In this chapter, research methods are described as following. Include research framework, research method, research procedure, research sample and measurement of variables.
Research Framework
The framework of this study was constructed from the previous literature review. The framework is aligning with the purpose of the study and investigating possible antecedents that affect the employee intention of Gen Y. Learning intention is selected as the independent variable, while the employee retention is the dependent variable. In addition, career adaptability as mediator of learning intention and employee retention.
The research framework of this study was shown as Figure 3.1.
Figure 3.1. Research framework Learning Intention
Employee Retention H1
H2
Career Adaptability H3
Research Hypothesis
Based on the reviewing of literature the hypotheses of this study were proposed as below:
Hypothesis 1: Learning intention positive influences career adaptability of Gen Y.
Hypothesis 2: Learning intention positive influences employee retention of Gen Y.
Hypothesis 3: Career adaptability positive influences employee retention of Gen Y.
Research Procedure
The Figure 3.2. as below describe all the steps of the research procedure 1. Review literature
The research reviewing literature and to gain conceptive recognize of Gen Y. During reviewing, researcher browsing exist problem and research scarcity of Gen Y workforce in Taiwan.
2. Identify Research Topic
Based on the previous step, determine research topic is a significant step to the following step. In this step, researcher described the research background, variables, definition of key terms and the purposes of the study.
3. Develop research questions and framework
After reviewing literature and identify research purpose, research conducted the framework of research to determine the variables employed. Meanwhile, research questions were proposed in accordance to research purposes and framework.
4. Develop Research Instrument
Based on the variables conducted in research framework, instruments of variables are in
reference to Knydt (2009), Savickas & Porfeli (2012), and Govaerts et al. (2011) to test each variables connection between each others.
5. Conduct Expert Review and Pilot Test
To confirm the validity and reliability of research, an expert review and pilot test were conducted.
6. Analysis Data
After the context and instruments review by expert and go through pilot text, the data will collect and analyze.
7. Conclude Research Finding
As analysis result, conclude the finding and provide concrete suggestion to managers will be conduct in this step.
Figure 3.2. Research procedure
1. Review Literature
2. Identify Research Topic
3. Develop Research Question and Purpose
4. Develop Research Instruments
5. Conduct Expert Review and Pilot Test
6. Analyze Data
7. Conclude Research Finding
Research Sample
This study focuses on employee learning intention, career adaptability and retention intention of Gen Y in Taiwan, to approach target population, convenience sampling were conducted via online questionnaire link. Target population was identified by requirements as below:
1. Taiwanese and have full time job (permanent and contingent), exclude part-time job.
2. Aged 22-35 (born between 1980-1993)
Data Collection
Data collection was conducted throughout the months between March and May, 2016.
Firstly, researcher share the questionnaire link in Google Form to people who is Gen Y in the researcher’s personal network, second, a snowball sampling technique was employed by asking people to share the link to their personal network they can reach. This research collected total 397 online questionnaires, however, 48 questionnaires were regarded as invalid because respondents were not matching up the criteria of age or employment type.
The valid questionnaires are 349 to data analysis.
Measurement
The measurements conducted in this research describe as below. The completed questionnaire and all each of measurement items can be seen in the Appendix B:
Questionnaire.
Demographic Factors
Personal dimension included six items, gender, born year, position, type of employment, educational level, and industry.
Gender. Gender is conducted as a variables of distinguishing male and female differences on
learning intention, career adaptability and employee retention.
Born year. A blank column for respondent to fill out their born year, it is the first criteria for
researcher to identify the target population (1980-1993), respondent who out off the range is identified as invalid questionnaire.
Position. Position is to indicate respondents’ current position in company, which is including
two categories, management and non-management.
Type of employment. Type of employment is the second criteria to identify target population.
Which is including three categories, permanent, contingent, and part-time job. Respondent who employed part-time job is identify as invalid questionnaire.
Educational level. To indicate respondents educational level: High School or below,
Bachelor or College level and Masters Degree or above.
Industry: To indicate respondents’ industry of current company, which is include education,
technology, service, finance and insurance and others.
Information related to work tenure. It is included three items, to identify low performance of retention.
Seniority. To show how long the respondents have work for the current job. The data is
measured in a monthly basis. It is an anchor to identify respondents’ degrees of retention wiliness.
Total working experience. Total working experience is to show how long the respondents
start to work.
Amount of companies have respondents worked for. The information is collected to show
respondents work for ho many companies.
Learning Intention
Learning intention is an independent variable in the research. The instrument for learning intention was integrated Govaerts et al. (2011) learning intention and Kyndt et al.
(2009) learning attitude. Sample items stated as following: “I Intend to talk with persons in my surroundings about job-related courses or training that I could follow.” Learning attitude sample items stated as following: “If I get the chance to learn, I will definitely take it. ”All of the items are rated on a 5-point Likert scale ranging (1=strongly disagree , 5=strongly agree.) The higher score means higher learning intention.
Career Adaptability
Career adaptability is a mediator in this research. Career adaptability was widely employed to plenty research and various target population. The instrument of career adaptability in this research is derived from Career Adapt-Abilities Inventory (CAAI) - International Version 2.0, which is developed by an international work team that gathered 18 countries’ researcher and employed in Savickas and Porfeli (2012).
CAAI version 2.0 was included four dimensions: career concern, career control, career curiosity and career confidence. These four dimension provide a clear acknowledge of
workforce occupational transitions, developmental tasks, and work traumas. Some sample item statements are: “Thinking about what my future will be like,” ”Making decisions by myself ,” ” Observing different ways of doing things,” and ” Taking care to do things well”.
Respondent based on their different strength when they build up their career to rate each items in dimensions from 1 represent “Not strong” to 5 represent “Strongest”. The higher score means the stronger competency of career adaptability.
Each dimension consists of six items, thus, 24 items totally this variable. In order to avoid social desirability bias, these four dimensions of career adaptability were designed in non-sequential order.
Employee Retention
To assess retention intention, the measurement developed by Kyndt et al. (2009) is employed in this study. Sample item stated as below: “I’m planning on working for another company within a period of three years.” and “Within this company my work gives me satisfaction.” All of the items are rated on a 5-point Likert’s scale ranging (1=strongly disagree, 5=strongly agree.) The higher score represented the higher level of employee retention.
Table 3.1.
Items Numbers of Career Adaptability Dimensions
Dimensions Item No.
Career concern 24, 28, 32, 36, 40, 44 Career control 25, 29, 33, 37, 41, 45 Career curiosity 26, 30, 34, 38, 42, 46 Career confidence 27, 31, 35, 39, 43, 47
Data Analysis
The statistic software of IBM SPSS 22 for iOS was used in this study to analyze the data.
The analysis method included as below. Also, the questionnaire was list in appendix B.
Demographic Information Analysis
Descriptive analysis provides the frequency distribution and percentage number for examining participants’ distribution pattern of demographic in detail. Descriptive analysis will be the criteria to identify the target participant and find out errors, outlier, and the distribution of the data.
T-Test
T-test was employed to compare the differences between two groups’ means and standard deviation. In this research, gender was tested to find the differences among learning intention and employee retention.
One-Way ANOVA
The one-way analysis of variance (ANOVA) is observed to determine whether there are any significant differences between the means of three or more independent (unrelated) groups. In this research, One-Way ANOVA is employed to test the educational level differences on learning intention.
Person Correlation Analysis
The analysis of correlation was conducted to examine the strength of the relationship between learning intention, career adaptability and employee retention, researcher determine the relationship of variables by observing correlation coefficients.
Structural Equation Modeling (SEM)
Structural Equation Modeling (SEM) is a statistical technique widely applies in the behavioral science, it also considered as a combination of factor analysis and regression or path analysis among variables. Path diagram were produced along with beta coefficient and model fit to examine the relationship between observed variables.
In this research, AMOS (Analysis of Moment Structure) was employed to carrier out CFA and path analysis in order to verifying hypothesis. Outputs for this research included X2/df, RMR, GFI, AGFI, RMSEA, AVE and CR to examine measurement model’s goodness of fit. The detailed introduction of each index of good fit as Table 3.2.
Table 3.2.
Index of Good Fit
Good Fit Acceptable Fit Sources
X2/df 2-5
>.90 >.80 Tabachnick & Fidell (2007) Torkzadeh (1994)
Adjusted goodness of fit (AGFI)
>.90 >.80 Tabachnick & Fidell (2007) Torkzadeh (1994)
Root Mean Square Residual (RMR)
<.05 Kline (2005)
Validity and Reliability
The test validity and reliability of measurement, this study went through a process of expert review, Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) and Cronbach’s Alpha. The purpose is to ensure the content validity, construct validity and reliability of measurements before main study. The detail described as following context.
Experts Review
An experts review was employed to confirm content validity of the measurements. In this research, there are two experts from academic field to review the items.
There are some suggestion is accepted, for example, “The work I’m doing is very important to me” is deleted because it is considered as an inconsistent item. Moreover, the Chinese translation of “I have checked out a job in another company previously ” and
“Taking responsibility for my actions.” is revised in order to be more comprehensive to respondents.
However, some suggestions are not accepted due to the different situation of this research. First, the measurement of learning intention is recommended as not suitable to apply because the original measurement have low reliability and validity, however, the current research still remain the measure due to the different respondent of that questionnaire.
Second, experts suggested that the item of age change open text to categorize into groups
“<20, 20~29, 30~39, 40~49 and >50” however, this research planed to target Gen Y, aged below 35, thus, the open text is remaining. Details refer to Appendix A.
Exploratory Factor Analysis (EFA)
Exploratory Factor Analysis (EFA) was carried out all measurements to confirm that question items fall under appropriate dimensions, EFA was conducted with all variables to see if there were cross factor loading and low factor loading among all variables in this study.
In this research, there were one hundred respondents were employed .KMO and Bartlett’s Test was tested to determine if these items are suitable for doing factor analysis.
According to Kaiser (1974), designates levels as follows: A measure >.9 is marvelous, >.8 is meritorious, >.7 is middling, >.6 is mediocre, >.5 is miserable, and <.5 is unacceptable.
Bartlett“s Test of Sphericity is a measure of the multivariate normality of set of distribution and should be significant (p<.001). In this research, the value of KMO Measure of Sampling Adequacy was .704, means middling, and Bartlett's Test of Sphericity was significant (Sig=.000>.05).
EFA was conducted with all variables to see if there were cross factor loading and low factor loading among all variables in this study. According to Fabrigar, Wegener, MacCallum, and Strahan (1999), in a simple factor structure each item has a relatively strong loading one factor (e.g., > .5). Therefore, in this study, items with factor loading lower than .5 and cross-loaded were deleted. Table 3.3. showed all the factor loadings of all the variables.
Table 3.3.
Confirmatory Factor Analysis (CFA)
Confirmatory Factor Analysis (CFA) was employed to confirm the data have same result from the resources of measurement. In order to ensure the instrument met construct validity, 249 participants were conducted to CFA, which is excluded the portion conducted to EFA.
Regarding to process of conducting CFA, the index in model fit was observed and if the majority of index out off acceptable range, regression weight and modification indices were examined for selection of possible items to delete. Afterward, the modified items was rerun in CFA, and then check the index again, the process is repeated until the index of measure satisfied measurement fit. Following context showed the CFA and index of each measurement respectively.
CFA for learning intention. The measurement run though Exploratory Factor Analysis remaining 7 items for CFA. All 7 items of learning intention run through confirmatory Factor Analysis but the index were under satisfactory model fit. Therefore, the table of Squared Multiple Correlations and Modification Indices was examined to identify which items need to
Table 3.3. (continued)
LI8 Delete Low factor loading
LI9 0.774
LI10 Delete Cross loaded
LI11 0.733
LI12 Delete Cross loaded
be deleted in order to have an ideal model fit. Items that have lowest factor loading or highest value in modification indices were removed, and remaining items underwent CFA to confirm the construct validity again. After deleting 1 item, the result of remaining 6 items showed satisfactory model fit and therefore underwent a measurement model fit test. A summary of model fit for original and modified output was showed as Table 3.3. The original and deleted items of learning intention refer to Table 3.5. The CFA model of learning intention in standardized regression weights was illustrated at Appendix C.
As the Table 3.4. showed, the model fit better after modified the structure model. Not only X2/df, RMR and RMSEA all decreased but also GFI and AGFI increased to met satisfactory model fit. Plus, CR remain the same after modified. However, AVE did slightly increased but remained below the adequate figure of .50 for goodness of fit, it’s becoming a limitation of this study.
To ensure the modified measurement model has cross-validity, a multi-group comparison of the randomly split sample was conducted. As Table 3.6. showed, measurement weights, structural covariances and measurement residuals were examined, the insignificant p-value indicate constraining random split samples into two equal part does not significantly
Table 3.4.
LI Model Fit Summary
LI Variable X2 df P X2/df RMR GFI AGFI RMSEA AVE CR Full items
list (7) 74.839 14 .000 5.346 .042 .911 .821 .132 .413 .821 Modified
Items(6) 26.543 9 .000 2.949 .028 .964 .916 .089 .45 .821
weaken the model fit, which means the modified measurement model was successfully cross validated with two random sample groups.
Table 3.5.
Multi-Group Comparison of Cross Validation of LI
Model df CMIN P NFI IFI RFI TLI
Delta-1 Delta-2 rho-1 rho2 Measurement weights 12 12.117 .436 .004 .004 -.005 -.006 Structural covariances 16 16.912 .391 .005 .006 -.007 -.007 Measurement residuals 32 40.604 .142 .013 .014 -.011 -.011
Table 3.6
Learning Intention Variable Items
Items Description Status
LI1
I intend to talk with persons in my surroundings about job-related
courses or trainings that I could follow. Deleted LI2 I intend to participate in a job-related learning activity within the
next year.
LI3
I intend to look for information about job-related courses and learning activities that I could participate in.
LI4 Sometimes I think about following a job-related training within the next year.
LI5 I intend to talk with my executive about job-related courses or trainings that I could follow.
LI6 If I get the chance to learn, I will definitely take it.
LI7 I think it is important to learn throughout my life. Deleted LI8 When I want to learn something that can be useful in the
workplace, I take the initiative. Deleted
LI9
To gain insight into a complex problem, I let my imagination run
free, even when a solution does not seem to be close at hand. Deleted LI10 I love undertaking activities on my own initiative. Deleted
LI11
Some subjects that arise during work are so interesting that I investigate them further, even when it is not necessary for my work.
LI12 I love to accept complex and challenging tasks. Deleted
CFA for career adaptability. The measurement of Career Adaptability run through Exploratory Factor Analysis remaining 21 items in four dimensions for Confirmatory Factor Analysis.
All 21 items of learning intention run through confirmatory Factor Analysis but the index were under satisfactory model fit. Therefore, the table of Squared Multiple Correlations and Modification Indices was examined to identify which items need to be deleted in order to have an ideal model fit. Items that have lowest factor loading or highest value in modification indices were removed, and remaining items underwent CFA to confirm the construct validity again. After deleting 5 item, the result of remaining 16 items showed satisfactory model fit. A summary of model fit for original and modified output was showed as Table 3.7. The CFA model of career adaptability in standardized regression weights was illustrated at Appendix D.
As the Table 3.7. showed, in over all, the model fit is slightly improve after modified, X2/df, RMR and RMSEA all decreased, on the other hand, GFI, AGFI all increased, however, it yet above the satisfactory model fit, it is considered as limitation in this study. Moreover, AVE slightly improved and CR decreased but remained acceptable.
Again, Multi-group comparison result for career adaptability cross validation of measurement model is shown in Table 3.8. P-value also revealed that the constraining randomly split samples into two equal does not significantly worsen the model fit, the modified measurement has cross validity with two sample groups. A list of full item and deleted item showed as Table 3.9.
Table 3.7.
CA Model Fit Summary
LI Variable X2 df P X2/df RMR GFI AGFI RMSEA AVE CR Full items list
(21) 675.69 189 .000/ 3.575 0.049 0.788 0.741 0.102 0.54 0.96 Modified Items
(16) 327.807 104 .000/ 3.152 0.4 0.855 0.811 0.093 0.57 0.952
Table 3.8.
Multi-Group Comparison of Cross Validation of CA
Model df X2 P NFI IFI RFI TLI
Delta-1 Delta-2 rho-1 rho2 Measurement weights 12 12.117 .436 .004 .004 -.005 -.006 Structural covariances 16 16.912 .391 .005 .006 -.007 -.007 Measurement residuals 32 40.604 .142 .013 .014 -.011 -.011
Table 3.9.
Employee Retention Variable Items
Items Description Status
Concern
1 Thinking about what my future will be like. Deleted 2 Realizing that today’s choices shape my future. Deleted 3 Preparing for the future.
4 Becoming aware of the educational and vocational choices
that I must make. Deleted
5 Planning how to achieve my goals.
6 Concerned about my career.
Control
1 Keeping upbeat. Deleted
2 Making decisions by myself. Deleted
3 Taking responsibility for my actions.
4 Sticking up for my beliefs.
5 Counting on myself.
6 Doing what’s right for me. Deleted
Curiosity
1 Exploring my surroundings.
2 Looking for opportunities to grow as a person.
3 Investigating options before making a choice. Deleted 4 Observing different ways of doing things.
5 Probing deeply into questions I have.
6 Becoming curious about new opportunities.
Confidence
CFA for employee retention. The measurement of Employee Retention runs through Exploratory Factor Analysis remaining 9 items for Confirmatory Factor Analysis.
All 9 items of learning intention run through confirmatory Factor Analysis but the index were under satisfactory model fit. Therefore, the table of Squared Multiple Correlations and Modification Indices was examined to identify which items need to be deleted in order to have an ideal model fit. Items that have lowest factor loading or highest value in modification indices were removed, and remaining items underwent CFA to confirm the construct validity again. After deleting 3 item, the result of remaining 6 items showed satisfactory model fit and therefore underwent a measurement model fit test. A summary of model fit for original and modified output was showed as Table 3.10. The CFA model of employee retention in
All 9 items of learning intention run through confirmatory Factor Analysis but the index were under satisfactory model fit. Therefore, the table of Squared Multiple Correlations and Modification Indices was examined to identify which items need to be deleted in order to have an ideal model fit. Items that have lowest factor loading or highest value in modification indices were removed, and remaining items underwent CFA to confirm the construct validity again. After deleting 3 item, the result of remaining 6 items showed satisfactory model fit and therefore underwent a measurement model fit test. A summary of model fit for original and modified output was showed as Table 3.10. The CFA model of employee retention in