This chapter explains the methods and instruments that will be adopted or adapted in this study. In this chapter, research framework, research procedure, research design, data collection and measurement will also be introduced.
Research Framework
Figure 3.1 illustrates the relationships between the variables in this study. Students’
motivation to join an EMI program represents independent variables which include intrinsic (including IM to know, IM toward accomplishment and IM to experience stimulation) and extrinsic motivation (including job advantage, job need, family pressure, peer pressure and social perception). Different types of motivation were proposed to have significant effects on the dependent variable, OCB (including helping behavior and organizational loyalty).
Figure 3. 1. Research framework
Research Hypothesis
According to the research framework, two main hypotheses and 8 sub-hypotheses were generated and stated as follows:
Hypothesis 1: Extrinsic Motivation will positively affect OCB.
Hypothesis 1a: EM toward Job Advantage will positively affect OCB.
Hypothesis 1b: EM toward Job Need will positively affect OCB.
Hypothesis 1c: EM from Family Pressure will positively affect OCB.
Hypothesis 1d: EM from social Pressure will positively affect OCB.
Hypothesis 1e: EM from Peer Pressure will have the most significant positive effect on
Extrinsic Motivation
Job Advantage Jon Need Family Pressure Peer Pressure Social Pressure
Intrinsic Motivation
IM to know
IM toward accomplishment IM to experience stimulation
Organizational Citizen Behavior (OCB)
Helping Behavior Organizational Loyalty
OCB than the other four types of EM.
Hypothesis 2: Intrinsic Motivation will positively affect OCB.
Hypothesis 2a: Intrinsic Motivation To Know will positively affect OCB.
Hypothesis 2b: Intrinsic Motivation To Experience Stimulation will positively affect OCB.
Hypothesis 2c: Intrinsic Motivation Toward Accomplishment will have the most significant positive effect on OCB than the other two types of IM.
Hypothesis 3: Intrinsic Motivation will have a more significant and positive relationship with OCB than extrinsic motivation.
Research Procedure
The researcher first identified a few topics of interest and reviewed literatures on possible topics to narrow down to a more specific topic. Then, the researcher reviewed literatures on the chosen topic to figure out possible variables to investigate the issue. After identifying the variables, researcher developed research topic and defined purpose of the study. Next, researcher moved on to develop research framework and hypotheses by reviewing journal articles and few e-books. In order to measure the variables, part of measurement were adopted and adapted from existing studies.
Figure 3. 2. Research procedure
Review Literature
Identify Research Questions
Select Topic of Interest
Develop Research Framework and Hypotheses
Develop Research Design
Adapt/Adopt Instruments
Present Research Findings Pilot Study
Collect Data
Analyze Data
Research Design
In this section, the researcher talks about the overall research design. A quantitative study by using survey questionnaire was adopted and used to collect data for measuring participants’ motivation and organizational citizenship behavior. Participants were asked to fill out the questionnaire to reflect their extrinsic, extrinsic motivation and organizational citizenship behavior. Statistical analysis tools were used to help generate the findings and conclusions of this study.
Sampling and Data Collection
For this study, quantitative method was chosen to identify the relationship between students’ motivation to join an EMI program and their organizational citizenship behavior,
and survey questionnaire was used to collect the data. The sample of this study included students who join(ed) an English-as-a-Medium-of-Instruction program in Taiwan. Survey questionnaires were distributed to students whose native languages are not English and join(ed) an EMI program. Participants who join an EMI program in Taiwan were chosen and reached from the list of English-as-a-Medium-of-Instruction programs in 2014 from the Ministry of Education (Ministry of Education, Republic of China, 2014) that includes the EMI programs in Taiwan. However, a few of the programs on the list informed the researcher that they were not English-taught programs and some refused to participate in this research.
In the end, this study reached local and international participants from 53 different EMI programs in Taiwan. To be more specific, there were a total of 181 programs on the lists and there were 2 which were not on the list but were discovered to be EMI programs through the
website of the universities. On the list, 53 programs agreed to help sending out the questionnaires to their students; 18 programs replied that they are not entirely English-taught programs; 5 programs were not counted as EMI programs; 4 programs thought it was not appropriate for their students to be participants in this study; 1 program had stopped recruiting new students; 99 programs did not respond (refers to appendix B).
Data collection process included two ways to distribute questionnaires. Hard copy survey questionnaires were delivered to target sample and online questionnaires (Google Survey Questionnaire) were also distributed through emails to target institutes. The 53 programs were accessed through sending mails to the secretary or the chairmen of the institutes. Through the mails, the researcher provides either hard copies or e-version of the questionnaire for the programs to choose which form they prefer. Later, the researcher sent the mails for inviting EMI program students to participate in this research for 5 weeks (once a week) according to the list of English-as-a-Medium-of-Instruction programs in 2014 from the Ministry of Education. Except for 1 program which requested hard copies of the questionnaire, 10 programs helped to send the questionnaires to their students and 9 programs either refused or were not whole English-taught programs during the first week. During the second week, 13 programs replied they have helped sending the mails to their students and 4 programs wither refused or were not EMI programs. For the third week, there were just 2 programs replied that they sent the questionnaires to their students. There were 8 programs replied helped sending the questionnaires to their students during the fourth week. As for the fifth week, 10 programs informed that they helped sending the questionnaires and 1 program was not an EMI program.
For this study, there were 53 programs that the research was able to reach. After getting the number of students of the 53 programs on the websites of the universities, response rate was calculated without including those which did not response or were not EMI programs.
261 questionnaires were returned and the response rate was 19.5%. A total of 215 questionnaires were useable for this study. 46 responses were excluded due to missing data, or did not pass all of the screening questions (do not or did not receive English-taught program in Taiwan or mother tongue is English and so on) to fit the sample criteria of this study. Due to the time limitation and limited resource, a t-test was done to test differences in response between early and late respondents in order to examine the representativeness of the 215 samples in this study. T-test result is shown in Table 3.1. The t-test result shows there is no difference between the early and late respondents on the main variables (intrinsic motivation, extrinsic motivation and organizational citizenship behavior) of this study. Thus, the 215 samples are still quite representative even though the sample size is limited in this study.
Table 3. 1.
Independent Samples T-test Result Summary
Respondent Group
Dependent Variable Early Respondents Late Respondents
M SD M SD P
Social Desirability 5.79 2.17 5.89 2.07 .744
English Proficiency 5.03 1.26 5.01 1.47 .915
Intrinsic Motivation 5.61 0.9 5.56 1.13 .696
Extrinsic Motivation 5.26 1.00 5.23 1.13 .813
Organizational
Citizenship Behavior 3.99 0.58 4.01 0.65 .825
Sample profile.
After reviewing the 215 questionnaires, the following are the demographics of the samples in this study. A majority of respondents at the rate of 65.1% join EMI program now and the others joined EMI program 1-3 years ago. Responses came mostly from female at the rate of 59.1% and the age of the participants ranged from 18 to 48 years old. Most of the respondents join(ed) EMI program for a master degree at a rate of 61.4%, 32.6% for a bachelor degree and 6% for a doctoral degree. A rate of 53.6 reported to join the program for duration of 1-2 years, 22.8% for 3 months-1 year, 15.8 for 3-4 years, 7% for more than 4 years and 1.9% for less than 3 months. All of the respondents came from 53 different EMI programs. Although participants are from 39 different countries, the majority of them are from Asia (Taiwanese at a rate of 44.7%, while 31.6% from other parts of Asia), 12.6% from Central/South America, 5.6% form Europe, 3.3% from Africa, 1.9% from Oceania and 0.5%
from North America (Mexico).
Table 3. 2.
Descriptive Statistics on Sample Characteristics
Item Frequency Percentage Item Frequency Percentage
1. When enroll in the program 4. Region
Now 140 65.1 Asia 164 76.3
1-3 years ago 75 34.9
Central ＆
South America 27 12.6
Europe 12 5.6
2. duration of enrolling in the
program Africa 7 3.3
less than 3 months 4 1.9 Oceania 4 1.9
(continued)
Method of Data Analysis
Statistical software IBM SPSS v.22 and AMOS were utilized in this study.
Descriptive statistics.
Since there is a specific target sample (EMI program students) and random sampling was not applied in this study, the main purpose of descriptive statistics is to observe the characteristics of the collected samples such as mean, medium, standard deviation, demographics and sample distribution of the sample. Descriptive statistics summarized and characterized the important features of the data collected (Cleveland, 2015).
Factor analysis.
Exploratory factor analysis (EFA) in SPSS and confirmatory factor analysis (CFA) in AMOS is conducted to make sure the validity of the measurement. EFA was used in the pilot study to examine whether problematic items existed while CFA was used to test whether the collected data fits the theoretical measurement model (Jackson & Purc-Stephenson, 2009).
Table 3.2. (continued)
3 months-less than 1 year 49 22.8 North America 1 0.5
1-2 years 113 52.6 5. Gender
3-4 years 34 15.8 female 127 59.1
more than 4 years 15 7 male 88 40.9
3. Education Level 6. Age
bachelor 70 32.6 18-22 44 20.5
master 132 61.4 23-27 119 55.4
doctoral 13 6 28-32 38 17.7
33-37 7 3.3
38-42 4 1.8
Correlation analysis.
Pearson’s correlation analysis was used to test the relationship between variables in this study. This analysis provides correlation coefficients that enable the researcher to observe that when a variable changes, whether the other variables would change. Besides, it also indicates the strength and significances of the relationship among variables, demographics and control variables (Cleveland, 2015).
Structural equation modeling (SEM).
Structural Equation Modeling (SEM) is a statistics process that is used to test casual relationship (Jackson, Gillaspy Jr, & Purc-Stephenson, 2009). SEM method applies statistical data and casual assumptions.
In AMOS, a sort of Structural equation modeling (SEM), confirmatory factor analysis (CFA) and path analysis were conducted which provide results that enable the researcher to answer research questions and hypotheses of this study. CFA of the three variables in this study were independently analyzed in SPSS AMOS and the path analysis was run to test the hypotheses of this study. The indices in AMOS of the outputs for this study included 𝑥2/df, SRMR, CFI, NNFI (TLI), RMSEA, AVE and CR. The above indices were used to examine the goodness of the model fit in this study.
𝑥2/df indicates that Chi-square divided by the degree of freedom which is also called relative chi-square and shows the fit of the data to the model. The standardized root mean square residual (SRMR) refers to “the square root of the difference between the residuals of the sample matrix and hypothesized covariance model” (Hooper, Coughlan, & Mullen, 2008, p54). Goodness of fit (GFI) is used to see the proportion of variances that explain the
projected population covariance. Adjusted goodness of fit (AGFI), which is related to GFI, is affected by the degree of freedom. Nevertheless, McDonald and Ho (2002) argued that the often seen indices reported are CFI,GFI,NFI and the NNFI(TLI). Moreover, Hu and Bentler (1999) suggested to report two index including SRMR with NNFI (TLI), RMSEA or the CFI.
Therefore, the researcher reports CFI and NNFI (TLI) instead of GFI and AGFI in this study.
Root mean square error of approximation (RMSEA) explains how well the model fits to the population covariance matrix (Hooper, Coughlan, & Mullen, 2008). There is a summary of model fit indices criteria of 𝑥2/df, SRMR, CFI, TLI and RMSEA in Table 3.3.
Table 3. 3.
Indices of Model Fits
Good fit Acceptable fit
𝑥2/df 2-5 <5
RMSEA <.08 .08-.1
CFI >.96
NNFI (TLI) >.96
SRMR <.05 ≤.08
Note. Summary according to Hooper, Coughlan, and Mullen (2008)
𝑥2= chi-square; df= degrees of freedom; RMSEA= root mean square error of approximation;
CFI= comparative fit Index; NNFI (TLI) = Non-Normed Fit Index; SRMR= standardized root mean square residual.
Composite reliability (CR) indicates the reliability of constructs in measurement and the structural model (Bacon, Sauer, & Young, 1995) and average variance extracted (AVE) presents the average sum of the squared standardized loadings of a construct. According to
Fornell and Larcker (1981), the acceptable criterion for CR is above .80 and for AVE .50.
Hierarchical regression analysis.
Hierarchical Regression Analysis estimates the coefficient of linear equation that helps predicting the value of dependent variable and it is able to involve more than one independent variable. In hierarchical regression, independent variables are able to be cumulatively entered by the purpose and logic of the research. This analysis enables the research to test the hypotheses in this study through observing the strength and significances of the relationship among variables, demographics and control variables (Cohen & Cohen, 1975).
Measurements
The questionnaire was divided into three sections. The first section was the 14 items for intrinsic motivation that includes questions for the three types of intrinsic motivation (IM to know, IM toward accomplishment and IM to experience stimulation). The second section was the 15 items for extrinsic motivation (job advantage, job need, family pressure, peer pressure and social perception). The last section was the 15 items for Organizational Citizenship Behavior (Helping Behavior and Organizational Loyalty). In addition, the questionnaires were in bilingual form to help Chinese-speakers having better understanding on the questions in this survey.
Intrinsic Motivation
IM is defined as when a person is intrinsically motivated, the individual would participate in an activity out of enjoyment or the sensation brought by the activity itself (Deci
& Ryan, 1985). The questionnaire of IM was adapted from Academic Motivation Scale (AMS) which uses 7-point-Likert scale ranging from 1 (Does not correspond at all) to 7 (Corresponds exactly) (Vallerand et al., 1992, 1993). The constructs of three types of intrinsic motivation including IM to know, IM toward accomplishment and IM to experience stimulation were drawn from the AMS. Two items of intrinsic motivation to experience stimulation were removed because they do not fit the sample in this study. In addition, a construct composed of four items from EMI choice, which uses a 7-point-Likert scale ranging from 1 (does not apply at all) to 7 (fully applies) (Lueg & Lueg, 2015) was added to replace part of the questions of IM to experience stimulation. The Cronbach’s alphas of the three constructs in AMS were all above .70 (see Table 3.4) which shows good reliability of the AMS subscale. As for the construct from EMI choice, the Cronbach’s alpha is .74 (Lueg
& Lueg, 2015).
Table 3. 4.
Reliability of AMS
Construct Alpha Sample 1 Alpha Pretest Sample 2
Note. Adapted from “The Academic Motivation Scale: A Measure of Intrinsic, Extrinsic, and Amotivation in Education,” by Vallerand et al. 1992, Educational and Psychological
Measurement, 52(4), p.1014. Copyright 1992 by the Psychological Measurement.
Extrinsic Motivation
External motivation exists when behavior is performed not because of inner interest but other forces such as instrumental reasons (Deci & Ryan, 2000a). For extrinsic motivation, the questionnaire was adopted from EMI choice, which uses a 7-point-Likert scale ranging from 1(does not apply at all) to 7 (fully applies) (Lueg & Lueg, 2015). Five constructs (job advantage, job need, family pressure, peer pressure and social perception) that are related to extrinsic motivation of joining an EMI program were adopted in this study. For job advantage, there are five questions and the Cronbach’s alpha for this construct was .860; four items for job need and the Cronbach’s alpha was .888; five items for family pressure and its Cronbach’s alpha was .770. Then, there are three items for peer pressure and four for social perception. Cronbach’s alphas for these two constructs were .647 and .725 (Lueg & Lueg, 2015).
Organizational Citizenship Behavior
Organ (1988, p.4) gave OCB the definition of "individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and that in the aggregate promotes the effective functioning of the organization." In order to measure the dependent variable, Organizational Citizenship Behavior-Helping (OCBH) (Podsakoff, Ahearne, &
MacKenzie, 1997; Podsakoff P. , MacKenzie, Moorman, & Fetter, 1990) was adapted as part of the questionnaire for measuring OCB in this study. In addition, Organizational Citizenship Behavior-Loyalty (OCBL) (Moorman & Blakely, 1995) was also adapted to measure the dependent variable in this study. OCBH and OCBL use a 5-point-Likert scale ranging from
1(strongly disagree) to 5(strongly agree). There are 11 items for OCBH and 5 items for OCBL. In the previous study, the reliability coefficients of OCBH were .81 and .86 for OCBL. A sample item for OCBH would be “I help my classmates if someone falls behind in his/her work.” and “I defends the organization when outsiders criticize it” for OCBL.
Control Variables
Demographics.
A total of 9 questions were used to collect related demographic information. One question for confirming the participant has or had joined an EMI program, one for the time participants join(ed) EMI program, one for the country the participants join(ed) EMI program, one for the duration the participants join(ed) EMI program. Other demographics questions include age, gender, program’s name of participants, education level, and country of origin.
English Proficiency.
According to Lueg & Lueg (2015), whether students would choose an EMI program depends on the English proficiency that students perceived by themselves. Thus, in addition to the above demographics questions, there were 7 questions on English proficiency by Lueg
& Lueg (2015) applied into the questionnaire as one of the control variables.
Social Desirability.
Since there was a concern that a social desirability bias may affect self-rated organization citizenship behaviors (OCB), a validated scale measuring social desirability was included into this study. Social desirability responding refers to participants’ tendency to show others the favorable image of themselves (Van de Mortel, 2008). Furthermore, the
author states that social desirability responding has a high possibility to occur when participants respond to questions that are socially sensitive and social desirability responding bias may affect the validity of the instrument. In this study, the social desirability measurement which is a short version of Marlowe-Crowne Social Desirability Scale was adopted from Strahan and Gerbasi (1972). There are 10 true or false items included in the measurement and the Cronbach’s alpha of this scale was .85. This scale basically consists of items that are social desirability but not true for most of the people such as “I am always willing to admit when I make a mistake.” or socially unfavorable but often seen behaviors like “I have never deliberately said something that hurt someone’s feelings.” The score was calculated by counting the number of “true” after reversing the negatively phrased items. A
person with a high score indicates that one has high social desirability while a person with low score means one has low social desirability.
Validity and Reliability
In order to ensure the reliability and validity of the questionnaire, the researcher has done literature reviews as the basis support for this study. A panel of experts verified the content validity through examining the measurement items. Furthermore, a pilot study was conducted to confirm the face validity and reliability of the questionnaire. The researcher also conducted the exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) to examine the factor structure for construct validity.
Pilot Test Study Sampling and Data Collection
For pilot study, quantitative method was used to collect data to examine the validity and reliability of the measurement and research design of this study. Online survey questionnaires were chosen to collect the data. Survey questionnaires were distributed to students whose native languages are not English and join(ed) an EMI program in the Graduate Institute of International Human Resource Development in National Taiwan Normal University. 50 questionnaires were sent and there were 44 people responded and the response rate was 88%.
Nonetheless, among the 44 questionnaires, 3 invalid questionnaires ones were discovered. As for the instrument, it was consisted of 3 variables and there were total of 53 questions. In addition, 15 questions for demographics were also included (Yeh & Tsai, 2016).
Exploratory Factor Analysis (EFA)
Exploratory Factor Analysis (EFA) was conducted to examine the factor structure of measurement items in the pilot study of this research and the results are shown in Table 3.5, 3.6 and 3.7.
For the EFA of IM measurement items, the Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO) value was .85 and the Bartlett’s test of sphericity was significant. There
were 3 factors extracted with the eigenvalue of larger than 1. The 7 components explain 80.47% of the variances. As for the communality, there was only one item, IM10 (.69), under.70, which indicated that the items represent the components quite well. According to
were 3 factors extracted with the eigenvalue of larger than 1. The 7 components explain 80.47% of the variances. As for the communality, there was only one item, IM10 (.69), under.70, which indicated that the items represent the components quite well. According to