• No se han encontrado resultados

CAPÍTULO 2 DISEÑO E IMPLEMENTACIÓN DE UNA PMU DE BAJO COSTO

2.2 DESARROLLO DE SOFTWARE

2.2.1 SOFTWARE PARA EL SISTEMA DE CONTROL

2.2.1.1 Programa para la estimación de sincrofasores, componentes

2.2.1.1.1 Subrutinas de configuración CPU1

The results of the literature review and data obtained from the interviews formed the basis of an extensive questionnaire, which was distributed to key personnel at Telstra. 3.3.1 Survey (questionnaire)

The questionnaire, presented at Appendix E (Questionnaire), used a seven-point Likert Scale which ranges from 1 (one) = disagree to 7 (seven) = agree, where a score below 4 (four) indicates degrees of dissatisfaction and scores above 4 (four) indicate increasing degrees of satisfaction. This scale was used to measure the different attributes and provide sufficient variance for analysis.

The survey was divided into different sections, the first of which dealt with respondents, details of business units and demographic profiles. The second section related to the respondents’ perceptions of the service quality actually provided by various industries in Australia, switching costs and finally the respondents’ satisfaction levels. The questionnaire instrument for the survey totalled 15 pages and spanned 178 items. The questionnaire covered the following points:

 respondent;

 company respondent worked for;

 resources;

 details of contracts;

 reason for outsourcing;

 benefits;

 issues with outsourcing partner;

 impact of outsourcing;

 relationship with vendor and outsourcing partner;

 outsourcing partner and benefits;

 switching costs; and

 setup and sunk costs.

This survey was distributed within the same Telstra business units where the initial interviews were conducted.

3.3.2 Data analysis

Data was collected by questioning a sample population of employees, managers and executives who were directly involved in outsourcing and managing the outsourcing relationship.

Questionnaires were distributed within Human Resources, financial business units and IT support services and were mailed back to the researcher anonymously. Thirty-three completed questionnaires were collected from the 100 that were personally distributed along with stamped and addressed return envelopes, giving a return rate of 33%.

The Statistical Package for Social Science (SPSS) was used to analyse the quantitative data collected. A frequency distribution was used to describe the sample. The mean and standard deviations of the attributes were also computed. Finally, paired ‘t-tests’were used to test the significant difference between sample means, as outlined in Figure 3.2. The t-test as designed by Moore et al. (1998) assesses whether the obtained mean of two groups is statistically different betweeneach group. This analysis is appropriate to compare the mean of two groups and especially appropriate as the analysis for the post-test only, two-group randomised experimental design (Wrigley, Drury & Farhoomand 1997).

Researchers have recommended that as an approximation, at least 10 sources of information per prediction are required (Chin & Lee, 1998; Kinnear & Gray, 2008). The sample size was adequate to perform the necessary analysis, although a larger and more varied response would have been preferred (Chin and Newsted, 1998; Gefen et al., 2000). In ad hoc discussions, it emerged that the main factor in the failure to answer and return questionnaires was the length and complexity of the survey. The factors in return of questionnaires were size, time to complete and complexity of questionnaires, as were identified by (Baroudi et al., 1983). This survey was based on SERVQUAL to give a service dimension of research. SERVQUAL is perceived as allowing a holistic mechanism to empower decision-making teams because it is capable of enhancing advanced outsourcing approaches and provides great opportunities for future research. SERVQUAL methodology was used to measure IT service quality based on a comparison between the expected and realised levels of the five attributes of service (Buttle 1996; Gi-Du Kang 2002):

1. tangibles; 2. responsiveness; 3. reliability; 4. assurance; and 5. empathy.

As highlighted by Buttle (1996), ‘SERVQUAL data can take several forms:

 item-by-item analysis (e.g. P1 – E1, P2 – E2);

 dimension-by-dimension analysis (e.g. (P1 + P2 + P3 + P4/4) – (E1 + E2 + E3 + E4/4), where P1 to P4 and E1 to E4 represent the four perception and expectation statements relating to a single dimension); and

 computation of the single measure of service quality ((P1 + P2 + P3 …+ P22/22) – (E1 + E2 + E3 + … + E22/22)), the so-called SERVQUAL gap’.

The major tool set used for analysing results was the SPSS 16 statistical software that includes capabilities for data analysis, data management and programming-enabled analysis. Before evaluating test data such as t-tests, the data was checked for abnormalities such as extreme values or skewed distributions by calculating the mean, median and standard deviation. The book SPSS 15 Made Simple by Kinnear & Gray (2008) was utilised as the primary SPSS reference.

Questions were collated into the SPSS toolset (SPSS AMOS) where complete Structural Equation Modelling (SEM) analysis was performed. Factor analysiswas used to uncover the latent structure (dimensions) of the set of variables. As discussed by Garson (2009), SEM does not draw causal arrows in models or resolve causal ambiguities. Theoretical insight and judgement by the researcher are still required.

The basic factor analysis steps used are:

 data collection and generation of the correlation matrix;

 extraction of the initial factor solution;

 rotation and interpretation; and

 construction of scales or factor scores to use in further analyses. Steps in analysis included the following tests:

1. Reliability comparison (Cronbach’s Alpha) was used to calculate the reliability for each scale. A further check of Reliability Comparison (Cronbach’s Alpha) was also used to measure reliability. Refer Figure 3.3 for Cronbach’s Alpha Formula.

2. The probability models such as F-Test for Analyses of Variance

(ANOVA) were used as a measure of how different the means are relative to the variability within each sample. Convergent validity was evaluated against items in the refined model for three first-order factors and one second-order factor analysis as detailed in Appendix K (Factor Analysis). 3. Interpretations of the categorical measurement seven-point Likert scale

code are interpreted as:

1 = Strongly disagree (Disagree) 2 = Medium disagree

3 = Neutral disagree 4 = Neutral agree 5 = Agree

6 = Medium agree

7 = Strongly agree (Agree).

The seven-point Likert scale was selected to allow direct comparison of previous work by Whitten (2004) and Goles (2001) and because it yields measurement accuracy superior to that of three- and five-point scales (Malhotra et al., 2009).

4. Mean: If the expected mean values for tangibles, reliability, responsiveness and assurance are above 5 respectively, then these values reveal that the respondents who expressed average expectation agree with the question. If expectation of empathy has a mean value below 5, it would indicate that the respondents have expressed an average expectation of neutral towards the question.

5. Median: The expectations for tangibles, reliability, responsiveness and assurance have a median above the value of 5, which indicates that ‘agree’

is the median opinion of the respondents. The expectation for empathy has a median value of 4.00, which indicates that ‘neutral’, is the median opinion of respondents.

6. Mode: The expectations for tangibles, reliability, responsiveness, and assurance have a mode value of above 5, which indicates that ‘Agree’ is the mode expectation of respondents. The expectation of empathy has a mode value of 4, which indicates that ‘neutral’ is the mode expectation of respondents.

7. The Standard Deviation: The expectations for tangibles, reliability, responsiveness, assurance and empathy have a standard deviation ranging from 0.739 to 0.873.

8. Variance: The expectations for tangibles, reliability, responsiveness, assurance and empathy have variance scores ranging from 0.006 to 0.002, which reveal that these variables have variations in the respondents’ expectations.

9. Range: The expectations for tangibles, reliability, responsiveness, assurance and empathy have a range which indicates that these variables have differences in respondents’ expectations, and respondents have expressed all types of opinions towards the study questions.

T-tests were used to evaluate the responses from different pairs based on whether the means are statistically different. They also look at differences between two groups by comparing their means relative to the spread or variability of their scores.

Cronbach’s Alpha measures how well a set of items (or variables) measures a single, one-dimensional latent construct. Cronbach’s Alpha is not a statistical test but a co-efficient of reliability based on consistency. It should be noted that a reliability co- efficient of 0.70 or higher is considered ‘acceptable’ in most social science research situations (Baroudi & Orlikowski 1988).

Figure 3.2: t-tests (Overview Diagram)

The formula for the standardised Cronbach’s Alpha is shown in figure 3.3 where: N is equal to the number of items, c-bar is the average inter-item covariance among the items and v-bar equals the average variance (Ives, Olson & Baroudi, 1983).

Figure 3.3: Cronbach’s Alpha Formula

Documento similar