Research Framework
The research framework is constructed from the literature review. Trait anxiety is proposed to be positively associated with job performance. In addition, political skill plays a moderating role in the relationship between trait anxiety and job performance. Figure 3.1. is the research framework of this study.
Figure 3.1. Research framework
Research Hypotheses
H 1. Trait anxiety is positively associated with job performance.H 2. Political skill positively moderates the relationship between trait anxiety and job performance.
H 1 H 2 Political Skill
Trait Anxiety Job Performance
Control variables -Gender
--Education level
Research Design
A predictive correlational study was conducted. The data of this study was collected at the different points in time. Job performance in this study reflects market share in the second one-half of 2014. The measure of market share was evaluated by objective data calculated by the securities firm. The Other two measures, anxiety and political skill, were assessed after market share was calculated.
Questionnaires were mailed to different branches located in the central area of Taiwan.
The address of each branch was obtained from a director working in one of the branches.
Managers in different branches facilitated subject selection and data collection. Securities specialists working in selected branches were entirely selected as subjects. Subjects received a package in the mail with the state-trait anxiety questionnaire, the political skill questionnaire and the performance questionnaire. These questionnaires were sent together with prepaid return envelopes.
The research design tried to reduce several errors of survey research in order to enhance the validity of this study. The threats to external validity including frame error, selection error and non-response error were controlled with assistance from managers. However, sampling error could not be controlled, since this study conducted purposive sampling. Additionally, in order to reduce measurement errors, this study selected reliable and valid scales developed by scholars, and use accurate outcome to measure job performance. Furthermore, subjects were insured that data was treated confidentially, and that this research not affected their employment.
Subject Selection
The subjects in this study were securities specialists working in the same firm but different branches located in the central area of Taiwan. This target securities firm has approximately 30 branches in the middle of Taiwan. According to the data from TMOL (2013), the age range is from 21 to 50, and a high school diploma is required as well. People who would
like to be a securities specialist in Taiwan are required to have a professional certificate after graduating from high school. The certificates of securities specialists have two different types including the securities specialist and the senior securities specialist. The distinction for securities specialist between two types of certificate is that the senior securities specialist can provide more services to clients. Job duties of specialist securities comprise several dimensions.
For instance, assist clients to open new accounts, develop and maintain long term client relationships. Securities specialists in Taiwan seldom ask for leave or take vacation during the working period, since they are frequently worried about losing clients. The combination of their salary, in general, is a low base salary and high bonus, and the average bonus has 67% of their total salary (TMOL, 2013). Therefore, clients are very important for them in this compensation system.
This study conducted convenience sampling, since market share of securities specialists was confidential data for the securities firm which was hard to be obtained. Managers were major channels to contact subjects. The permission of managers of different branches was demanded, since the market share data of securities specialists was generally confidential for the securities firm. The subject criteria was securities specialists who were current employees working in different branches of the target securities firm located in the central area of Taiwan.
The target population was securities specialists in the central area of Taiwan. The accessible population was securities specialists working in different branches of the target securities firm located in the central area of Taiwan. According to the statistic data of a director working in the target securities firm, accessible population was approximately 460.
Samples were securities specialists working in different branches located in the central area of Taiwan where managers were willing to provide the market share data of subjects. The sample sizes were around 210 which are calculated by a 95% confidence level, a 5%
size was based on a formula. (Krejcie & Morgan, 1970).
Measures and Variables
This study used reliable and valid scales to measure anxiety and political skill, and the precision of measurements was confirmed through the reliability analysis. The degree of precision was indicated by Cronbach's alpha illustrating a coefficient of internal consistency.
The accepted value of Cronbach's alpha has to be greater than 0.7 (Nunnally, 1978). In addition, this study used the individual market share in the second one-half of 2014 to develop the response categories to measure objective performance.
Anxiety
The State-Trait Anxiety Inventory (STAI) developed by Spielberger et al. (1983) is selected. The STAI consisted of two domains to measure state anxiety and trait anxiety. Each domain consisted of 20 items. The state anxiety inventory measured whether individuals feel anxious at the present moment. The trait anxiety inventory measured whether the anxious state for individuals is caused by their inherent anxiety-proneness. The trait anxiety inventory is employed in this study. Statements used include the following: "I feel pleasant." and "I wish I could as happy as others seem to be." All 20 items are measured on a 5-point Likert scale (1= not at all; 5= very much so), and summing score on each dimensions overall would produce interval data with a high score indicating a high level of anxiety. Additionally, Cronbach's alpha for the trait anxiety inventory is 0.82 (Spielberger et al., 1983).
Since subjects prefer to respond to the Chinese questionnaire, this study refers to The State-Trait Anxiety Inventory in Chinese translated by Chong and Lung (1984) to conduct translation. Cronbach's alpha for the trait anxiety inventory was 0.86 (Chong & Lung, 1984).
Political Skill
This present study selected The Political Skill Inventory (PSI) developed by Ferries et al.
(2005). Political skill referred to utilizing persuasion, manipulation and negotiation to exercise influence (Mintzberg, 1983). The PSI inventory is comprised of 18 items. The items
asked included: "I always seem to instinctively know the right thing to say or do to influence others." and " It is easy for me to develop good rapport with most people." All 18 items were measured on a 7-point Likert scale (1 = strongly degree, 7 = strongly agree), and summing score on each dimensions overall would produce interval data, with a high score indicating a high level of political skill. Additionally, Cronbach's alpha of the PSI is 0.9 (Ferries et al., 2005).
Job Performance
This study used the accurate and objective outcome, individual market share, to measure job performance of securities specialists. Although securities specialists sell various products, such as stocks, financial derivatives, futures and various securities, market share of individual only represents the percentage of the total stock transactions value in the securities market.
For securities specialists, market share is a tangible incentive from outside that is directly rewarded by customers (Siders, George, & Dharwadkar, 2001), which represents securities specialists need to interact with customers frequently. For managers, market share of the securities specialist is one of the objective measures to evaluate performance.
Since the objective data of the market share was confidential to the securities firm, the director suggested reporting the interval data instead of the actual data. Therefore, in order to develop the measurement of job performance, a seminar on defining the interval of the market share is held to discuss with the director. After the seminar, one item measurement was developed. The item is "Market share you had in the second one-half of 2014?" This one item is measured on a 5-point Likert scale. (1 = below 0.00003, 2 = 0.00003 and above to below 0.00005, 3 = 0.00005 and above to below 0.00007, 4 = 0.00007 and above to below 0.00009, and 5 = 0.00009 and above), with higher scores indicating higher job performance.
Subjects were self-reported their market share in the second one-half of 2014 through this measurement. Since the measure of objective job performance was derived from objective data,
Control Variables
This study controlled two variables to ensure that the results of hypothesis test were fair and valid. Previous research has found that gender (Dwyer & Shepherd, 1998； Lyness &
Thompson, 1997), and education level (Ng & Feldman, 2009) have impacts on job performance. Thus, these variables were served as control variables in the analyses.
Gender.
Previous studies found that the effect of sex on job performance existed in organizations (Dwyer & Shepherd, 1998； Lyness & Thompson, 1997). Specifically, research demonstrated that in a sales setting, individuals tended to sell products to customers who were the same gender, which may constrain their selling boundaries and performance (Dwyer & Shepherd, 1998). Thus, subjects would be asked to indicate their gender.Dummy variable was coded as female (0) or male (1) to measure sex.
Education level.
Education level refers to "the academic credentials or degrees an individual has obtained"
(Ng & Feldman, 2009, p.91). Higher education level is likely to provide greater abilities and more knowledge for an individual, which makes a highly educated individual may have higher job performance. In addition, vocational or nonvocational education might lead to different levels of job performance (Ng & Feldman, 2009). For instance, vocational or technical schools might provide individuals with a specific skill to complete a particular job. In Taiwan, a high school diploma or vocational high school diploma is required for an individual who would like to be a securities specialist. Thus, education level was coded as 1 (high school diploma or vocational high school diploma), 2 (associate degree), 3 (bachelor degree), 4 (master's degree) or 5 (doctoral degree). When the data of education level was entered into the regression analysis, the data was recoded into a dummy variable. The responses of high school diploma or vocational high school diploma and associate degree were coded as 0；the responses of bachelor degree, master's degree and doctoral degree were coded as 1.
Questionnaire Design
The questionnaire used scales developed in English with established reliability and validity. Since the first language of subjects was Chinese, the questionnaire was translated into Chinese. In order to confirm preciseness of statement of each item and to confirm the face validity, the process of translation was reviewed by two Chinese translators including a high school English teacher and an assistant professor receiving Ph.D. in the U.S.
Apart from confirming the validity of the questionnaire, the questionnaire was also designed to reduce common method variance (CMV) which might make the systematic measurement error (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). First of all, the answers of respondents were confidential to encourage subjects to respond as honestly as possible.
Second, the questionnaire used different scale formats including two 5-point Likert scales and one 7-point Likert scale to reduce CMV. Third, negative items were reverse coded in the questionnaire of the anxiety measure to avoid response sets.
Conditions of Testing
The connection with securities specialists was through managers. This study initially contacted one director working in one of branches located in the central area of Taiwan.
Through this director, the connection with other managers was extended. The target securities firm has approximately 30 branches located in the central area of Taiwan. 215 questionnaires were mailed to 12 branches located in the central area of Taiwan after communication with other managers by the director.
Managers of each branch not only facilitated the survey process and data collection, but shared instructions to encourage subjects to respond as honestly as possible. For example, the data collection is for research and only reports as grouped data. The managers distributed questionnaires to securities specialists and then mailed questionnaires back after collecting the data, with this process taking approximately two to three weeks.
Data Analysis
This study was a quantitative study conducted in questionnaires. The statistical software of SPSS v18.0 was used in this study to analyze the data. The analysis method includes:
Pearson Correlation Analysis
The Pearson correlation analysis was used in this study. Before using the Pearson correlation analysis, the scatterplot was used to confirm for a monotonic relationship between the two variables which indicated the direction of correlation. A monotonic relationship between the two variables on scatterplot refers to positive or negative correlation which means either the one variable value increases, the other variable value increases, or as one variable value increases, the other variable value decreases. Pearson's r ranged from -1.0 to +1.0. Table 3.1. demonstrates a standard that is used for describing the magnitude of the correlations.
Table 3.1.
Summary of the Describing Magnitude
r Adjective
1.0 0.70-0.99 0.50-0.69 0.30-0.49 0.10-0.29 0.01-0.09
Perfect Very High Substantial Moderate Low Negligible
Note. Adopt from "Elementary survey analysis," by Davis, J. A. 1971. Copyright 2010 by Englewood Cliffs.
Descriptive Statistics
This study used descriptive statistics to describe the characteristics of subjects. The frequency distribution and the percentage of numbers were used to determine the distribution
patter of subjects in age, gender, work experience, organization tenure, goal achieved, and education level.
Confirmatory Factory Analysis (CFA)
CFA was used to determine how well a measurement model generalizes across groups or time (Brown, 2006). In other words, CFA was used to examine whether an existing measure that was developed in the past is appropriate for the current population. According to Hair, Tatham, Anderson and Black (2010), fit indices of CFA model can be identified as three categories: absolute fit indices, incremental fit indices, and parsimony fit indices. In addition, three to four fit indices are typically adequate to provide evidence of model fit. Generally, at least one absolute index and one incremental index as well as chi-square (χ²) results should be reported to assess model fit.
Chi-square (χ²).
Chi-square is a fundamental statistical measure to assess the statistical probability that the observed and predicted covariance matrices are actually equal in a current population. This statistical probability is represented by the p-value. Although the p-value for the Chi-square test is less meaningful as sample sizes become large or the number of observed variables becomes large, this Chi-square value for a model does provide researchers with an evidence to judge about the fit of model.
Absolute fit.
Absolute fit indices assess how well a theory fits the sample data. The root mean square error of approximation (RMSEA) is widely used and recommended from this category. The RMSEA can indicate how well a model fits a population .The RMSEA that is relatively sensitive to sample size is a population-based index. RMSEA values of 0.0 indicate a perfect fit, and RMSEA values very approaching 0.0 imply good model fit (Brown, 2006).
Furthermore, the standardized root mean square residual (SRMR) is widely used from this
discrepancy between the correlations in the input matrix and the correlations predicted by the model. SRMR values of 0.0 indicate a perfect fit, and SRMR values close to 0.0 suggest better model fit (Brown, 2006).
Incremental fit.
Incremental fit indices are used to "assess how well the estimated model fits relative to some alternative baseline model (p.650). The comparative fit index (CFI) is one of the most widely used indices from this category. CFI Values that range between 0 and 1 closer to 1.0 indicate good model fit (Brown, 2006).
Acceptable model fit.
Although multiple guidelines for assessing acceptable model fit exist, this study will use guidelines developed by Hair, Tatham, Anderson and Black (2010). Table 3.2 provides cutoff values based on different sample sizes and model complexity. N indicates the number of respondents, and m indicates the total number of observed variables.
Table 3.2.
Note. Adapted from “Multivariate Data Analysis (7th Edition),” by Hair, J. F., Tatham, R. L.,
Anderson, R. E., & Black, W. 2010, p.654. Copyright 2010 by Pearson Education, Inc.
Regression Analysis Linear Regression Analysis
Linear regression analysis was used to examine the relationship between trait anxiety and job performance. Two steps were conducted. In the first step, control variables were entered. In the second step, the independent variable was entered to test the effect of the variable on the dependent variable.
Hierarchical Regression Analysis
Hierarchical linear regression analysis (Baron & Kenny, 1986； Cohen, Cohen, West, &
Aiken, 2003) was used to examine the moderating role of political skill in the relationship between trait anxiety and job performance. In order to reduce the multicollinearity problem, before calculating an interaction term (trait anxiety x political skill), trait anxiety and political skill were centered by subtracting the mean from the original ones (Aiken & West, 1991).Three steps were conducted to test the moderating effect of political skill. In the first step, control variables were entered. In the second step, the independent variable and the moderator were entered spontaneously to test the impact of the two variables on the dependent variable. Lastly, the third step was to enter multiplication item to test the interaction effect on the dependent variable.
Pilot Study
The pilot study was conducted to confirm the reliability of measurements, examine relationships between variables, and describe the characteristics of the samples. Although the larger samples are better for the precision of estimates, a reasonable minimum sample size for a pilot study to do a preliminary survey is 30 (Johanson & Brooks, 2010). Therefore, the sample size was 30 collected from two branches of the target securities firm located in the central Taiwan. Through the assistance of the director, the period of data collection was finished in one day.
Table 3.3. summarizes the descriptive statistics analysis described the frequency distribution and the percentage of numbers to understand the demographic information of the samples. Most of the respondents who were manager (23.3%) and senior manager (20.1%) were female (90%) with an associate degree (50%). Over half of the respondents (56.7%) achieved their performance goal set by the firm in the first one-half of 2014. The respondents predominantly were 46 to 50 years old (26.7%) and 51 to 55 years old (30%), and had 16 to 20 years of work experience (33.3%) and 21 to 25 years of work experience (33.3%) and 16 to 20 years of organizational tenure (43.3%) and 21 to 25 years of organizational tenure (40%).
In the pilot study, the reliability and the Pearson correlation analysis were also conducted to confirm the reliability of the measurements and to preliminary understand the relationship among each variable. Table 3.4 presents the mean, standard deviation, correlations, and reliability. First, for the reliability analysis, the accepted value of Cronbach's alpha has to be greater than 0.7 (Nunnally, 1978). Cronbach's alpha for the scale of political skill was 0.884, and Cronbach's alpha for the scale of trait anxiety was 0.896. Second, the result of the correlation analysis indicated that political skill exhibited a significant and negatively
moderate correlation with trait anxiety (r = .48, p < .05). Furthermore, market share exhibited an insignificant and negatively low correlation with trait anxiety (r = -.25). Contrary to the expected relationship between market share and trait anxiety, market share exhibited an
insignificant and positively moderate correlation with political skill (r =.31).
Table 3.3.
Descriptive Statistics for Pilot Study (N=30)
Item Frequency Percentage Item Frequency Percentage
1. Age
Table 3.4.
Mean, Standard Deviation, Correlations, and Reliability for Pilot Study (N = 30)
Note. Two-tailed test. The Cronbach's alpha estimates are in parentheses.
**p < .05.
Mean S.D. 1 2 3 4 5
1. Gender .10 .31
2. Education Level 2.10 .71 .11
3. Political Skill 5.06 .56 .10 .19 (.884)
4. Trait Anxiety 2.40 .43 .19 .10 -.45** (.90)
5. Market Share 2.63 1.10 -.30 .09 .31 -.25 (-)