural equation models are proposed.

First explore the 12 explanatory variables "comments", "likes", "hot comments", "popular clicks", "posts", "new followers", "shared live streams", "contributions" "Number of times on the list", "number of new friends", "viewing duration", "number of live broadcasts watched", "number of live broadcast channels" and three interpreted variables "cold start count", "end broadcast rate", "DAU / MAU ", this article establishes a multi-dependent linear regression model based on stepwise regression. The results of the regression model show that not all indicators can finally enter the regression model of the dependent variable. For the three explained variables described, the" The most significant related indicators of "Leng Qi Count" are "Number of New Friends", "Watching Time", "Number of Contributions to the Ranking", "Hot Reviews", and "Number of Posts". The metrics that are significantly related to "End Rate" are "Watching Time", "Postings", "Shared Live Streams", "Comments", "New Followers", "Watched Live Streams", and "Hot Reviews" ". The indicators that are significantly related to "DAU / MAU" are "Number of Posts", "Length of Watching", "Number of Newly Followed Users", "Hot Reviews", "Number of Contributions to the Ranking", "Number of Live Views", and "Comments" number". Among them, "viewing time" and "number of publications" play a relatively important role in the three explained variables.

Further, analysis conducted by Amos software shows that in the model that affects user usage behavior, "flow experience" acts as an intermediary, while "network interaction" and "depth of use" indirectly affect user usage behavior through "flow experience" "Social impact" directly affects the effectiveness of "user usage behavior" rather than through intermediary variables. Path analysis shows that "flow experience" and "social impact" are most closely related to "user usage behavior".

These belong to the "flow experience" and "social impact" categories, which are consistent with the conclusions drawn by previous multi-dependent linear regression models.

Therefore, in entertainment mobile short video, the company considers "social impact", that is, enhancing user recognition, establishing a positive product image, guiding users to use it in a healthy way, and making short video useful. Increasing the confidence of users and user acquaintances can bring better profits to the enterprise. From the perspective of user experience, the company needs to focus on the user's "flow experience" to increase the user's immersion and user pleasure when using product features. It can also achieve a win-win situation while optimizing the user experience. Based on three quantitative regression equations, the user's willingness to use can be predicted by detecting user data. This helps the platform tap potential users. In addition, by detecting abnormal values to review violations and abnormal behaviors, the healthy development of the pan-entertainment mobile short video platform is maintained.