STRUCTURAL EQUATION MODELING There are two steps that need to be done in testing

The 2015 International Conference of Management Sciences ICoMS 2015, April 23, UMY, Indonesia | 187 findings. Based on the indicators that make up the second factor, the second factor is still relevant naming named competencies of employees. Factor 3 benefit and salary The third factor consists of three indicators statement. Based on the pattern of statements that make up the third factor, it leads to benefits and compensation earned by the employees if it is able to perform the duties and functions of their job well, such as career development, job promotion opportunities, and income in kind with the work load. In addition, employees also expect the benefits provided by the company as working hours. Based on the indicators that make up this third factor, considered relevant given the name benefits and salaries. This indicator is also relevant to the findings of previous studies. Factor 4 work pressure The fourth factor is formed by four indicators of measurement. Measurement indicators that make up the four factors include: public perceptions of the public service agencies is still relatively low, there is thought to come out of the current work in the service industry, the size of the working pressure but not the service consumer. This question is relevant indicator called pressure of work because the underlying question asking about the picture illustration on the working pressure in public institutions. Basically working pressure in public institutions, especially in Indonesia is still very low. The low pressure in the work of public institutions in Indonesia has created a culture of service that is stiff, weak, and less serve. Employees in public institutions began to realize the importance of working pressure after the government gives to the employee reward system based on performance. The fourth factor is considered relevant given the name work pressure. Factor 5 service mindset and transferability The fifth factor is also formed by three indicators statements, which include the level of service orientation of public organizations, adoption services operating instructions and operating instructions transferability services adopted from the west into the service industry organizations. Basically statement adopted in the empirical literature on these previous studies used in a business setting and not a public organization. Therefore, although the naming of these five factors are consistent with the findings of previous research, exploration efforts in the context of specific research investigations on public organizations still needed. Factor 6 preferences employees The sixth factor is formed by two indicators statement. Indicators of statements that make up the sixth factor are related to the efforts of front-line employees in providing the best service to customers. Statements made relating to whether there is a certain preference of employees in serving customers, relating to consumer education background, and reputation of the company. Two questions that represent the second factor are appropriate given the name preferences employees. Factor 7 Courtesy The seventh factor is composed of two indicators of questions. Indicators of questions that make up the seven factors include respect for employees delivered to customers. Indicators of statements that make up these seven factors include the reluctance of employees to serve customers are demanding something that does not make sense, the desire of employees to serve customers who polite. Basically, the front-line employees the company is trained individuals who are able to control their emotions in order to face different types of consumers. Front-line employees must be able to show respect to customers. Philip Kotler in his various texts was written that customers should be treated like a king. Therefore, the front-line employees the company must be able to show respect to employees. The seventh factor is a new factor that is found in a manuscript researcher of this study. The seventh factor is also relevant to the results of interviews and focus group discussions conducted by the author in the complete repertoire of the findings of this study. Factor 8 feeling proud and honored The eighth factor is also formed by two indicators of questions, which include a feeling of pride to work in public institutions. Sense of pride shown by the strong support given to family members if they want to work in the same agency or working in Public Sector Service Company. The eighth factor also seems appropriate given the name of feeling proud and honored.

5. STRUCTURAL EQUATION MODELING There are two steps that need to be done in testing

structural equation models, namely 1 the development of the measurement model, and 2 testing structural equation models. 5.1 Measurement Model Measurement model used to test how a set of items represents a large number of constructs Hair et al. , 2010, h.695. SEM applications in order to assess the contribution of each indicator variables to represent the underlying constructs and measures how well the combination of a set of indicators that represent the construct reliability and validity. From the test results, and Cronbach alpha reliability constructs can be concluded that there are variables in the models which do not meet the criteria of reliability, which is a variable working pressure, employee preferences, respect, and feeling proud and honored. Especially for variables feeling proud and honored to have the value of Cronbachs alpha were very small, namely 0.358 and the value of the critical ratio is very large and showed irregularities, in the amount of 91.319. Automatically based on the calculation of the reliability and The 2015 International Conference of Management Sciences ICoMS 2015, April 23, UMY, Indonesia | 188 convergent validity of the test earlier, this dimension will be eliminated from the data and are not included in the subsequent analysis. In addition, the low reliability value in the variable working pressure, employee preferences, and respect have a low reliability is influenced their indicators have a low coefficient of weighting factors in this variable. Judging from convergent validity testing in Table 1 it is known that there is an indicator variable that has a working pressure of 0.373 the value of the weight factor, there is an indicator on the employees who have a preference variable weighting factor value 0.453, and there is an indicator variable that has a value of respect coefficient weighting factor of 0.487. But throw measurement indicators used also become problematic issues for employees’ preference and courtesy, because the underlying variable has only two indicators measuring only. Removing indicator used the same measurement does not include these variables for further analysis. To overcome this problem, this study uses a single composite indicator on the next stage of analysis. Table 2 Construct Reliability, Cronbach alpha, and AVE Dimensions Construct Reliability Cronbach alpha AVE Relationship employee’s – customer 0.942148 0.936 0.673712 Employee ’s Competency 0.864019 0.845 0.564184 Benefit and Salary 0.841894 0.841 0.640027 Work pressure 0.657562 0.641 0.410377 Mindset service and transferability 0.730628 0.709 0.481517 Preference of employee ’s 0.63522 0.570 0.489805 Respect 0.668027 0.600 0.522685 Feeling Proud and Honored 91.31976 0.358 117.6426 Service Innovation 0.950454 0.950 0.73378 Organizational Performance 0.924267 0.933 0.709818 Source: data analysis 5.2 Structural Model Testing structural model is intended to test the suitability of the model-based hypothesis to theory-based empirical research data. Illustration testing structural models in the form of goodness of fit and the path diagram. Structural model testing in the early stages do not get results goodness of fit indices were good. Modification index of output results AMOS 16.0 advised to do respesifikasi structural model. Referring to the opinion of Anderson and Gerbing 1982, then re-specification of the models carried out, in order to obtain an alternative model that best fits the overall sample data obtained. Respecification is based on the index modification produced by the model tested in the overall sample. Modification index ≥ 4.0 gives an indication that if the coefficients are estimated, there will be a diminution or impairment of a significant χ2. Overall model modification indices in Table 3 indicate that the Chi-square value obtained can be improved by eliminating some appraisals based on the advice of the index modification in output Amos 16.0. The results showed that the improvement of structural models proposed structural model has had an index value of a good model fit. Thus, the model has to be interpreted. Table 3 Goodness of fit index Source: data analysis Table 4 Correlation on Structural Model Covariances Estimate S.E C.R P The relationship employee- consumer → Service innovation -0.103 0.140 -0.740 0.460 Employee competency → Service innovation 0.624 0.197 3.167 0.002 Benefit and salary → Service innovation -0.196 0.121 -1.618 0.106 Work pressure → Service innovation -0.159 0.107 -1.490 0.136 Mindset service and transfer- ability → Service innovation 0.147 0.138 1.060 0.289 Service innovation → Organiza tional performance 0.421 0.081 5.191 0.000 Source: data analysis

6. CONCLUSION AND LIMITATION This study having two objectives that wants to achieve,