This chapter is composed by three sections which are research framework, research procedure, research hypotheses and research methods.
Research Framework
The research framework, as shown in Figure 3.1, is established based on the aim of the study and the previous reviewed literature. First, the levels of educational attainment could impact on the possibility of education-employment mismatch (overeducation and mismatch), the length of jobsearch duration, and perceived work values. In the study, the levels of educational attainment were restricted to bachelor degree and master degree.
Second, people might be willing to be overeducated (receiving additional years of overeducation) or mismatched (between job and major), as their needs can be satisfied by that job, namely, their work values were met. Also, people might be overeducated or mismatched, if they could not get jobs for a long time. Thus, the length of jobsearch duration and work values might be associated with education-employment mismatch.
Finally, work values might affect the length of the jobsearch duration as well. As mentioned before, due to the preference for the high salary, men would take time to find jobs with good pay, thus causing that men have longer jobsearch duration. Therefore, the preferences for certain work values might be correlated with jobsearch duration.
Figure 3.1. Research framework
Research Hypotheses
There are six hypotheses generated according to the research questions and the reviewed literature, and the research hypotheses were also demonstrated in Figure 3.1.
Question 1: Is there a relationship between educational attainment and education-employment mismatch?
Hypothesis 1: Educational attainment is significantly related to education-employment mismatch.
Hypothesis 1.1: Master’s graduates are significantly overeducated than bachelor’s graduates.
Hypothesis 1.2: Bachelor’s graduates are significantly mismatched than master’s graduates.
Question 2: Is the jobsearch duration of master’s graduates shorter than that of bachelor’s graduates?
Hypothesis 2: The jobsearch duration of master’s graduates is significantly shorter than that of bachelor’s graduates.
Question 3: Are the perceived work values significantly different between bachelor’s and master’s graduates?
Hypothesis 3: The perceived work values of bachelor’s graduates are significantly different from that of master’s graduates.
Question 4: Is there a positive relationship between jobsearch duration and education-employment mismatch?
Hypothesis 4: Jobsearch duration is significantly and positively related to education-employment mismatch.
Hypothesis 4.1: Jobsearch duration is significantly and positively related to overeducation.
Hypothesis 4.2: Jobsearch duration is significantly and positively related to mismatch.
Question 5: Is there a positive relationship between work values and education-employment mismatch?
Hypothesis 5: The certain perceived work values are significantly and positively associated with education-employment mismatch.
Hypothesis 5.1: The certain perceived work values are significantly and positively associated with overeducation.
Hypothesis 5.2: The certain perceived work values are significantly and positively associated with mismatch.
Question 6: Is there a relationship between work values and jobsearch duration?
Hypothesis 6: Work values are significantly associated with the length of jobsearch duration.
Research Procedure
The procedure of the study is showed in Figure 3.2 and the detail was described as following:
1. Identify the objectives of the research.
The first step is to clarify the aim of the study. After discussing with the advisor, the researcher determined the objectives of the study, and headed toward next step.
2. Write down the research questions.
Since the research questions are generated from the aim of the study, after answering these questions the purpose of the study can be achieved.
3. Review relevant literature.
After searching relevant studies, literature review was conducted to understand the findings and discussion of other research regarding the topic the research.
4. Develop research hypotheses.
Based on the reviewed literature and the aim of the study, the researcher could develop the hypotheses.
5. Collect the data.
The researcher applied to Taiwan Higher Education Database for the survey results.
6. Analyze the data.
After obtaining the data, the research could utilize statistical software to analyze the data, and to test the hypotheses.
7. Represent the results.
Then the statistical results were presented in accordance with the hypotheses.
8. Draw the conclusion and provide suggestions.
Consequently, the researcher drew the conclusion or implications from the results, and then proposed some suggestions based on the findings.
Figure 3.2. Research procedure
Research Methods
This section includes data collection, and data analysis. The sample, instrument, and measures in the research would be illustrated in the subsection of data collection, whereas the used statistical methods would be demonstrated in the subsection of data analysis.
Data collection
The data was from a survey conducted by Taiwan Higher Education Database (THEDS).
Due to the support from the Ministry of Education, THEDS could collect the information of students graduated from the junior colleges, colleges, universities and technological colleges and universities in Taiwan. Hence, THEDS could carry out a graduate census to investigate the flow of graduates one year after graduation. After obtaining the information about graduates, THEDS sent e-mail to inform them, irrespective of their majors, to complete the questionnaires on line. Consequently, the questionnaires were delivered to 262,743 bachelor’s graduates and 88,694 questionnaires were returned, thus the response rate was 33.76%. On the other hand, the questionnaires were delivered to 47,283 master’s graduates, and 11,912 questionnaires were returned, thus the response rate was 25.19% (Taiwan Higher Education Database, 2008).
Sample
The participants of the survey were bachelor’s and master’s graduates who graduated in 2006. In addition, the sample in the study was drawn as follows. First, the graduates who took in-service education program or in-service training were excluded. Second, THEDS divided the sample into 18 groups by their majors. However, due to the aim of the study, only graduates majored in the engineering and technology would be selected in the study. In order to extract the sample required for the study from these 18 groups, the principles established by Ministry of Education and National Science Council to classify the majors were followed.
Further, the researcher specifically focused on mathematics and computation. Finally, graduates who had a part-time job or an intern job were excluded, that is only graduates who
had a full-time job were included. Moreover, an individual who had a full-time job but ran one’s own business would be eliminated.
Instrument
As mentioned, the questionnaire was used to survey the flow of graduates. The questionnaire covered four topics including current situation, employment situation, preparation for further study, and personal background. And since information concerning education-employment mismatch (overeducation and mismatch), jobsearch duration, and the perceived work values could be obtained in the employment situation section, the survey was suitable for the study. As a result, the survey to investigate the flow of graduates one year after graduation was employed in this study. Last, this survey was conducted from June 2007 to February 2008.
Measures
Employment-Education Mismatch. As discussed previously, employment-education mismatch was defined in two perspectives, one is overeducation (in terms of the additional years of overeducation), and the other one is mismatch (between job and major), so respondents were asked to answer the following questions. And self-assessment method was used to measure overeducation.
(a) What extent is your current job related to your field of education? Four response alternatives were provided:
Not related
Somewhat related
Mostly related
Closely related
(b) What extent educational level do you think is required to perform your current job?
Six response alternatives were provided:
Senior High school
Senior Vocational school
Junior college
University and college
Master degree
Therefore, according to their answers, the respondents could be divided into two groups, matched people and mismatched people. Respondents who answered not related or somewhat related to the first question were described as mismatched people. Since the first question is scored on a four-point likert scale ranging from 1 (Not related) to 4 (Closely related), a lower score reflect the mismatch.
Moreover, if the educational level required for the job was below one’s actual educational level, he or she was also called overeducated. Supposing that the respondent with university or college degree not answered university and college degree or master degree to the second question, s/he was overeducated. On the other hands, if the respondent with master degree not answered master degree to the second question, s/he was overeducated.
Additionally, the extent of overeducation can be calculated by comparing the number of years of education required by one’s job (required schooling years) with the number of years of the education level one attained (schooling years). Providing that one’s schooling years was larger than required schooling years, one was overeducated. And Table 3.1 present the schooling years of each level of education (Ministry of Education, 2007).
Table 3.1.
The schooling years of each level of education Level of education Schooling years Below junior high school 9
Senior High school 12
Senior Vocational school 12
Junior college 14
University and college 16
Master degree 18
Doctorial degree 21
Source: Ministry of Education, 2007.
Jobsearch duration. Jobsearch duration was measured by this question. There are no options for respondents to choose, and they would write down the number on the space.
How many months did you spend on looking for this job?
Work values. The respondents were asked to evaluate the importance of the nine items of work values. These nine items are compensation, benefit, security, work location, advancement, work challenge, work responsibility, work independence, and contribution to the society.
(a) While choosing an occupation, how important is each one to you? Each item has four options for respondents to check.
Very unimportant
Somewhat unimportant
Important
Very important
Hence, a four-point Likert scale was used to calculate the score of the importance (1 = very unimportant, 2 = somewhat unimportant, 3 = important, 4 = very important).
Data Analysis
SPSS 15.0 for window was used for data analysis to test the research hypotheses.
Following are the definitions of variables and statistical methods used in the study to test the hypotheses.
First, variables used in the study were identified as following. Additionally, for regression purpose, some categorical variables must be converted to dummy variables (see Table 3.2).
1. Independent variables
The independent variables is educational attainment (bachelor = 1; master = 2), while jobsearch duration, and the nine work values served as independent variables in hypothesis four and five respectively.
2. Dependent variables
The dependent variable was employment-education mismatch. As mentioned, there were two items to evaluate employment-education mismatch, one item is mismatch, which is the relation between job and major, and the other one is overeducation, which is the additional schooling years of overeducation, namely, one’s actual schooling years (educational level) is higher than required schooling years (required educational level) for the job. And employment-education mismatch was dichotomized as 0 = matched/not overeducated and 1 = mismatched /overeducated.
3. Control variables
According to the previous reviewed literatures, gender, marital status, and school type might be related to overeducation. Also, in Chen’s study (2006), the result showed that sector difference was associated with overeducation. Therefore, these four variables served as control variables in this study. And the codings of four control variables were as follow:
gender (male= 1; female = 2), marital status (single = 1; married = 2), school type (public school = 1; private school = 2), and sector type (public sector = 1; private sector = 2; other =
3; voluntary sector was the reference group). And the public sector consists of government, public enterprise, and public school, while private sector consists of private enterprise, and private school; voluntary sector consists of non-profit organizations (Wikipedia).
Table 3.2.
Variables coding system
Variables Coding system
Educational attainment 0 Bachelor degree 1 Master degree
Employment-education mismatch 0 Matched / Not overeducated 1 Mismatched / Overeducated
Gender 0 Male
1 Female
Marital status 0 Single
1 Married
School type 0 Public school
1 Private school
Sector type 0 (100) Public sector
1 (010) Private sector 2 (001) Other sector 3 (000) Voluntary sector
Then, here are the statistical methods used in the study.
1. Descriptive statistics
Description statistics provide the summaries concerning the measures. The information concerning number, mean, and standard deviation of the variables such as education-employment mismatch (overeducation and mismatch), jobsearch duration, and work values could be identified. Additionally, the rank of work values of bachelor’s graduates and master’s graduates would be demonstrated.
2. Independent sample T test analysis
Independent sample T test analysis was utilized to test the hypothesis one, two, and three.
Whether there was a significant difference in employment-education mismatch (overeducation and mismatch), jobsearch duration, and work values among bachelor’s graduates and master’s graduates can be confirmed.
3. Multiple regression model
Hierarchical regression model. There are four levels in the study. The first level was composed of demographic variables (educational attainment and four control variables, gender, marital status, school type and sector type). Then, the second level included the five intrinsic work values and control variables, and the third level included all work values and control variables. There were nine work values variables denoted as following: X1 = compensation, X 2 = benefit, X3= security, X4 = work location, X5 = advancement, X6 = work challenge, X7 = work responsibility, X8 = work independence and X9 = contribution to the society. And these nine items fall into two categories, one group contains intrinsic work values and the other one contains extrinsic work values based on the Minnesota Satisfaction Questionnaire (MSQ; Weiss, Dawis, England & Lofquist, 1967). As a consequence, intrinsic work values cover security, work challenge, work responsibility, work independence and contribution to the society while extrinsic work values cover compensation, benefit, work location, and advancement. Finally, the fourth level was composed of above variables, and
jobsearch duration.
Thus, the relationship between employment-education mismatch (overeducation and mismatch) and educational attainment /jobsearch duration / these nine work values could be explored through this model, and employment-education mismatch could be predicted.
Further, multiple regression model could be utilized to test the hypothesis six, the relationship between work values and jobsearch duration, so which work values was significantly related to jobsearch duration could be identified. In addition, the collinearity diagnostics would be conducted to examine whether there is a high correlation between independent variables.
4. Logistic regression model
There were two items to measure employment-education mismatch (overeducation and mismatch). Firstly, according to the answer to the question (a), respondents who answered not related or somewhat related were mismatched, while people who answered mostly related or closely related were matched.
Likewise, according to the answer to the question (b), people who were overeducated could be identified. For bachelor’s graduates, people whose answers were below junior high school, senior high school, senior vocational school or junior college were classified as overeducated, whereas master’s graduates whose answers were not master degree were classified as overeducated.
While dependent variable employment-education mismatch is a categorical variable, the logistic regression model was utilized. Therefore, the relationship between employment-education mismatch and educational attainment /jobsearch duration / the nine work values could be explored as well. Also, the hierarchical regression model is adopted.