t- Test Partial Test F

49 or absence can be done by looking at whether there is a specific pattern on a scatterplot graph between SRESID and ZPRED wherein Y is a Y axis that has been predicted, and the X axis is the residual prediction Y - Y in fact who have in-studentized. With the analysis if there is a specific pattern of regular wavy, widened and then narrowed, then identifying been going heteroskedasticity and if there is no clear pattern, as well as the points spread above and below the number 0 on the Y axis, then there is no heteroskedasticity Ghozali, 2011:139.

F. Hypothesis Test

1. t- Test Partial Test

To determine whether the independent variables partially individual have a significant influence on the dependent variable. The statistical test T basically shows how far the influence of the independent variables individually in explaining the variation of the dependent variable Ghozali, 2011:98. The t-test was used to test the partial each variable. T test results can be seen in the table on the column sig coefficient significance. If the t value or significance probability 0,05, it can be said that there are significant independent variable on the dependent variable partially. 50 However, if the probability value or significance t 0,05, it can be said that there is no significant effect of each variable on the dependent variable Besas. t test formula: to = Where: to = t value bi = coefision regression Sbi = standart error Hypothesis based Significance namely: a. If the number sig. 0,05, then Ho is accepted b. If the number sig. 0,05, then Ho is rejected

2. F

– Test Simultaneous Test This test aims to prove whether the independent variables X simultaneously together have an influence on the dependent variable Y Ghozali, 2011:88. If F count F table, then Ho is rejected and Ha accepted, which means that the independent variable has a significant effect on the dependent variable using a significant level of 0,05 if the value of F count F table then together all independent variables affect the dependent variable. Additionally, you can also see the value of probability. If the probability value less than 0,05 for a significance level of = 0,05, the independent variables jointly 51 affect the dependent variable. Meanwhile, if the probability value is greater than 0,05, the independent variables simultaneously has no effect on the dependent variable. Formula F test F = ⁄ ⁄ Where: R 2 = multiple correlation coefficient squared n = number of sample Then it will be known whether this hypothesis simultaneously rejected or accepted, while the form of simultaneous hypothesis is: H : β1 = β2 = β3 = 0 ; service quality, sales promotion, customer satisfaction simultaneously does not affect the customer loyalty. H : β1 ≠ β2 ≠ β3 ≠ 0 ; service quality, sales promotion, customer satisfaction simultaneously influence the customer loyalty. G. Multiple Linear Regression 1. Similarity Multiple Linear Regression Analysis method in this research is a multiple linear regression that is used to test service quality, sales promotion and customer satisfaction toward customer loyalty. The equation of multiple linear regressions is as follows: Y = a + β 1 X 1 + β 2 X 2 + β 3 X 3 + e 52 Where: Y = Customer Loyalty a = Constanta e = Error sampling X 1 = Service Quality X 2 = Sales Promotion X 3 = Customer Satisfaction β 1 , β 2 , β 3 = Regression coefficient

2. Coefficient of Determination Adjusted R