The influence of promotions on job satisfaction and the difference between men and women
Master Thesis: MSc Human Resource Management University of Groningen
Faculty of Economics and Business
Date: 17.02.2019
Marie Juliane Weinke S3502058
Vijfhoek 141, 8011NZ Zwolle m.j.weinke@student.rug.nl
Supervisors:
Dr. Peter van der Meer Dr. Tim Vriend
The influence of promotions on job satisfaction and the difference between men and women
Abstract
This thesis researches the difference of job satisfaction between men and women after promotion. According to literature, promotion leads to an increase in job satisfaction. However, it might affect women and men differently. Although women are just as likely to receive a promotion as men, they still perceive lower chances to be promoted. This suggests that if they do get promoted, women ́s job satisfaction increases more intensely than men´s. However, women experience smaller salary gains after promotion. That entails a feeling of unfair treatment and would lead to even lower job satisfaction. To measure this gender difference in the increase of job satisfaction, a difference in difference analysis is done on the data of the German Socio- Economic Panel study (GSOEP). The findings prove that women experience a smaller increase of job satisfaction after promotion than men do, both after internal and external promotion.
Keywords
Promotion; Job satisfaction; Gender
TABLE OF CONTENT
1. INTRODUCTION... 3
2. THEORY ... 5
2.1PROMOTION ... 5
2.2JOB SATISFACTION ... 7
2.3PROMOTION AND JOB SATISFACTION ... 8
2.4GENDER DIFFERENCES... 8
3. DATA & METHODS ... 12
3.1DATA SET AND MAIN VARIABLES ... 12
3.2METHOD ... 15
4. RESULTS ... 18
4.1DESCRIPTIVE STATISTICS... 18
4.2DIFFERENCES BETWEEN GENDERS ... 18
4.3RELATION BETWEEN VARIABLES ... 19
4.4RELATIONSHIP BETWEEN JOB SATISFACTION AND JOB CHANGE/(INTERNAL/EXTERNAL) PROMOTION ... 21
4.4.1 Regression analysis (1 to 3) ... 21
4.4.2 Difference in difference analysis (1A to 3A) ... 24
5. DISCUSSION ... 27
5.1MAIN FINDINGS ... 27
5.3THEORETICAL AND PRACTICAL IMPLICATIONS ... 30
5.4STRENGTHS, LIMITATIONS AND FURTHER RESEARCH ... 31
6. CONCLUSION ... 33
REFERENCES ... 34
APPENDIX ... 38
LIST OF TABLES TABLE 1:DESCRIPTIVE STATISTICS ... 18
TABLE 2:GENDER-DIFFERENCES OF SELECTED VARIABLES ... 19
TABLE 3:CORRELATIONS OF MODEL 1 ... 20
TABLE 4:CORRELATIONS OF MODEL 2 ... 20
TABLE 5:CORRELATIONS OF MODEL 3 ... 21
TABLE 6:EFFECTS OF CONTROL VARIABLES ON THE RELATIONSHIP BETWEEN JOB SATISFACTION AND THE INDEPENDENT VARIABLES ... 22
TABLE 7:COEFFICIENTS OF THE DIFFERENCE IN DIFFERENCE ANALYSIS ON THE RELATIONSHIP BETWEEN JOB SATISFACTION AND THE INDEPENDENT VARIABLES ... 25
1. INTRODUCTION
Almost every employee has been or is expecting to be promoted at some point in his or her career. It is an aspired movement in everyone’s working path because people naturally like to be rewarded. They need a compensation from time to time to stay motivated and feel valued.
Promotions are seen as a reward for employees. Thus, they influence one’s satisfaction with the job and work surroundings. Therefore, a received promotion has a positive effect on job satisfaction as it aligns with the expectations of an employee. Employees are more satisfied with the job they are doing after they have been promoted. Also expected promotions in the future motivate and satisfy as the chance of advancement becomes more realistic for employees. On the other hand, if a promotion was expected but did not occur, the satisfaction level with one’s job is harmed and thus decreases. Hence, received promotion is positively affecting job satisfaction.
(Francesconi, 2001; Kosteas, 2011; Lindquist, 2010; Pergamit & Veum, 1999)
The positive influence of promotion on satisfaction might, however, differ between men and women. Gender plays an important role in this relation as male and female employees do not face promotion the same way. Firstly, perceived promotion possibilities vary as women on average feel like they have less chance to be promoted. Women´s expectations are lower, whereas men expect to be promoted more often (Wynn, 2017; Kosteas, 2011). Secondly, women however are just as likely to be promoted as men when the situational constraints align (Booth, Francesconi & Frank, 2003; Kosteas, 2011; Wynn, 2017). And thirdly, women experience smaller salary increase after promotion than men (Booth, Francesconi & Frank, 2003;
Francesconi, 2001).
Although promotion is generally satisfying for both genders (Kosteas, 2011), the different expectations and experiences of men and women might influence how satisfying a promotion actually is. Women could have a larger increase in job satisfaction because they experience lower perceived promotional possibilities (Wynn, 2017; Kosteas, 2011) and actually receiving a not expected promotion could boost their satisfaction even more. Whereas men did expect to be promoted, thus might not be as overwhelmed. On the other hand, women could have a lower increase in job satisfaction because they gain lower salary increases than men after being promoted (Booth, Francesconi & Frank, 2003; Francesconi, 2001). This could nurture the
impression of unfair treatment resulting in lower job satisfaction (Witt and Nye, 1992). Men are receiving a larger salary increase so their enhanced job satisfaction after promotion is not harmed by feeling treated unfairly.
To my knowledge, there has not been a research until now which looks into gender as a moderation of the positive effect of promotion on job satisfaction. We do not know yet if the job satisfaction of women is differently affected by promotion than men´s job satisfaction. Thus, this thesis will research the questions: Does the relation between promotion and job satisfaction differ for men and women? Is the job satisfaction of women higher or lower, compared to men, after having received a promotion?
2. THEORY
The theory section provides a theoretical framework for this research and discusses the concepts used. Firstly, the independent variable promotion will be covered as I am studying the influence of promotions on job satisfaction. Secondly, the concept of job satisfaction is introduced. Thirdly, the relationship of promotion and job satisfaction is covered with regard to gender differences. It ends with elaborating multiple hypotheses.
2.1 Promotion
Promotions are profitable for employees and for employers. Both parties gain individual advantages when receiving or granting promotions. Promotions are a way of rewarding and motivating. Employers are using promotions not only because of the increase in employees´ job satisfaction but to provoke more effort and a raise in productivity as a result. It is an incentive for most employees because the top of the job ladder is not yet reached in many occasions, so employees strive for mobility that improves their situation. Promotion is a very realistic and actual option to move upwards (De Souza, 2002; Kosteas, 2011; Pergamit & Veum, 1999).
Generally, a promotion is a vertical career advancement. All other lateral position changes are summarized as development opportunities. This counts for both internal and external position changes. Promotions place the employee in a new position that is generally associated with an increase in remuneration and other positive extras. The job title will change according to the rank enhancement, and for instance, the employee will receive additional benefits like a company car or a new office. As an improvement in the career of an employee, promotions into higher functions often also come with non-materialistic changes such as extensions of job characteristics. For instance, growing autonomy, responsibility and assigned supervision of other employees will be added to the field of duties. Additionally, the employee moves up ranks that internally places them higher in the hierarchy and contributes to their personal career. Thus, employees might value and strive for promotions for various reasons. Non-monetary rewards are just as an important part of the reward package as monetary rewards are (Cobb-Clark & Dunlop, 1999; Francesconi, 2001; Kosteas, 2011; Pergamit & Veum, 1999; Rosenbaum, 1984).
Hence, promotions are seen as upward movements within the internal labour market at the same employer as well as career advancements on the external labour market by changing
employers. Internal promotions are often preferred because of the better-established fit between employee and organization. Existing employees have great firm-specific human capital that lets both employee and employer favour the ongoing relationship over other possibilities. Especially, the replacement of tenure employees is expensive for the employer. Thus, the expected performance of the external candidate has to surpass the internal candidate’s performance to such an extent that the expenses like recruitment and training costs pay off in the long run. (Chan, 1996; DeOrtentiis, Van Iddekinge, Ployhart & Heetderks, 2018)
However, internal promotion is only an option if a suitable internal candidate is available.
Exclusively promoting from within might also lead to complications associated with the Peter Principle. If employees are continually being promoted based on their performance, they might reach a leading position that is demanding different experience and training. They end up being unsuccessful because a current good performance does not imply a good performance in the higher position too. In those cases, an external candidate might be a better fit. (Peter & Hull, 1969)
External candidates are not always changing employers voluntarily because they strive for a promotion. Reaching out for new opportunities and a position with more responsibility and larger salary is only the reasoning for some job changes. Terminations, for instance, also force employees into other jobs and career paths that might come with a decrease in rank and pay.
Some employees also chose to start a position that is associated with lower rank to strike a new path or be able to enter an organization that is more favourable on long-term. (Mincer, 2012) However, this research focuses on vertical career advancements that include an improvement in rank and thus count as a promotion.
On the other hand, the strive for promotion can also be a tournament. Employers use such tournament approach to evoke competition between employees. This increases productivity as employees strive to be better than a colleague who is also considered to be promoted. Expected promotion can thus be an incentive which leads to an increase in effort and performance.
However, competing for a promotion, but not receiving one can be demotivating and thus decreasing job satisfaction. According to Lindquist (2010), the tournament approach leads to a feeling of unfair treatment for employees resulting in two effects. The negative impulsive effect (decrease of effort) and the positive strategic effect (increase of effort to enhance later chances).
Nonetheless, the negative effect dominates in her research. Promotion can therefore lead to an expectation problem too. (Lindquist, 2010)
2.2 Job satisfaction
The most common definition of job satisfaction was established by Locke (1976) who states that job satisfaction is “a pleasurable or positive emotional state resulting from the appraisal of one´s job or job experiences”. As job satisfaction is the positive feeling an employee has when liking their job, Locke (1976) continues to discuss variables like one´s needs, values, expectations and perceptions that determine such positive feeling and together form a person´s job satisfaction. According to him, “job satisfaction results from the perception that one´s job fulfils (…) one´s important job values, providing and to the degree that those values are congruent with one´s needs” (Locke, 1976). Thus, if an employee feels like all their expectations, needs and values are being met, he or she is satisfied with a job.
The utility function established by Clark and Oswald (1996) looks into satisfaction as a larger concept. It captures both job satisfaction (u) and the satisfaction gained by other variables apart from employment (). The function of overall well-being is: v = v (u (y, h, i, j), ). Thus, the sub-utility function u for the utility from work is: u = u (y, h, i, j) with y as the income, h as the hours worked, i as the vector of person-specific characteristics and j as the vector of job- specific characteristics (Clark & Oswald, 1996). This indicates that multiple aspects of work but also personal input account for overall job satisfaction. Such job characteristics determine job satisfaction in a way that they design a job to motivate an employee and account for satisfaction.
Hence job characteristics, according to Hackman and Oldham´s Job Characteristics Model (1980), that influence one´s internal motivation are skill variety, task identity, task significance, autonomy and feedback from the job. Skill variety refers to the skills and talents a job requires from the employee that he or she can learn or develop. Task identity refers to the feeling of ownership of tasks that have a visible and contributing outcome. Task significance refers to how impactful one´s work or task is for the organisation or society at large. Hackman and Oldham group these first three core job characteristics under the critical psychological state
“Experienced Meaningfulness” of one´s job as it is highly valuable to employees to follow a meaningful purpose with their work. Autonomy refers to how independent and free the employee is when deciding over the pace, order and ways to perform their tasks. The level of autonomy
relates to the degree to which the employees feels personally responsible for their results. As the fifth core job characteristic, job feedback refers to the communication of clear and immediate responses about how effective the employee was in completing their task. This concerns the knowledge of direct task success rather than oral feedback from a manager later on.
Consequently, all five core job characteristics play a role in determining motivation and hence job satisfaction of an employee. If the job characteristics improve, so does job satisfaction.
(Hackman, 1980)
2.3 Promotion and job satisfaction
Kosteas (2011) found that promotion has a positive effect on the job satisfaction of the promoted employee. According to him, job satisfaction increases after having received or expecting to receive a promotion. Even after controlling for salary increases that could be a possible source of larger satisfaction at the job, Kosteas (2011) finds that satisfaction can largely be put down to non-monetary rewards like promotion and its advantages.
De Souza (2002) also found in her research among managers that participants were positively influenced by the promotion they received and experienced an increase in their satisfaction about work. Promotion as a component of a career path improves the employee’s human capital by developing and using abilities and skills further. This motivates and empowers the employee, thus performance will increase, as well as satisfaction and commitment (De Souza, 2002).
2.4 Gender differences
However, following an upward career path can differ between genders. The influence of promotions on job satisfaction is expected to be different for men and women because of various reasons. First of all, the expectation of promotion varies since more men move up the job ladder and high positions are filled gender-unequal. In her research, Wynn (2017) states that women are facing the dilemma of either being the main family caretaker or a career aspirer.
The cultural expectations that women deal with interfere with their perceived and actual chances of receiving a promotion. Especially mothers have very low perceived chances for promotion. Due to part-time work or no possibility for over hours (because of family responsibilities), mothers feel like they cannot advance in their career, unlike fathers who´s
expectations are not restricted by a work-family context. On the other hand, women without children expect the same promotion possibilities than men (Wynn, 2017; Gerson, 2009). Chiu (1998) states in her research that professional woman expect the same promotion possibilities as their male colleagues. However, professional women, in Chiu´s research female lawyers, report lower job satisfaction because of inequality in promotional opportunities.
In contrast, others (Booth, Francesconi, & Frank, 2003; Kosteas, 2011; Wynn, 2017) found that women overall are just as likely to be promoted as men and their job satisfaction is also similar to men’s (Sousa-Poza & Sousa-Poza, 2000). However, this is not the case for all countries, thus researchers disagree about gender differences in job satisfaction. Sousa-Poza and Sousa-Poza (2000) highlight in their research that in the US, Great Britain and Switzerland women show higher job satisfaction than men. But in most countries, men and women experience similar job satisfaction. As this research is using German panel data, I focus on the result Sousa-Poza and Sousa-Poza (2000) found for Germany. German men have slightly higher overall job satisfaction than women. Hauret and Williams (2017) found in their research that German women are more likely to rate their job satisfaction as very positive. Men tend to use less positive ratings although they are slightly more satisfied overall. This would evoke the assumption that women are even less satisfied than indicated because they tend to rate their experience more positive but still score lower than men.
Nonetheless, to my knowledge the effect of promotion on job satisfaction has only been researched in the US by Kosteas (2011). According to him, promotion has a positive effect on job satisfaction1. Based on Kosteas’ results, I assume the job satisfaction of men and women in Germany to also increase after promotion.
Hypothesis 1: Promotion has a positive effect on job satisfaction.
Yet, the circumstances of promotion differ between men and women. Although both genders are expected to experience an increase in job satisfaction, there might be a difference in how large the increase eventually is. According to Wynn (2017), women expect less, so a received promotion might cause astonishment and feeling of success, hence their satisfaction
1 However, he did not find a difference in job satisfaction between US men and women like Sousa-Poza and Sousa- Poza (2000). Thus, the level of job satisfaction was the same for men and women before the promotion and raised the same after the promotion. Gender was used as a control variable here.
raises outstandingly. Men already anticipate their promotion, thus, fulfilled expectations do not produce a surpassing rise of job satisfaction. Additionally, Chiu (1998) states that professional women experience lower job satisfaction because of missing possibilities for influence and promotion. They are dissatisfied with their opportunities. Assumingly, after being granted such desired opportunity for advancement, their job satisfaction would increase extensively. In fact, this assumption is also supported by research done in the field of psychology. Generally, women tend to feel either positive or negative emotions in a stronger way, meaning that both negative and positive influences affect females more intensely than males (Diener, Suh, Lucas, & Smith, 1999).
Based on this assumption, I hypothesize:
Hypothesis 2: The relation between promotion and job satisfaction differs for men and women.
Hypothesis 2a: The relation between promotion and job satisfaction is more positive for women than for men.
In contrast, Booth et al. (2003) showed by establishing a “sticky floors” model that women’s salary increase after promotion is lower compared to men. They receive a smaller gain in their monetary reward than men do. According to Booth et al. (2003), this is due to the extent of outside opportunities. Especially in higher ranks, women have less opportunities on the labour market that are more interesting than or similar to an internal promotion, hence companies do not need to compete as much for women as they do for men. Because of the amount of outside opportunities, companies want to make sure a male employee stays with the firm, so they offer salary incentives in line with the market. Women find themselves in a position that does not give such bargaining power because there are little outside offers that companies would have to align with or generally, family commitments would restrict women in taking such offers. (Booth et al., 2003; Francesconi, 2001)
Regarding the underpayment of women, older findings propose that women tend to value non-monetary rewards more than monetary ones because they appreciate aspects like interpersonal harmony and social achievement more (Kahn, Krulewitz, O'Leary, & Lamm, 1980). This would suggest that women do not find their salary as such an important factor of
their reward package and hence, their job satisfaction will not be affected by the underpayment (Witt and Nye, 1992). However, Witt and Nye (1992) find in their research about perceived fairness of promotion and pay that there is no gender difference existent. Men and women are equally affected by unfairness. In fact, male and female employees were found to both be dissatisfied about unfair promotion or pay. This shows, that women do find salary an important aspect of their reward package because their satisfaction level is influenced by it just as the satisfaction level of men is (Witt and Nye, 1992). In particular, salary increases that usually go hand in hand with promotions, only have a positive effect on the employee if such increase is not lower than the pay increase of colleagues. The relative income thus works as the reference point for employees to be satisfied with their wage increase or not. The increase in job satisfaction might be restrained because of a relatively small wage gain in comparison to a larger gain of peers (Diriwächter & Shvartsman, 2018).
Women, on average, receive smaller salary gains after promotion than men and are dissatisfied with such inequality. Thus, even after being promoted, their job satisfaction does not increase as much as men’s because they feel unfairly treated (Booth et al., 2003; Witt and Nye, 1992). I hypothesize:
Hypothesis 2b: The relation between promotion and job satisfaction is less positive for women than for men.
3. DATA & METHODS
3.1 Data set and main variables
To research the influence of promotion on the job satisfaction of men and women in general, a large data set is needed. Additionally, it is necessary to have information over multiple years to correctly capture previous promotions and their impact. Thus, for this research, I am using the German Socio-Economic Panel study (GSOEP) to look into job changes of surveyed participants over multiple years and their related job satisfaction. The GSOEP is a longitudinal study of 11,000 private households in Germany with more than 20,000 respondents. It is a multi- stage random sample with regional clusters. A pre-tested questionnaire is used to report information from all individuals within one household, either during an interview or by filling it out individually. The GSOEP data is published every year since 1984. The latest available individual questionnaire is from year 2016 (Welle 33). This questionnaire is also used as reference in this study, hence the question numbers following in the next sections refer to the questionnaire of 2016. The GSOEP data set that is used here is part of the long study and is called pl.sav. It contains most variables needed for the question research in this study.
Additionally, variables from the data set pgen.sav are added. This research will focus on the samples of the years 2006 to 2016 to receive findings that are the most recent but still cover at least 10 years back.
Research population
The sample will be resized to fit the research question. Only participants between the ages of 25 to 60 that are participating in the labour market are being included. Participants between the age of 25 and 60 are being selected based on their year of birth between 1946-1956 to 1981-1991 (variable code ple0010) to include all respondents that were 25 to 60 between 2006 and 2016.
To ensure an employment, question 31 “Are you currently employed? Which one of the following applies best to your status?” has to be answered with either “Employed full-time” or
“Employed part-time” (variable code PGEMPLST).
The GSOEP data set includes 115,552 observations that fit the research population of this study.
Independent variable: Job satisfaction
The level of job satisfaction is surveyed every year with question 1.3 “How satisfied are you today with your job?”. On a scale from 1 (completely dissatisfied) to 10 (completely satisfied) participants rate their job satisfaction (variable code plh0173).
Dependent variable: Promotion
The GSOEP continues asking about the current job. Question 40 “Have you changed jobs or started a new one since December 31, 2014?” (variable code PGJOBCH) has to be answered with “Yes”. Because of the panel structure and the yearly questionnaire capturing job changes within the last year or two of the questionnaire since 2006, the data used in this study includes all job changes done between 2005 and 2016.
Question 43 “What type of occupational change was that?” (variable code plb0284) has to be answered with either “I started a new position with a different employer” or “I changed positions within the same company”. This selects the participants that underwent an internal or external position change that might have been a promotion. Additionally, it deselects the participants that just entered the job market and did not have a chance to receive a promotion yet as well as returning employees (parental leave, sickness etc.) and self-employed participants.
Until now, I have identified 14.607 cases that report a job change and are presumably a promotion. Of which 9.488 are an external job change and 1.684 are internally.
To determine if a respondent had a promotion (vertical advancement), the income of each respondent will be compared to the year before. Thus, the necessary data about promotions can be found in the panel structure of the study because the GSOEP does not specifically survey if a respondent received a promotion. The indication of the salary as a direct measure will be used to estimate an increase to the year before. A salary increase thus indicates a promotion (as a promotion comes with salary increases). This way of filtering for promotional job changes was done by Clark & D´Ambrosio (2015) before in their study about the social context of promotions also using GSOEP data. To distinguish a salary increase due to a promotion from a salary increase because of other reasons like seniority, labour contracts or collective agreement changes of trade unions, the average yearly salary increase indication of the German Federal Statistical Office (Destatis) will be used to set a boundary. According to Destatis, real earnings increased by 0.85% on average between 2006 and 2016 with the highest increase of 2.4% from 2014 to
2015. Therefore, in this study I will use either 0.85% as the margin for all salary increases in the years of a lower increase than 0.85% or the indicated increase for each year to make sure all salary increases are indeed due to promotion. For instance, between 2011 and 2012 salaries increased by 0.5%. Here, I will use 0.85% as the boundary as it is the average increase in my sample. However, between 2009 and 2010 salaries increases by 1.4%. For this year, I will use 1.4% as the boundary to not include salary raises due to other reasons than promotion. The net and gross income of each participant is asked for in question 81 “What did you earn from your work last month?” (variable code gross income PGLABGRO). The calculation will be based on the respondent´s gross income so the actual salary increase received from the employer is measured, not a net income that was already influenced by different individual tax regulation because of tax categories.
Dependent variable: Gender
The gender of the participant is recorded in the personal details section and can be answered with either male or female. In the panel data gender is coded with 1 indicating male and 2 indicating female (variable code pla0009). For my analysis, I recoded gender into male=0 and female=1.
Control variable: Age
Age is recorded in the personal details section at the beginning of the questionnaire (variable code ple0010). The participant has to provide their month and year of birth. Based on their birth year, a new variable “Age” was computed. In the difference in difference analysis, age drops out. To keep controlling for age, a new squared variable is introduced: Age². The lag of Age² subtracted by Age² is not a constant and can therefore be included.
Control variable: Education
The level of education is surveyed based on the highest level of school-leaving qualification the respondent obtained. This variable uses the International Standard Classification of Education of 1997 as indication for the different levels of vocational school, higher education or still in school (variable code PGISCED97).
Control variable: Job characteristics
To control for job characteristics and their possible change, respondents’ indication of work pressure, development possibilities and autonomy at work will be included based on Hackman and Oldham´s Job Characteristics Model (1980). Work pressure is surveyed with the agreement to question 66 “There is often high time pressure due to the larger volume of work”
(variable code plb0607) and “My workload has increased steadily over the last two years”
(variable code plb0609). Those variables will be combined to measure work pressure as they are highly positively correlated (r=.914, p<0.01). Development possibilities will be determined by also combining variables “bad promotion possibilities” (variable code plb0610) and “appropriate chances of occupational advancement” (variable code plb0615). The correlation between those variables is also significantly positive (r=.789, p<0.01).
Additionally, occupational autonomy (variable code PGAUTONO) is included.
3.2 Method
The analysis will consist of two steps to test my hypotheses. Firstly, I will measure if promotion increases job satisfaction for both men and women (Hypothesis 1). Secondly, I will measure the increase to see the difference between genders (Hypotheses 2, 2a and 2b). During the analysis, I use control variables for age, education and job characteristics to obtain a non- biased and non-distorted result. Based on literature review, I expect factors such as age, education and job characteristics to have an influence on job satisfaction, therefore they need to be controlled for.
Additionally, in both steps of testing Hypothesis 1 and 2, I will include equations on the influence of general job change (model 1) and specifically on internal and external promotion (model 3) to provide a more distinguished result. Model 2 tests the influence of promotion and gender on job satisfaction. The variables gender*job change, gender*promotion etc. capture the influence of the interaction effect of those variables on job satisfaction. The regression equations are as follows:
(1) 𝐽𝑆𝑖 = 𝛼 + 𝛽1 𝐽𝐶ᵢ + 𝛽2 𝐺ᵢ + 𝛽3 𝐺𝐽𝐶ᵢ + 𝛽4 (𝐸ᵢ + 𝐴ᵢ + 𝐴2ᵢ + 𝑂𝐴ᵢ + 𝑊𝑃ᵢ + 𝐷𝑃ᵢ) (2) 𝐽𝑆𝑖 = 𝛼 + 𝛽1 𝑃ᵢ + 𝛽2 𝐺ᵢ + 𝛽3 𝐺𝑃ᵢ + 𝛽4 (𝐸ᵢ + 𝐴ᵢ + 𝐴2ᵢ + 𝑂𝐴ᵢ + 𝑊𝑃ᵢ + 𝐷𝑃ᵢ)
(3) 𝐽𝑆𝑖 = 𝛼 + 𝛽1𝐼𝑃ᵢ + 𝛽2𝐸𝑃ᵢ + 𝛽3𝐺ᵢ + 𝛽4𝐺ᵢ(𝐼𝑃ᵢ + 𝐸𝑃ᵢ) + 𝛽5(𝐸ᵢ + 𝐴ᵢ + 𝐴2ᵢ + 𝑂𝐴ᵢ + 𝑊𝑃ᵢ + 𝐷𝑃ᵢ)
where JSᵢ is the job satisfaction of one individual i. JC indicates job change, P promotion, IP internal promotion, EP external promotion, G gender, E education, A age, A² age squared, OA occupational autonomy, WP work pressure and DP development possibilities. GJC stands for the gender*job change interaction term, GP for the gender*promotion interaction term, GIP for gender*internal promotion and GEP for gender*external promotion.
The first regression analysis, however, does not control for “unobserved heterogeneity”.
The assumption of standard models is that all respondents have been exposed to the same outside effects and hence, experienced the same influence on their responses. Because the data used in this research is a panel study, respondents were surveyed at different locations and different years of the last decade. Therefore, the data includes variation among cases. To overcome the
“unobserved heterogeneity”, one could estimate a fixed effects panel model that includes dummies for each individual and thus controls for individual differences. Another possibility to drop the individual effects is to perform a difference in difference analysis. A difference in difference analysis only measures variances within one respondent and not, like the standard model, compares between respondents.
Therefore, a difference-in-difference analysis will be conducted (1A to 3A) to be able to measure both the increase of job satisfaction and the difference in genders within that increase before and after a promotion without having to include an individual effects model. Variables will be lagged and used for the deduction of the variables at different times. In this analysis, age and gender disappear as either they are constant, or their difference is a constant. The following regression equations are estimated:
(1𝐴) 𝐽𝑆𝑡− 𝐽𝑆𝑡−1
= 𝛼 + 𝛽1(𝐽𝐶𝑡− 𝐽𝐶𝑡−1) + 𝛽2(𝐺𝐽𝐶𝑡− 𝐺𝐽𝐶𝑡−1) + 𝛽3((𝐸𝑡− 𝐸𝑡−1) + (𝐴²𝑡− 𝐴²𝑡−1) + (𝑂𝐴𝑡− 𝑂𝐴𝑡−1) + (𝑊𝑃𝑡− 𝑊𝑃𝑡−1) + (𝐷𝑃𝑡− 𝐷𝑃𝑡−1))
(2𝐴) 𝐽𝑆𝑡− 𝐽𝑆𝑡−1
= 𝛼 + 𝛽1(𝑃𝑡− 𝑃𝑡−1) + 𝛽2(𝐺𝑃𝑡− 𝐺𝑃𝑡−1) + 𝛽3((𝐸𝑡− 𝐸𝑡−1) + (𝐴²𝑡− 𝐴²𝑡−1) + (𝑂𝐴𝑡− 𝑂𝐴𝑡−1) + (𝑊𝑃𝑡− 𝑊𝑃𝑡−1) + (𝐷𝑃𝑡 − 𝐷𝑃𝑡−1))
(3𝐴) 𝐽𝑆𝑡− 𝐽𝑆𝑡−1
= 𝛼 + 𝛽1(𝐼𝑃𝑡− 𝐼𝑃𝑡−1) + 𝛽2(𝐺𝐼𝑃𝑡− 𝐺𝐼𝑃𝑡−1)+ 𝛽3(𝐸𝑃𝑡− 𝐸𝑃𝑡−1) + 𝛽4(𝐺𝐸𝑃𝑡− 𝐺𝐸𝑃𝑡−1) + 𝛽5((𝐸𝑡− 𝐸𝑡−1) + (𝐴²𝑡− 𝐴²𝑡−1) + (𝑂𝐴𝑡− 𝑂𝐴𝑡−1) + (𝑊𝑃𝑡− 𝑊𝑃𝑡−1)
+ (𝐷𝑃𝑡− 𝐷𝑃𝑡−1))
4. RESULTS
4.1 Descriptive statistics
The research population in total contains 115,552 cases of 28,266 different respondents.
The mean of the respondent´s birth year is 1966,58, meaning the average respondent was 49 in 2016. Only slightly more respondents are male (52.7%) and more than 75% are employed full- time. The largest group of respondents has middle vocational education (46%), the second largest group with 29.4% reports the highest possible ISCED classification (higher education).
The average gross income is 2911.27€ per month. Most respondents are married (64.5%), others are living alone (22%) or together with an unmarried partner (13%)
Table 1: Descriptive statistics
Variable N (%) or Mean (SD)
Gender Male 60,904 (53%)
Female 54,648 (47%)
Age 49,42 (SD=10,245)
Employment Full-time 86,843 (75,2%)
Part-time 28,709 (24,8%)
Education Middle vocational 53,130 (46%) Gross monthly
income
2911,27€ (SD=2336,47€)
Partner Indicator Married 74,507 (64,5%)
4.2 Differences between genders
The differences between genders of the main variables can be found in table 2. Women report a slightly higher level of overall job satisfaction than men. Interesting to note is that women change jobs more often than men. They also receive promotions somewhat more often although almost half of the female respondents work part-time (47%), which is a much larger amount than part-time working men (5%). Additionally, the share of women in internal as well as external promotions is slightly higher. Though, the correlation of external promotion is not significant.
Important to note is the gender difference in the income increase after having received a promotion. According to the T-Test performed on this data set, the increase of income is larger for women as for men.
Table 2: Gender-differences of selected variables
Male Female
Variables Mean (SD) Mean (SD) Job satisfaction 6.69* (2.813) 6.77* (2.2708) Job change .114* (.317) .141* (.348) Promotion .029* (.169) .031* (.174) Internal promotion .005* (.068) .006* (.079) External promotion .019 (.138) .02 (.139) Employment 1.05* (.215) 1.47* (.499) Income increase
after promotion
1.66* (24.53) 2.05* (30.08)
Method of analysis: Independent Samples T-Test *significant correlation (p<.01)
4.3 Relation between variables
The correlations between the dependent variable job satisfaction and the independent variables are reported in table 3-5. Table 3 shows the correlations for model 1 on the independent variables job change, gender and gender*job change controlling for age, education and job characteristics. The correlation between job satisfaction and job change is significantly positive (r=.021), as is the correlation between job satisfaction and gender (r=.021), job satisfaction and gender*job change (r=.032) and gender and job change (r=.033). These correlations are quite weak compared to gender and gender*job change (r=.28) and job change and gender*job change (r=.691).
Table 3: Correlations of model 1
Job satisfaction Job change Gender Gender*Job change Job satisfaction 1
Job change .021* 1
Gender .021* .033* 1
Gender*Job change .032* .691* .28* 1
Method of analysis: Partial Correlation. *significant correlation (p<.01)
The correlations for model 2 are shown in table 4. Job satisfaction is significantly positive correlated with promotion (r=.042), with gender (r=.021) and with gender*promotion (r=.028). The correlation between promotion and gender*promotion (r=.69) is much larger than between gender and gender*promotion (r=.129). The correlation between gender and promotion however is not significant (r=.004).
Table 4: Correlations of model 2
Job satisfaction Promotion Gender Gender*Promotion Job satisfaction 1
Promotion .042* 1
Gender .021* .004 1
Gender*Promotion .028* .69* .129* 1
Method of analysis: Partial Correlation. *significant correlation (p<.01)
The correlations of model 3 are reported in table 5. Most of the correlations are significantly positive, except the significantly negative correlations between internal promotion and external promotion (r=-.013) as well as internal promotion and gender*external promotion (r=-.013), external promotion and gender*internal promotion (r=-.013) and gender*external promotion and gender*internal promotion (r=-.012). The correlation between gender and external promotion is the only one non-significant (p=.745).
Table 5: Correlations of model 3
Job
satisfaction
External promotion
Internal promotion
Gender Gender* external promotion
Gender*internal promotion Job satisfaction 1
External promotion
.038* 1
Internal promotion
.009* -.013* 1
Gender .021* -.001 .013* 1
Gender* ext.
promotion
.026* .686* -.009* .101* 1
Gender*internal promotion
.005* -.01* .739* .058* -.007** 1
Method of analysis: Partial Correlation. *significant correlation (p<.01) **significant correlation (p<.05)
4.4 Relationship between job satisfaction and job change/ (internal/external) promotion
4.4.1 Regression analysis (1 to 3)
The first analysis of the relationship between the dependent variable job satisfaction and the independent variables are reported in table 6. It includes the means and standard deviations (SD) and is split in 3 models each representing one of the regression equations (1) to (3). The control variables are age, education and the job characteristics work pressure, occupational autonomy and development possibilities. In model 1, the findings for the relationship between job satisfaction, job change and gender can be found. Model 2 shows the relationship between job satisfaction, promotion and gender. In model 3, the relationship between job satisfaction, internal/ external promotion and gender is given.
Table 6: Effects of control variables on the relationship between job satisfaction and the independent variables
Means Model 1 Model 2 Model 3
Parameter (SD) Estimate Std. Error Estimate Std. Error Estimate Std. Error
Intercept 5.158* .151 5.213* .15 5.227* .15
Control
Development possib. .51 (1.59) .168* .012 .168* .012 .167* .012 Work pressure .53 (1.65) -.086* .012 -.088* .012 -.087* .012 Occ. autonomy 2.88 (1.09) .348* .009 .343* .009 .343* .009
Education 4.03 (1.5) -.043* .006 -.043* .006 -.042* .006
Age 44,79 (9.8) .034* .007 .031* .007 .031* .007
Age² 2104 (880) .00* .00 .00* .00 .00* .00
Main
Job change .13 (.332) .039 .036
Gender .47 (.499) .072* .017
Gender*job change .07 (.249) .281* .049
Promotion .03 (.171) .749* .066
Gender .115* .016
Gender*promotion .01 (.121) -.15 .094
Int. promotion .01 (.73) .533* .164
Ext. promotion .02 (.139) .82* .081
Gender .116* .016
Gender*int. promotion .00 (.054) -.342 .221
Gender*ext. promotion .01 (.096) -.137 .116
N 115.552 115.552 115.552
Adjusted R² .019 .02 .02
F statistic 253.052 266.514 215.522
Std. Error of Estimate 2.738 2.736 2.737
*significant correlation (p<.01) **significant correlation (p<.05) ^significant correlation (p<.1), Dependent variable: job satisfaction
The findings presented under model 1 in table 5 indicate that job change has a small positive but non-significant effect on job satisfaction (B=.039, p=.27), meaning that job changes do not significantly influence job satisfaction. The gender*job change interaction however has a
significant positive effect on job satisfaction (B=.281, p<.01). This interaction between gender and job change remains the only significant interaction term of the analysis reported in table 5 and indicates that job change only increases job satisfaction for women, not for men.
Furthermore, model 2 represents the relation between job satisfaction, gender and promotion.
Promotion has a strong positive and significant effect on job satisfaction (B=.749, p<.01). This means, that receiving one or multiple promotions increases job satisfaction. Gender also has a significantly positive effect on job satisfaction (B=.115, p<.01) which confirms the finding on job satisfaction and gender reported in table 2. Women have a higher level of job satisfaction in general. The gender*promotion interaction however, has a negative and non-significant effect (B=-.15, p=.111). Model 3 shows the finding regarding the relationship between job satisfaction, gender and internal/ external promotion. Both internal promotion (B=.533, p<.05) and external promotion (B=.82, p<.01) have a significant positive effect on job satisfaction which corresponds with the findings of promotion and job satisfaction. The effect of gender is positive and significant (B=.116). The effect of the interaction of gender*internal promotion is negative and non-significant (B=-.342, p=.122) as well as the interaction effect of gender*external promotion (B=-.137, p=.238).
The control variables influence the relationships between the dependent variable and the independent variable differently. Work pressure has a small significantly negative effect (B=- .086/-.088/.087, p<.01) as well as education (B=-.043/-.043/-.042, p<.01). Occupational autonomy has the largest effect on all relationships (B=.348/.343/.343, p<.01). The other control variables that have a significantly positive effect are development possibilities (B=.168/.168/.167, p<.01) and age. Age records a rather weak positive and significant effect (B=.034/.031/.031, p<.01) that is quite unusual as age normally has a negative effect resulting in a curvilinear graph. However, the positive coefficient found here indicates a linear line for the relationship between age and job satisfaction, meaning job satisfaction increases as age increases.
To summarize, model 1 indicates that job changes influence men and women differently.
Although the estimate of job change is not significant, the gender*job change interaction is.
Women experience an increase in job satisfaction, whereas a job change has no effect on men´s job satisfaction. However, model 2 shows that promotion has a significant and positive effect on job satisfaction (B=.749, p<.01). Additionally, both internal and external promotion positively
affect job satisfaction. External promotion records a slightly higher coefficient, meaning that external promotion increases job satisfaction slightly more. To confirm the first hypothesis, I will include the second regression analysis, though. Nonetheless, the non-significant effect of the gender*promotion interaction term indicates no differences between men and women regarding the effect of promotion on job satisfaction. The interaction effects of gender*internal promotion and gender*external promotion are not significant either. This could be due to missing individual effects that were not included in this regression model, as mentioned in the methodology. To account for such individual effects, I will continue with a difference in difference analysis.
4.4.2 Difference in difference analysis (1A to 3A)
Table 7 provides the means, standard deviations (SD), regression coefficients and standard errors (SE) of the difference in difference analysis. The relationships between the differences in time of the dependent variable and the independent variables can be found here.
These variables have been newly estimated by deducting the variables data at time t-1 from the data at time 1. This was done to eliminate the individual variances of respondents and to be able to only compare observation within one individual.
Table 7: Coefficients of the difference in difference analysis on the relationship between job satisfaction and the independent variables
Means Model 1A Model 2A Model 3B
Parameter (SD) Estimate SE Estimate SE Estimate SE
Intercept .214* .024 .081* .024 .086* .024
Control
Develop. poss. .63 (1.742) .105* .012 .105* .012 .104* .012
Work pressure .66 (1.83) -.055* .012 -.055* .012 -.053* .012
Occ. autonomy .02 (.611) .167* .015 .149* .015 .152* .015
Education .02 (.259) .135* .035 .135* .035 .150* .035
Age² 97.7 (39.6) .00 .00 .001* .00 .001* .00
Main
Job change -.02 (.384) .788* .034 Gender*JC -.01 (.278) -.126* .047
Promotion .01 (.251) 1.059* .05
Gender*promo. .00 (.175) -.216* .072
Int. promotion .00 (.109) .598* .123
Gender*int. promo. .00 (.08) -.28^ .167
Ext. promotion .01 (.202) 1.251* .062
Gender*ext. promo. .00 (.139) -.213** .089
N 87.286 87.286 87.286
Adjusted R² .014 .011 .011
F statistic 176.111 143.2 109.675
Std. Error of Estimate
2.668 2.671 2.672
*significant correlation (p<.01) **significant correlation (p<.05) ^significant correlation (p<.1), Dependent variable: job satisfaction
The findings under model 1A indicate a significantly positive relationship between the change in job change and the change in job satisfaction (B=.788, p<.01). The effect of the change of the gender*job change interaction on the change of job satisfaction is however significantly negative (B=-.126, p<.01). The analysis of model 2 shows that the change of
promotion has a very strong significantly positive effect on the change in job satisfaction (B=1.059, p<.01). On the other hand, the effect of the gender*promotion change is significantly negative (B=-.216, p<.01). Model 3 shows, that the relationship between the change in internal promotion and the change in job satisfaction significant and positive is (B=.598, p<.01). The effect of the change in gender*internal promotion on the change in job satisfaction is negative (B=-.28) and moderately significant (p<.1). The change in external promotion affects the change in job satisfaction significantly and strongly (B=1.251, p<.01). The change in gender*external promotion has a significant but negative effect (B=-.213, p<.05) on the change in job satisfaction.
These findings indicate that there indeed is a positive significant relationship between job change and job satisfaction (B=.788, p<.01) that is larger for men than for women. Additionally, the results of this second analysis also show a significantly positive and strong relationship between promotion and job satisfaction (B=1.059, p<.01). Hence, Hypothesis 1 “Promotion has a positive effect on job satisfaction” can be accepted. Although the first analysis indicated no gender difference within that relationship, the difference in difference analysis corrects that assumption. There is a -.216 difference between genders, indicating that promotions increase job satisfaction less for women than for men. Hence, men´s job satisfaction after promotion rises more than women´s. Based on this result, Hypothesis 2 “The relation between promotion and job satisfaction differs for men and women” as well as Hypothesis 2b “The relation between promotion and job satisfaction is less positive for women than for men” can be accepted.
Therefore, Hypothesis 2a ”The relation between promotion and job satisfaction is more positive for women than for men” is rejected because it functions as the opposing assumption.
The results indicate similar relationships for internal and external promotion. The effect of external promotion on job satisfaction is very strong (B=1.251, p<.01) and also stronger for men than for women as indicated by the interaction term gender*external promotion (B=-.213, p<.05). Less strong but still significantly positive is the effect of internal promotion on job satisfaction, with a stronger effect for men too.
