Tesis ini membahas tentang analisis pengaruh variabel similarity, brand reputation, perceived risk, perceived innovativeness dan original category loyalty terhadap brand extension acceptance khususnya di kategori produk penataan rambut sebagai original category dan perawatan wajah sebagai kategori hasil brand extension pada merek Gatsby. Alat analisis yang digunakan adalah Structural Equation Modelling (SEM) dengan menguji dua model, yaitu yang tidak menggunakan mediating variable dan yang menggunakan mediating variable. Hasil penelitian ini menunjukkan bahwa pada model tanpa mediating variable, similarity dan original category loyalty berpengaruh signifikan secara statistik dan positif terhadap brand extension acceptance. Sementara pada model dengan mediating variable, brand reputation dan original category loyalty berpengaruh secara statistik dan positif terhadap perceived risk. Sementara itu, similarity berpengaruh secara statistik dan positif terhadap perceived innovativeness. Namun, dalam model dengan mediating variable tidak terbukti bahwa mediating estimation lebih baik dibandingkan dengan direct estimation. Dari hasil penelitian ini juga ditemukan bahwa model yang tidak menggunakan mediating variable lebih baik dilihat dari indikator nilai-nilai Uji Kecocokan Model (goodness of fit test) dibandingkan model yang menggunakan mediating variable.

---

The focus of this study is to analyze the impact of similarity, brand reputation, perceived risk, perceived innovativeness and original category loyalty variables to brand extension acceptance, specifically in hairstyling category as original category and face wash category as extended category for Gatsby brand. The statistical analysis of this study is Structural Equation Modelling (SEM) to test two models; without and with mediating variables. This study find that similarity and category loyalty variables are statistically significant and positive to influence brand extension acceptance of Gatsby brand in face wash as extended category in model without mediating variable. Meanwhile, in model with mediating variable, brand reputation and original category loyalty are statistically significant and positive to influence perceived risk, while similarity also statistically significant and positive to influence perceived innovativeness. However, in model with mediating variable, there is no evidence to accept that mediating estimation is better than direct estimation. The result of this study also find that model without mediating variable is better than model with mediating variable that is reflected by comparing of goodness of fit test result.