Validity and Reliability Test

50 cause. Internal validity is ability of the research instrument to measure what it is purposed to measure. External validity is when an observes causal relationship can be generalized across persons, settings, and times Donald Cooper, 2006:312. According to Ghazali 2005:45 the validity test is used to measure the legality of a questionnaire. Test validity used to measure the legal valid or invalid of a questionnaire. A questionnaire is said valid if the questions on the questionnaire are able to reveal something that will be on the questionnaire measure. The total score on test validity of said valid if the number of scores 0.30 Sugiono, 2007:178. With calculating the correlation between each question with a total score using the product moment. With r = 0.03 then the question said to be invalid. b. Reliability If a measure has been declared invalid, then the next step is to measure the reliability of the instrument tool. According to Cooper and Schindler 2006:352, a characteristic of measurement concerned with accuracy, precision, and consistency; a necessary but not sufficient condition for validity if the measure is not reliable, it cannot be valid. Reliability is concerned with estimates of the degree to which a measurement is free a random or unstable error. Reliability can be defined as the extent to which measures are free from random error Maholtra, 2009:315. If a measurement tool has been declared invalid, then the next step is to measure the reliability of the tool. As a 51 measure which shows the consistency of the measuring instrument in measuring the same phenomenon on the other occasion. To see the reliability, then the calculated cronbach alpha each the instrument. The variable is said to cronbach alpha have greater value 0.60 Ghozali, 2006:42. According to Ghazali 2005:42 reliability measurements can be done in 2 ways: 1 Measurements repeated: here someone will be given the same questions at different times, and then see if he remains consistent with the answers. 2 One shot or one-time measurement: here measurement only once and then the result were compared with another question or measure the correlation between answers to questions. SPSS provides facilities to measure the reliability with statistical unit Cronbach alpha α. A construct or variable is said Cronbach reliable if the value of alpha 0.60. This thesis will use one time measurement that using Cronbach alpha test α. A variable is said to provide reliable if the Cronbach alpha values 0.60 Sugiono, 2005. The score of reliability is differentiated on each of every variable to interpret low or high of the reliability instrument, as directive is base on certainty as follows: 52 Table 3.2 Reliability Instrument Scale Interval Coefficient Level of Reliability 0,200 0,200-0,399 0,400-0,599 0,600-0,799 0,800-1,00 Very Low Low Sufficient High Very High Source: Sugiyono, Strategi Jitu Memilih Metode Statistik Penelitian Dengan SPSS, 2005:48

2. Descriptive Statistics

The objective from this descriptive analysis is to make the formulation or picture factual systematically and real actual fact, and behavior Sugiyono, 2005:29.

3. Classic Assumption

Multiple linear regression model can be termed as a good model if the model meets the assumptions of normality of data and free from assumptions of classical statistics, whether it multicollinearity, autocorrelation, and heteroskedastisitas. a. Normality Test The normality test is a test of the most widely performed by parametric statistical analysis. The use of normality tests because there is a parametric statistical analysis, the assumptions that must be owned by data is that the data are normally distributed Ghozali, 2005:110. There are two ways to detect normal distribution the graph analysis and statistical 53 tests. One of the easiest ways to see the normality of residuals is to look at a histogram graph comparing observational data with the distribution of near-normal distribution. Normal distribution will create a straight line diagonal and plotting residual data will be compared with the diagonal line. If the residual data distribution is normal, then the line that describes the actual data will follow the diagonal line Ghazali, 2005:110. b. Multicolinearity Test According to Ghozali 2005:91 stated that multicolinearity test aimed to test whether regression model is founded correlation among independent variables. To detect the presence or least multicolinearity in the regression model is as follows: 1 The value of R 2 is generated by an empirical regression estimates are very high, but individually variable, independent variables are many that do not affect the dependent variable. 2 Analyzing the correlation matrix of variables-the independent variable. If there is a correlation between independent variables is quite high usually above 0.90, then this is an indication of multicollinearity. If below 0.90, the absence of multicollinearity. 3 Multicollinearity also can be seen from the value of tolerance and Variance Inflation Factor VIF. Both these measures indicate each independent variable that is explained by other independent variables. Tolerance measures the independent variables were selected that are not explained by other independent variables. Low tolerance value