基于多分Logistic回归的旅游局官博转发影响因素研究

Factors Influencing the“Forwarding”of Tourism Administration Official Microblogs Based on Multinomial Logistic Regression Model

微博拓宽了旅游信息的传播渠道,也为旅游局开展目的地营销与宣传提供了优势平台与便捷路径。微博的转发是信息扩散程度的主要体现,也是检验旅游局官博运营价值的标准之一。文章以我国省级旅游局在新浪注册的官方微博为研究对象,根据相关文献和长期观察,提取了影响微博转发的3类特征变量,即用户特征变量、文本特征变量和内容特征变量,并利用多分Logistic回归方法,探究各个因素对微博转发水平的影响、作用方向和作用程度。研究结果表明:(1)旅游局官博不同转发水平的影响因素不尽一致,应有针对性地设计信息、开展营销。(2)用户特征、文本特征以及内容特征均会对信息转发产生重要影响,但各因素的影响显著度和作用方向存在差异。(3)对于转发水平,影响力度相对更大的因子主要是微博活动信息、粉丝数和图片。(4)变量筛选过程中剔除的地理分区、链接、表情、微栏目和视频等因子与转发水平没有显著关联。

Microblogs widen the dissemination channels of tourism information, and provide advanced platform and convenient path for Tourism Administrations to market their destinations. Whether a microblog message is forwarded reflects the degree of information dissemission, and is also one of the standards to test the operation value of Tourism Administration Official Microblogs. This paper focuses on official microblogs of provincial tourism administrations in China. In order to make the study sample representative, data were collected from April 20th to May 6th in 2012, a period before and after May Day holiday, which is a peak period that provincial tourism administrations post official microblogs. During this period, 1657 pieces of original official microblogs were posted by 19 provincial tourism administrations. Based on pertinent literature and a long time observation experience, this paper confirmed three types of characteristic variables, including the characteristic of user, text and content, and 15 influential factors. User characteristics include geographical zone, and the number of followers, followees, and average daily microblogs; text characteristics include the microblogging date and time, mention, URL, emoticon, topic, micro-column, picture, video and audio; content characteristics are divided into 14 content types according to tourism six elements theory, pertinent literatures and the analysis of study sample. The premised hypothesis of this study is that different ＂forwarding＂ level is affected by different influential factors. Therefore, we used Multinomial Logistic Regression Model to study the influence of these factors in terms of its direction and degree on microblog ＂forwarding＂ level. The study results indicate that there are different factors affecting the ＂forwarding＂ level of officials microblogs of tourism administrations. The ＂forwarding＂ level here is divided into senior, ordinary and none forwarding. The premised hypothesis of this study is proven to be true. For official microblogs of provincial tourism administrations, combining different factors will make different propagation effect. As a result, it is necessary to design tailored messages for marketing purposes. Second, the characteristics of user, text and content have impact on microblog ＂forwarding＂, but there are differences in the significance and influence direction of each factor. The reasons causing different results are related to the identities of factors and preferences of information receivers. Third, the factors influencing the ＂forwarding＂ level are mainly microblog activity information, pictures, and number of followers. Fourth, there is no significant relationship between the ＂forwarding＂ level and other factors （i.e., geographical location, URLs, emoticon, micro-column and videos）. It is a new and effective method for management organizations of tourism destinations to promote destinations using microblogs, which deserves deep research. This research enriched the theory of microblog marketing and played an important role in tourism microblog operation and tourism destination marketing. However, it is not a long time for the provincial tourism administrations to use official microblogs for information dissemination, and the application of official microblogs and such studies are still in their infancy; therefore, less research can be used for reference. Additionally, the number of study samples used in this paper is also limited. For these reasons, the research results need to be tested in practice and improved in future studies.