In this chapter, research framework, research method, research measurement, research sample, data collection, data analysis, validity and reliability of the study, and research procedure would be discussed.
Research Framework
Based on previous literature review, the framework aligns with the purposes of this study was constructed as figure 3.1. Job satisfaction was selected as independent variable, while retention intention was the dependent variable. In addition, age, education level, position and tenure were viewed as demographic variables in this study.
Figure 3.1. Research framework
Hypotheses
Based on literature review, the hypotheses of this study were proposed as below:
Hypothesis 1. Job satisfaction has positive influence on retention intention.
Hypothesis 1-1. Pay has positive influence on retention intention.
Hypothesis 1-2. Promotion has positive influence on retention intention.
Hypothesis 1-3. Supervision has positive influence on retention intention.
Hypothesis 1-4. Fringe benefits have positive influence on retention intention.
Hypothesis 1-5. Contingent rewards have positive influence on retention intention.
Job Satisfaction
Hypothesis 1-6. Operating procedures have positive influence on retention intention.
Hypothesis 1-7. Coworkers have positive influence on retention intention.
Hypothesis 1-8. Nature of work has positive influence on retention intention.
Hypothesis 1-9. Communication has positive influence on retention intention.
Hypothesis 2. Age has a significant interaction effect on the relationship between job satisfaction and retention intention.
Hypothesis 3. Education level has a significant interaction effect on the relationship between job satisfaction and retention intention.
Hypothesis 4. Position has a significant interaction effect on the relationship between job satisfaction and retention intention.
Hypothesis 5. Tenure has a significant interaction effect on the relationship between job satisfaction and retention intention.
Research Method
This study conducted a quantitative approach to measure the relationship between job satisfaction and employee retention based on the research method of previous researchers (Enu-Kwesi et al., 2014; Kyndt et al., 2009; Medina, 2012; Mitchell & Albright, 1972; Ronra &
Chaisawat, 2011), since using quantitative method would test hypotheses and be more structured, reliable and objective.
The method of this study used two versions of questionnaires (paper and online) to collect data. The author employed Google form to create online questionnaire.
Measurement Design
The measurements were designed separately in this research as below. The integrated questionnaire includes 47 items which consists of 11 questions of employee retention and 36 questions of job satisfaction and demographic information (see appendix). In addition, to avoid the inappropriate targets, there is a filter showed as “Are you a cabin crew on active duty in Taiwan airlines?” on first page.
Demographic Information
Demographic information includes five items:
Gender. Gender was separated into two sexes: male and female. Because of the nature of work in Taiwan airlines, there was a wide gap between male and female, approximately one to nine (Association of Wage-Earners, 2000). In hence, gender was merely conducted to analyze the frequency in this study.
Age. A column was blank for respondents to fill out their born year. By calculation of qualities, the age would be divided into 4 ranges, the age of 23 to 25, 26 to 28, 29 to 31 and above 32.
Education level. Education level consisted of five levels: high school or below, associate, bachelor, master and doctor.
Position. Position represented respondents’ current position in the company, including two categories: management and non-management.
Tenure. To identify the precise tenure, respondents must fill in the specific numerals of working length in current company. By calculation of qualities, the tenure was composed of four ranges, below 1 year, 1 to below 3 years, 3 to below 7 years and above 7 years.
Job Satisfaction
Job satisfaction was an independent variable in this study. The instrument of job satisfaction was Job Satisfaction Survey which was developed by Spector in 1985.
The Job Satisfaction Survey (JSS) was a 36-item questionnaire consisting of nine-dimension scales to assess employees’ attitudes about the job and aspects of the job (Spector, 1985). Each scale comprised four items written in positive and negative direction (see table 3.1). Different from Spector (1985), The scoring of each item employed 5-point Likert scale, ranging from 1 (very dissatisfied) to 5 (very satisfied). For the 4-item subscales with a range from 4 to 20, scores of 4 to 10 are dissatisfied, 14 to 20 are satisfied, and between 10 and 14 are ambivalent. For the 36-item total where possible scores range from 36 to 180, the ranges are 36 to 90 for dissatisfaction, 120 to 180 for satisfaction, and between 90 and 120 for ambivalent.
Table 3.1.
Operating procedures 26 17, 35, 42
Coworkers 18, 36 27, 45
Nature of work 28, 38, 46 19
Communication 20 29, 37, 47
Note. Job Satisfaction Survey. Adapted from “Measurement of human service staff satisfaction:
Development of the Job Satisfaction Survey,” by Spector, P. E., 1985, American Journal of Community Psychology, 13(6), 693-713.
Retention Intention
The instrument of retention intention used Kyndt et al.’s questionnaire (2009). Eleven items stated including whether the respondents would like to keep working on their current jobs, whether they wanted to change jobs or functions and whether they had future prospects.
For instacne, “I’m planning on working for another company within a period of three years.”
and “I see a future for myself within this company.” The total items used 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The total scoring of all items ranges from 5 to 55. The higher score represented higher employee retention.
Research Samples
This study focuses on all cabin crews in Taiwan airlines. To approach appropriate targets, the samples should be a current cabin crew working at Taiwan airline, excluding foreign airlines.
Before conducting the formal questionnaire, the pilot test must be executed since this would manifest some particular problems or conceal issues that should be considered in future research (Connelly, 2008). According to annual report of Civil Aeronautics Administration (CAA) in 2016, the total number of cabin crews in different airlines was 7,183. As for the sample size of pilot test, there were several distinct discussions. Connelly (2008) declared that the sample size of pilot test should be around 10 percent of the population size, that is, approximately 720 respondents. In addition, Herzberg (2008) mentioned that sample size from
10 to 40 per group is adequacy of evaluating and estimating a variety of possible aims.
Furthermore, Gay and Diehl (1992) asserted that at least 30 samples are required to build a relationship in correlational research. On the other hand, to analyze sample size requirements for structural equation models, Boomsma (1982) declared that a minimum sample size was 100.
Moreover, 5 or 10 observations per estimated parameter was necessary (Bentler & Chou, 1987;
Bollen, 1989; Hair, Anderson, Tatham, & Black, 1998) as Thompson (2000) affirmed that sample size were at least 10 or 15 times observation items. In accordance with previous studies about sample size of pilot test (Bentler & Chou, 1987; Bollen, 1989; Boomsma,1982; Connelly, 2008; Gay & Diehl, 1992; Hair et al., 1998; Herzberg, 2008; Thompson, 2000), the sample size of the pilot test was 10 times observation of retention items, that is, 110 in this study.
As for the sample size of official questionnaire, this study collected 283 questionnaires in total; however, 57 questionnaires were considered as invalid since respondents did not match the criteria or provide complete information. In hence, 226 samples were employed to be analyzed.
Data Collection
Data collection was conducted from May to July in 2017. Through convenient sampling technique, the online questionnaire was delivered via social software, such as Facebook and Line to request people to assist to complete the survey if they fit the criteria of the research samples. Moreover, by snowball sampling, if respondents were willing to share online or paper questionnaires, they would get the link or paper to forward or deliver to their personal network who fits the qualification. As for collected samples, since paper questionnaires were delivered by a cabin crews of a local airline and the samples were limited by convenient and snowball sampling, the result of samples mainly sourced from two major local airlines and most of the respondents are female.
Data Analysis
To analyze the data, the statistic software of IBM Statistical Product and Service Solutions (SPSS) 22 for Win 7 were used in this study. The analysis method of each variable included as below.
Descriptive Statistics
To recognize the basic features of the data, descriptive statistics was used to measure in the study and present quantitative descriptions in a manageable form or graphics. It provides simple summaries about frequency distribution of samples and measures, such as, mean, standard deviation and so on.
Pearson Correlation
Correlation is a technique for probing into the relationship between two quantitative, continuous variables. The most common measure method is Pearson product moment correlation (PPMC), called Person Correlation for short. The value of correlation coefficient shows the strength of a linear association between two variables, varying between plus one (namely perfect degree of positive correlation) and minus one (namely perfect degree of negative correlation). As the correlation coefficient value is zero, it represents there is no correlation between two variables. The direction of the relationship between the variables is either positive or negative correlation. In this study, each dimension of job satisfaction was analyzed to test the correlation with retention intention separately.
Linear Regression
Linear regression is the most basic type of regression, attempting to explain the relationship between one dependent variable and one or more independent variables by fitting a linear equation to observed data. The direction of coefficient value and significance determine positive or negative relationship between the variables. In this study, the relationship between each dimension of job satisfaction and retention intention were analyzed.
Hierarchical Regression
Hierarchical regression is an approach to compare several regression models and explain a statistically significant amount of variance in one dependent variable after all other variables.
The improvement in R2 shows the proportion of explained variance in dependent variable by the model. The result determines whether adding newly variables increase or decrease the variance of R2. In this study, age, education level, position, and tenure were considered as control variables to analyze the relationship between job satisfaction and retention intention.
In addition, the interaction effect of each demographic variable on the relationship between job satisfaction and retention intention was examined through hierarchical regression as well.
Validity and Reliability of the Study
To test the validity and reliability of the study, the process includes experts review, confirmatory factor analysis (CFA) and Cronbach’s alpha which respectively examine content validity, construct validity and reliability. The information described in detail as follows.
Validity of the Study
Validity refers to the credibility or believability of an instrument, implying precise and accurate results acquired from the data collected. There are four typical types of validity: face validity, content validity, criterion validity (includes concurrent and predictive) and construct
validity (includes convergent and discriminant). To confirm the validity of the study, content validity and construct validity were employed.
Content validity. Content validity, named as logical or rational validity as well, refers to the measurement representing each aspect of a construct. In this study, five experts from several professional field reviewed and assisted to check the translation between English and Mandarin.
Construct validity. Construct validity refers to the degree to which an instrument measures its construct by using multiple indicators and has two subtypes: how well the indicators of one construct converge (namely convergent validity) and how well the indictors of different constructs diverge (namely discriminant validity). In this study, analysis of moment structure (AMOS) was employed to conduct confirmatory factor analysis. The criteria and result of confirmatory factor analysis showed as table 3.2.
Table 3.2.
Model Fit Indices Criteria and Results of 47 items
χ2/df RMSEA GFI AGFI CFI CR AVE
As for the criteria of internal model fit, the value of CR should be larger than 0.7, whereas the value of AVE should be larger than 0.5 (Hair et al., 1998). The result showed that most value of CR and AVE reached the criteria of acceptable range, which represented the construct of the instrument is reliable and valid. On the other hand, as for the whole model fit, although the value of RMSEA of pay and operating conditions subscales was not fit for the acceptable
criteria, the value of GFI (> .09) showed good model fit (Hu & Bentler, 1999). Several scholars explained that if the number of samples was less than 200, the value of RMSEA would be high (Bentler & Yuan, 1999; Boomsma, 1982; Marsh, Hau, Balla, & Grayson, 1998). On the whole, the value of each criterion item was acceptable, and the instrument of each subscale was valid to measure.
Reliability of the Study
Reliability refers to the degree to which an instrument produces stable and consistent quality of results. There are four general types of reliability, each of which measures reliability in a different approach, including inter-rater or inter-observer reliability, test-retest reliability, parallel-forms reliability and internal consistency reliability (Phelan & Wren, 2005). Internal consistency reliability, a common way to test reliability, were employed in this study.
Internal consistency. Internal consistency reliability is a measure of reliability used to evaluate the consistency of results across items within a test. On purpose of examining the internal consistency of the questionnaire, Cronbach’s alpha was conducted to show the results.
According to several researchers, if the Cronbach’s alpha value is below .4, the items should be deleted (Leech, Barrett, & Morgan, 2005). The Cronbach’s alpha value below .6 should not be acceptable (DeVellis, 1991; Leech et al., 2005). Data is reliable, as the Cronbach’s Alpha value is equal or more than .6 (Ghozali, 2013), or should be at least above .7 (DeVellis, 1991;
Leech et al., 2005; Nunnally, 1978). As for the reliabilty analysis in the pilot test, the Cronbach’s alpha value of employee retention is .818, whereas total job satisfaction is .904, showing the instrument of this study is reliable. The Cronbach’s alpha value and description of each subscale showed as table 3.3.
Table 3.3
The Description and Cronbach’s Alpha Value of Each Dimension of JSS
Scale Alpha Description
Pay .63 Pay and remuneration
Promotion .45 Promotion opportunities Supervision .66 Immediate supervisor
Fringe Benefits .76 Monetary and nonmonetary fringe benefits
Contingent Rewards .73 Appreciation, recognition, and rewards for good work Operating Procedures .56 Operating policies and procedures
Coworkers .67 People you work with
Nature of Work .77 Job tasks themselves
Communication .58 Communication within the organization
Total .90 Total of all facet
Note. N=110
In accordance with previous results of validity and reliability of this study, each item of the questionnaire was keep to become the official questionnaire after pilot test.
Research Procedure
Due to the violent strike of CAL in 2016, this aroused author’s motivation to explore what factors affecting cabin crews to stay or to leave in Taiwan airlines. Through reviewing relevant literatures and studies, the research topic, the definition of key terms, research background and son on gradually emerged from plenty of factors and industries. Since a small number of researchers probed into aviation industry, especially on cabin crews, it showed the importance to research and to develop further discussion.
After reviewing literatures, the author developed research questions and hypotheses in accordance with research purposes and framework. To measure specific variables in this research, instruments were developed and chosen subject to the reliability and validity. Before delivering the official questionnaire, the experts should inspect the quality of the study to confirm the content and measurement of the research could be conducted and pilot test was employed in the continued step.
Going through above steps, the samples could be collected and analyzed. The last but not the least, the research results were described in the conclusion, providing discussion and concrete suggestions for future research. The research procedure was illustrated as figure 3.2.
Figure 3.2. Research procedure Motivation development
Literatures review
Research topic dertermination
Research questions and
framework development
Research instruments development
Experts review
Pilot test and adjustment
Data collection and analysis
Research finding conclusion and
suggestions