Amsterdam Business School
Bachelor Economics and Business Specialization Business Administration
The Relationship Between Higher Education and Labor Market Success To What Extent GPA Affects Starting Salaries – A Meta-Analysis
BSc Thesis by
Vivien Andrea Boros 10860843
Supervisor: Vladimer Kobayashi Amsterdam, 26th
Statement of Originality This document is written by Student Vivien Andrea Boros, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
Statement of Originality ... 2 Abstract ... 4 Keywords ... 4 1. Introduction ... 5 2. Literature Review ... 7 3. Theoretical Framework ... 9 4. Research Design and Conceptual Framework ... 12 4.1. Meta-analysis ... 12 4.2. Variables ... 13 5. Findings ... 15 5.1. GPA ... 15 5.2. Moderator Analysis ... 17 6. Discussion ... 25 7. Limitations ... 29 8. Future Research ... 30 9. Conclusion ... 31 10. Appendices ... 33 10.1. Appendix 1: Results of the Literature Review ... 33 10.2. Appendix 2: Data used in the SPSS Analysis ... 34 11. Bibliography ... 35Abstract: The extent to which grade point average (GPA) influences salaries, specifically starting salaries, has been a long researched and debated issue. After conducting an extensive literature review, it has been hypothesized that a higher GPA has a positive effect on starting salaries. Subsequently, carrying out a meta-analysis based on 12 studies with 43 637 participants, has yielded the result that GPA has a significant positive relationship with salaries (r=0.3582, p<0.01). This qualifies as a medium effect size, which was also heterogeneous. After implementing a moderator analysis on sample size, publication year of the study, the country the research was conducted in, age and gender, there was no support provided that any of them act as moderating variables. In conclusion, a better GPA seems to have a positive effect on salaries, however, more research needs to be conducted while including several moderating variables for the sake of getting a more saturated picture of what factors matter the most during the hiring process after graduation. Keywords: GPA, labor market success, salary, higher education, starting salary, meta-analysis
1. Introduction
The relationship between education and labor market success has been a long debated topic. There has been a considerable amount of research conducted, but according to Kupfer (2011) not enough emphasis has been put on how higher education and academic achievements are related to labor market success. Labor market success can be defined as an individuals ranking on certain jobs ‘desirability’, and how well they have done in the competition for these jobs, which other workers have also ranked as desirable (Jencks, Perman, &, Rainwater, 1988). Desirable job characteristics include salary, working hours from an economic point of view; and non-monetary characteristics such as job satisfaction, number of vacation weeks, risk of job loss, on the job training and educational requirements.
In previous research, emphasis has been put on a lot of distinct variables that have an effect on career accomplishments, such as cognitive and non-cognitive skills, motivation, work experience during studying and the structure of social networks. Duncan and Dunifon (2012) have emphasized the necessity of ‘soft skills’ for long-run labor market success. They claim that besides ‘basic new skills’ such as relatively good understanding of math, good communication and technology skills, ‘soft skills’ are equally, if not more, important. ‘Soft skills’ include the ability to work in groups and building social-capital relationships (Duncan & Dunifon, 2012).
Not only how to measure labor market success has been a heavily debated issue, but the question of how to operationalize academic achievements and success has also generated a lot of discussion amongst academics. Roth and Clarke (1998) have put the emphasis on work experience, claiming that working students have much more experience to use in the field as soon as they start and this should be valued, hence they are ought to receive higher starting salaries. Sandvig et al. (2005) also found that in case of business administration graduates, internship experience associated with a higher starting salary (β = 0.38, p < 0.001), and they determined that this is the most influential factor of predicting starting salaries.
Wise (1975) explained that a higher GPA could be an indicator for employers that someone is more motivated, productive, has better cognitive abilities and may be better at exercising their ‘soft skills’. Hence executives are more inclined to hire employees with a higher GPA because it could suggest a greater level of labor market success.
Other than cognitive abilities, Gleason (1993) has focused on the effect of having a job while studying at university. He concluded that the right amount of work can have a positive effect on GPA, however, working too much can also delay graduation, have a detrimental effect on grades and may even cause dropping out of the education program. Even though he has observed both positive and negative effect of work on GPA, he has not analyzed whether this change in GPA also has a further impact on job market achievements, specifically on salaries.
Therefore, it can be seen that current academic literature has not focused on how GPA affects employment opportunities after graduating, when a student most likely does not have professional experience to rely on yet. The aim of this paper is to focus on the relationship between university GPA and the effect it has on labor market success, specifically the extent to which it influences starting salaries.
This paper is going to be structured in the following way: first, a literature review about the current state of academic opinions on whether GPA affects starting salaries will be provided. Next, in the theoretical framework it will be explained why GPA and starting salary are adequate measures respectively of higher education outcomes and labor market success. Following, the research design and conceptual framework are going to be described. In the next section the research findings are going to be presented, followed by a discussion of the results. Finally, the limitations of the research are going to be discussed, suggestions for possible future research are going to be presented, and a conclusion is going to be drawn.
2. Literature Review
There are conflicting views on whether GPA is an effective predictor when talking about labor market success or if it even has a relationship with long-run labor market success. The existing academic literature is going to be analyzed based on findings, whether they supported the hypotheses that GPA is correlated to labor market success.
Many studies have established a relationship between grade point average and job market accomplishments. According to Wise’s human capital theory (1975), ‘an individual chooses the occupation and level of education that maximize the present value of his expected lifetime earnings. In his research he found that the estimated rates of salary are highly influenced by grade point average, in addition to college selectivity. Furthermore, Jones and Jackson (1990) conveyed that grades might indicate a productive capacity to employees before they start investing in employee training, which supports Wise’s human capital theory.
Some studies found that GPA may be a more valid predictor than thought; the general findings were that the predicting validity of GPA on salaries was higher after one year on the job (Roth, BeVier, & Schippmann, 1996). Roth and Clarke (1998) found that the correlation between GPA and starting salary was modest, but it was more meaningful when looking at to what extent it affects current salary. They also pointed out that the field might have an effect how GPA is evaluated prior to hiring – for example, in the medical field it might be taken seriously and affect starting salaries.
To sum up, research which find GPA important when predicting job market accomplishments is mainly based on human capital theory and claims that GPA could be a forewarning for employees about prior productivity and engagement with tasks.
On the other hand, many academics have different findings on the relationship between GPA and labor market success.
Bretz (1989) said that for businesses it is of vital importance to try to tell about someone whether they will be successful employees prior to hiring and assessing their GPA could help in this. However, after conducting research he reached the conclusion that GPA is a poor predictor of adult work-related achievement. He found that GPA might be a sufficient predictor for starting MBA salaries, but the results were inconsistent amongst groups. Another study on academic and occupational outcomes has reached similar conclusions. Samson, Weinstein and Walberg (1984) conducted a quantitative analysis within the field of medicine, military, nursing; engineering and business, and they found that GPA obtained in the 2nd and 3rd year could be correlated to job success, but their
findings were also irreconcilable between across groups.
When analyzing the relationship between educational predictors and pay, Ferris (1982) has reached a more consistent conclusion. He initiated that having a graduate degree is more important than the grades obtained, hence more monetary value is attached to education than GPA. He also found that the effect of grades on salary level declines after a year and at the senior level it is insignificant (Ferris, 1982).
Ultimately, considering the studies that have found that GPA is not affecting labor success or salaries, is based on inconsistent quantitative analysis’ results. Moreover, researchers who claimed that GPA has no relation to labor market success noted that there are many other variables that contribute to the achieved grade point average, which should be considered, such as extracurricular activities, university ranking, field of study and elected courses (Bretz, 1989). Hence, the explanation for their results may be the fact that they compared GPAs being composed of distinct measures.
As it can be seen, there are a lot of conflicting results on how GPA affects labor market outcomes and whether it has a relationship with (starting) salaries. Many findings are inconclusive, inconsistent or contradictory when looking at different academic literature. Research has confirmed that GPA is somewhat correlated to the level of salaries, but the results are ambiguous regarding at what point in one’s career it has effect. What this means, is that Roth
et al. found that the effects of GPA are higher after one year on the job, whereas Ferris concluded the effect diminishes after a year. Therefore, it would be interesting to conduct meta-analysis on the research question to what extent GPA has an effect on starting salaries.
3. Theoretical Framework In this section the meaning of human capital theory, GPA and starting salary are going to be discussed and it is going to be established why the latter two are sufficient variables to describe the relationship between higher education achievements and labor market success. According to the human capital theory, people make rational choices regarding investments in their own human capital, trying to maximize the return on all of their expenditures. Education is one of the most important investments when talking about labor market success; since the earnings, salary progression and developmental opportunities of more educated people are usually above average (Becker, 1994). It has been suggested that firms themselves only invest in employee skills when those are justifiable in terms of future value creating capabilities. Hence GPA could be an indicator whether a job applicant could be a potential profitable investment when their skills are identified, developed and deployed (Lepak & Snell, 1999). According to Wayne et al. (1999), human capital investment and its effect on labor market success could be measured by comparing the differences between salaries of those who obtained a Bachelor’s/Master’s/PhD degree. In this paper, human capital theory is what helps to explain the relationship between academic performance and starting salaries, since it is expected that someone will try to acquire a high GPA in order to look more appealing as a job candidate and companies will also look for those who performed better in their academic career.
Grade Point Average (GPA) is defined as the average value of the accumulated final grades earned in different courses over the period of obtaining a degree. Loury and Garman (1995) found that amongst personal factors, GPA and the choice of major had the most significant effects on earning when starting a job, hence GPA could be a competent predictor of starting salary. It has been statistically shown in several studies that a more expensive and highly ranked university does not necessarily generate the biggest returns on investment. When making hiring decisions, employers found GPA a more significant indicator of performance, therefore, in relation to human capital theory, it seems that with a smaller amount of investment in education the same results in salaries may be
achieved (Jimeno et al., 2016; James et al., 1989). Donhardt (2004) also said that the human capital theory holds that cognitive skills, which are established at university and are reflected in academic achievements; are the same ones leading to labor market success. He said that GPA is the best measure of academic achievements because in line with the human capital theory, it reflects the presence and quality of cognitive skills when there is not a vast amount of information available about an entry-level job candidate’s skills yet; and it should have a positive relationship with job market accomplishments. Saks and Waldman (1998) found that undergraduate GPA was significantly correlated with performance and promotability in case of newly hired entry-level accountants and played a role in determining their earnings.
Career or labor market success has been defined objectively and subjectively. Objective career success includes ‘observable career achievements which can be measures, such as pay and promotion rates’. Subjective labor market success is characterized as ‘an individual’s feelings of accomplishment and satisfaction with his career (Wayne, Liden, Graf, & Kraimer, 1999). Starting salary in this case is operationalized as the payment received when acquiring a particular type of job for the first time after obtaining a degree from a higher education institution. Paglin and Rufolo (1990) stated that the amount of a person’s human capital can be held accountable for their efficiency and earnings and they also established that GPA has a powerful ability of success within fields, specifically when looking at earning-maximization. Attention has been brought to the fact that while GPA influences the amount of starting salaries, however, disparities may arise because of the differences between grading standards or regional labor markets and other factors such as internship experience play a particularly strong role in determining starting salaries (Sandiv, Tyran, & Ross, 2005). In Pfeffer’s (1977) research it was identified that GPA and socioeconomic background mainly influence starting salaries, but the effect of GPA becomes less significant after acquiring a master’s degree. When conducting research on the starting salaries of computer science graduates, Dowlatshahi (1994) found that GPA is a significant indicator for employers, but his most important finding was that with the same degree and same GPA females received a lower starting salary
than males, hence gender is going to be included as a moderating variable in this study. Overall, based on previous academic research it can be said that GPA is a valid instrument to measure academic achievements and starting salaries are an adequate construct to measure labor market success directly after graduation to answer the research question to what extent GPA influences starting salaries.
4. Research Design & Conceptual Framework 4.1. Study Selection Process & Inclusion/Exclusion Criteria After conducting a literature review, search terms have been identified which are related to GPA and starting salary. The search strategy was to look for literature conducted between 1975 and 2017 written in English. The search engines used were: University of Amsterdam online library, Google Scholar, Web of Science and Science Direct. These databases were chosen because they accommodate published studies and Peer-reviewed journals, which constitute as reliable sources. Moreover, they also include unpublished studies that help eliminate the file drawer problem, which is the ‘selective reporting of scientific findings’; referring to the fact that mostly researches with statistically significant results get published (Franco, Malhotra, & Simonovits, 2014). In Figure 2, the search terms and the number of results they yielded for each database are presented. Database Number of Results Search Term UvA Online Library 3196 GPA* Starting Salary* Web of Science 37 917 Grade Point Average* OR GPA* Starting Salary* OR Starting Wage* Google Scholar 24 900 Grade Point Average* OR GPA* Starting Salary* Science Direct 265 Grade Point Average* OR GPA* Starting Salary* Total 66 278 Figure 2. - Number of Results and Search Terms
The inclusion criteria were that the study must be written in English, has to be a quantitative study, and has to include the Pearson correlation coefficient as a measure of the relationship between GPA and starting salary. Literature reviews and qualitative analyses were excluded, as well as studies that did not focus on starting salaries but rather salary growth or salary in general. A visual representation of how the meta-analysis resulted in consisting of 12 studies can be seen in Figure 3. Figure 3. – Flow Diagram of the Selection Process
4.2. Meta-analysis This research is going to be based on an extensive literature review followed by a meta-analysis, in which 12 studies (N=43 637, between 1977 – 2016) will be included for final coding to help systematically analyze and evaluate the outcomes in order to help to answer the research question whether GPA has an effect on starting salaries. More than 40 published research papers were coded, which was executed with the help of a coding sheet, and in the final analysis only studies, which used Pearson correlation (r) to observe the effect of GPA on starting salaries, were included. The meta-analysis will be performed based on a step-by-step procedure, which observes the effect size and inverse variance weight of each study included in the study (Lipsey & Wilson, 2001).
Effect sizes, which measure the magnitude of the impact of GPA on starting salary; were coded based on the Pearson correlation coefficient (r), which measures the strength of the relationship between two continuous variables, in this case GPA and salary (Rosenthal, 1979). The effect sizes are interpreted using Cohen’s (1977) system: 0.1 is small, 0.3 is medium and 0.5 is large. In order to overcome the problematic standard error formula, the Pearson correlations were transformed with a Fisher’s Zr-transformation. Subsequently, inverse variance weights are calculated, because it is assumed that an effect size based on a larger sample size tends to be a more precise predictor, therefore, studies with a large number of participants should carry more weight in the meta-analytical process (Lipsey & Wilson, 2001). Lipsey and Wilson (2001) also proposed that if a meta-analysis is based on Zr-transformed correlation coefficients, the weight of each study included should equal w=n-3.
Thereafter the data was organized first into an Excel file with the necessary data for the analyses (Appendix 2), later on into an SPSS file and with the help of Wilson’s macros for meta-analysis. For overall analysis, the random effects model is going to be considered because of the small sample sampling error must be taken into account and the assumption of population heterogeneity, since each research included in the meta-analysis provides information about effect sizes in a different population (Field & Gillet, 2010). In
the analysis a 95% confidence interval is going to be used (p<0.05). In order to analyze the effect sizes, the SPSS macros of Wilson are going to be used (MetaES, MetaF, MetaReg).
4.2. Variables
The dependent variable in the analysis is starting salary, which is measured as the euro/dollar amount received per month. The independent variable is Grade Point Average (GPA), which is most commonly measured in percentages, on a 1-4 scale in the USA, 1-10 in the Netherlands, and A-F scale in the UK.
Moderating variables, which positively or negatively facilitate the existing relationship between the dependent and independent variable are Age (quantitative) and Gender (Male or Female). In this study the available moderating variables based on the studies included in the meta-analysis are: Gender, Age Sample Size of a Study (N), Country of Research (USA or Europe) and Publication Year of Study. In the dataset, age, gender, sample size and publication year are treated as continuous variables and country of research as a categorical variable. The average age was coded in each study for the moderator ‘Age’ (N = 5), and the percentage of males for ‘Gender’ (N = 7).
Figure 2 is a visual conceptualization of how the different variables relate to each other. In Appendix 1, the most commonly used and significant variables are summarized based on previous studies. Based on the literature review and theoretical framework, the following hypothesis has been constructed:
Figure 1. - Research Model of the Relationship between GPA and Starting Salary
5. Findings
In this section the findings of the research to the question to what extent GPA influence starting salary are going to be discussed and other variables, which potentially could have a significant effect on starting salary, are also going to be presented. The results of the meta-analysis are going to be explained based on the outcome of 12 studies (N=43637) from 1977 to 2016. Based on these findings; a questionnaire is going to be proposed in Appendix 3, which could be used for future research to gather up-to-date data.. 5.1. GPA First, the relationship between GPA and starting salary was analyzed and based on existing academic research it was hypothesized that a higher GPA has a positive effect on salaries. In Table 1, the outcomes of the analysis are presented.
Distribution Description
N Min ES Max ES Weighted SD
12 -0.009 0.730 0.138
Fixed & Random Effects Model
Mean ES -95% CI +95% CI SE Z p-value
Fixed 0.2914 0.2820 0.3008 0.0048 60.8535 0.0000 Random 0.3582 0.2670 0.4493 0.0465 7.6985 0.0000 Random Effects Variance Component v = 0.023785 Homogeneity Analysis Q df p-value 835.2259 11 0.0000 Table 1. - Basic statistics of the relationship between GPA and starting salary
From the table, it can be seen that the meta-analysis based on 12 correlations and effect sizes presents a considerable amount of variation, the lowest effect size being equal to -0.009 and the largest is 0.730, which according to Cohen, falls into the ‘large’ category. As mentioned before, the outcomes of the random effects model are analyzed, since it cannot be assumed that studies with identical populations are used for the meta-analysis (Borenstein, Hedges, Higgins, & Rothstein, 2009). Hence, the average effect size of the random effect model is 0.3582, which indicates a medium effect size. Looking at the homogeneity analysis, if the null hypotheses is that homogeneity is assumed, based on Q=835.2259 and p<0.01, the null hypotheses can be rejected. Since the null hypothesis of homogeneity is rejected, heterogeneity can be presumed which indicates that within studies, moderating variables should be considered which help to better explain the relationship between GPA and starting salaries. In the following section possible moderators are going to be inspected.
5.2. Moderator Analysis In this section, a moderator analysis is going to be conducted to further examine whether the country of research, age, gender, sample size and the year in which the research was conducted alleviate the relationship between GPA and starting salaries. 5.2.1. Country
The first moderator analyzed is going to be the country the research was conducted in, the outcomes of this ANOVA analysis can be found in Table 2. Table 2. - ANOVA test on the moderating effects of the Country where the research was conducted The country of research was categorized into two groups: (0) USA or (1) Europe. From Table 2, it can be concluded that there is no significant variance between the means of the studies conducted in the two different countries (p=0.6844). The mean effect sizes between the two groups are also not significantly different (USA=0.3454, Europe=0.4008). Analog ANOVA Table Q Df p Between 0.652 1 0.6844 Within 9.9331 10 0.4464 Total 10.0983 11 0.5216 Q by Group Group Qw Df p-value 0 6.2471 8 0.6196 1 3.6860 2 0.1583 Effect Size Results Total
Total Mean ES SE -95% CI +95% CI Z p-value k
0.3597 0.597 0.2428 0.4766 6.0292 0.0000 12
Effect Size Results by Group
Group Mean ES SE -95% CI +95% CI Z p-value k
0 0.3454 0.0693 0.2096 0.4811 4.9863 0.0000 9
5.2.2. Sample Size
Based on the outcomes of the moderator analysis for sample size, it appears that it does not moderate the relationship between GPA and starting salaries (p>0.05). Based on the homogeneity analysis it can be inferred that the model is not significant (p=0.3973, >0.05) and only a very small part is explained by the sample size. However, this could be due to the fact that only 12 studies were included with a broad range of sample sizes, with the smallest being 55 and the largest being 13 984. Descriptives Mean ES R- Square k 0.3691 0.0570 12 Homogeneity Analysis Q df p-value Model 0.7166 1 0.3973 Residual 11.8456 10 0.2955 Total 12.5621 11 0.3229 Regression Coefficients
B SE -95% CI +95% CI Z P Beta
Constant 0.3965 0.0693 0.2606 0.5324 5.7180 0.0000 0.0000 N 0.0000 0.0000 0.0000 0.0000 -0.8465 0.3973 -0.2388 Table 3. – Moderator Analysis on Sample Size (N)
5.2.3. Age
The third moderator tested was age, based on the average age in a sample. Out of the 12 studies included in the meta-analysis, only 5 had information included about the mean age of participants. Due to the high p-value, (p = 0.4589, >0.05) it appears that age does not moderate the association between grade point average and starting salary. From the homogeneity analysis it can be seen that the model is not significant (p=0.4589, >0.05) and most of it is not explained by ‘Age’, only 0.5486 of the total 5.6139. In Table 4, the results of the moderator analysis are presented. According to Saks and Waldman (1998), who observed entry-level accountants, the results were antagonistic. They argued that in every case, the employer should research and understand why an employee may be applying to at younger or older age for a position with a certain age stereotype. A positive reason could be that someone spent a lot of time working in internship positions, which has added a lot to their professional experience, despite entering the job market at a later age. The number of previous jobs could be negatively related to age, since having a large number of jobs at a relatively young age could indicate that someone is unable to adapt to a firm’s organizational culture or they are unsure of their career goals. Therefore, it seems that several aspects have to be considered when including ‘Age’ as a moderating variable and interpreting the results. Descriptives Mean ES R- Square k 0.3624 0.0977 5 Homogeneity Analysis Q df p-value Model 0.5486 1 0.4589 Residual 5.0654 3 0.1671 Total 5.6139 4 0.2299 Regression Coefficients
B SE -95% CI +95% CI Z P Beta
Constant -0.0856 0.6139 -1.2888 1.1175 -0.1395 0.8890 0.0000 Age 0.0169 0.0229 -0.0279 0.0618 0.7407 0.4589 0.3126
5.2.4. Gender
Running Wilson’s macros on ‘Gender’ to see whether it moderates the relationship between GPA and starting salaries has not yielded significant results. The model is insignificant with a high p-value of 0.6176 and ‘Gender’ only explains 0.2493 out of 7.1341. With a p-value of 0.6176, it appears that gender does not mitigate the relationship between grade point average and salaries. In Table 5, the outcomes of the analysis on ‘Gender’ as a moderating variable are shown. Descriptives Mean ES R- Square k 0.4313 0.0349 7 Homogeneity Analysis Q df p-value Model 0.2493 1 0.6176 Residual 6.8848 5 0.2294 Total 7.1341 6 0.3086 Regression Coefficients
B SE -95% CI +95% CI Z P Beta
Constant 0.5826 0.3136 -0.0320 1.1972 1.8578 0.0632 0.0000 Gender -0.2064 0.4135 -0.0169 0.6040 -0.4993 0.6176 -0.1869 Table 5. – Moderator Analysis on Gender
5.2.5. Publication Year
Amongst the four other moderators (Gender, Age, Sample Size, Country of Research), the year of publication seems to be the least significant. The low R2 of
0.0005 means that only 0.05% of the regression model is explained by the variable ‘Publication Year’. The model also has an extremely high p-value (p=0.9397) and only represents 0.0057 out of 11.8741. The p-value of the regression coefficient ‘Year’ is remarkably high p=0.9397), indicating that the year of publication of a certain study does not have an effect on how the relationship between grade point average and starting salary is perceived. The results of the moderator analysis on ‘Publication Year’ can be found in Table 6. Descriptives Mean ES R- Square k 0.3593 0.0005 12 Homogeneity Analysis Q df p-value Model 0.0057 1 0.9397 Residual 11.8684 10 0.2940 Total 11.8741 11 0.3732 Regression Coefficients
B SE -95% CI +95% CI Z P Beta
Constant 1.0531 9.1704 -16.9208 19.0270 0.1148 0.9086 0.0000 Year -0.0003 0.0046 -0.0093 0.0087 -0.0757 0.9397 -0.0220 Table 6. – Moderator Analysis on Publication Year To sum up the findings, it is clear that the relationship between GPA and starting salary is significant; however, moderating variables require further statistical analysis, since none of them have been suggestive of having an effect on this relationship.
6. Discussion In the following section, the implication of findings about the relationship between GPA and starting salaries and the role of moderating variables is going to be discussed. Based on existing academic research it is also going to be discussed what the underlying reasons could be that the relationship between GPA and starting salaries is not fully explained by the moderating variables included in this research.
Most studies point out that there are relatively few studies existing about the question how GPA influences starting salaries, hence more research needs to be done. First, findings that support the hypotheses that higher GPA leads to higher starting salaries are going to be evaluated. Roth and Clarke (1998) conducted research on how grades affect starting salaries and salary growth based on correlation analysis within an 80% confidence interval. One of the most important findings was that salary is correlated to field, but there are differences across industries, for example GPA is more relevant for engineering students (r=0.35, moderate) than for business students (r=0.17, low). They found a modest relationship between starting salary and GPA (r=0.20) and a negligible correlation between salary growth and GPA (r=0.05). This could indicate that that GPA is not an important factor besides when entering the job market; nonetheless the authors argue that it is an unfair judgment by an employer to assess someone’s future performance based on a 3-4 year period summarized by a single number. These results are mostly in line with the findings of the meta-analysis, however, the effect size between GPA and starting salary was found to be higher (r=0.3582).
The difference between the correlations could mean that the different fields of studies also have different moderating variables; hence this could be a reason why there was no support provided for the moderating variables in this analysis. In the future, it would be advantageous to group studies together which have been conducted in the same job market (e.g. business, engineering, medicine) to see if there are any differences between the significances of moderators between the disparate areas. Baird (1985) has confirmed this by conducting a research resulting in the conclusion that
engineering and medicine are more likely to be associated with GPA than other fields.
When considering what factors can contribute to a higher GPA, the role of giving access to higher paying fields (law, medical professions) has to be considered, hence there might be bias when grading students, therefore the university itself, the subject and the person who awarded the grade should be taken into account (Roth, & Clarke, 1998).
Another circumstance to assess when evaluating what contributes to a higher GPA is working while studying. Gleason (1993) found that 47% of 16-24 year old university students were working for an average of 20 hours per week. Working only had a negative effect on GPA when a student worked for more than 40 hours a week, which is considered as a full-time job besides studying. Students who worked between 11-20 hours had the highest average GPA of 2.94, as opposed to non-working students with a GPA of 2.64. Mounsey, Vandehey and Diekhoff (2013) had similar findings, and they said that employers evaluate the GPA of students who were working while studying in two ways: either as more motivated and experienced future employees or they think that they took easier classes and studied less because of time constraints. Hence, the number of hours worked while completing a university degree could be added as a moderating variable in the model observing the relationship between GPA and starting salaries.
When looking for evidence for the human capital theory and screening hypothesis (in case of asymmetric information, the employer attempts to balance this asymmetry by trying to learn as much about a potential employee as he can) with regards to starting salary, a significant positive relationship has been established. According to the screening hypotheses, a higher GPA would indicate better inherent abilities and employers convey that a higher GPA means a faster rate of skills acquisition (Jones, & Jackson, 1990). In a sample of 811 already employed business administration graduates GPA was confirmed to be a significant determinant of earnings, a 1 point increase in GPA meant an 8.9% increase in annual earnings regardless of a firms level or worker investment (Jones, & Jackson, 1990). Weisbrod and Karpoff (1968) also completed research on human capital theory, and they found that return on investment in education
is going to be the largest when one graduates on top of their class. They found a $30 000 annual salary difference between students who finished in the top 1/10 and bottom 1/3 of their class. Wise (1975) had a similar finding; he said that workers with a bachelor’s degree had a higher introductory salary than those who entered the job market before obtaining a degree, however, a master’s degree only has an effect on the level of salary if the student graduated at least in 1/3 of their class. Therefore, it seems like achieving a higher GPA had higher returns on investment.
However, Sirin (2005) conducted a meta-analysis on how social status influences academic achievements, and it was concluded that socio-economic background has a major influence on higher educational performance, hence on GPA as well. As a consequence, the role of human capital theory is questionable, since there might be people who would be performing excellently academically, but their socioeconomic status simply does not let them make an investment in education, which could be a hindrance for them when trying to achieve labor market success. This theory is validated by the findings of James et al. (1989), who found that GPA acts as an intermediary for human capital and a higher GPA can lead to an annual 9% rise in earnings, nevertheless, people who could not invest in higher education need a different assessment method to judge their abilities.
Dowlatshahi (1994) observed that even though females performed better at university, they still received lower salaries than their male counterparts in the same job position, on average $1397 less per month. Garcia, Hernández and Lopez-Nicolas (2001) have conducted research on the gap between gender-wage differences and they have reached the conclusion that personal characteristics and the participation decision of women account for more in achieving labor market success, rather than academic achievements, such as grade point average. The earnings difference between genders is an immensely debated topic nowadays; hence ‘Gender’ appearing as insignificant in this research, this construct should be further analyzed as a moderating variable in the relationship between higher education outcomes and labor market achievements.
Besides ‘Gender’, ‘Race’ appears to be a moderating variable, for which unfortunately there was not enough data provided to include the variable in the moderator analysis. Loury and Garman (1995) have concluded extensive research on what determines starting salaries, and they found support that white males earned more than black males in the same positions, they achieved higher years of schooling mostly due to having higher parental income. This finding is in line with the above-mentioned theory (Sirin, 2005), that socioeconomic background is also a big contributor to the level of job market achievements. Many researchers, Becker (1994) being one of them, argued that going to university does not increase productivity and motivation is not reflected by the GPA obtained, because students who decide to pursue an academic career are already more motivated than their peers who finish their education after high school. He also said that university GPA does not convey much about a student because the atmosphere highly differs by country, for example in the Unites States it is very flexible, individualistic and rather undisciplined compared to most European universities. He also claimed that companies put more emphasis on capabilities and achievements in the context of working rather than academic achievements, and nowadays on-the-job training is taken so seriously that the investment in that might match the employee’s investment in education (Becker, 1994). Consequently, based on Becker’s findings, higher educational achievements are not a valid indicator of future job performance and they are also not determinants of labor market success, therefore new dimensions are needed to measure that. In 1975, Wise has already observed that physical ability, persistence, creativity, leadership capability are all valid constructs for future employers to better decide whether the applicant would thrive in the position they are competing for.
Griffith and Rask (2014) observed the peer effects in higher education, namely the effect of roommates on the performance of university students. They investigated how academic ability, academic characteristics and a roommates ability influences the performance of a student and their mental ability to perform well. At a 10% significance level in a large sample of two different universities (N=3886 and 3334) they found that a roommates skills had a significantly positive effect on males but not on females, and they also
concluded that aided minority students performance was impacted the most by their roommates academic skills and achievements. Schmidt and Hunter (2004) have also conducted research on to what extent general mental ability (GMA) influences job performance and labor market success. They pronounced that general mental ability is a better predictor than work experience and most other measures, such as GPA, so it would be an interesting area to research how GMA as a non-quantitative performance indicator is related to starting salaries.
When considering what could determine labor market success other than higher education outcomes, Woo (1983) found that the amount of earnings related to an organizations size and structure. He found that initially a degree may help in finding a better starting position, but it does not enhance a workers productivity and having a master’s degree only has a 1-2% influence on earnings, and it varies by the type of employer. This difference is also known as the ‘size-wage effect’ (Sandvig et al., 2005) and is an interesting and potentially influential circumstance to observe; hence in future research it could be included as either a moderating or controlling variable. To sum up, there is still a lot of discussion going on whether GPA is a valid and reliable construct to measure an employee’s skills. According to the results of this research, GPA is perceived as a credible benchmark when determining starting salaries, however, there are several moderating variables that should be taken into account. For example, Gender and Age have appeared to be insignificant based on the meta-analysis that has been carried out; nonetheless previous research has shown that depending on the professional field being analyzed, they could have a significant effect. In conclusion, it appears that there is a positive relationship between grade point average and starting salaries, however, there are many new non-quantitative and non-categorical aspects that have to be taken into consideration when examining the connection between higher educational achievements and labor market success nowadays, such as general mental ability, cognitive skills, motivation and leadership characteristics.
7. Limitations
Although the research has reached its aims, due to the scope of this study, there are some limitations that need to be addressed. First of all, there are not many in-depth studies about the relationship between GPA and salaries, so the small sample size (N=12) used in this research could be a drawback when trying to generalize the results. Another limitation could be that some of the studies are not very recent, which makes the results of them questionable, whether they are still applicable today. This limitation could be overcome by trying to find more studies from a longer time period; so more results can be compared in order to get a more detailed answer. The studies analyzed were all published papers, so according to Rosenthal (1997), publication bias could occur, which means that studies with non-significant statistical findings often end up unpublished, which could lead to an inordinately positive association about the relationship between two variables.
The fact that the model used only focused on salary as a labor market outcome and success measure could also limit the scope of the research. As Jencks et al. (1989) suggested, non-monetary characteristics have to be taken into account when assessing labor market success, therefore a more complex model should be developed which includes other significant measures of labor market success, for example job satisfaction, flexibility, individualism, salary growth, promotions, job benefits. Moreover, other suggested variables such as work experience, age and field of study should be addressed.
A third possible limitation is that only studies from Europe and the United States of America were analyzed, hence it should be considered that there might be underlying differences between countries in order to make the results more generalizable.
In several studies, Age and Gender have appeared as significant moderators. Adding more studies to the sample and analyzing whether that adjusts the results could extend this research.
8. Future Research
While conducting meta-analysis can provide meaningful insights into a certain topic, more research should be done in order to triangulate the results. Based on the questionnaire constructed, real-time data could be obtained and a more complex regression model could be constructed as well. In a more complex regression model more mediating variables should be included, for example the country of origin, different fields or industries, ranking and type of university; as well as moderating variables, such as age and gender. A comparison could be made across different types of jobs, observe the level of job, type of courses, tuition fee taken and see how these variables affect starting salaries.
Conducting a longitudinal study could provide useful understanding of how GPA is relevant through one’s career, to what extent it influences starting salaries at different workplaces and what other factors determine labor market success. Besides longitudinal studies, experiments could be designed to observe how different employers perceive GPA in the hiring process and how it influences their decisions when evaluating recent graduates’ CVs. Several levels of GPA, different studies, experience could be included and it could be tested whether GPA is an aspect that significantly influences the outcomes of a job interview. It could also be an interesting option to explore the differences between bachelor and master degree students’ starting salaries. With these methods, primary studies can be created which are more up-to-date, so it could help to triangulate results, which are based on secondary data.
9. Conclusion
The aim of this research paper was to examine the relationship between grade point average and starting salaries and determine possible moderating variables. In conclusion, based on the meta-analysis conducted and existing academic literature, it can be assumed that there is a positive relationship between a higher GPA and starting salaries (r= 0.3582), but sample size, the year of publication, age and gender do not have a moderating role in this association. However, there are some conflicting outcomes in previous research; and even the studies included in the meta-analysis have shown a great variance in effect sizes, ranging from -0.009 to 0.730. After conducting a moderator analysis on available variables, it can be concluded that the year of publication of a study, the sample size, age, gender and the country the study was conducted in does not have an effect on the relationship between GPA and starting salaries. Nevertheless, after analyzing and coding several studies, it could be assumed that work experience, age, gender and the field of study should be included in forthcoming research as moderating variables.
Despite the limitations mentioned above (small sample size, publication bias, generalizability), this study has contributed to academic research by meta-analyzing several studies from a wide timeframe (1977-2016) and resulting in the conclusion that a higher GPA is positively associated with higher starting salaries. Furthermore, it has given prominence to the fact that in future research, either different definitions of higher educational outcomes and labor market success are needed, or different moderating variables have to be included in the models being tested.
10. Appendices
10.1. Appendix 1: Results of the Literature Review
Author(s) Main Variables Conclusions
Supportin g Wise (1975) Only white males hired after 1945, Salary, Environment, Socioeconomic background, Position in the firm, College attended, Non-academic activities Human Capital Theory – Education is an investment and employers take GPA into account Jones & Jackson (1990) Gender, Job Characteristics (Location, Establishment Size, Occupation within the Firm), Human Capital (Experience, Graduate Degree Status) GPA can be a signal of productivity and motivation, therefore employers pay close attention to it Roth, BeVier & Schippmann (1996) Education Level, Years between obtained GPA and performance, Source of information (Boss, Expert, Supervisor), Type of organization The actual effects of GPA were more relevant after one year on the job as opposed to when only starting Roth & Clarke (1998) Field, Number of Universities, Type of Salary, Type of Degree, Source of Information The effect of GPA might be highly dependent on the field; their research confirmed a modest correlation between starting salaries and GPA. O p p osi n g Bretz (1989) Undergraduate GPA, Graduate GPA, School, Age, Hours worked while in school Inconsistent results amongst groups, the most he found that GPA could be affecting salaries after completing an MBA degree. Samson, Weinstein & Walberg (1984) Area of Study, Geographic Area within US, Gender, Major field in college, College Type, Source of Data Analyzed different fields of work, but their results were conflicting amongst the different groups. Ferris (1982) Only used men working as auditors, Level of Educational Attainment, Educational Institution The result was that having a degree is more important and meaningful than the actual GPA obtained.
10.2. Appendix 2: Data Used in the SPSS Analysis
Author(s) Study N Year Country r Zr W Gender Age
Griffith, A. L., & Rask, K. N. (2014). 1 3886 2014 USA (.0094) (.0094) 3883 49% M 51% F X Jimeno, J. F., Lacuesta, A., Martínez-Matute, M., & Villanueva, E. (2016) 2 9678 2016 Europe 0.225 0.229 9673 47.4% M 52.6% F 35.5 Mason, G., Williams, G., & Cranmer, S. (2009) 3 3589 2009 Europe 0.345 0.361 3586 100% M 23.56 Roth, P. L., & Clarke, R. L. (1998) 4 1238 1998 USA 0.20 0.203 1235 X X Roth, P. L., BeVier, C. A., Switzer III, F. S., & Schippmann, J. S. (1996) 5 13984 1996 USA 0.27 0.277 13981 X X Jones, E. B., & Jackson, J. D. (1990) 6 811 1990 USA 0.321 0.333 808 69.9% M 30.1% F 24.83 Baird, L. L. (1985) 7 55 1985 USA 0.24 0.245 52 X X Bretz Jr, R. D. (1989) 8 328 1989 Europe 0.73 0.929 325 70.7% M 29.3% F X Donhardt, G. L. (2004) 9 175 2004 USA 0.56 0.633 173 X X Pfeffer, J. (1977) 10 371 1977 USA 0.46 0.497 368 100% M X Saks, A. M., & Waldman, D. A. (1998) 11 1200 2005 USA 0.52 0.576 1197 X 23.71 Sandvig, J. C., Tyran. C. K., & Ross, S. C. (2005) 12 8322 1986 USA 0.47 0.510 8319 75% M 25% F 24.6
11. Bibliography Sources marked with an * have been included in the meta-analysis *Baird, L. L. (1985). Do grades and tests predict adult accomplishment?. Research in Higher Education, 23(1), 3-85. Becker, G. S. (1994). Human Capital: A theoretical and Empirical Analysis with Special Reference to Education. The University of Chicago Press, 2(3), 15-28. Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Fixed-Effect Versus Random-Effects Models. In Introduction to Meta-Analysis (pp. 77–86). John Wiley & Sons, Ltd. *Bretz Jr, R. D. (1989). College grade point average as a predictor of adult success: A meta-analytic review and some additional evidence. Public Personnel Management, 18(1), 11-22. Cohen J. (1977). Statistical power analysis for the behavioral sciences. New York: Academic Press. *Donhardt, G. L. (2004). In Search of the Effects of Academic Achievement in Postgraduation Earnings. Research in Higher Education, 45(3), 271-284. Dowlatshahi, S. (1994). A Statistical Analysis of Graduates’ Starting Salary and Employers’ Selection Criteria. International Journal of Value Based Management, 7, 127-139. Duncan, G. J., & Dunifon, R. (2012). “Soft-Skills” and long-run labor market success. In 35th Anniversary Retrospective (pp. 313–339). Retrieved from https://books.google.nl/books?hl=en&lr=&id=Q-sDKpIGesAC&oi=fnd&pg=PA313&ots=ud8Vxyn2vR&sig=K6lYgMaZgyz5v CuxoumAfp-69F0#v=onepage&q&f=false
Ferris, K. R. (1982). Educational predictors of professional pay and performance. Accounting, Organizations and Society, 7(3), 225-230. Field, A. P., & Gillett, R. (2010). How to do a meta-analysis. British Journal of Mathematical and Statistical Psychology, 63(3), 665–694. Franco, A., Malhotra, N., & Simonovits, G. (2014). Publication bias in the social sciences: Unlocking the file drawer. Science, 345(6203), 1502-1505. Garcia, J., Hernández, P. J., & Lopez-Nicolas, A. (2001). How wide is the gap? An investigation of gender wage differences using quantile regression. Empirical economics, 26(1), 149-167. Gleason, P. M. (1993). College Student Employment, Academic Progress, and Postcollege Labor Market Success. Journal of Student Financial Aid, 23(2), 5–14. *Griffith, A. L., & Rask, K. N. (2014). Peer effects in higher education: A look at heterogeneous impacts. Economics of Education Review, 39, 65-77. Huedo-Medina, T. B., Sánchez-Meca, J., Marín-Martínez, F., & Botella, J. (2006). Assessing heterogeneity in meta-analysis: Q statistic or I² index?. Psychological methods, 11(2), 193. James, E., Alsalam, N., Conaty, J. C., & To, D. L. (1989). College quality and future earnings: where should you send your child to college?. The American Economic Review, 79(2), 247-252. Jencks, C., Perman, L., & Rainwater, L. (1988). What is a good job? A new measure of labor-market success. American Journal of Sociology, 93(6), 1322–1357.
*Jimeno, J. F., Lacuesta, A., Martínez-Matute, M., & Villanueva, E. (2016). Education, Labour Market Experience and Cognitive Skills: A First Approximation to the PIAAC Results. OECD Education Working Papers, 146. OECD Publishing: Paris. *Jones, E. B., & Jackson, J. D. (1990). College grades and labor market rewards. The Journal of Human Resources, 25(2), 253-266. Kupfer, A. (2011). Towards a theoretical framework for the comparative understanding of globalisation, higher education, the labour market and inequality. Journal of Education and Work, 24(1-2), 185-208. Lepak, D. P., & Snell, S. A. (1999). The Human Resource Architecture: Toward a Theory of Human Capital Allocation and Development. Academy of Management Review, 24(1), 31-48. Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Sage Publications, Inc. Retrieved from http://doi.apa.org/psycinfo/2000-16602-000 Loury, L. D., Garman, D. (1995). College Selectivity and Earnings. Journal of Labor Economics, 13(2), 289-308. *Mason, G., Williams, G., & Cranmer, S. (2009). Employability skills initiatives in higher education: what effects do they have on graduate labour market outcomes?. Education Economics, 17(1), 1-30. Mounsey, R., Vandehey, M., & Diekhoff, G. (2013). Working and non-working university students: Anxiety, depression, and grade point average. College Student Journal, 47(2), 379-389. Ng, T. W. H., Eby, L. T., Sorensen, K. L., & Feldman, D. C. (2005). Predictors of Objective and Subjective Career Success: A Meta-Analysis. Personnel
Psychology, 58(2), 367–408. http://doi.org/10.1111/j.1744-6570.2005.00515.x Paglin, M., & Rufolo, M. (1990). Heterogeneous Human Capital, Occupational Choice, and Male-Female Earnings Differences. Journal of Labor Economic, 8(1), 123-144. *Pfeffer, J. (1977). Effects of an MBA and Socioeconomic Origins on Business School Graduates’ Salaries. Journal of Applied Psychology, 62(6), 698-705. Rosenthal, R. (1979). The file drawer problem and tolerance for null results. Psychological Bulletin, 86(3), 638. *Roth, P. L., & Clarke, R. L. (1998). Meta-analyzing the relation between grades and salary. Journal of Vocational Behavior, 53(3), 386-400. *Roth, P. L., BeVier, C. A., Switzer III, F. S., & Schippmann, J. S. (1996). Meta-analyzing the relationship between grades and job performance. Journal of Applied Psychology, 81(5), 548-556. *Saks, A. M., & Waldman, D. A. (1998). The Relationship between Age and Job Performance Evaluations for Entry-Level Professionals. Journal of Organizational Behavior, 19(4), 409-419. *Sandvig, J. C., Tyran. C. K., & Ross, S. C. (2005). Determinants of Graduating MIS Students Starting Salary in Boom and Bust Job Markets. Communications of the Association for Information Systems, 16(29), 604-624. Samson, G. E., Graue, M. E., Weinstein, T., & Walberg, H. J. (1984). Academic and occupational performance: A quantitative synthesis. American Educational Research Journal, 21(2), 311-321.
Schmidt, F. L., & Hunter, J. (2004). General mental ability in the world of work: occupational attainment and job performance. Journal of personality and social psychology, 86(1), 162. Sirin, S. R. (2005). Socioeconomic status and academic achievement: A meta- analytic review of research. Review of educational research, 75(3), 417-453. Wayne, S., Liden, R. C., Graf, I. K., & Kraimer, M. L. (1999). The Role of Human Capital, Motivation and Supervisor Sponsorship in Predicting Career Success. Journal of Organizational Behavior, 20, 577-595. Wise, D. A. (1975). Academic achievement and job performance. The American Economic Review, 65(3), 350-366. Woo, J. H. (1986). Graduate Degrees and Job Success: Managers in one U.S. Corporation. Economics of Education Review, 5(3), 227-237.
