Qualitative Analytical Methods Quantitative Analytical Methods

Questionnaire is a pre-formulated written set of questions to which respondents record their answers, usually within rather closely defined alternatives Sekaran, 2000. Questionnaire is a list consist of questions which will be asked to respondent consisted of lines and columns to be filled with answers Supranto, 2003. The questionnaire contains questions related to variable that are measured. Various statements becoming indicators for each variable collected from various theories, journals, and books. Continuous Scale applied to measure respondent perception by distributing questionnaires over respondent. Questions in questionnaire created by using scale 1 to 5 represent respondent opinion. Respondents will give their answer by giving a mark on the line scale then it is measured to get the data Ferdinand, 2006.

3.4.2. Literature Study

Literature study was done by collecting information from books, journals, magazines, and internet which has correlation with research.

3.5. Analytical Techniques

3.5.3. Qualitative Analytical Methods

Qualitative analysis is analysis which is not applies mathematical model, statistic model, econometrics or other certain models. Data analysis done limited to data processing technique, such as data and tabulation checking -in this case simply reading tables, charts, or numbers which is available, then do the breakdown and interpretation- Hasan, 2002.

3.5.4. Quantitative Analytical Methods

Quantitative analysis is analysis utilizing analyzer having the character of quantitative. Analyzer having the character of quantitative is analyzer using models, like mathematics model, statistical model, and econometrics. The result presented in the form of numbers then explained and interpreted in a description Hasan, 2002.

3.5.4.1. Validity Tests

Validity test applied to measure validity or invalidity of a questionnaire. A questionnaire is valid if questions at questionnaire are able to explain thing which will be measured by the questionnaire. The way of measuring the validity by calculating correlation between scores of each question with total score Ghozali, 2009. If the level of significant is less than 0.05, hence the question is not valid. Question that is not valid might be released from questionnaire then calculated again with correlation calculation. The decision bases to test validity of questionnaire item are: If r result positive and r result r table hence the variable is valid. If r result is not positive and r result r table hence the variable is not valid.

3.5.4.2. Reliability Tests

Reliability test is equipment to measure a questionnaire which is variables indicator. A questionnaire expressed reliable if someone’s answer to questions was consistency or stable from time to time. Measurement of reliability can be done with one shot measurement then the result will be compared with other question or measures the correlation between question and answers. Reliability test is done with help of SPSS giving facility to measure reliability with statistic test cronbach alpha α. A variable is reliable if it is 0.60 Ghozali, 2009. 3.5.4.3. Classic Assumption Tests 3.5.4.3.1. Multicollinearity Tests The purpose of multicollinearity test is whether at a regression model found the existence of correlation between independent variables. If it happened, the correlation hence it named multicollinearity problem. It shouldn’t happen in a good regression model Ghozali, 2009. Guidances of a regression model which has multicollinearity free are VIF Variance Inflation Factor value 10, or Tolerance value 1.

3.5.4.3.2. Heteroscedasticity Tests

Heteroscedasticity test does in a regression model whether un-equality of variance happened from residual out of one respondent to another. The equality of variance and residual out of one observation to other observation hence called as homoscesdasticity, and otherwise it called as heteroscedasticity. Heteroscedasticity wouldn’t happen in a good regression model Ghozali, 2009. The existence of heteroscedasticity can be detected by seeing the Scatterplot. The decision making base is: a. If there are certain patterns like points forming a regular pattern surging, wide, then narrows, hence heteroscedasticity was already happened. b. Otherwise, if there is no clear pattern of points which disseminating above and below number 0 at axis of the ordinate, hence heteroscedasticity was not happened.

3.5.4.3.3. Normality Tests

The purpose of normality test is to test whether in a regression model, dependent variable, independent variable or both having normal distribution or comes near to normal Ghozali, 2009. Normality detection is done by seeing normal chart of Probability Plot. The decision making base is as follows: a. If data disseminates around the diagonal line and follows the direction of diagonal line, hence the regression model fulfills normality assumption. b. If data disseminates far from diagonal line and doesnt follow the direction of diagonal line, hence the regression model doesnt fulfill assumption of normality.

3.5.4.4. Data Analysis Techniques

Two regression models applied in order to test hypothesis without moderator and hypothesis with moderator.

3.5.4.4.1. Linear Regression Analysis

Data analysis applies linear regression to know independent variables of CRM, that are congruency, duration, resource invested, and management involvement influence to dependent variable that is brand loyalty. Equation model of regression which will be tested is:

3.5.4.4.2. Linear Regression Analysis with Moderator

Regression model with moderating variable is a conditional model which one or more independent variables influence one dependent variable, on the condition that the influence will become weaker or stronger if another variable come up as a moderating variable. Data analysis applies linear regression with moderation to know CRM variables, that are congruency, duration, resource invested, and management involvement influence to brand loyalty if it is moderated by consumer involvement. Equation model of regression which will be tested is: Y = α + β1 X 1 + β2 X 2 + β3 X 3 +β4 X 4 +e Y = α + β1 X 1 + β2 X 2 + β3 X 3 +β4 X 4 + β5 X 5 + β6 X 1 X 5 + β7 X 2 X 5 + β8 X 3 X 5 + β9 X 4 X 5 +e Description: Y = brand loyalty X 1 = congruency X 2 = duration X 3 = resource invested X 4 = management involvement X 5 = consumer involvement X 1 X 5 = congruency moderated by consumer involvement X 2 X 5 = duration moderated by consumer involvement X 3 X 5 = resource invested moderated by consumer involvement X 4 X 5 = management involvement moderated by consumer involvement e = disturbance error 3.5.4.4.3. Goodness of Fit Test According to Ghozali 2009, Goodness of Fit Test is used as an accuracy function of a regression in appraising its actual value which can be measured from goodness of its fit. Statistically, at least it can be measured from statistic value of t, statistic value of F, and its coefficient of determination. F Test basically shows whether all independent variables packed into model altogether having influence to dependent variable Ghozali, 2009. Criterions applied are: If probability 0,05 hence Ho is received. If probability 0,05 hence Ho is refused. Statistic t Test applied to measure how far the influence of each independent variable individually toward dependent variable. The coefficient of determination R 2 used to measure models ability in explaining independent variable variation. Coefficient of determination value is between zero and one Ghozali, 2009.

CHAPTER IV RESULTS AND DATA ANALYSIS

4.1. Research Object Description 4.1.1. Company Information