Chapter 4: Research method
4.6 Plan of data analysis
This section describes the data analysis used to test hypotheses in this thesis. It identifies relevant techniques of factor analysis and the reliability coefficient for the research constructs. A detailed analysis plan for each research question is then revealed.
Multiple Regression is the main analytical technique used in the current study. Prior to reporting the results of hypothesis testing, factor analysis is also presented to examine the reliabilities for the scales used in this study. This is desirable because the scales for measuring the study constructs have been modified to suit the research needs. Normality assessment for the hypotheses testing is also conducted in order to ensure that the assumptions of multiple regression analysis are met.
It is worth noting that the principal component analysis (PCA) technique is employed in the current study for factor analysis. PCA attempts to produce a smaller number of linear combinations of the original variables in a way that captures most of the variability in the pattern of correlations. It helps the original variables to be transformed into a smaller set of linear combinations with all of the variables being
used. According to Stevens (1996, p. 363), PCA is psychometrically sound and simpler mathematically, and it avoids some of the potential problems with ‘factor indeterminacy’ associated with factor analysis. Tabachnick and Fidell (2001, pp. 610- 611) commented that PCA is a better choice for an empirical summary of the data set. Furthermore, Cronbach’s coefficient alpha is used to report the scale reliabilities. Reliability concerns the extent to which a measurement provides stable and consistent results (Carmines and Zeller, 1979), and contains one important dimension: internal consistency (Zikmund, 1995). It refers to the degree to which the items that make up the scale correlate with each other to measure the same variable (Zikmund, 1995). According to Pallant (2005), Cronbach’s (1951) coefficient alpha is regarded as the most popular reliability coefficient of internal consistency between items in a unidimensional scale, and it estimates the degree to which the items in a scale represent the domain of the construct being measured.
Having discussed the factor analysis and reliability coefficient techniques, the following section details the analysis plan for each hypothesis identified from relevant literature review in relation to the research questions. As mentioned previously, research question 1, 2 and 3 each involve a single hypothesis. However, to attain an answer to research question 4, three related hypotheses need to be tested. Similarly, as research question 5 involves mediation testing, a number of relevant hypotheses based on the mediational model theory proposed by Baron and Kenny (1986) are required there too.
Research question 1: Is trait EI related to the service performance of casino key account representatives?
This question is addressed by hypothesis 1: Trait EI is positively related to the service performance of casino key account representatives. Prior to analysing the proposed relationship, trait EI is subjected to factor analysis to examine its underlying dimensions, as suggested by Petrides and Furnham (2000). The identified dimensions are employed to analyse the hypothesis, while multiple regression analysis is conducted with the dimensions of trait EI as predictors, and the service performance of casino key account representatives serving as criterion variable in the regression equation model.
Research question 2: Is the FFM related to the service performance of casino key account representatives?
This research question is tested by hypothesis 2: the service performance of casino key account representatives is positively related to Conscientiousness, Extraversion, Agreeableness and Openness, while negatively related to Neuroticism. Multiple regression analysis is conducted to examine the proposed relationship. The five personality factors are entered in the regression equation, with the service performance of casino key account representatives serving as criterion variable in the regression analysis.
Research question 3: Does trait EI explain additional variance of the service performance of casino key account representatives over the effects of the FFM of personality?
This research question is answered by hypothesis 3: trait EI explains additional variance in the service performance of casino key account representatives over the FFM of personality. To test this hypothesis, a hierarchical regression analysis is conducted, with the FFM of personality entered at Step 1, the identified dimensions of trait EI entered at Step 2, while the service performance of casino key account representatives serving as the criterion variables in the regression equation.
Research question 4: Are customer orientation and adaptability related to the service performance of casino key account representatives?
To answer this research question three hypotheses, namely hypotheses 4a, 4b and 4c, need to be evaluated. The first two hypotheses test if customer orientation and adaptability are positively related to the service performance of casino key account representatives respectively. The final hypothesis tests if adaptability is a better predictor in the service performance of casino key account representatives. Prior to testing the relationships, both customer orientation and adaptability are subjected to factor analysis to identify their underlying factor structure. Multiple regression analyses are conducted to test hypotheses 4a and 4b, with the identified dimensions of customer orientation and adaptability entered in the regression equation, and with the service performance of casino representatives serving as the criterion variable.
Multiple regression analysis is also used for investigating hypothesis 4c. However, in this regression equation, the total scores of customer orientation and adaptability are used, as the underlying structure of each construct is not the focus of this test.
Research question 5: Do customer orientation and adaptability respectively mediate the relationships between basic personality traits composed of trait EI and the FFM of personality with the service performance of casino key account representatives? Does the model with customer orientation and adaptability as the mediation variables account for a great proportion of variance in performance rating than does a direct model without mediation?
This research question involves a few hypotheses to test the mediation relationships between each of the independent variables and the dependent variable. Two mediators: customer orientation and adaptability are involved in this study, and they are analysed separately. Furthermore, the unidimension of the two constructs is used for the mediation analyses. For the hierarchical relationship of the two mediators between trait EI and the service performance of casino key account representatives, a unidimensional trait EI is used.
The hypotheses are tested on the basis of the four procedures or steps (hereafter the steps) recommended by Baron and Kenny (1986). The four steps are most commonly used in mediating testing (Preacher & Hayes, 2004). The first step is to show that the independent variable X is correlated with the dependent variable Y, using Y as the criterion variable in the regression equation and X as a predictor. The second step is to show that the independent variable X is correlated with the presumed mediator M, using M as the criterion variable in the regression equation and X as a predictor. The third step is to show that the presumed mediator affects the outcome variable, using Y as the criterion variable in a regression equation and X and M as predictors. The last step is to establish that the presumed mediator M mediates the X-Y relationship with
factors and trait EI respectively. M refers to customer orientation and adaptability respectively. Y referred to the service performance of casino key account representatives. The statistics analysis techniques for testing each step of the mediation models is detailed in the next chapter.
This form of analysis is widely acknowledged as the most rigorous test of mediation effects (Baron & Kenny, 1986; Williams & Attaway, 1996). Using this method in the analysis allows for the effects of control variables and main-effect variables to be excluded from the final result so that the variance due to the mediator variable is clear. At the end of Chapter 5, based on the results of hypothesis testing, post-hoc analyses are presented, which include a t-test to examine the perceived differences between men and women in emotional intelligence, customer orientation, adaptability and the service performance of casino key account representatives. ANOVA is also used to examine the effect of different age groups, education groups and tenure groups with emotional intelligence, customer orientation, adaptability and the service performance of casino key account representatives respectively.
It is worth noting that, some of the categories need to be recoded to make them suitable for ANOVA analysis. The original 7 categories of the education variable are collapsed to 4 categories, as very few cases were reported above master degrees. For those who reported master degree and above, master degrees are relabelled to be bachelor degree above. The four categories of the age variable are recoded to three categories, as only 2 cased reported in last category. Hence, the last two categories are combined to be one and labelled to be 36 years and 36 years above. The seven categories of the tenure variable are collapsed to 5 categories. The last three
categories are combined to one and became the fifth category, labelled as 10 years above.
Prior to the hypothesis analyses, the means and standard deviations for all the independent and dependent variables are obtained through the SPSS command, as well as the correlation matrix pattern for all variables in the current study.
4.7 Summary
This chapter has presented the research framework of the thesis. It began by summarising the research questions and the hypotheses constructed to answer those questions. The following section described the methods and characteristics of sampling and surveying used. In particular, the reasons for sampling just one casino and for selecting the particular casino used in this thesis have been detailed. In the instrument section, the rationale for choosing particular measures were elaborated. This was followed by a detailed discussion of the data analysis techniques employed in the hypothesis testing.
Chapter 5: Analysis and results