3. Methodology & Results
We divide the research into 2 studies based on babyloania.com customer transaction. Study 1 was
designed to explore hypothesis 1 and hypothesis 2 while Study 2 was designed to explore hypothesis 3,
hypothesis 4, and hypothesis 5. We collect all transaction data from babyloania.com customer and control
them with several criteria. The first criterion is the total number of transactions, which in this case is set to
a minimum of 2. This allows us to measure the interval between first and second purchase. The second
criterion is the date of first purchase. We set a cut off date of 20 March 2015 to ensure all samples in the
dataset have a lifetime of at least 4 months at the time of this analysis. Implementing the filters produces
266 transaction data points eligible for study. Since both study 1 and study 2 use multiple regression to
predict the outcome of dependent variables, the minimum sample size to reach good prediction level for 2
independent variables with R2 = 0.232 (study 1) and 3 independent variables with R2 = 0.656 (study 2) is
100 and 21 [37]. Therefore, 266 transaction data that we prepare for this research fulfill the criteria.
Study 1 – Initial perceived value for saving benefit positively affects to transaction frequency while
initial recency negatively affects to transaction frequency
Method
Both perceived value and recency has direct effect to repeat purchase. In this study about Product-Service
System, perceived value is defined as benefit savings incurred when renting instead of purchasing an item
at retail price. These two prices are shown in every product page at the babyloania.com website.
Regarding the recency, we give much attention to the time interval between a customer’s first and second
purchase. Then, we defined recency as initial recency. Meanwhile, repeat purchase was defined as
transaction frequency in the first 4 months of customer participation in babyloania.com platform. We
adapt the length of 4 months period for analyzing customer behavior from previous study [38]. Therefore,
for study 1, we use multiple regression by using SPSS 22 with initial perceived value for benefit savings
and initial recency as independent variable. On the other hand, transaction frequency was placed as
dependent variable. We use the method as an approach for modeling their relationship [39]. Beside
multiple regression, we use descriptive statistics to explain the phenomenon of customer transaction
frequency.