基於迴歸模型探討影響職業賽事球迷進場數之研究─以美國職棒中區為例

美國職棒發展至今已超過百年，如今已是一個具有30支球隊的商業聯盟，票房收入則是支持整個聯盟的關鍵，而觀眾數則會直接影響到整體的票房收入，因此探討影響觀眾數變化的因素對於職業球團而言即是一個重要的課題。本研究旨在探討並找出可能影響美國職棒賽事觀眾數的因素，取2012年至2019年共8年的數據以觀眾數為中心，搭配三種類型的研究變數，分別為地理統計變數、球團相關變數以及團隊成績變數。研究中以球迷分布圖（Fanbase map）作為資料統計區，再將年度資料以每年4月1日至7月31日、8月1日至8月31日以及9月1日至10月第一周分為三個時間段，與年度資料各別進行多元迴歸分析建立出四個模型。研究除了找出有哪些因素對觀眾數具影響力外，更要比較不同時間區段所劃分出的四個模型之間影響變數會出現什麼樣的差異。分析結果顯示，在討論觀眾數變化時，人均收入、白人人口比例、鄰近職業球隊數以及團隊薪資是重要的影響指標，其中前三者呈現顯著負相關，團隊薪資則呈現顯著正相關。此外會發現不同時間中變數的顯著性會產生變化，鄰近職業球隊數於4月1日至7月日31之間呈現不顯著，而其餘時間段皆呈現顯著負相關，表示我們在討論觀眾數的變化時，不只要考慮變數的性質，更要討論變數在時間上的差異。本研究除了呈現影響觀眾數的變數外，更可以根據分析結果對球團經營方向和未來研究提出建議，以供球團或後續研究者一個參考依據。

關鍵字

職棒觀眾數；美國職棒；空間關係；多元迴歸分析

並列摘要

MLB (Major League Baseball) has developed for hundreds of yeas, now, it is a league with more than 30 teams. Attendance revenue which depends on audience is the key of the league. Therefore, it is an important topic for the league to know what kind of factors will impact the attendance. This research aims on the factors which impact the attendance. We use the statistics from 2012 to 2019, and discuss attendence by three types of variables, which are geography, decision from team officials and team stat. In the research, in view of the Fanbase map, we divide the annual data into three period: April 1 to July 31, June 1 to June 31 and Sep 1 to the first week of October. These three statistics are cooperated with annual data by multiple regression analysis to build four modules. This research discovers not only which factors impact the attendance, but also comparing the differences between the four modules. According to the analysis, when it comes to attendance, per capita income, the white population rate, the numbers of professional team nearby and payrolls become an important index. The first three index are significantly negatively correlated to attendance while the payrolls are positive correlation. In addition, we find out variable varies in different period. For example, the numbers of professional team nearby do not show any obvious impact on the attendance from April 1 to July 31. However, it shows significant negative correlation in other periods. This means when we discuss about the change of attendance, we have to consider not only variable itself, but also the period affect variable. This research not only points out how variable affects the attendance, but also provides suggestion to the future development of team officials. We hope this research can be a powerful tool to team officials and follow-up investigators.