In this chapter, research method conducted in this study is introduced. It starts by presenting the framework, then, the hypotheses are listed according to the literature review.
Furthermore, it provides information on the research procedure of this study. The research design and samples used to test the hypotheses are also explained in this section. Definition and measurement of variables are introduced at the end of the chapter.
Research Framework
The framework of this study was constructed from the previous literature review. This framework is applied for the purpose of investigating the possible antecedents that affects the turnover intention of Generation Y. Person-organization fit is selected as the independent variable, while the intention to leave is the dependent variable, to test P-O fit as the possible cause of the intention to leave of Generation Y. In addition, this study proposes perceived supervisor support and perceived peer support as moderators in the relationship between person-organization fit and intention to leave.
The following Figure 3.1 is the research framework of this study.
Figure 3.1. Research framework
Research Hypothesis
Based on the previous literature reviews, the hypotheses of the study were proposed as following:
Hypothesis 1: Demographic factors have a significant relationship with intention to leave of Generation Y.
Hypothesis 1a: Gender has a significant relationship with intention to leave of Generation Y.
Hypothesis 1b: Age has a significant relationship with intention to leave of Generation Y.
Hypothesis 1c: Education level has a significant relationship with intention to leave of Generation Y.
Hypothesis 1d: Number of dependents has a significant relationship with intention to leave of Generation Y.
Demographic Factors Gender
Age
Education level
Number of dependents Tenure
Person-Organization Fit
Perceived Supervisor Support
Perceived Peer Support
Intention to Leave H2 (-)
H3 H4
H1
Y.
Hypothesis 2: P-O fit has a negative relationship with intention to leave of Generation Y.
Hypothesis 3: Perceived supervisor support has a moderating effect on the relationship between P-O fit and intention to leave of Generation Y.
Hypothesis 4: Perceived peer support has a moderating effect on the relationship between P-O fit and intention to leave of Generation Y.
Research Procedure
The research procedure in this study includes eight steps that researcher followed to complete this study. Figure 3.2 provides all the steps in the order of the research procedure.
At the beginning, the researcher was interested in one specific topic, therefore, the researcher started to review literatures and information that related to this topic. After identifying the topic, the researcher discussed with the advisor in order to determine the purpose and research questions of this study specifically. Next, the researcher started to develop the research framework based on the literature review and also develop the research instrument of this study. The entire instrument went through an expert review and was pilot tested to ensure the validity and reliability. After the instruments were tested, the data was collected and analyzed. Lastly, the findings of this study were reported and conclusions made based on the results of data collection and literature.
Figure 3.2. Research procedure
Research Design
Based on the characteristics of Generation Y and antecedents of turnover, this study chose person-organization fit as the independent variable, intention to leave as the dependent variable, and perceived supervisor support and perceived peer support as moderators on the relationship between the independent and dependent variables. Hypotheses were proposed according to previous research findings and theories. A quantitative approach was conducted in this research using a survey questionnaire as the instrument for data collection. Before the formal data collection, the questionnaire was pilot tested to ensure the validity and reliability of all the measurement scales. Inferential statistics were applied to test study hypotheses.
Identify Research Topic Review Literature
Develop Research Purpose and Questions
Develop Research Framework
Develop Research Instrument
Conduct Expert Review and Pilot Test
Analyze Data
Conclude Research Findings
Sample
The research target population was Generation Y workers in Taiwan. Because there is no list of this target population, this study used the convenience sampling method. Online questionnaire links were sent to sample participants that the researcher could approach. There were 377 questionnaires in total collected in this study. The sample participants were born between 1977 to 1997 (Tapscott, 2008), equivalent to the age range of 17 to 37 years old in 2014. Besides, to qualify as the sample of this study, participants were employed in an organization within two years and the working experience needs to be over three months.
Most of the participants were female, from 21 to 30 years old, and single. Moreover, most of them had a bechelor degree, no dependent to take care of, worked below 3 years in their current company, and had a monthly salary ranged from 20,001 NT dollars to 40,000 NT dollars. (see Table 4.1. Descriptive Statistics on Sample Characteristics)
Data collection
Data collection was conducted throughout the months between February and April, 2014.
Online questionnaire was utilized to collect data in this study. Firstly, the researcher sent the questionnaire link to members of Generation Y in the researcher‟s personal network, and then used the snowball sampling technique to ask people to send the link to other members of Generation Y they can reach. The researcher collected 404 online questionnaires. However, 27 questionnaires were regarded as invalid questionnaires because the participants failed the sample criteria of age or tenure. The number of valid questionnaires to enter into the analysis is 377.
Questionnaire Design
The questionnaire was designed to collect data on P-O fit, intention to leave, perceived supervisor support, perceived peer support, and demographics. The questionnaire was
research population is Generation Y in Taiwan whose official language is Mandarin Chinese, the questionnaire was translated from English to Chinese. To ensure the Chinese translation matched the original meaning of all statements in English, back translation was conducted by native Mandarin speakers who have good English skills (e.g. scores above 90 in iBT TOEFL Test or above 860 in TOEIC Test). The final translated version of the questionnaire was pilot tested using 70 participants from the sample of Generation Y, which showed good internal consistency reliability.
Measurement
The measurements used in this study were listed below. The complete questionnaire which shows the instructions and all items of each measurement scale can be seen in the Appendix: Measurements and Questionnaire.
Person-Organization Fit
Kristof (1996) defined P-O Fit as the compatibility between individuals and organizations. Chatman (1989) defined P-O Fit as the equivalence between the norms and values of organizations and the values of individuals. According to the definitions, employees may choose to stay the organizations which fit their own value and once unmet satisfaction occurs, intention to leave may arise afterwards. Therefore, the study utilizes the psychological factors proposed by Kickul and Lester (2001) who categorized these factors into two components, intrinsic outcomes (that relate to the nature of the job itself) and extrinsic outcomes (that relate to the consequence of completing the job). Among these two components, there are four factors totaling 19 items. Kickul and Lester (2001) regarded these items as what employees care and concern about among their organizations. All of the 19 items are rated on a 5–point Likert-type scale ranging from „1‟ “strongly disagree” to „5‟
“strongly agree.” Some sample item statements are: “My organization provides me the level
perform my job,”, “My organization provides me safe work environment.” The score of P-O fit is an aggregate of the 19 measurement items. A high score represents higher degree of P-O fit.
Intention to Leave
Miller et al. (1979) defined intention to leave as an overall behavior of thought of quitting and intention to search other possible jobs. Tett and Meyer (1993) defined turnover intention as the conscious and deliberate willingness to leave an organization. The instrument to measure intention to leave is adopted from Wayne et al. (1997), and Botsford and King (2012). Five items are rated on a 7–point Likert-type scale ranging from „1‟ “strongly disagree” to „7‟ “strongly agree.” Respondents indicated their level of agreement or disagreement on items such as, “I often question whether to stay at my current job”, “I am looking for a change from my current job”, “I am actively looking for a job outside my current company” and, “As soon as I can find a better job, I'll leave my current job.” A high score from each of the four items described above represents a higher level of intention to leave. The fifth item “I think I will be working at the current company five years from now”
is a reverse-scored statement. The score of intention to leave is an aggregate of the five measurement items.
Perceived Supervisor Support
Lynch et al. (1999) indicated that perceived supervisor support is the feelings between employees and supervisors and the level of concerns that supervisors give to each employee. Perceived supervisor support was measured in the work of Eisenberger, Stinglhamber, Vandenberghe, Sucharski, and Rhoades (2002), where in their study they used the items chosen by Eisenberger et al (1986). Nine items were adapted by replacing the term organization with the term supervisor. These items are rated on a 5–point Likert type scale ranging from „1‟ “strongly disagree” to „5‟ “strongly agree.” Respondents will indicate their
not I achieve my career goals.”, “My supervisor keeps me informed about different career opportunities for me in the organization.” A high score for these items represented a higher level of perceived supervisor support. The score of perceived supervisor support was aggregated from the nine measurement items.
Perceived Peer Support
Peer support is social emotional support which is mutually offered or provided by persons sharing similar mental health condition (Solomon, 2004). This study utilized seven items from the Teacher Collegiality Scale (TCS) from the Demonstrating mutual support trust (DMS) dimension, which was developed by Shah (2011). Seven items were adapted by replacing the term teachers with the term colleagues. These items are rated on a 5–point Likert type scale ranging from „1‟ “strongly disagree” to „5‟ “strongly agree.” Respondents indicated their level of agreement or disagreement on items such as, “There is a feeling of trust and confidence among colleagues in my current company.”, and “My colleagues provide strong social support to me.” The score of perceived supervisor support was aggregated from the seven measurement items.
Demographic Factors
Gender, age, education level, number of dependents, and tenure are chosen to be the demographic factors based on previous studies and literature.
Gender.
Gender refers to the socially constructed roles, behaviors, activities, and attributes that a given society considers appropriate for men and women (Bem, 1981; Laws, 1979). Men and women have different values, so they may have different turnover intention. As such, the researcher uses gender as a demographics factor because gender may affect Generation Y‟s intention to leave. Gender is measured with one item: What is your gender? The options were dummy-coded as male = 0 and female = 1.
Even though age range in this study is limited by the definition of Generation Y, this study still includes age as a demographics factor because of abundant evidences of its effect in the literature. Age data is collected through an open-ended question: Which year were you born? (e.g., 1989). This data is later recoded as age by subtracting the birth year from current year (2014).
Education level.
Federico et al. (1976) found that higher education is associated with lower tenure, thus education level is included as a demographic factor. Education level is measured with one item: What is your education level? The options were coded as, high school degree and below
=1, bachelor degree = 2, master degree = 3, and doctoral degree = 4.
Number of dependents.
Based on the literature review family responsibilities is a main factor that may influence turnover intention (Steel & Lounsbury, 2009; Mobley et al., 1979; Price & Mueller, 1986).
Therefore, number of dependents is included as a demographic factor. Number of dependents is measured with one item: What is your current number of dependents? The options were no dependent =1, one dependent = 2, two dependents = 3, three dependents = 4, four dependents
= 5, five dependents or more = 6.
Tenure.
Tenure is one of the main personal factors that have an impact on turnover intention, which has been studied by many scholars (Steel & Lounsbury, 2009; Mobley et al., 1979;
Jackofsky, 1984; Gerhart, 1990). Thus, tenure is included as a demographic factor. Tenure data is collected through an open-ended question: How long have you worked in this company (e.g., 1 year 2 months)? This data is later recoded into number of months.
Data Analysis
The statistic software of IBM SPSS 21 for Windows was used in this study to analyze the data. The analysis method included:
Descriptive Analysis
The SPSS was applied for analyzing respondent information and acquiring descriptive data report, which analyzed the demographics of the sample. Descriptive analysis was used to check for errors, outliers and the distribution of the data. Moreover, descriptive analysis was used to examine the mean and standard deviation of all the variables in this study. The frequency distribution and percentage number were used to identify the participants‟
distribution pattern in all demographic information.
Pearson Correlation Analysis
Pearson correlation analysis was used to examine initial relationships on the hypotheses and to understand the degree of linear relationship between two variables. Also, correlation was utilized to understand the strength of the relationship between the variables.
Specifically, correlation analysis was adopted to understand the relationships among demographic factors (i.e. age, numbers of dependents and tenure), P-O fit, intention to leave, perceived supervisor support, and perceived peer support. If the correlation coefficient shows +1, it means there is a positively perfect correlation between two variables, instead, a correlation of -1 means there is a perfect negative correlation between two variables.
T-test
T-test was conducted to test the differences between two groups of participants by comparing means and standard deviations of each group. Gender, one of the demographic factors of this study, was examined by t-test to find if there is a difference between male and female participants on their levels of intention to leave.
Hierarchical Regression Analysis
using hierarchical regression analysis. Hierarchical regression analysis was used to examine the hypothesis in this study to see how the independent variables influence the dependent variables under the moderating effect. This study constructed the hierarchical regression model using four steps to test the moderating effects of perceived supervisor support and perceived peer support on P-O fit and intention to leave. The first step was to enter the demographic variables, and the second step was to enter P-O fit, perceived supervisor support, and perceived peer support simultaneously into the regression model to test the three variables‟ influence on intention to leave. The third step was to enter the two interaction terms of P-O fit times PSS, and P-O fit times PPS to test the moderating effect of PSS and PPS on intention to leave. Before calculating the interaction terms, P-O fit, PSS, and PPS were all mean-centered by subtracting their mean from the original scores in order to reduce the multicollinearity problem.
Validity and Reliability
SPSS was used to conduct exploratory factor analysis (EFA) to check for the factor structure of each research construct to ensure construct validity. Also, the variables‟
Cronbach‟s alpha was observed to assess the reliability of the measures in this study.
Exploratory Factor Analysis
Before testing the factor analysis, KMO and Bartlett‟s Test was tested to determine if these items are suitable for doing factor analysis. According to Kaiser (1974), when the value of KMO Measure of Sampling Adequacy is between 0.8~1.0 means good and Bartlett's Test of Sphericity should be significant. The value of KMO Measure of Sampling Adequacy in this study was 0.94 and Bartlett's Test of Sphericity was significant (p<.001).
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,
one factor (e.g., > .5). Therefore, in this study, items with factor loading lower than 0.5 were deleted. Table 3.1 shows all the factor loadings of all the variables.
Moreover, according to Podsakoff and Organ (1986) Common Method Variance (CMV) problem may affect the results of this study since the data for the independent and dependent variables were collected from a single source. In order to minimize CMV, different scale points were utilized for dependent and independent variables for this study. Furthermore, Harman‟s one-factor test was conducted to examine the magnitude of CMV in this study. The result shows the largest un-rotated factor only explained 36.83% of the total variances, which indicates a minimal CMV problem with the measurements since it is lower than the 50%
threshold as suggested by Podsakoff, Mackenzie, and Podsakoff (2003).
Table 3.1.
Exploratory Factor Analysis Results
Variables Item Factor Loading Reason for Deleting Item Component
Person-Organization Fit
POF8 deleted Loaded below 0.5
POF9 0.70
3
POF10 0.51
POF11 deleted Cross loaded
POF12 0.68
POF17 deleted Cross loaded
POF18 deleted Cross loaded
POF19 0.67
Table 3.1. (continued)
According to the result, P-O fit was divided into four factors, which was different from the two components (i.e. intrinsic and extrinsic) listed in literatures. Therefore, the researcher categorized the four factors extracted from EFA result into two components. The items for each factor and component were shown in Table 3.2.
Table 3.2.
Intrinsic and Extrinsic Components of P-O Fit
Components Factors Items Variables Item Factor Loading Reason for Deleting Item Component
Perceived Peer Support
Cronbach’s Alpha Reliability
The reliability of this study was assessed by observing the variables‟ Cronbach‟s alpha.
According to Nunnally (1978), the Cronbach‟s alpha score should be about .70. Table 3.3 shows the Cronbach‟s alpha of all variables in this study.
Table 3.3.
Cronbach’s Alpha
Variables Cronbach’s Alpha
Person-Organization Fit 0.903
Intention to Leave 0.840
Perceived Supervisor Support 0.934
Perceived Peer Support 0.854