In this chapter, the research framework, hypotheses and the methodology were presented. It outlined the research procedure, sample of the study, instrument used, data collection procedure and statistical methods for data analysis.
Research Framework
The research framework was developed on the basis of the purpose of this study, and the literature reviewed in the previous chapter. The figure presented in Figure 3.1. shows the variables of this study. According to the hypotheses aforementioned, this study aimed to explore the relationship between career plateau and turnover intention, and demonstrate how people with diverse career anchors react differently when facing a plateaued situation in their career.
Figure 3.1. Research framework.
Two hypotheses were derived from the research framework of this study.
H1: Career plateau will be positively significantly related to employees’
turnover intention.
H2: Career anchor profiles will moderate the positive relationship between career plateau and turnover intention.
Research Procedure
The research procedure included the eight steps shown in Figure 3.2. below. At the beginning of this research, the researcher was interested in one specific phenomenon. The researcher then reviewed the previous studies and came up with a topic for this study. After the research topic was determined, the researcher thought about the purpose and significance of this research, trying to find out the questions and answers the study aimed to discover. Next, the researcher developed the research framework. Based on the framework, the researcher then searched for appropriate instruments developed from previous researchers and adopted them in the present study. The research procedure was then designed. The selected instruments were compiled and organized into a questionnaire, which went through the expert review and pilot test to insure the validity and reliability. After the instruments were tested, the questionnaires were distributed to the research targets for data collection. Finally, the hypotheses were tested and the results were presented at the end of the study.
Figure 3.2. Research procedure.
Conclude Research Results Analyze Data
Conduct Data Collection
Conduct Expert Review and Pilot Test Develop Research Instrument
Design Research Procedure Construct Research Framework
Develop Research Purpose and Questions Conduct Literature Review
Determine Research Topic
Research Design
This study adopted the quantitative research method to examine the relationship among career plateau, turnover intention and the moderating effect of career anchors.
A questionnaire was used in this research to obtain data from employees in different industries in Taiwan. Before the data collection process, the questionnaire was examined by experts to insure content validity of the scales used in this study.
Finally, statistical analysis methods were adopted to insure construct validity and reliability of the measurement and to test the abovementioned hypotheses.
Research Sample and Data Collection
The targets of this study were the current employees in Taiwan. They worked in all kinds of industries were full-time employees who had been working in private sectors for at least one year in the same position in the organizations. During the data collection, the researcher constantly monitored the distribution of the collected data through the demographics to make sure that the numbers of responses received from all industries were similar so as to present more general results that apply to all industries.
The data collection process started from April 17th to April 30th and was conducted through online questionnaires. The link of the questionnaire was sent through email to people that the researcher has personal contact with. These people were either current employees in the company or the managers who assisted in the distribution of the online questionnaires. In addition, the link was posted on several social media websites to reach a wider range of targets. The total number of questionnaires collected was 533 and the final number of the valid responses was 412 (valid rate: 77%).
Sample Profile
After reviewing the final valid responses, the following are the demographics of the sample in this study. A total of 224 (54.4%) responses were from females and 188 (45.6%) from male participants. Moving on to the industries the participants come from: 76 (18.4%) of them were from the real estate industry, 58 (14.1%) service, 52 (12.6%) manufacturing, 48 (11.7%) financial and 178 (43.2%) other different industries. Regarding participants’ education level, 182 (44.2%) had a bachelor degree as their highest education level, 107 (25.9%) a master degree or above and the rest (123, 29.8%) a degree of vocational school, high school or under.
As to respondents’ position within their company, 190 (46.1%) were at the entry level, 74 (18%) low-level managers, 72 (17.5%) middle-level managers and 37 (9%) top-level managerial positions. Finally, 244 (59.2%) of the respondents were married and 145 (35.2%) were single. Detail information is shown in Table 3.1.
Table 3.1.
Demographics of the Sample (n=412)
Demographics Category Frequency Percentage
(%)
Education Level High School, Vocational School 123 29.9
Bachelor Degree 182 44.2
Master Degree 104 25.2
PhD Degree 3 0.7
(Continued)
Table 3.1. (Continued)
Demographics Category Frequency Percentage
(%)
Table 3.1. (Continued)
Demographics Category Frequency Percentage
(%)
Industry Real estate 76 18.4
Other services 58 14.1
Manufacturing 52 12.6
Financial and insurance service 48 11.7
Education 30 7.3
Professional, scientific and technical service 24 5.8
Wholesale and retail trade 21 5.1
Information and communication 21 5.1
Construction 19 4.6
Human health and social work 19 4.6
Accommodation and food service 12 2.9
Agriculture, forestry and fishing 9 2.2
Arts, entertainment and recreation 9 2.2
Transportation and storage 8 1.9
Administrative and support service 3 0.7 Electricity, gas, steam and air conditioning
supply 2 0.5
Mining and quarrying 1 0.2
Questionnaire Design
The questionnaire included the measurement that measured the variables in the research framework. It consisted of four sections, three research variables parts and one demographic section. These measurements were found through the previous studies and were originally developed in English. However, the targets of this study were Taiwanese employees who speak Chinese, so the questionnaire was translated into Chinese. After the translation, the Chinese version was examined by experts who had expertise in Chinese and English to back translate and make sure the meanings of the items remained the same. After finalizing the questionnaire, the pilot test was conducted to make sure the initial validity and reliability.
According to Podsakoff and Organ (1986) the common method variance (CMV) might affect the result of this study because the instruments used to measure all variables were self-reports from the same source. In order to minimize the effect of CMV, 7-point and 5-point Likert-type scale and a 6-point frequency scale were utilized on career plateau, turnover intention and career anchors respectively.
Meanwhile, the order of instruments was re-arranged so that the respondents answered the outcome variable first than the independent variable. The moderating variable and personal information were answered at last.
Measurement
The measurements used in this study are described below. The complete questionnaire can be seen in the Appendix: Questionnaire.
Career Plateau
Milliman (1992) first developed a two-dimensional instrument (Job content plateau and Hierarchical Plateau) to measure an individual’s perceived career plateau. The version this study utilized was the adapted version of Milliman (1992) which was presented in Allen et al. (1999). The scale consists of two dimensions:
job content plateau and hierarchical plateau. In each dimension, six items were rated on a 7-point Likert scale ranging from “1” “Totally Disagree” to “7” “Totally Agree”
in order to measure an individual’s self-perception of career plateau. A sample item for Job Content Plateau is: “My job tasks and activities have become routine for me.”
A sample item for Hierarchical Plateau is: “I am unlikely to obtain a much higher job title in the Organization.” Some items were reverse coded to insure the reliability of the responses. A sample reverse coded item is: I expect to be constantly
plateau were 0.83 and 0.85 respectively in Allen et al.’s (1999) research.
Turnover Intention
To measure the respondents’ intention to leave their current job, the scale from Wayne, Shore and Liden (1997) was adopted. There were five items in total and each item was rated on a 5-point Likert scale ranging from “1” “Strongly Disagree”
to “5” “Strongly Agree”. A sample item is: “I am actively looking for a job outside [company name]”. The Cronbach Alpha reported in Wayne et al. (1997) was 0.89.
Career Anchors
The instrument used to test the moderating effect in this framework was the career orientation inventory developed by Schein (2006). There were eight dimensions and each contained 5 items. The following shows the eight categories and a sample item for each.
Technical /Functional Competence: “I want to be so good at what I do that others will always seek my expert advice.”
General Managerial Competence: “I will feel successful only if I become a high-level general manager in some organization.”
Autonomy/Independence: “I will feel successful in my career only if I achieve complete autonomy and freedom to define my work.”
Security/Stability: I would not stay in an organization that would give me assignments that would jeopardize my job security.
Entrepreneur Creativity: “I dream of starting up and building my own business.”
Service/Dedication to a Cause: “I dream of being in a career that makes a real contribution to humanity and society.”
Pure Challenge: “I prefer work opportunities that strongly challenge my
problem-solving and competitive skills.”
Lifestyle: “I have always sought out work opportunities that minimize interference with my personal and family concerns.”
The 40 items from eight dimensions were rated based on a 6-point self-reported frequency ranging from “1” “Never True for me” to “6” “Always True for me”. The use of an even number scale was to avoid the neutralization of responses so as to have a more distinguished categorization. The Cronbach Alpha of this instrument was reported from 0.77 to 0.81 by Coetzee and Schreuder (2011)
Control Variables
The following demographic variables are selected from the previous studies.
They serve as the control variables in the current study.
Gender. From the past studies, Valcour and Tolbert (2003) found that females tend to have higher turnover rate over males. Therefore, in this study, the respondents were asked to answer their gender.
Age. According to the previous literatures, employees’ age seems to influence their turnover intention (Hayes, 2015). Therefore, in this study, the respondents were asked to fill in their birth year (e.g.,1992) so as to calculate their age for the analysis.
Education Level. An individual’s education level affects their career plateau and turnover intention as well even though the effect on turnover intention was inconclusive (Mobley, 1982; Cotton & Tuttle, 1986). Hence, the researcher intended to collect the information of the participants’ education level so as to do further analysis. The respondents were asked to choose among “Under High School or vocational school”, “Bachelor’s degree”, “Master’s degree” and “Doctorate degree”
to indicate their education level.
Marital Status. Koh and Goh (1995) indicated that the marital status and the number of dependents have great impact on individuals’ intention to leave. They proposed that people who were married were less likely to quit their jobs. Therefore, in this study, the researcher asked the respondents to answer whether they were
“married”, “Divorced”, “Single”, “Widowed” and “Cohabitate”.
Number of Dependents. According to Steel and Lounsbury (2009), family responsibility was one of the main factors that influence employees’ intention to leave the organization. Participants were asked to answer how many dependents they had.
Demographic Variables
Tenure of Current Job. An individual’s tenure of current job might influence their perception toward career plateau. Therefore, the respondents were asked to fill out an open-ended question of the years they have been working at the present job and throughout their lives.
Total Working Years. The total working years of an employee may affect their intention to leave the organizations. The longer time people work, their perception of career plateau and their intention to leave might be changed.
Industry type. In order to examine the relationship among career plateau, turnover intention and career anchors in different industries, the respondents were asked to report the industries they were currently working in. The 19 industry categories: Agriculture, forestry and fishing, Mining and quarrying, Manufacturing, Electricity, gas, steam and air conditioning supply, Water supply; sewerage, waste management and remediation activities, Construction, Wholesale and retail trade;
repair of motor vehicles and motorcycles, Transportation and storage, Accommodation and food service activities, Information and communication,
Financial and insurance activities, Real estate activities, Professional, scientific and technical activities, Administrative and support service activities, Public administration and defense; compulsory social security, Education, Human health and social work activities, Arts, entertainment and recreation and Other service activities are from the Directorate-General of Budget, Accounting and Statistics, Executive Yuan in Taiwan (2016).
Validity and Reliability Tests
Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were both used to test the construct validity of the research. EFA was first utilized during pilot testing. According to the EFA result, the item translation and the sequence of measures in the questionnaire were slightly modified before data was collected for the main study.
After the final valid responses were determined, CFA was conducted. However, since the CFA resulted in less-than-satisfactory fit, minor modification to the measurement was performed. The researcher relied on EFA to modify the measurement, but cross validated the modification using CFA with data from a different sample. To complete this procedure, the sample was randomly split into half. The first half of the data was analyzed through EFA in SPSS. According to Costello and Osborne (2005), a factor loading above .4 is acceptable. However, in order to retain only those items that better represent the construct, the criterion of the factor loading for an item was set at the minimum of .65 and no cross loading to ensure the quality of the scales was good. After the modification, the new measurement model went through a cross-validation test by entering the second data set from the split sample in a CFA.
Both career plateau and turnover intention measurement successfully went through cross validation process mentioned above. However, since the career anchors scale was designed as a formative measurement instead of a latent one, and that the AMOS CFA technique does not handle well on formative measures, the validity of the career anchors scale was tested using only EFA. The results are as followed.
Exploratory Factor Analysis (EFA) Result
Table 3.2. Table 3.3. and Table 3.4. show the EFA result of turnover intention, career plateau and career anchors. The finalized dimensions and the item deleted are indicated in the tables.
Table 3.2.
Exploratory Factor Analysis (EFA): Turnover Intention
Item
Factors
1 Final Dimension Items Deleted
TI3 .911 TI
TI4 .831 TI
TI1 .812 TI
TI2 .798 TI
TI5 .533 Deleted
Note. Extraction Method: Principal Component Analysis. Rotation Method:
Varimax. TI: Turnover Intention.
Table 3.3.
Exploratory Factor Analysis (EFA): Career Plateau
Item
Factors
1 2 Final Dimension Items Deleted
JCP6 .904 JCP
JCP2 .881 JCP
JCP5 .873 JCP
JCP1 .855 JCP
JCP4 .738 JCP
HP2 .631 .439 Deleted
HP6 .500 .499 Deleted
HP5 .899 HP
HP4 .845 HP
HP1 .805 HP
HP3 .781 HP
JCP3 .503 Deleted
Note. Extraction Method: Principal Component Analysis. Rotation Method:
Varimax. JCP: Job Content Plateau; HP: Hierarchical Plateau; TI: Turnover Intention.
The Kaiser-Meyer-Olkin Measure of Sampling Adequacy for EFA of turnover intention and career plateau were .799 and .835. The Bartlett’s tests of sphericity were both significant. These indicate that the data was suitable for the EFA analysis.
One factor in turnover intention and two factors in career plateau were extracted with the eigenvalue larger than 1, which fit the original designs of the scales: one
dimension for turnover intention and two for career plateau. Those items with a factor loading lower than .65 or with cross loading problems were deleted; therefore, the remaining numbers of items were four for turnover intention; five for job content plateau and four for hierarchical plateau.
Table 3.4.
Exploratory Factor Analysis (EFA): Career Anchors
Item
Table 3.4. (Continued)
Table 3.4. (Continued)
Note. Extraction Method: Principal Component Analysis. Rotation Method:
Varimax. CHA: Pure Challenge; ENT: Entrepreneurial Creativity; SER:
Service/Dedication to a Cause. STA: Stability/Security; TEC: Technical/Functional Competence; LIF: Lifestyle; AU: Autonomy/Independence; GEN: General Management Competence.
As to career anchors, the Kaiser-Meyer-Olkin Measure of Sampling Adequacy for EFA was .929. The Bartlett’s test of sphericity was also significant. Nine factors were extracted according to the result. However, TEC4 itself stands alone in the ninth factor and was deleted. In addition, items with factor loading lower than .65 or cross loaded were deleted as well. The final eight-factor structure was the same as the initial development of the scale and the number of the remaining items was 30.
Confirmatory Factor Analysis (CFA)
Base on the EFA result of career plateau and turnover intention, the remaining items were used to conduct CFA with data from the other half of the randomly split
sample. The purpose was to see whether the modified measurement model did have a satisfactory model fit.
The modification indices in AMOS output that this study selected include x2/df, SRMR, CFI, RMSEA. Base on Hooper, Coughlan and Mullen (2008), these indices were appropriate indicators that should be able to point out the goodness of model fit in this study. The criteria of good model fit indices are presented in Table 3.5.
Table 3.5.
Indices of Model Fits
Fit Indices Good Fit Acceptable Fit
x2/df 2~5 <5
SRMR <0.05 ≤0.08
CFI ≥0.95
GFI ≥.90 ≥.80
AGFI ≥.90 ≥.80
RMSEA <0.08 <0.1
CR >.7
AVE >.5
Note: Adapted from “Structural Equation Modelling: Guidelines for Determining Model Fit,” by D. Hooper, J. Coughlan and M. Mullen, 2008, Electronic Journal of Business Research Method, 6(1), p. 53-60. Copyright 2008 by the Academic Conferences Ltd. and “Multivariate Data Analysis” by Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L., 1998., fifth ed. Prentice Hall, New Jersey.
Figure 3.3. Career plateau and turnover intention model.
CFA for career plateau and turnover intention
Base on the EFA result, nine items measuring the career plateau and four for turnover intention were used as the measurement model. The model and the standardized regression weights are shown in Figure 3.3. The goodness of fit indices of career plateau and turnover intention model is presented in Table 3.5.
Table 3.6.
Career Plateau and Turnover Intention Model Fit Summary
Model X2/df CFI GFI AGFI SRMR RMSEA
CP and TI 2.785 .936 .87 .812 .0929 .095
Composite Reliability (CR) Average Variance Extracted (AVE)
Job Content Plateau 0.929 0.727
Hierarchical Plateau 0.923 0.750
Turnover Intention 0.904 0.703
After examining the model fit summary in Table 3.6. and the criteria for good model fit in Table 3.5., the model adopted in this research had an acceptable fit.
Therefore, the following statistical analysis utilized the measurement model proposed above.
Common Method Variance (CMV)
Because all instruments used in this research were self-reported, the potential problem of common method variance needed to be inspected. Harman’s one factor analysis was examined to make sure there was no serious CMV problem. The total variance explained in the first component of Eigenvalue greater than one was 24.89%, which was far below the 50% criteria. Therefore, the CMV problem was not a potential threat to this study.
Cronbach’s Alpha
After the model was set, Cronbach’s alpha reliability test was performed to make sure that the measurement scales adopted in this research were reliable.
According to Nunnally (1978), the Cronbach’s alpha above .70 was considered reliable. The Cronbach’s alpha is demonstrated in Table 3.7. and the results were all above .70, which mean that the scales utilized in this research were reliable.
Table 3.7.
Cronbach’s Alpha
Measurement Scale Coefficeint Alpha
Career plateau .76
Job Content Plateau .92
Hierarchical Plateau .88
Turnover Intention .87
Career Anchors .93
Autonomy / Independence .83
Security / Stability .85
Technical / Functional Competence .82
General Managerial Competence .86
Entrepreneurial Creativity .93
Service / Dedication to a Cause .90
Pure Challenge .93
Lifestyle .86