SEM and Moderation effect test
Table 10 shows the model SEM pathway between Job Autonomy and Workplace
wellbeing and moderation effect of poverty in whole data, the self-employed group and
employee group. All the models goodness of fit shows good results (see in Table11).
Generally, within the group of the self-employed and the group of employees, Job
autonomy is significantly related to workplace wellbeing (job satisfaction, job anxiety and
job depression) (p<0.001), which confirm the H1: In general, job autonomy has a positive
or negative relationship with workplace wellbeing.
In Table 8, for the self-employed, job autonomy manifests a strong predicting power
on job-wellbeing. However, the poverty does not show a significant relationship with
workplace wellbeing. Moreover, the interaction term job autonomy* poverty does not
present its significance among the self-employed as well, which suggesting that
poverty does not exert moderating effect between job autonomy and job anxiety. For
the employees, and the whole dataset, job autonomy* poverty has significantly impact
on relationship between job autonomy and job anxiety. Thus, the moderation effect of
poverty is verified (p<0.01).
Table Summary of Goodness of Fit
Moderation model
χ2 df CFI TLI RMSEA SRMR
Whole dataset 1183.985 103 0.98 0.973 0.038 0.034
The self-employed group 356.104 103 0.98 0.977 0.032 0.030
Employee group 1013.226 103 0.97 0.965 0.043 0.043
In conclusion, in the employee group, when the incomes of individuals are below the
poverty line, job autonomy can weaker its magic powder on predicting workplace
84
wellbeing, and the moderating models support our third hypothesis H3b: poverty
moderates the relationship between job autonomy and workplace wellbeing in
employee group. However, this moderating effect cannot be found in the self-
employed groups. Therefore, H3a: poverty moderates the relationship between job
autonomy and workplace wellbeing in the self-employed group cannot be supported.
Discussion and Conclusions
Building on Maslow’s hierarchy of needs and SDT, this study test the mean
differences of job autonomy and workplace wellbeing between the self-employed and
employees when they live in poverty. Consequently, the findings are concluded as
below: Firstly, the self-employed are always experiencing higher job autonomy and
workplace wellbeing than employees even their incomes are below the poverty line.
Secondly, the pattern of job autonomy related with workplace wellbeing has been
verified between the self-employed and employees. Thirdly, poverty exerts
moderating effect in the relationship between job autonomy and workplace wellbeing
only in employees group, that job autonomy’s predicting power on workplace
wellbeing is weaker in the employee observations with lower income. For the self-
employed, the autonomy’s power has no different no matter how much the self-
employed earned. Our data size (n=20626) is large enough to provide sufficient
empirical data evidence to answer our two main research questions.
This research makes contributions on theoretical, empirical and methodological
aspects. On the theoretical perspective, Maslow’s hierarchy of needs and SDT are
classical theories, but to date a comprehensive empirical and theoretical development
of two theories still is still far to adequacy in the entrepreneurship research, indicating
that little can be said about the validity of the model within the field. By testing the
interplay between Maslow’s hierarchy and SDT, this study shed some light on both the
85
psychological and self-employment theories. This study focus on specific two factors
in the context of working and their effects on workplace wellbeing: job autonomy
from SDT theory and personal income from Maslow’s hierarchy. It is found that
income constrains job autonomy’s magic power on workplace wellbeing in employee
group rather than the self-employed’s group. These findings respond the current
research interest in the psychological field of studying the interplay between the
theories of SDT and Maslow’s hierarchy of needs (López-Rodríguez and Hidalgo,
2014; Tay and Diener 2011and Chen, 2015).
Moreover, empirically, this study does not only confirm the results consistent with
previous research which implies that the self-employed are happier than employees
(Benz and Frey, 2004), which largely due to the extent of job autonomy that the self-
employed owned (Le Blanc et al., 2001). It also pays attention to the income status. In
this study, more than 30% the self-employed live in poverty, while the employee is
13%. Indeed, the self-employed have a greater chance to stay in poverty status.
However, we found that, regarding the self-employed, poverty cannot constrain job
autonomy’s magic power on workplace wellbeing among the self-employed but can
be found significantly among employees. This confirmed many entrepreneurial
researchers’ inference that non-monetary rewards of the self-employed can be a
distinguished motivator of starting and maintaining their own business even to some
extent exceeding the significance of financial reward (Merz 2007; Shane, 2008). This
enlightens policymaker to emphasise more on workplace wellbeing as non-monetary
rewards while promoting entrepreneurial activities.
In light of methodology, matching approach is applied in this study as an approach to
update the sample. It is believed that by applying this approach, controlling for the
demographic differences among groups before multiple-group analysis can be more
86
accurate for researchers dealing with group differences via CFA. This method reduces
the complex multiple regression control process while measuring mean differences of
latent variables (most multiple-group analysis papers does not have the process of
controlling for demographic differences, but just constraining the measurement
invariance, however, measurement invariance constraint cannot control for those
group differences, especially those highly relate with the measured variable).
Therefore, this study wishes this method can shed some light on the research
approach for the further multi-group research.
The limitations of this study are similar to all the common studies, firstly, this study have
relied on self-reported measures of workplace wellbeing derived from answers to
subjective questions that may be perceived differently by people with different
background and personality(Kristensen and Johansson2008; Le´vy-Garboua and
Montmarquette 2004). Many of the patterns this study uncovered were intricate and
cannot be explained simply by broad response variables such as social desirability or
acquiescence. The measures by self-reported used in the Understanding Society survey
were undoubtedly less than optimal regarding reliability, owing to the need for brevity
and simplicity in a large survey of this type conducted in the UK. With better measures, it
is expected that the associations would have been stronger. Furthermore, the current
analysis does not allow isolating directions of causality. Although the sample is a notable
strength of our study, there are limitations as well to the methods used in the study.
Because our sample is cross-sectional, it cannot be certain of causal direction. People with
higher workplace wellbeing might be more likely, for example, to have a prosperous
motivation to increase financial income and own the psychological capability of
managing autonomy over the work. Thus, it is suspected that many of the associations this
study uncovered have bidirectional causality. In the future, by taking
87
the advantage of experimental, quasi-experimental, and longitudinal approach, a strong
case can be made that the needs this study involved do in fact cause increases in
workplace wellbeing.
