Word Of Mouth Effect

Mouth of Word Effect study based on online shopping: Evidence from TAOBAO
Yang Chen
Abstract
With the rise of online shopping, the impact on word of mouth effect is increasingly significant. This paper focuses on Taobao buyers’ online store rating system and evaluates and analyzes this system to indicate its reasonability, analyzing consumer word of mouth effect and time-varying correlation, to give consumers reasonable advices for choosing products based on the reputation of online shops and also for the improvement of the online shops.
Research background and objective
In the 21st century, the Internet has become a necessity of life, according to the China
Internet Network Information Center (CNNIC) statistics, as of the end of June 2008

In many categories, we first select the two for studying, they are jewelery and digital cameras category. There are three reasons:
First, the number of stores of both types of goods is more than 2%; Second, the number of stores of both types of goods is more appropriate, getting the data is relatively less time consuming and more accurate. Third, these two types of goods are totally different in nature, and they have more value.
From the data we get, Taobao’s online rating system includes three parts:
Correspondence with products score ; Seller attitude score；logistic speed score from different time. Before the analysis of the relationship of them, we select the representative and meaningful data from the two categories and remove the errors finally we have the data with the number of goods is more than 50 and the total rating times is more than 200

= 0.9061，F = 2873.72 we can get that the model is significant , but the P values of log(y02) and log(y03) are greater than 0.05, they are not significant, so we get rid of them, taking recent 6 months’ socre(x3) and the score before 6 months’ score( x4) as the independent variables related to time.
(2) Fix model
The same we use the score in 6 months(x3), the score before 6 months(x4), correspondence with products score(x5) and seller attitude score(x6) as independent variables and the dependent variable is the total score wy. The result is as following: wy = 5.13830 − 0.02656x! + 0.3542x! + 0.6649x! the model is significant,but the p value of x3’s coefficient is greater than 0.05, thus it is not significant,the total socre is just related to the socre of 6 months ago. Whats more ,VIF values are less than 2, it shows that there is no colinearity of these parameters, they independently contribute to the total score. We use the same measure mentioned to test the heteroscedasticity, we find that they have homogeneity of variance. 6.Explainations for the both models
From the results we can see that the rating model for them are very different: