In this chapter, it demonstrates the concept, procedure, and statistical techniques of this study to make sure the process meet adequately with the purpose. The headings below include research framework, hypothesis, research procedure, questionnaire design, sampling and data collection, measurement, and validity and reliability.
Research Framework
The following framework demonstrates the relationship among four variables in this study. Self-efficacy is the independent variable, and in-role performance and extra role performance are the dependent variables. Mentoring function serves as a direct effect on the outcome variables, as well as the moderator, which will affect the relationship between self-efficacy and performances.
Figure 3.1. Research framework
24
Hypothesis
H1: Self-efficacy has a positive relationship with in-role performance.
H2: Self-efficacy has a negative relationship with helping.
H3: Self-efficacy has a positive relationship with voice.
H4a: Career development mentoring function will moderate the positive relationship between self-efficacy and in-role performance.
H4b: Psychosocial support mentoring function will moderate the positive relationship between self-efficacy and in-role performance.
H4c: Role modeling mentoring function will moderate the positive relationship between self-efficacy and in-role performance.
H5a: Career development mentoring function will moderate the positive relationship between self-efficacy and helping behavior.
H5b: Psychosocial support mentoring function will moderate the positive relationship between self-efficacy and helping behavior.
H5c: Role modeling mentoring function will moderate the positive relationship between self-efficacy and helping behavior.
H6a: Career development mentoring function will moderate the positive relationship between self-efficacy and voice behavior.
H6b: Psychosocial support mentoring function will moderate the positive relationship between self-efficacy and voice behavior.
H6c: Role modeling mentoring function will moderate the positive relationship between self-efficacy and voice behavior.
H7a: Career development mentoring function has a positive relationship with in-role performance.
H7b: Psychosocial support mentoring function has a positive relationship with in-role performance.
25
H7c: Role modeling mentoring function has a positive relationship with in-role performance.
H8a: Career development mentoring function has a positive relationship with helping.
H8b: Psychosocial support mentoring function has a positive relationship with helping.
H8c: Role modeling mentoring function has a positive relationship with helping.
H9a: Career development mentoring function has a positive relationship with voice.
H9b: Psychosocial support mentoring function has a positive relationship with voice.
H9c: Role modeling mentoring function has a positive relationship with voice.
26
Research Procedure
Firstly, the researcher identified the topic of this research, which was based on personal experience when serving internship in a company. Because of a lack of operational experience, the researcher had a lower level of self-efficacy while doing the tasks. For the reason, this research tried to investigate the issue on early-career employees, in order to help them in workplace. Secondly, the researcher conducted the literature review, through reading articles to gain knowledge about the history and definition. Literature review is one of the most important and essential process in research procedure, in order to get substantial knowledge in professional domain. Next step, the research developed the research questions, for the reason to ensure the direction of the research. The next step was to develop the framework, by establishing the independent variables and criterion variables, and the effect of the moderator with them, and to create the hypotheses. Then, the researcher developed the measurement for research variables. Through literature review, the researcher sought out some relevant scales which can be used to assess the variables, checked the citation and Cronbach alpha reliability of the scales, for the sake of measurements’ reliability and validity. After all of the above process were done, this study collected data from the target sample. Once all the needed data were collected, data were analyzed to test the hypothesis, and findings were interpreted. Finally, the researcher provided recommendations and conclusions based on the result.
27
Figure 3.2. Research procedure
Research Design
This study uses a quantitative approach to statistically test the hypothesis. Quantitative research is a systematic investigation via statistical and computational techniques for observing the particular phenomena (Bhawna & Gobind, 2015). By quantitative approach, the researcher can get massive data to analyze the relationship between self-efficacy and performance and also predict the influence of mentoring function. Early-career workers are the sample of this study.
Sampling and Data Collection
The sample of this research is 395 participants of early-career employees whose tenure were below 5 years and with a working mentor in past 6 month. The target comprises employees from various industries with different professional backgrounds. This study used convenience sampling method to collect data by online questionnaires. The online survey was designed using google forms, which is free and time saving. The questionnaire
28
was distributed via technological means, like email, social media (e.g, ptt, Facebook) websites, etc.
Table 3.1. provides a demographical sketch of the sample. Base on the descriptive statistics, it shows that most of the participants were 21 to 30 years old (89.6%) and most of them had a bachelor degree (77.7%). This study surveyed the early-career employees, especially focusing on the employees who had below 5 years of working experience.
According to the survey of ministry of labor of R.O.C in 2016, it showed that 60.47% of people started to work when they were 22-25 years old. The result was the same as the data of this study. University education is very common in Taiwan. As a result, a bachelor degree has become a standard, critical for looking for a job in Taiwan. Most of the students will try to study in university in order to fulfill the social expectation. The sample data shows the same distribution with the phenomenon. Moreover, most of their tenure were lower than 2 years (46.5%). For the year after graduation from last education, participants were evenly distributed among still studying (21.6%), below 1 year (24.4%), and 1~2 years (22.1%). Over 60 percent of respondents reported that they were in their first job experience (60.7%). And approximately 70% of respondents do not have similar working experience before. Although this study only collected the data from a small sample of a large population, the sample distribution shows these participants can be considered as representative of early career employees.
29
Table 3.1.
Descriptive Statistics on Sample Demographics (N = 395)
Questionnaire Design
The questionnaire is composed of four parts, and most items were adapted from the previous studies and translated from English to Chinese. Minor revisions on the language
Feature Categories Frequency Percentage (%)
Age Below 21 24 6.1
Bachelor’s Degree 306 77.7
Master Degree 71 18.0
30
may have been performed to suit the target. The self-efficacy measurement was adopted from previous studies, which provided several versions in different languages of the questionnaire. The first part of questionnaire asks about the mentoring function, which has 15 items, and is about how you feel about your mentor, the behavior, and relationship of you and your mentor. The second part includes the items which focus on personal in-role performance and extra-role performance at work, with a total of 24 items. The third part measures self-efficacy with 10 items, focus on individual’s belief. And to the last part, deals with demographic questions about the respondents, including age, gender, education, working age, and personal working-related experience. All of the items used in the questionnaire are listed in the Appendix.
Measurement
This study used previously published scales to measure each variable of interest to ensure validity and reliability. All of the measurements are self-report. The details of each measurement is described below.
Self-efficacy
Schwarzer, Bäßler, Kwiatek, Schröder, and Zhang (1997) regarded self-efficacy as people’s belief of they can solve a problem instrumentally. As the result, people are more intended to take the challenge and also they feel more committed to the decision they’ve made. This study adopted the self-efficacy scale from their publication. They compared three versions of self-efficacy scales, in order to examine if the theoretical construct of perceived self-efficacy is universal, and at the same time, test if the scale can be used with different language populations. In the measurement, it used 5-point Likert scale with 10 items in one dimension, focusing on individual’s general self-efficacy. The original version is from German, and also been proved to be reliable and valid in various studies (Schwarzer, 1993). Chinese and English version were adopted from the original translated items. There were 959 Costa Rican, and 293 Chinese university students participated in the research. As
31
the result, the Chinese version for the Hong Kong sample, had excellent Cronbach, α = .91;
and the English version, tested at Costa Rica, got the Cronbach α = .81.
This study focuses on Taiwanese early-career employee, hence, adopts Chinese version of the self-efficacy scale, using 5-point Likert scale. After data collection, the items were aggregated to form an overall value, the higher the value, the higher is the respondent’s self-efficacy.
Mentoring Function
Scandura and Ragins (1993) asked respondents about if they had a working relationship which significantly affected their career mobility in their organization. Based on 20-item scale, investigating 608 employees who had a mentor, this measurement was adapted by Scandura (1992) which assesses mentoring function. After that Hair, Anderson, and Tatham (1987) deleted five items because of their factor loading lower than .50. The measurement included three dimensions. “Psycho-social support” reflects the friendship and psychological support of mentor. It includes five items. The Cronbach α estimate of internal consistency was .81. “Career development”, which describes a set of 6 items to assess the work-related support by mentor, the Cronbach α estimate was .75. “Role modeling” uses four items to reflect whether the protégés behavior were affected by mentors; the Cronbach α was .70 (Hair et al., 1987).
In this research, mentoring function measurement was adapted from Scandura and Ragins’s (1993) version, and adjusted the particular words to fit the target of this research, for example, for the item “I respect mentor’s knowledge of the accounting profession.”, this research changes “accounting profession” into “specific profession”, in order to distribute to employees in various industries. This research used 5-point Likert scale as evaluation for respondents.
In-role performance
This study adapted the items from Van Dyne and LePine (1998), which conducted two
32
pilot studies with different participants in their research. The first pilot included 48 MBA students, using self-report; the Cronbach α was .89. Another pilot test had 321 participants, which were subordinates from several organizations. The Cronbach α was .90. Both of the studies’ Cronbach α results were good. Hence, this research adapted the measurement.
Furthermore, because most of the study sample were Taiwanese, this research translated the measurement into Chinese and used the back-translation process to ensure that the translation did not change the meaning of items. The items were evaluated using a 5-point Likert scale. The higher the overall value, the higher is the respondent’s in-role performance.
Helping
The helping behavior evaluation scale came from Van Dyne and LePine research (1998), which was adapted from Organ and Konovsky (1989) and Smith, Organ, and Near (1983). Van Dyne and LePine (1998) tested in-role performance and extra performance at the same time, and they tested helping variable with 7 items with a Cronbach α of .85.
This study adapted the measurement for the study and translated the scale into Chinese, utilizing back translation method to ensure the accuracy of items. Five-point Likert scale was used in the measurement.
Voice
According to Van Dyne and colleagues (1994) and Whithey and Cooper (1989) researches, voice was measured with six items by Van Dyne and LePine (1998). The measurement was tested by different population, included self, peer, and supervisor. In the result, Cronbach α were .88, .95, and .94, respectively.
This research adapted the measurement from Van Dyne and LePine (1998), and translated it into Chinese. The items were assessed using a 5-point Likert scale.
Demographics and Control Variables
A total of 7 variables were intended as controls in the analyses, which also included
33
the demographic controls. There are several reasons underlying the inclusion of controls.
First, since the outcome variables of in-role performance, helping and voice behaviors, were self-reported, they may be associated with personal background variables including age, gender and education. Secondly, mentoring is an interaction between the mentor and the protégé in the workplace, therefore, contextual factors of the job/task such as tenure, job experience were considered to be controlled. Measurement items of all of these variables were included in the last part of the survey, the background information section.
Personal background. Three variables are related to personal information of the
respondents. The variables are below:
Gender. Gender may have different effect on the preference of individual behavior and thinking. Because of the discrepancy between male and female in physiological structure and cognition, it might cause the different result.
Age. Age might influence personal thinking, feeling and experience. This research distinguished respondents into four groups. The age from below 20 years old, 20-29 years old, 30-39 years old and above 40 years old. Only four options were provided because this study focused on early career employees, who normally will not age above 40.
Education. Education has potential influence on personal self-efficacy and performance. It was measured with an ordinal question with five options from below high school diploma, high school degree, bachelor degree, master degree, to doctoral degree.
Working context. The variables were designed to understand the context of the
respondents’ working experience. In order to investigate the potential effect of working experience on self-efficacy and performance.
Working tenure. The working tenure was attempted to understand respondent’s length of working years, which has a potential impact on respondents’ career self-efficacy and performance. The questionnaire provided six options, which were: less than 1 year, 1- less than 2 years, 2- less than 3 years, 3- less than 4 years, 4- less than 5 years, 5 years and
34
above. This research focused on the employees whose working tenure were less than 5 years. As the result, the option of above 5 years was treated as a screening question, for filtering those who were not qualified.
First job. First job is measured to find out if participants have previous job experience.
Practice makes perfect. Work experience has potential influence on employees’ career outcomes. This variable was measured by a yes-or-no question, to distinguish participants’
condition.
Validity and Reliability
The questionnaire in this study was based on the literature review by previous researches. The most important thing for a measurement is its validity and reliability. This research performed the exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) to examine the measurement validity. Furthermore, this research conducted reliability testing using Cronbach's Alpha.
Pilot Test
The pilot test was conducted using 77 early-career employees who have mentor experiences in the past six months. All of the participants were selected by convenience sampling and snowballing method through the internet, such as Facebook and social media.
According to Table 3.2., it shows that most of the sample (90.9%) were 21 to 30 years old. There were more male than female respondents participated in the pilot test, which is 57.1% and 42.9% respectively. Most of the respondents were bachelor degree (90.9%) and had a working experience below two years. Approximately 56 % of respondents claimed that the experience they responded in the survey was their first job experience. In addition, most of them (70%) did not have a similar job experience before.
Furthermore, the variables’ reliability was observed with Cronbach’s Alpha value. In Table 3.3., all of variables showed a good result. Most of the variables’ Cronbach’s Alpha were approximately .90. The result of pilot test showed an acceptable outcome, hence, this
35
research utilized the measurement directly to collect the data.
Table 3.2.
Descriptive Statistics on Sample Demographics (n = 77)
Table 3.3.
Reliability Analysis of Pilot Test (n = 77)
Scales Number of Items Cronbach’s Alpha
Self-efficacy 10 1.00
Mentoring Function 15 .95
Career Development 6 .92
Psychosocial Support 5 .88
Role Modeling 4 .89
In-role Performance 4 .90
Extra role Performance 13 .90
Helping 7 .85
Voice 6 .87
Feature Categories Frequency Percentage (%)
Age Below 21 7 9.1
Year after Graduation Still studying 33 42.9
Below 1 year 15 19.5
36
Main Study Confirmatory Factor Analysis (CFA)
In this study, Confirmatory Factor Analysis (CFA) was utilized to ensure the construct validity. By using AMOS to check the CFA of measurement model, the CFA should meet the satisfactory fit which is provided below, Table 3.4. According to previous study, if the value of Chi-square divided by degree of freedom test (2/df) is under 3, it indicates good model fit (Gefen, Straub, & Boudreau, 2000). Goodness of Fit Index (GFI), Normed Fit Index (NFI), Comparative Fit Index (CFI), Root Mean Square error of approximation (RMSEA), Non-Normed Fit Index (TLI), and Incremental Fit Index (IFI) are some indices which will determine the overall model fit.
Table 3.4.
Summary of Goodness of Fit Criteria
Index Good fit References
χ2/df < 3 Gefen et al., 2000
GFI > .080 Byrne, 1994
NFI > .90 Byrne, 1994
CFI > .90 Byrne, 1994
RMSEA < .08 McQuitty, 2004
NNFI(TLI) > .90 Bagozzi and Yi, 1988
IFI > .90 Hu and Bentler, 1999
37
The original CFA model included 42 items, 10 items measuring self-efficacy, 6 items for career development, 5 items for psychosocial support, 4 items for role modeling, 4 items measuring in-role performance, 7 items of helping behavior and 6 items for employee’s voice behavior. Refer to Figure 3.3, it shows the 42 items with different variables. All of the items’ factor loading are above .60. And the result of model fit is listed in Table 3.5., which shows two of these indices (GFI = .766, NFI = .852) are not acceptable according to the criteria. Furthermore, CFI = .899, TLI = .891, are slightly lower than indicated in the criteria above. As a result, this study tried to modify the items in order to improve the fit to satisfy the CFA criteria.
Table 3.5.
Model Fit Summary – Original CFA Model (N = 395)
χ2 Df χ2/df GFI NFI CFI RMSEA TLI IFI Original
42 items
2204.9
55 798 2.763 .766 .852 .899 .067 .891 .900
38
Figure 3.3. Research CFA model
39
Table 3.6.
Confirmatory Factor Analysis Result (N = 395)
Item Estimate
Self-efficacy
SE1 I can always manage to solve difficult problems if I
try hard enough. .835
SE2 If someone opposes me. I can find means and ways
to get what I want. .769
SE3 It is easy for me to stick to my aims and accomplish
my goals. .799
SE4 I am confident that I could deal efficiently with
unexpected events. .884
SE5 Thanks to my resourcefulness, I know how to handle
unforeseen situations. .885
SE6 I can solve most problems if I invest the necessary
effort. .870
SE7 I can remain calm when facing difficulties because I
can rely on my coping abilities. .894
SE8 When I am confronted with a problem, I can usually
find several solutions. .874
SE9 If I am in a bind, I can usually think of something to
do. .870
SE10 No matter what comes my way, I’m usually able to
handle it. .855
Career Support
CS1 Mentor takes a personal interest in my career. .785 CS2 Mentor has placed me in important assignments. .858
CS3 Mentor gives me special coaching on the job. .871
CS4 Mentor advised me about the promotional
opportunities. .751
CS5 Mentor helps me coordinate professional goals. .838 CS6 Mentor has devoted special time and consideration to
my career. .803
(continued)
40
Table 3.6. (continued)
Item Estimate
Psychosocial Support
PS1 I share personal problems with mentors. .778
PS2 I socialize with mentor after work .746
PS3 I exchange confidence with mentor. .754
PS4 I consider mentor to be a friend. .836
PS5 I often go to lunch with mentor. .707
Role Modeling
RM1 I try to model my behavior after my mentor. .766
RM2 I admire mentor’s ability to motive others. .849
RM3 I respect mentor’s knowledge of the specific
profession. .845
RM4 I respect mentor’s ability to teach others. .902
In-role Performance
IR1 I fulfill the responsibilities specified in my job
description. .799
IR2 I perform the tasks that are expected as part of the
job. .878
IR3 I meet performance expectations. .838
IR4 I adequately complete responsibilities. .850
Help
Help1 I volunteer to do things for the cooperative. .693 Help2 I help orient new employees in the cooperative. .739 Help3 I attend functions that help the cooperative. .614 Help4 I assist others with their work for the benefit of the
cooperative. .770
cooperative. .770