線上消費者評論縱貫性分析

Abstract

消費者在作出購買決策時越來越依賴線上評論與評分，藉由徵詢與聆聽素昧平生的其他消費者關於產品或服務之意見與經驗，其中尤以經驗性產品因其品質在購買之前通常難以判斷，以至於更加需要仰賴線上評論來評估品質。雖然現存許多關於線上評論相關研究的文獻，但絕大多數皆關注於評論數量、評論效價、評論長度、評論變異數及評論數量與效價間交互作用等線上評論中結構化數值屬性對產品銷售量的影響。卻無研究探討隨著時間推移在消費者線上評論與評分是否會產生差異？倘若有差異，造成此一差異的原因又可能為何？ 線上消費者評論與評分隨時間的改變也隱含著評論目標在一段期間內的消費者知覺服務滿意是否穩定，如果線上評論評分有顯著改變，也就意謂評論目標在此段時間內知覺服務滿意度的不穩定。以往文獻在探討服務滿意度變化議題時大多使用實驗設計或問卷分析方法來進行研究。本研究則利用電腦輔助內容分析方法嘗試以生活中實際線上消費者評論數據來探討一段時間內評論目標服務滿意度的穩定性和改變原因。 本研究在萃取消費者線上評論中非結構化文字內文的各別情感屬性時，不再只是存粹討論傳統定義上的正向與負向情感屬性，本研究藉由General Inquirer情感辭典將評論文字內文拆解成情緒上與非情緒上的正向和負向情感屬性。研究結果也顯示無論情緒上或非情緒上正向的情感屬性改變量皆與評論評分改變量呈正向顯著關係，表示正向情感屬性分數改變量上升則評論評分改變量亦會同方向上升。此外，當線上評論中情緒上負向的情感屬性改變量增加，飯店評論評分就會下降，亦表示消費者對目標服務產生情緒上負面的情感，進而導致消費者知覺服務滿意度降低。 本研究提供服務產業管理階層一個嶄新的方向，就是從線上消費者評論字裡行間中讀懂顧客的內心想法與知覺服務滿意度變化，可以藉由分析線上評論來瞭解「顧客說什麼？」。整體來說，本研究說明可藉由消費者實際線上評論非結構化文字內容中所萃取而來的消費者知覺服務滿意度資訊，結合電腦輔助內容分析方法與既有的情感辭典，來充分解釋消費者知覺服務滿意度的變化。此研究提供學術界與實務界一個簇新的研究方法來探索顧客服務滿意度之相關議題。

Consumers rely heavily upon online reviews and ratings to seek out opinions and experiences on the Internet from people they might be unfamiliar with or even have never met before. This is expressly true for experience goods with uncertain quality, because they are more difficult to evaluate. Although there are abundant studies on online reviews in the existing literature, there are scant studies exploring the issue of how online reviews of a specific product or service dynamic change over time. While most studies in the literature use experimental methods when analyzing the issue of service satisfaction change, this present research executes computerized content analysis of service satisfaction changes with an established semantic dictionary. We investigate if and how online ratings change over time on specific online review platforms. As it is very important for service management researchers and practitioners, our study aims to address these understudied but managerially relevant issues. In our empirical analysis we utilize sentiment analysis and the General Inquirer dictionary to identify and tag sentiment categories of words in the textual content after collecting actual online consumer reviews of hotels. The results reveal that when the score of emotionally or non-emotionally positive sentiment increases over time, a hotel’s review rating will also increase. Conversely, when the score of emotionally negative sentiment increases over time, the review rating of hotel will decline. Our study offers managers in the service industry with a new lesson to learn about reading customer minds through online review textual content. We show that managers can analyze online review in order to understand “what customers are saying”. Overall, by adopting a systematic computerized content analysis approach and utilizing an established semantic dictionary, our results demonstrate that information extracted from review textual content is able to substantially explain such differences. We thus offer a new research method and opportunities for marketing scholars and practitioners to explore changes in service satisfaction.