Data Collection Method RESEARCH METHODOLOGY

41 In gretl, the simplest way to tackle heteroscedasticity problem is to use least squares to estimate the intercept and slopes and use an estimator of least squares covariance that is consistent. This is the so- called heteroscedasticity robust estimator of covariance Adkins, 2011:176. Meanwhile, in eviews, Heteroscedasticity can be eliminated through White’s cross-section standard errors, if heteroscedasticity caused by the cross section or White’s period standard errors, if heteroscedasticity caused by the variability over the time. If the heteroscedasticity caused by both cross section and time series, thus the it can eliminate through Whi te’s diagonal standard errors.

c. Multicollinearity Test

The independent variables which contain of multicollinearity make the coefficient of regression become unsuitable with the substances, thus the interpretation become inappropriate Fadhliyah, 2008. According to Wibowo, 2012: 87, one way to detect multicollinearity in SPSS is to use a test tools that called Variance Inflation Factor VIF. According to Nachrowi and Usman 2006 in Fadhliyah 2008, the strong multicollinearity has value 0.8. 42

d. Autocorrelation Test

According to Ariefianto 2012:30 the commonly uses testing method to test the autocorrelation is through Durbin-Watson test DW tests. The decision making based on the Durbin-Watson test can be categorized into: 1 4 – d 1 DW 4 ; indicates the negative autocorrelation 2 4 – d u DW 4 – dl ; indicates the indeterminate 3 2 DW 4 – d u ; indicates that there is no autocorrelation 4 d 1 dw d u ; indicates the indeterminate 5 0 DW d L ; indicates the postive autocorrelation The value of du and dl acquired from Durbin Watson statistic table. According to Gujarati 2004:475, if a research is using Generalized Least Square GLS model, then a model contains of autocorrelation problem in a panel data, thus, the output will be free from autocorrelation problem.

4. Panel Data Regression

Panel data also known as longitudinal or cross-sectional time-series data is a dataset in which the behavior of entities is observed cross section across time. Table 3.1 will show the difference between cross sectional and time series data