Is general self‐esteem a predictor of job
satisfaction?
Bachelor thesis of Demi Florence den Daas Student # 10350438 Finance and Organization Faculty Economics and Business University of Amsterdam The purpose of this paper is examining the relationship between job satisfaction and general self‐esteem. In a sample of over a six thousand working American citizens aged between 40 and 50 a strongly significant coefficient of ‐.038 was found. Ordered probit techniques were used to account for ordinarily of job satisfaction. The negative effect of general self‐esteem on job satisfaction does not confirm expectations but could be explained by the age distribution of the sample, role‐ orientation of individuals and negativity in general. Supervisor: dr. S. Dominguez Martinez February 20162
Table of Contents
1 Introduction ... 3 2 Conceptual framework ... 4 2.1 Definition and conceptualization ... 4 2.2 Core evaluations ... 6 3 Variable and measurement ... 10 3.1 Job satisfaction ... 10 3.2 Self‐esteem ... 10 3.3 Gender ... 12 3.4 Hours worked ... 12 3.5 Occupation ... 13 4 Hypothesis ... 14 5 Method ... 14 5.1 Data selection ... 14 6 Modeling ... 18 6.1 Model 1 ... 18 6.2 Model 2 and 3 ... 18 6.4 Model 4 ... 19 7 Results ... 19 7.1 Results model 1 ... 19 7.2 Results model 2 ... 21 7.4 Results model 4 ... 22 8 Conclusion ... 23 9 Discussion ... 24 References ... 26 APPENDIX ... 29
3
1 Introduction
The basis of relationship between self‐esteem and job satisfaction was found in early 90’ when Hapak (1935) defined a person’s needs to be satisfied physically, psychologically and socially to attain job satisfaction. Research in line with measurement of an individual self‐worth became popular and Rosenberg designed in 1965 his 10‐item scale to measure self‐esteem. During the seventies Korman (1970, 1976) published several papers focused on employee self‐esteem. The central idea was that confident workers outperform those without confidence. Implications of his findings were that self‐esteem, formed around work and organizational experiences could define employee’s motivation (as cited in Pierce and Gardner, 2003).
For a long time the relationship between job satisfaction and self‐esteem has been examined with as ultimate purpose to learn about improving workers productivity. Studies reveal that satisfied employees are more likely to have low absenteeism and turnover (Locke 1987, Carsten and Spector 1987, Tett and Meyer 1993), which is positively related to performance (Cavender 1984). That self‐esteem plays an important role in predicting employee's behavior and attitude was confirmed in the work of Brockner (1988) and Judge and Bono (2001). Fitts (1972) states low self‐esteem individuals, as compared to high and medium, perform less efficient under stress and failure, have lower aspirations and expectations of success, exhibit poorer social skills and less sociability (Rosenberg 1965). Later Pritchard found that self‐esteem was highly correlated with quantity of output (1980). Further relevance for examining the relation between self‐esteem and job satisfaction is found in the field of job selection, were it is the goal to hire workers who have a higher likelihood to perform the given job more successfully. As illustrated above, high self‐esteem employees have other qualities then low self‐ esteem individuals. This may result in performing a certain job better (which require for example more stress‐resistant personalities). Requiting employees with a desired high self‐esteem level could be useful1. 1 Testing implications of requiting on base of self‐esteem is beyond the scope of this paper but evidence can be found in Ellis and Taylor, Role of self‐esteem within job search process (1983)
4
During the nineties the focus was placed on the research for ‘organizational based self‐esteem’ (OBSE) ‐ a conceptualization of self‐esteem determined by individuals’ work‐ and organizational experiences (Pierce and Gardner, 2004). When examining this relation the individual is seen as an employee and asked to rate job satisfaction when only business environmental aspects are taken into account. In my research I will treat the individual not as employee but I will relate general self‐ esteem to job satisfaction, as scholars have argued that individuals may have different perception of their self‐worth when preforming different roles (Korman, 1970 and Rosenberg, 1995). The relationship between Organizational Based Self‐ esteem and job satisfaction could therefor suffer from in‐role bias and may result in overestimation. In this paper I will bring the focus back on the relation between general self‐esteem with job satisfaction instead of OBSE to control for this effect.
To summarize my research, I have developed the following research question: Does general self‐esteem influences the job satisfaction level of an individual? To answer this question I first place self‐esteem and job satisfaction in the perspective of core evaluations of the self, then I summarize relevant literature to explain the relationship and general implications of and between self‐esteem and job satisfaction. Second, hypothesizes are presented in which I expect this relationship to be positive. Data covering the subject were examined, a discussion of results follows and a conclusion is drawn.
2 Conceptual framework
2.1 Definition and conceptualization
Job satisfaction. The concept of Job satisfaction is most commonly defined by Locke, “a pleasurable or positive emotional state, resulting from the appraisal of one’s job or job experiences” (1976 p. 1304). Later a review of Public Personnel Administration satisfaction as an “affective or emotional response toward various facets of one’s job” (Kim, 2005, p. 246) or most recent as “how an individual feels about his or her job and various aspects of it usually in the sense of how favorable— how positive or negative—those feelings are” (Rainey, 2009, p. 298). The conditions under which an individual will perform its job can be comfortable, easy, difficult or
5
even dangerous. Facets that will determine if an employee will be satisfied with his or her job. These facets could be divided into different categories as external factors, those were an employee cannot change (climate) subjective factors as gender and self‐esteem and organizational related as working hours and occupation. In my research I include subjective and organizational related determinants.
Self‐esteem. Self‐esteem refers to an individual overall self‐evaluation of his/her competencies or the degree to which the individual sees him/herself as a competent, need‐satisfying being (Rosenberg, 1965). It is an attitude towards the self to which the individual beliefs to be capable, successful and worthy. Over time self‐esteem is divided into three mayor groups; global (a general evaluation), specific (to a situation or role; such as work) and task specific (competence in an activity) self‐esteem. Work role specific self‐esteem was shown most effective in relation with job satisfaction and this study turned into organizational based self‐esteem (OBSE) in the late twenties. OBSE differs from general conceptualizations of self‐ esteem because it refers to one’s belief about his or her self‐worth and competence as an organizational member. OBSE is expected to yield relatively stronger relationships with work‐related variables than general self‐esteem (Chen et al., 2004; Pierce et al., 1989; Tang & Ibrahim, 1998). Scholars have argued that individuals may have different perception of their self‐worth when preforming different roles (Korman, 1970 and Rosenberg, 1995) and therefor comparing OBSE and general self‐ esteem has totally different implementations. In a meta‐analysis by Bowling et all, a correlation of .49 between general self‐esteem and OBSE was found. OBSE was significantly associated with job satisfaction with a correlation coefficient of .57, compared to general self‐esteem with .29 (Bowling, 2010), and .26 establish in a meta‐analysis by Jugde and Bono(1997). Although higher correlation between job satisfaction and OBSE is expected, it is important to also understand the relation between general self‐esteem and Job satisfaction. It is stated by Brockner (1983) that low‐esteem individuals are more influenced by the environment than those with high self‐esteem. As OBSE is created in organizational‐environment, an individual who reports a low OBSE is therefor expected to report a lower job satisfaction score as well. But contrary to general self‐esteem this effect may not
6
appear (Brockner, 1983). For example, someone could report a low self‐esteem, because he values his self as a bad friend/husband/father (not formed around or within the organization) but is satisfied with his job so that he reports a high job satisfaction (Korman, 1970 and Rosenberg, 1995)
2.2 Core evaluations
Although Job satisfaction and self‐esteem are one of the most widely studied subjects in history, there exist a lack in linking both subject to each other. Job satisfaction and self‐esteem happen to be complex subjective phenomena’s (measure what people say or think, rather than what people do) and various concepts have been used describing it. Definitional issues in job satisfaction and measurement issues in job satisfaction and self‐esteem are the most important reasons for variation in papers. Before I explain why I will examine the specific relation, it is important to place the relation in the larger perspective of core self‐ evaluations.
The most recent, cited and relevant work about the relation between job satisfaction and self‐esteem is done by Jugde (1997), were he hypothesized that core evaluations would be direct and indirect related to job satisfaction and are the best predictors of job satisfaction. The core evaluations are fundamental, bottom‐line evaluations that people make of themselves (1997). Late twenties his work was renewing because self‐esteem was no longer seen as a single predictor of job satisfaction but as one of the four determinants (traits) of core‐evaluations, how people value their self. The four traits include self‐esteem, generalized self‐efficacy, locus of control and low neuroticism (emotional stability). With respect to job satisfaction Judge and Bono found in their meta‐analysis a true correlation of .26 for self‐esteem, .45 for self‐efficacy, .32 internal locus of control and .24 for Neuroticism (2001). Judge, Locke and Durham (1998) provide a confirmatory factor analysis of the effect of core evaluations on job satisfaction by showing significant results in three independent samples testing the four traits onto job satisfaction.
7
Figure 1 Core‐evaluations in correlation with job satisfaction
Not only qualitative reviews were established, various papers provide reasons to expect a positive relationship between job satisfaction and given traits. Starting with Kormans’ self‐consistency theory were a persons with high self‐esteem choose occupations in line with their interest (1970) and therefor more likely to be more satisfied, which was supported by Tharenou (Tharenou, 1979). Later Locke, McClear and Knight (1996) suggest that a person with a high self‐esteem is more likely to view a challenging job as a deserved opportunity which he can master and benefit from, whereas a person with low self‐esteem will view it as an undeserved opportunity or chance to fail (Locke, McClear and knight, 1996). And further Dodgeson and Wood noted that individuals with high self‐esteem maintain optimism in the face of failure, which makes further success (and thus future satisfaction) more likely (Dodgeson, 1998).
Hypothesis Ia: There will be a positive correlation between job satisfaction and general self‐esteem
This paper tries to establish if self‐esteem measured by the 10‐point scale of Rosenberg is a determinant of ones level of job satisfaction. Above described core evaluation traits show a factor‐loaded correlation between each of the core evaluation traits ranches from .55 to .85, which suggest they indicate a common concept (Judge, 1997). jobsatisfaction Selfesteem (0.26) Neuroticism (0.24) Self‐efficacy (0.45) locus of contol (0.24)
8
Notable is that self‐esteem is not as highly correlated with job satisfaction as for example generalized self‐efficacy (GSE), which refers to an individual's belief in his or her capacity to execute behaviors necessary to produce specific performance attainments (Bandura, 1977, 1986, 1997). Judge (1992) argued that GSE should affect job satisfaction because its association with practical success of a job. Gist and Mitchell add onto that high self‐efficacy individuals deal more effectively with difficulties and persist in face of failure (1992). Although the high correlation is found between self‐efficacy and job satisfaction, social cognitive theorists (e.g., Bandura, 1986, 1997; Mischel & Shoda, 1995; Stajkovic & Luthans, 1998) have argued that the utility of GSE for both theory and practice is low, because GSE is seen as a construct distinction from self‐esteem (Stanley and Murphy, 1997).
Second important finding of Bono and Judge (2001) is the correlation between lotus of control and job satisfaction is also higher than with self‐esteem. Lotus of control refers to the belief someone has, that he/she can control events affecting him, or that things happen because of luck or forces he/she cannot control (Rotter 1966). Because the correlation between lotus of control and job satisfaction is higher it might be more interesting examine this relationship. However due measurement problems of lotus of control, validation of the variable is hard to establish. Literature is not unambiguous about which scale would give least biased results in different field or research2 and if locus of control belongs even to the CSE framework (Bono and Jugde, 2003).
Although lotus of control and generalized self‐efficiency have higher correlations with job satisfaction, self‐esteem as predictor for job satisfaction variable obtains most valid support from literature (Tharenou, 1979). Questions about validity of the core evaluations have risen because they happen to be complex social attitudes and measurements are rather subjective. Most commonly used scales to measure individual core evaluation traits are: Rosenberg 10‐point Self‐ esteem scale (1965), 10‐point generalized Self‐Efficacy scale (Jerusalem and
2 Discussion of different scales and validations of Lotus of Control could be found in The
9
Schwarzer, 1992), Lotus of Control scale (Levenson, 1981) and Neuroticism scale (eysenck and Eysenck, 1968).
Most recent work about the positive correlation between self‐esteem and job satisfaction of Jugde, Heller and Klinger (2008) is presented in table 2.1 and 2.2. Correlations among the core evaluation traits are measured with the scales named above. Table 2.1 Correlation among measures of Core‐evaluations and Jobs satisfaction (Jugde, Heller, Klinger, 2008) Variable 1 2 3 4 5 1. Self‐esteem .88 2. Locus of Control .70** .70 3. Generalised self‐efficacy .80** .52** .87 4. Neurotism .85 .41** .65** .85 5. Job satisfaction .49** .20* .39** .35** .89 Notes: *P<.05 (two‐tailed); **P< 0.01 (two‐tailed) Table 2.2 Correlation of core evaluation traits with job satisfaction (Jugde, Heller, Klinger, 2008)
Variable Job satisfaction time 1 Job satisfaction 6 months later
Self esteem .55 (.49)** .39 (.35)**
Lotus of control .25 (.20)** .35 (.28)**
Generalised self‐efficacy .44 (.39)** .30 (.27)**
Neurotism .40 (.35)** .38 (.34)**
Notes: correlations are corrected for measurement error based on internal consistency reliabilities reported in table 1. Time 1 had a response rate of .81. Time 2 ,6 months measured after time 1, had response rate of .77% *P<.05 (two‐tailed); **P< .01 (two‐tailed)
From table 2.1 and 2.2 we see correlation between core evaluations and job satisfaction differ substantially and are significant (Judge, Heller, Klinger, 2008), but the two most consistent correlations with job satisfaction were self‐esteem and neurotism. When trying to predict job satisfaction due the core traits, self‐esteem
10 happens to be the most respectable and proved determinant in literature (Judge and Bono, 2001) Hypotheses Ib: Self‐esteem is a significant determinant of Job satisfaction
3 Variable and measurement
3.1 Job satisfaction Job satisfaction is the dependent variable and was measured with single‐item ‘I feel satisfied with my job’ and 4‐point measurement scale with a .67 Cornbach alpha. 1 classified totally dis‐agree; 2, disagree; 3, agree; 4, totally agree. Job satisfaction scores are ranked from 1 = low to 4 is high.3.2 Self‐esteem
Self‐esteem is measured by the 10‐point Rosenberg Self‐Esteem scale (RSE), which is still by far the most widely used scale for self‐esteem measurement (Blascovich & Tomaka, 1991; Gray‐Little, Williams, & Hancock, 1997). The RSE comprises 10 items to stress out a multidimensional and overall global evaluation of positive or negative attitudes towards the self. Items as ‘on the whole, I am satisfied with myself’ and ‘I feel I am a person of worth’ were rated on a 4‐point scale. Points are obtained as follow: totally dis‐agree = 0, disagree = 1, agree = 2, totally agree = 3. Respondents could score up to 30 points in total. The RSE makes use of reverse wording of items to get a more valid responds. This means that the items are negative presented ‘I do not feel I have anything to be proud of’ and points are obtained as follow: totally dis‐agree = 3, disagree = 2, agree = 1, totally agree = 0. The full list of items and explanation is presented in appendix 1.0 The RSE reached Cronbach alpha coefficients between 0.87 and 0.89.
What both self‐esteem and job satisfaction have in common is that they both try to define a feeling, which makes them a subjective variables, which leads to complexities due to their dependency on psychological states (Freeman, 1977). Job satisfaction and self‐esteem are a phenomenal construct, which means
11
measurement is comprised3 or indicated by various factors. Because the variables are both latent and cognitive, a broad definition of both concepts in literature is observed. Use of different casual measurement techniques are the reason, external validation is hard to establish which results in inconsistency of observed correlation (positive or negative) between job satisfaction and self‐esteem. Figure 2 Job satisfaction (or self‐esteem) as formative variables
Measurement of overall attitude toward ones job could cause a positive casual relation between specific satisfactions (or measurement criteria) when measured by a multiple model. For example when In figure 3.1 there is a high positive correlation observed between X1 and X2, job satisfaction would suffer from multicollinearity, which creates internal validation problems. Therefor in this research, job satisfaction is measured by the single item model. Wanous, Reichers and Hudy found a reliably for the single‐item model of .67. A single‐item measure eliminates item redundancy and therefore reduces the fatigue, frustration, and boredom associated with answering highly similar question (Wanous, Reichers and Hudy, 1997). Although .67 is a respectable level of reliability, there should be taken into account that a multiple model yields higher reliability scores, but for simplicity I make use of the one item model.
Second, the existence optimistic and pessimistic individuals could lead to biased results in the measurement of self‐esteem (or job satisfaction). Although Optimism and pessimism are highly correlated with the core‐evaluations, self‐ esteem suffers less from this. The reason is that most people are optimistic or pessimistic regards a specific area of life (e.g. optimistic in relationship, but pessimistic in work‐related issues) (Hecht, 2013). Hoge and McCarthy found that a 3 Treatment of a concept as a manifest/formative variable were specific items cause the concept
Job
satisfaction
X2 X1 X3
12
measure that combined nine single‐item self‐concept ratings (e.g., ‘‘I am smart,’’ ‘‘I am good‐looking’’) weighted by the importance attributed to these self‐concepts (e.g., ‘‘How important is it to you to be very smart?’’) correlated less with self‐ esteem than the other core‐evaluations (Hoge and McCarthy, 1984). Such findings suggest limitations exist on the measurement of the core‐evaluation, but these limitations do not relate to general self‐esteem derivations.
3.3 Gender
Resnick, Fauble and Osipow (1970) expressed similarities when sexes were tested on self‐esteem and asked about career perspectives. High self‐esteem man and woman expressed more certainty about career chooses than low‐self‐esteem individuals, but there was no significant difference between sexes in self‐esteem levels (Resnick, Fauble and Osipow, 1970). In a more resent study of Clark, woman reported a higher job satisfaction score than man and therefore I expect a higher job satisfaction score among woman(Clark, 1997). Sex is recorded as dummy variable 0 = Male; 1 = Female. Hypothesis IIa: Woman report a higher job satisfaction score than men Hypothesis IIb: There is no difference in self‐esteem between men and woman 3.4 Hours worked
A negative relation between job satisfaction and hours worked would be expected from meta‐analysis preformed NG and Feldman (2008), were Bronschein confirmed a low job satisfaction was observed were long working hours were required (Bronschein, 2006). Others papers suggest the opposite and pointed out that income may mediate this relationship as higher income and longer hours are also correlated (Cabrita and Perista, 2006).
Hypothesis IIIa: There is no relation between hours worked and job
satisfaction
The distribution of working hours of the cohort is presented in table 4.3.1 were we can see up to 21.79% of respondents worked up to 39 hours a week, 46.01% worked 40 hours a week and 32.20% worked 40 hours or more.
13
Table 4.3.1 Tabulation of hours worked
Hrs Freq. worked Percent Cum.
1 3 0.05 0.05 10 28 0.45 2.10 20 130 2.08 6.30 30 201 3.22 13.06 39 16 0.26 21.79 40 2,872 46.01 67.80 49 13 0.21 80.39 50 504 8.07 88.47 60 210 3.36 96.07 70 67 1.07 98.45 125 1 0.02 100.00 Total 6,242 100.00 3.5 Occupation Scholars have argued that individuals may have different perception of their self‐worth when preforming different roles, so that persons may value their‐self high and competent as an employee but less valued in being a good friend (Korman, 1970, Rosenberg 1995). Because of this reason it would be valuable to control for occupation in an extra model. Johnson at all (2005) states, job satisfaction differs across occupational level and therefor I specify main occupation that are presented in my sample. Job satisfaction appears to be the lowest in occupations were high workload is combined with low decision latitude (Landsbergis, 1988) as health care occupations.
Hypothesis IVa: There is no difference in job satisfaction scores between occupations
14
4 Hypothesis
In table 4.1 a brief summary is presented with the hypotheses that I will test, the models fitting to the hypothesis and which results belong to that. Table 4.1 summary of hypotheses testedHypothesis Models Results
Ia There will be a positive correlation between job satisfaction and general self‐esteem 1, 2, 3 and 4 7.1.1 rejected Ib Self‐esteem is a significant determinant of Job satisfaction 1 7.1.2 accepted IIa Woman report a higher job satisfaction score than men 1 7.1.3 accepted IIb: There is no difference in self‐esteem between men and woman 4 7.4.1 IIIa There is no relation between hours worked and job satisfaction 1, 2, 3 and 4 7.2.1 accepted IVa There is a difference in job satisfaction scores between occupations 4 Appendix 3.0 accepted
5 Method
5.1 Data selectionThe data I use for my research are obtained from NLSY79 database, years 1972‐2012. The NLSY79 Cohort is a longitudinal project that follows the lives of a sample of American youth born between 1957‐1964 aged between 14 and 22 years old. For this paper all employed respondents in the year 2006 were included so respondents were aged 41 to 50 at this time. I excluded respondents that did not rate their job satisfaction. I also excluded respondents that did not notify their self‐ esteem score (5% of the sample).
Summary statistics of the total cohort are displayed in table 5.1. Respondents are all American citizens and reported a total net family income above poverty status
15
in 2005 which is 9,570$ for a one‐person household. For every additional family member + 3,260$ per year and the poverty line for a four‐person family is 19,350$ a year. In the sample there were a few non‐responses in hours worked per week (24) and occupational status (146). In table 5.2 outliers were deleted on the base of highly unlikely to be true; values for worked hours 0, 140 and 168 were deleted. Also all un‐codeble (12) observations for occupational status were deleted because these reported all 1 for job satisfaction, which is highly unlikely in the sample. When comparing means of all variables of table 5.1 and 5.2 no significant difference was found.
Table 5.1: Sample 1: Summery statistics of total cohort excluding non‐responds job satisfaction and self‐esteem.
Variable Observations Mean Std. Dev. Min Max
Gender 6439 0.489 .499 0 = M 1 = F
Self‐esteem 6439 23.814 4.301 5 30
Job satisfaction 6439 1.597 .694 1 4
Table 5.2 Sample 3: summary statistics of clean data set
Variable Observations Mean Std. Dev. Min Max
Gender 6242 .509 .500 0 = M 1 = F Self‐esteem 6242 23.820 4.289 5 30 Job satisfaction 6242 1.601 .696 1 4 Hours worked 6242 41.176 11.917 0 125 In my research I will use four different models and preform two different tests. The first test will be a ordered probit regression on model 1 with job satisfaction as dependent ordinal variable, the marginal effects will be tested and presented. The original sample as in table 5.1 is presented was used with hours worked as normally distributed variable. The remaining 3 models are presented as alternative models were model 2 and 3 will be tested as well with an ordered probit regression and job satisfaction as
16 dependent ordinal variable. Marginal effects are not presented, therefore interpretation is different than model 1, but I will focus on the sign of the coefficients. In model 2 and 3 hours worked are specified in groups and occupation is added to the regression. In sections 3.4 and 3.5 the relevance of controlling for working hours and occupational status is clarified. From the original sample some occupations have been deleted because less than 50 respondents were present in the group and for the others more than a hundred. I controlled if these deleted occupations had higher correlations with job satisfaction or self‐esteem than other occupations, but no significant differences were presented. Summary statistics of this sample are presented in table 5.3 and distribution of occupations are presented in appendix 2.0 Table 5.3 Summary statistics of sample without underrepresented occupations
Variable Obs Mean Std. Dev. Min Max
Jobsat 6055 1.601 .696 1 4 Self‐esteem 6055 23.812 4.28 5 30 Gender 6055 .509 .499 0=M 1=F Hrs40 6055 .460 .498 0 1 Hrshigh 6055 .320 .466 0 1 The second type of test is an ordinary probit with job satisfaction score 1 and 2 where I test model 4. I choose to make this distinction in models because the distribution of satisfied rates in the sample is skewed. Table 5.1 and 5.22 show the tabulation of dependent variable job satisfaction, where in table 5.2 the occupations that were under represented in the sample are deleted. As seen from these tables there exist almost no difference between the groups (only rounding difference) when the underrepresented occupations are deleted from the sample without making use of sample selecting. From these tables we see 92.58 (92.57) % of the respondents are (very) dissatisfied with their job. Therefor I will also run a normal probit regression with the new sample (left table) and job satisfaction score 1 as base level and job satisfaction score 2 as dummy 1. Job satisfaction scores 3 and 4
17 are deleted from the sample and in table 5.3 the new distribution of the sample is presented and used to test model 4. To test the models STATA13 is used. Table 5.4 Weighted distribution sample Job satisfaction sample 1
Jobsat Freq. Percent Cum.
1 3,105 49.74 49.74 2 2,674 42.82 92.58 3 312 5.00 97.58 151 2.42 100.00 Total 6,242 100.00 Table 5.5 : Weighted distribution sample job satisfaction reduced sample Table 5.3 Weighed distribution model 4.
Jobsat Freq. Percent Cum. 0 3,012 53.74 53.74
1 2,593 46.26 100.00
Total 5,605 100.00
Jobsat Freq. Percent Cum.
1 3,012 49.74 49.74
2 2,593 42.82 92.57
3 303 5.00 97.57
4 147 2.43 100.00
18
6 Modeling
6.1 Model 1 When testing the first model, job satisfaction is used as ordinal dependent variable I use the econometric model of job satisfaction of Blanchflower and Oswald (2004). (1) (1) Were Job satisfaction (JS) is assumed to be affected by level of self‐esteem reported, gender and hours worked. Gender is presented as dummy were 1 = female. Both hours worked and self‐esteem are normally distributed variables. Because job satisfaction is an ordinal variable measured on a 4‐point scale I run a ordered probit regression. 1 φ λ B BX φ λWere J has values 1 to 4, is defined as 1 when and φ is cumulative normal distributed. 6.2 Model 2 and 3 For model 2 and 3 the same model as model 1 is used only hours worked is divided into three different groups and occupations are added. Self‐esteem is a normally distributed variable and gender is presented as dummy were 1 = female. Hours worked are presented as dummy’s and with base level 40 or less, were hrs40 is exactly 40 hours weekly worked and Hrshigh represents more than 40 hours worked per week. 19 different occupations were tested for significance. The occupations are presented as dummy’s and added in model 3 (2)
19 6.4 Model 4 In Binary model 4 the dependent variable job satisfaction can only take value 0 (when job satisfaction =1) or 1 (when job satisfaction = 2). The interpretation is, when the coefficients are positive and presented, this leads to a higher job satisfaction score (2 instead of 1) ∗ ∗ ∗ (4) Were job satisfaction (JS) is assumed to be affected by level of self‐esteem reported (normally distributed), gender and hours worked and given determines in interaction with gender. Gender and hours worked are measured and can be interpreted the same way as in model 2 and model 3. To control for the effect suggested in section 3.3 (females are more satisfied with their job) interaction terms are included, to see where this effect comes from. Further justification behind including these interaction terms in the regression can be found in section 3.2 and 3.3
7 Results
7.1 Results model 1 To test Hypothesis Ia: ‘There will be a positive correlation between job satisfaction and general self‐esteem’, I first calculated correlation between all variables presented in the model. From table 7.1.1 we can see there is a negative correlation between job satisfaction and self‐esteem and therefor hypothesis Ia is rejected. This is confirmed in table 7.1.3 were the likelihood for job satisfaction = 1 is increases with a higher self‐esteem rate reported, while the likelihood to have job satisfaction =2, 3 or 4 is decreasing when self‐esteem increases.
20
Table 7.1.1 correlations between explanatory factors
Jobsat Self‐esteem Hours worked Gender Job satisfaction 1.0000 Self esteem ‐.139*** 1.000 Hours worked ‐.029** .084 1.000 Gender .021*** .025* 0.263** 1.000 Notes: *,**, *** means the coefficient is statically different from zero at 10, 5 and 1 percent two sided To test hypothesis Ib: Self‐esteem is a significant determinant of Job satisfaction’ I ran an ordered probit regression with model 1. Table 7.1.2 shows the coefficient of self‐esteem is significant at a = .01. If the employee reports one point higher on self‐esteem, the likelihood he or she will report a higher job satisfaction score will decrease with .0385177. Following this interpretation hypothesis Ib is accepted. Self‐esteem has a significant (negative) effect on job satisfaction. Table 7.1.2 Ordered probit estimation of job satisfaction
Variable Coef S.E.
Self‐esteem ‐.039*** .004 Hours worked ‐.003** .001 Gender .081*** .031 Notes: Dependent variable job satisfaction is a discrete variable based on self‐declared satisfaction rates (1; very dissatisfied onto 4; very satisfied). The model is estimated with order probit. Standard errors are heteroscedastic (robust).**, *** means the coefficient is statically different from zero at 5 and 1 percent two sided Because interpretation of normal probit results should be done carefully, in table 5 the marginal effects the determinants of job satisfaction scores are reported separately. From here we can see that when 1 point higher on self‐esteem is reported, the likelihood for reporting a job satisfaction score of 1 increases with .0151377. From here we assume a higher self‐esteem score results in a lower job
21 satisfaction score. The likelihood of reporting a higher job satisfaction score decreases when individuals of the separate groups report a higher self‐esteem score. The effect is significant for all job scores and also supports hypothesis Ib. Table 7.1.3 Marginal effect of ordered probit model
dy/dx Job sat = 1 Job sat = 2 Job sat = 3 Job sat = 4
Self esteem .015*** ‐.010*** ‐.003*** ‐.002*** Hours worked .001** ‐.001** ‐.002** ‐.000** Gender ‐.032*** .021*** .007*** .005*** Notes: Standard errors are heteroscedastic (robust).**, *** means the coefficient is statically different from zero at 5 and 1 percent, tested two sided. What we can see as well from table 7.1.3 is that when the respondent is a female the likelihood of reporting a high job satisfaction score is higher. Hypothesis IIa ‘woman report a higher job satisfaction score than men’ is accepted. 7.2 Results model 2 In table 7.2.1 summary results of models 1, 2 and 3 are presented where in appendix 3.0 all occupations are displayed. From table 7.2.1 we see that there is a difference in the effect of hours worked if we compare model 1 with model 2. Were in model 1 Hypothesis IIIa ‘a longer workweek leads to lower job satisfaction’ is rejected, is hypothesis IIIa in model 2 accepted and significant. Interpretation of hrshigh is if someone reports to work 40 hours or more, the coefficient is negative and suggests a lower job satisfaction score.
22 Table 7.2.1 summary of models presented Models (1) (2) (3) Self‐esteem ‐.038*** ‐.038*** ‐.035*** Gender ‐.003** .075** .009 Hours worked .0811*** Hrs40 .073** .015 Hrshigh ‐.052** ‐.064
Occupation NO NO YES
Notes: the full summary model is presented in appendix 3.0. **, *** means the coefficient is statically different from zero at 5 and 1 percent, tested two sided.
To test hypotheses IVa: ‘Occupational status does influence job satisfaction rates’, model 3 is used (the full version is presented in appendix 3.0). From alternative model 4, hypothesis IVa is accepted because certain (see appendix) occupations are significant influencers of job satisfaction score. Education, Training and Library occupations have reported to be least satisfied with their job. A significant coefficient of ‐.710 was observed. ‘Personal care and Service’ and ‘health care’ occupations flowed up with coefficients of ‐.595 and ‐.315 respectively. These findings confirm expectations from literature were is suggested service and health care occupations are associated with lower job satisfaction (Landsbergis, 1988) 7.4 Results model 4 In alternative model 4 I test the effect of gender, were I include interaction terms between given variables and gender. From this model we can see self‐esteem does not differ significantly across gender and therefor I accept hypothesis IIb ‘there is no difference in self‐esteem level between man and woman’. From the same table we see gender does play a significant role on hours worked. Female workers report a higher job satisfaction score when more than 40 hours worked.
23
Table 7.4.1 results model 4
Jobsat Coef. Std. Err.
Self‐esteem ‐.035*** .006 Gender .181 .201 Gender*Self‐esteem ‐.004 .008 Hrs40*Gender ‐.030 .095 Hrshigh*Gender .190* .104 Hrs40 .099* .055 Hrshigh ‐.220*** .068 Constant .696*** .137 Notes: Standard errors are heteroscedastic (robust).*, *** means the coefficient is statically different from zero at 10 and 1 percent, tested two sided.
8 Conclusion
Job satisfaction is a phenomenon that is hard to measure and depends on external, subjective and organizational based factors. In this paper I tried establish if self‐esteem has an effect on the level of an individual’s job satisfaction. My research confirmed expectations; self‐esteem has a significant effect on job satisfaction with an alpha of .001. Although the effect is confirmed, I observed a negative correlation and a negative estimation coefficient, which I did not expect at first. The effect of general self‐esteem on job satisfaction appears to be negative with an estimation coefficient of ‐.038. The negatively correlated between self‐esteem and job satisfaction does not confirm expectations but could have several reasons.
1) Individuals may have different perception of their self‐esteem when preforming different roles (korman, 1970 and Rosenberg, 1995). An individual could value their self for example good friend/husband/father and therefor report a high self esteem, but is not satisfied with his job so that he reports a low job satisfaction score. In this case a negative correlation is expected.
2) The respondents in the sample I used are more negative‐minded people. This conclusion is drawn because 92.5% of the respondents notified they (totally) disagreed with the sentence ’On a hole, I am satisfied with my job’.
24
3) The age of the cohort. Clark, Oswald and Warr (2011) provided evidence that the relation between age and job satisfaction is U‐shaped, which means individuals within the age of 30 to 50 are in general least satisfied with their job. In my regression I did not control for age because all respondents were between 41 and 50 years old. The on average low job satisfaction score could therefore be explained by the age of the cohort.
The gender/job satisfaction paradox of Kaiser (2002) or the paradox of the contented female worker of Bender et al (2005) is confirmed in my research. The paradoxes refer to the fact that women report higher job satisfaction than men despite a clearly disadvantaged position in the labor market in terms of earnings, recruitment/dismissals, promotions and career prospects. The effect is even higher when woman work more than 40 hours a week. Men who worked more than 40 hours report a lower job satisfaction. In this research there was no significant difference found in self‐esteem level between men and woman.
As expected, certain occupations do influence job satisfaction significant. The influences were negative, which was inline with literature.
9 Discussion
Interpretation of the results of model 2, 3 and 4 should be done carefully because margins are not taken into account. When we would control with margins, the effect of coefficient could be interpreted more precise. Also it should be taken into account some people are generally more negatively minded than others (Jugde, 1997) and probable the sample suffers from this too, but I could not control for this fact. For further research it would be relevant to examine the same cohort but in a different year. If the job satisfaction score differs over time, the U‐shaped age effect could be confirmed. As self‐esteem follows different patterns in relation with age it would be interesting control for this effect. Including extra explanatory variables, as income and job characteristics could be relevant in further investigation.
Job satisfaction may seem to be an easily understood subject but due the diversity of conceptualizations and measurement techniques a wide range of papers is presented on the subject. What could be learned form this paper is that the relation between general self‐esteem and job satisfaction is a bit outdated and
25
turned into Organizational based‐self‐esteem. Therefor I suggest further investigation should be related to this topic.
26
References
Abraham, R. (1999). The relationship between differential inequity, job satisfaction, intention to turn over, and self‐esteem. The Journal of Psychology, 133 (2), 205‐211.
Alavi, H. R. and Askaripur, M. R. (2003). The relationship between self‐esteem and job satisfaction of personnel in government organizations. Public Personnel Management, 32(4), 591‐598.
Bandura, A. (1977). Self‐efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191‐215.
Baumeister, R. F., Campbell, J. D., Krueger, J. I., and Vohs, K. D. (2003). Does high self‐esteem cause better performance, interpersonal success, happiness, or healthier lifestyles? Psychological Science in the Public Interest, 4, 1‐44.
Bowling, N.A. K.J Eschleman, Q. Wang, C. Kirkendal and G. Alarcon. (2010) A meta‐analysis of the predictors and consequences of organizational based self‐ esteem. Journal of Occupational and Organizational Psychology. 83, 601–626.
Brutus, S. M.N. Ruderman, P.J. Ohlott, C.D. (2000). McCauley Developing from job experiences: The role of organization‐based self‐esteem Human Resource Development Quarterly, 11 (4), 367–380.
Cabrita, J. and Perista, H., (2006) Measuring job satisfaction in surveys ‐ Comparative analytical report. EurWORK
Casper, E. S. and Fishbein, S. (2002). Job satisfaction and job success as moderators of the self‐esteem of people with mental illnesses. Psychiatric Rehabilitation Journal, 26(1), 33‐43.
Chen, G., Gully S.M. and Eden, D., (2001). Validation of a New General Self‐ Efficacy Scale Organizational Research Methods, 4, 62‐83Clark, A, (1997), Job satisfaction and gender: Why are women so happy at work? Labour economics, 4 (4), 341‐372.
Freeman, R. B. (1978) Job satisfaction as an economic variable, American Economic Review, 68(2), 135‐141
27
Fetai, B, Abduli B. Qirici, S. (2015) An Ordered Probit Model of Job
Satisfaction in the Former Yugoslav Republic of Macedonia. Procedia Economics and Finance. 33, 350–357.
Gray‐Little, B., Williams, V.S.L., & Hancock, T. D. (1997). An item response theory analysis of the Rosenberg Self‐Esteem Scale. Personality and Social Psychology Bulletin, 23, 443‐451. Halpert, R. and Hill, R. (2011). The Locus of control construct’s various means of measurement: A researcher’s guide to some of the more commonly used Locus of Control scales. Will to press beach haven, 8‐21 Hecht, D. (2013). The Neural Basis of Optimism and Pessimism. Experimental Neurobiology, 22(3), 173–199. Influence of individual and group identity salience in the global self‐esteem of youth. Hoge, D. R., McCarthy, J. D. (1984) Influence of individual and group identity salience in the global self‐esteem of youth. Journal of Personality and Social Psychology, 47(2), 403‐414.
Johnson, S., C. Cooper, S. Cartwright, I. Donald, P. Taylor, C. Millet, (2005). The experience of work‐related stress across occupations. Journal of Managerial Psychology, 20 (2), 178 – 187.
Landsbergis, P. A. (1988). Occupational stress among health care workers: A test of the job demands‐control model. Journal of organizational behavior, 9(3), 217‐ 239.
Maslow ,A. H. (1943). A theory of human motivation. Psychological Review, Vol 50(4), 370‐396.
Ng, T. W. H. and Feldman, D. C., (2008) Long work hours: a social identity perspective on meta‐analysis data. Journal of Organizational Behavior. 29 (7) 853– 880. Pierce, J. L., (2003) Self‐Esteem Within the Work and Organizational Context: A Review of the Organization‐Based Self‐Esteem Literature. Journal of Management 5, 591‐622. Rosenberg, M. (1965). Society and the adolescent self‐image. Princeton, NJ: Princeton University Press
28
Rotter, J. (1966). Generalized expectancies for internal versus external control of reinforcement, Psychological Monographs, 80 (1), 609.
Rosta, J., Gerber, A., (2008 ) Gesundheitswesen, Job satisfaction of hospital doctors. 70(8‐9), 519‐24
Resnick, H.; Fauble, M. L.; Osipow, S. H. (1970). Vocational crystallization and self‐esteem in college students. Journal of Counseling Psychology, 17(5), 465‐467.
Stanley, Keith D., and Michael R. Murphy. (1997). A comparison of general self‐efficacy with self‐esteem. Genetic, Social, and General Psychology Monographs 123.1 (79)
Tharenou, P., (1979). Employee Self‐Esteem: A Review of the Literature. Journal of Vocational Behavior, 15(3), 331‐46.
Wallston, K.A., Kaplan, G. D. and Maides S. A. (1976). Development and validation of the health Locus of Control scale. Journal of Consulting and Clinical Psychology. 44 (4), 580‐585. Wanous JP1, Reichers AE, Hudy MJ. (1997). Overall job satisfaction: how good are single‐item measures? Journal of Applied Psychology. 82(2), 247‐52. .
29
APPENDIX
Appendix 1.0: Rosenberg Self‐Esteem Scale items and scoring Item 1 I am a person of worth. 2 I have a number of good qualities. 3* I am inclined to feel that I am a failure. 4 I am able to do things as well as most other people. 5* I felt I do not have much to be proud of. 6 I take a positive attitude toward myself. 7 I am satisfied with myself. 8* I wish I could have more respect for myself. 9* I certainly feel useless at times. 10* At times I think I am no good at all. Scoring for items 3, 5, 8, 9, 10: Strongly agree=3 Agree=2 Disagree=1 Strongly disagree=0 Scoring for items 1, 2, 4, 6, 7 is reversed so that a higher score indicates higher self‐esteem. Scores of 10 items were summed. Total score could range from 0 to 30 points.
30
Appendix 2.0 Occupation distribution
Occupation Freq. Percent Cum.
Executive, Administrative and Managerial Occupations 718 11.86 11.86 Management Related Occupations 256 4.23 16.09 Mathematical and Computer Scientists 143 2.36 18.45 Engineers, Architects, Surveyors, Related Technicians 108 1.78 20.23 Life, Physical and Social Science Technicians 114 1.88 22.11 Teachers 264 4.36 26.47 Education, Training and Library Workers 137 2.26 28.74 Health Diagnosing and Treating Practitioners 159 2.63 31.36 Healt Care Technical and Support Occupations 281 4.64 36.00 Protective Service Occupations 167 2.76 38.76 Food Preparation and Serving Related Occupations 230 3.80 42.56 Cleaning and Building Service Occupations 287 4.74 47.30 Personal Care and Service Workers 179 2.96 50.26 Sales and Related Workers 452 7.46 57.72 Office and Administrative Support Workers 946 15.62 73.34 Construction Trade and Extraction workers 410 6.77 80.12 Installation Maintenance and Repairs Workers 409 6.75 86.87 Setters, Operators and Tenders 349 5.76 92.63 Transportation and Material moving workers 446 7.37 100.00 Total 6,055 100.00
31 Appendix 3.0 Full presentation of model 1, 2 and 3 including occupations models (1) (2) (3) Self‐esteem ‐.038*** ‐.038*** ‐.035*** Gender ‐.003** .075** .009 Hours worked .0811*** Hrs40 .073** .015 Hrshigh ‐.052** ‐.064 Executive, Administrative and Managerial ‐.303*** Management Related ‐.188** Mathematical and Computer Scientists ‐.229** Engineers, Architects, Surveyors, Related Technicians ‐.092 Life, Physical and Social Science Technicians ‐.228* Teachers ‐.397*** Education, Training and Library ‐.710*** Health Diagnosing and Treating Practitioners ‐.290*** Health Care Technical and Support ‐.317*** Protective Service ‐.034 Food Preparation and Serving Related ‐.095 Cleaning and Building Service ‐.121 Personal Care and Service ‐.595*** Sales and Related ‐.104 Office and Administrative Support ‐.058 Construction Trade and Extraction ‐.232*** Installation Maintenance and Repairs .036 Setters, Operators and Tenders .010
