Heteroscedasticity Autocorrelation Classic Assumption Test

58 Ho : β 1 ≠ 0 independent variable have significance influence to dependent variable 1 Defining Level of Significance Level of significance which is used about 5 or α = 0,05 2 Defining F value F Formula of F test in table 3.2: Table 3.2 Table ANOVA Source Sum of Squares Df Mean Squares F Value Regression SSR K MSR=SSRk Error SSE n-k-1 MSE= SSE n-k-1 F = Total SST n-1 Source : Nachrowi, 2006:18 Where: df : degree of freedom k : total amount of coefficient slope n : total of observation 9sample0 F- value table ={α; df= n-1, n-k } = {α; df= n-1, nT-n-k} 59 3 Defining accepting criteria and rejected Ho If F count F table = Ho accepted independent variable doesn’t influence dependent variable. If probability 0,05 = Ho reject F probability 0,05 Ho accepted

3. Adjusted

Determination coefficient Goodness of Fit, which donated with R 2, is an important measurement in regression, because can inform the good or not the regression model which estimated is. Or in other words, that number can measure how close the estimated regression line with real data. Determination coefficient R 2 shows the model ability to explain the relationship between independent variable and dependent variable. R 2 value will always above between 0 and 1. As close to 1, it means the ability for independent variable is bigger to explaining influence to dependent variable. R 2 with this formula: R 2 = = 1 - There is some problem with the usage of R 2 , is 1 R 2 the closeness between Y prediction and observed Y. If it is used to predicting data which is not or have not exist inside observation, it is maybe not suitable. 2 in 60 comparing 2 R 2 or more, dependent variable or regressand have to be 3 R 2 the value isn’t decreased if independent variable added to equity. In this research using Adjusted R 2 can overcome the weaknesses of R 2 ≤ R 2 shows that the addition of independent variable, can decreases Adjusted R 2 value. Adjusted R 2 value can be added if t value is absolute which is added is more than 1. However the Adjusted R 2 value is better than R 2 , but you have to remembered that dependent variable have to be the same among models which is being compared. Bigger the Adjusted R 2 value, so good the model Wing Wahyu W, 2007:4.21.