The self-employed and employees and Data and Methods
Based on the above hypotheses, this study is expected to find the differences in the
work-related stress coping mechanism between the self-employed and employees. As
noted before, the self-employed may experience the different levels of job satisfaction
(Lange,2012), self-efficacy(Bradley, and Roberts, 2004) and also work-related
stress(Hessels, Rietveld and Zwan,2017). Moreover, job characters are also different
from each other (Baron, Franklin andHmieleski, 2016). Role differences between self-
employment and employment are well established in the academia (Hoang and
Gimeno, 2009; Haynie and Shepherd, 2009). However, there is insufficient theory
linking these role differences to discuss the relationships among work-related stress,
job satisfaction and self-efficacy. Therefore, this study will base on the findings of
hypotheses tests between the self-employed and employees to suggest: if the findings
between the self-employed and employees are different from Hypothesis 2: work-
related stress negatively relate with job satisfaction among the self-employed Then
this study suggests Hypothesis2c: the self-employed moderate the relationship
between job satisfaction and work-related stress.
If the findings between the self-employed and employees are different from Hypothesis 3:
self-efficacy is positively related to job satisfaction, then it suggests Hypothesis3c: the
self-employed moderate the relationship between self-efficacy and job satisfaction. If the
findings between the self-employed and employees are different from Hypothesis
4: self-efficacy negatively moderates the relationship between work-related stress and
job satisfaction. Then it suggests Hypothesis4c: the self-employed moderate the
relationship between the interaction term of self-efficacy*work-related stress and job
satisfaction.
Data and Methods
Dataset
The sample is selected as the respondents claimed themselves as the self-employed or
employees in Understanding Society Questionnaire. Also, the observations need to
answer the relevant questions from self-efficacy module (only available in wave 5) and
workplace wellbeing module (available in wave 6) and other relevant questions. Thus,
this study merges dataset of Wave 5 and Wave 6 to obtain data for all needed variables.
171
After sample selection, the dataset consists of 12162 observations (Employees:
n=10481; self-employed: n= 1081).
4.3.2 Measurements
Job satisfaction, the dependent variable, job satisfaction is measured by the question
‘how dissatisfied or satisfied are you with your present job overall?’, the answers
scale ranges from 1 completely dissatisfied to 7 completely satisfied.
Self-efficacy, the moderator, is measured by a set of 10 questions, the scale is sourced
from Schwarzer and Jerusalem’s (1995) Generalised Self-Efficacy Scale (short form),
which has been applied widely in the psychological and social studies (Chiu,2014).
Exemplary questions include ‘Solve difficult problems if try hard enough’; ‘someone
opposes me can find ways to get what I want’; ‘Easy to stick to aims and accomplish
goals’; ‘Confidence can deal with unexpected events’. The answers are listed from 1
Not at all true to 4 exactly true. This study uses the principal factor analysis to
generalise a principle factor to represent self-efficacy, with higher score, individual
has higher level of self-efficacy (Cronbach’s a=0.87)
Work-related stress, the Independent Variable, Understanding Society dataset
automatically calculates two indexes to measure the two aspects of work-related stress:
job anxiety and job depression, which is measured by using the job-related wellbeing
scale developed by Warr (1990). This scale consists of two three-item subscales
measuring ‘job-related depression’ and ‘job-related anxiety’. Both scales use a Likert-
type response format and have demonstrated acceptable reliability and validity.
172
Here we reverse the score of the index to understand the stress easier. The reversed
scores mean that people with a higher score experience a greater level of job stress.
Kerr, McHugh, and McCrory (2009) and Rothmann (2008) have used this
measurement to test job stress and wellbeing.
Demographic variables include age, sex, marriage status, education and monthly job-
related income. For instance, women tend to report more psychosomatic symptoms
(Jamal and Badawi 1995). Similarly, Jamal (1997) noted that age might play an important
role: older people report more health problems than younger people do.
Moreover, higher education may enlarge individual’s employment
opportunity and
hence enhance the chance to find a more stratified job. The single person experiences
higher job satisfaction due to the lack of family-work conflicts (Kristof‐Brown,
2005). Therefore, consistent with previous research, all these Demographic variables are
included as control variables. Moreover, higher income can increase job satisfaction
(Kristof‐Brown, 2005). All these variables will be controlled in the matching
approach to establishing a comparable dataset, which will be explained in the next part.
4.3.3 Methods
To test the hypothesis, the paper will research three steps:
The first step is to establish a comparable data pool by using the Matching approach.
Observations are classified into two groups by the
respondents’ job (the
self-employed:1, employees:0). The second step is to assess the mean differences of
work-related stress, job satisfaction, self-efficacy between groups after matching. This
173
aims to test Hypothesis 1: Self-employed (solo self-employed and self-employed with
hiring employees) experience less work-related stress than individuals in wage work.
Moreover, Hypothesis 2: self-employed has a higher level of job satisfaction. The
third step is to test Hypothesis 3 and Hypothesis 4.This study will use hierarchical
regression models to test the moderating effects in the framework. Firstly, this paper
will test the direct effects of work-related stress and self-efficacy on job satisfaction
within the whole dataset and between the self-employed and employees respectively.
Then the interaction terms of the self-employed and work-related stress(job
anxiety*the self-employed; job depression* the self-employed); self-employed and
self-efficacy(self-efficacy*the self-employed) will be added into the model to test
group differences between the self-employed and employees on the relevant variables.
Lastly, the interaction term work-related stress and self-efficacy (work-related stress*
self-efficacy) will be added into the model to test the moderating effect of Self-
efficacy on the relationship between work-related stress and job satisfaction in both
occupational groups.
The matching approach
This study will use R software to run the Matching approach by applying the software
package of ‘Matchit’. To be more specific, this study control the rationale of the
treatment group population to control group population group as 1:1, so that is 1681 (self-
employed)：1681(employees) and after matching, the distributions of propensity scores
tend to be similar between the two groups (see in Figure 7). Also, it can be seen
174
that the demographical differences among control variables have been widely reduced. The
T-test p-values results are presented in Table 20. None of them displays a significant
difference in demographic variables between self-employed and employed.
