Effect of job level on the performance of human capital attainment: an exploratory analysis.

Choudhury, Askar ; Jones, James

INTRODUCTION

In recent years, researchers have devoted much of their effort in identifying factors that determine earnings differentials (Moore, Newman, and Terrell, 2007; Heitmueller and Inglis, 2007; Lauermann, 2006; Gottschalk, 1997; Hartog & Vriend, 1990; Hartog, 1988; Autor, Katz, and Krueger, 1998; Lord and Falk, 1980). Many factors have been cited (Shen & Deng, 2008; Buddeberg-Fischer, Stamm, Buddeberg, & Klaghofer, 2008; Ng, Eby, Sorensen & Feldman, 2005; Krueger, 1993; Judge & Bretz, 1994; Petersen and Saporta, 2004; Doms, Dunne, & Troske, 1997) as sources of earnings variations as a measure of career success; among these, human capital investment (Bassi & McMurrer, 2007; Carrera, Carmona, & Gutierrez, 2008) plays a very significant role. Becker (1962, 1975) suggested that inequality in income distribution may be explained by the investment in human capital. Human capital theorists' argument is that investment in education and training are important to improve individuals' earnings and thus enhance career success. Political economist Adam Smith believed that ultimate source of a nation's wealth is the quality of its labor force and the disparities in workers' earnings are due to the differences in their human capital investments. Therefore, the earnings disparities between job levels may be due to the differences in human capital investments. However, the differential effect of human capital may not be same at various job levels even with comparable human capital accumulation. In general, individuals at higher job level receive much greater return (Chang and Huang, 2005) on their investments in human capital. Therefore, the concern of income differential due to job level, tie in closely through the performance efficiency associated with the job-level. If this is the case, then we can postulate a hypothesis that comparatively individuals at higher job level perform more efficiently to attain human capital (such as, acquiring further education, certification, training, etc.) than others. In turn, this accelerates their career success and thrust them further upward in their career path and creates a domino effect. This specific nature of efficiency in performance exists primarily at the higher job level that goes beyond basic human capital and may be a result of managerial role motivation theory (Berman & Miner, 1985).

In this study, we propose a hypothetical model to examine the effect of various determinants on the attainment of human capital as a measure of performance efficiency. Specifically, we observe the differential effect due to job-level on the implementation of human capital attainment. This research primarily differs from other studies in that we are interested in finding if human capital is acquired efficiently by individuals at higher job level. Previous studies have tested the effect of human capital on earnings or other career success measures. Our objective is to examine, if performance efficiency is dependent on career success (i.e., job-level) rather than human capital itself. To our knowledge, no research has been done to test the effect of job-level on the performance efficiency. That is, if an individual is successful in their career and achieves higher job level, will that individual be more efficient by exhibiting upsurge in performance.

Chartered Property and Casualty Underwriter (CPCU)

Present study is based on data from the CPCU certified individuals. The system of CPCU (Chartered Property and Casualty Underwriter) professional examinations and designation is the most recognized system in the area of property/casualty insurance, which provides comprehensive integrated, skill and knowledge set in all areas of property/casualty insurance. As with professional designations in other fields, such as the CPA in accounting, the CPCU is awarded to individuals willing to go beyond the normal requirements of their profession. The American Institute for Chartered Property and Casualty Underwriters (AICPCU) confers the CPCU designation. The CPCU designation is earned through the successful completion of eight college-level courses with national essay examinations, an experience requirement, and an agreement to be bound by ethical standards. Curriculum includes risk management, insurance products, insurance operations, financial analysis, and legal and regulatory environment of insurance. Each course is accredited by the American Council on Education (ACE) for at least 3 college undergraduate credits and some for 3 graduate credits. The certification helps practitioners to make sound, ethical decisions in the complex environment of property and casualty insurance. An eight course program is tantamount to completing about 24 hours of college credits (per ACE).

Property/casualty insurance industry in the United States operates in a regulated environment, and within the evolving American culture, consumer markets, and labor force. Thus factors such as, overall educational trends, demographic, litigation, and consumerism influence the insurance industry. Therefore, the need for educated professionals, and ultimately the desire and ability of insurance industry individuals to seek and attain the CPCU designation for diverse knowledge to keep up with the dynamic change in the environment requires further investment to acquire additional human capital.

This paper thus, examines the effect of job level on the performance of CPCU certification a source of human capital. This study's purpose is to determine if the upper job level individuals are more efficient in attaining the CPCU designation. In particular, we examine the effect of job level on two different (faster and slower) categories of completion time (total time for completing the certification program) to observe the performance differential. We control for the age, gender and level of education. After controlling for demographic factors, we find that higher job-level (executive) is instrumental in enhancing the performance. This suggests that individuals at the executive level are more efficient in acquiring human capital. Our results provide solid support of positive contribution by higher job level individuals in performance efficiency. Therefore, our results contribute to the literature by documenting the constructive externalities of job-level differential, and associating systematic efficiency of executives' (job-level) performance with the success outcome.

METHODOLOGIES

A response variable "completion time" is initially created from the length of time that it takes to complete the certification program, sometimes known as "travel time" in the literature. This is further categorized into "success" and "failure" dichotomous variable as an outcome measurement. Success is identified as those 25% with lower completion time (i.e., below first quartile) and failure is associated with those 25% that has higher completion time (i.e., above third quartile). An individual is categorized as efficient in attaining human capital if the person falls in the "success" group. Therefore, if an individual is at higher job level (such as, executive) and also positively associated with "success" then we can assert that the higher job level induces a setting for efficient attainment of human capital.

Our sample period consists of about 3466 completed individuals (i.e., number of individuals who completed the program) record of data. Table-1 presents summary statistics of age, with respect to gender and education; and Table-2 presents percentage distribution of success outcomes by gender and education. Logistic regression analysis was applied to assess the significance of job-level on the performance outcome. Job-level is a categorical variable and is incorporated in the model as a dummy variable to assess the differential effect of human capital attainment. Job levels are classification of positions in an organization occupied by individuals who perform similar activities and are confronted with similar decision making problems. Therefore, individuals at different job levels will most likely exhibit heterogeneity in their performance. Rice and Shook (1990) reported that individuals at different job levels use different communication structure. Individuals at higher job levels are faced with more challenging tasks in their course of action (Hannaway, 1985). In addition to this primary predictor variable, our analysis also included three other independent variables: gender, age, and level of education as control variables. Gender is a binary variable, coded "1" for male and "0" for female. A number of prior studies have investigated the impact of gender as a predictor variable on academic performance. Two earlier studies found that female students performed better than males in accounting area (Mutchler, Turner, & Williams, 1987; Lipe, 1989), while others found males outperforming females in finance (Borde, Byrd, & Modani, 1996), Economics (Dale & Crawford, 2000; Heath, 1989), and in professional certification (Choudhury, Jones, Gamage, & Ostaszewski, 2008; Brahmasrene & Whitten, 2001; Zook & Bremser, 1982). Several studies in the computer arena found that, compared to male, females tend to display lower computer aptitude (Rozell & Gardner, 1999; Smith & Necessary, 1996; Williams, Ogletree, Woodburn, & Raffeld, 1993) and higher level of apprehension (Bozionelos 1996; Igbaria & Chakrabarti 1990). Since, most of the prior researches indicate that an individual's gender may play a role in producing differential results; gender was controlled for in our research.

Another factor that we have included in our study is the level of education to control for background knowledge. Vermunt (2005) observed that, education and learning patterns influence individuals' academic performance. This in turn may affect performance and efficiency of obtaining human capital. Many studies have found grade point average (a measure for intelligence) to be a significant factor for academic performance (Bagamery, Lasik, & Nixon, 2005). These include among others, in MBA (Gropper, 2007), accounting (Doran, Bouillon, & Smith, 1991; Eskew & Faley, 1988; Garcia & Jenkins, 2003), marketing (Borde, 1998), and economics (Bellico, 1974; Cohn, 1972; Dale & Crawford, 2000). In this study, the levels of education differ greatly among the individuals and because their performance on success may be influenced by their level of education, we therefore include education level in our analysis. Education is an ordinal (hierarchical) categorical variable and therefore, kept in its original format ranging from high school diploma to doctorate.

To observe the relationship between the response variable and job-level, we perform two separate analyses. First, we use basic summary statistics (Table 1) and percentage distributions (Table 2) to observe whether the gender difference or education level exhibit any systematic change. Then, we examine pair-wise correlations to assess the direction of association between variables. Second, we regress the success factor (dummy variable "1" or "0", see above for detail) on age (AGE), gender (GENDER), education level (EDUCATION), and job-level (indicator variable). Age is a continuous independent variable. In general, it is assumed that there is a difference between younger and older people in the performance of their human capital attainment. In addition, this factor also reflects most of the experience base knowledge differences. These differences may also relate to individuals' job-level and the experience that they bring with them to the human capital investment environment. Therefore, it is important to use this factor as a control variable to isolate and extract the differential effect of job-level on the attainment of human capital.

Statistical analysis was performed using logistic regression that utilizes maximum likelihood estimation method and the analysis was run using SAS software (SAS/STAT User's Guide, 1993) on these following predictor variables; age, gender, education level, and job-level. Job-level is used to measure the differential effect of job categories, in particular to determine if executives are more efficient than others when it comes to human capital attainment. This measure is designed to test the hypothesis that individuals at higher job level (specifically executives) are more efficient in the process of executing human capital investment.

In logistic regression, the dependent variable is a logit, which is the natural log of the odds, that is,

log(odds) = logit (P ) = ln(P/1 - P) ....................................(1).

So a logit is a log of odds and odds are a function of P, the probability of "success." Where "success" is coded "1" and failure is "0", such that, Prob. (Y=1) = P and Prob. (Y=0) = 1-P. In logistic regression, we find the logit mean response as,

logit (P) = [[beta].sub.0] + [[beta].sub.1][X.sub.1] + [[beta].sub.2][X.sub.2] + ..... + [[beta].sub.k][X.sub.k], ............ (2),

where [X.sub.i] is any predictor variable. Then the log-likelihood function can be expressed as,

[log.sub.e] L([beta]) = [n.summation over (i=1)][Y.sub.i]([beta]'[X.sub.i]) - [n.summation over (i-1)][log.sub.e][1 + exp([beta]' [X.sub.i])], ............(3),

where; [beta]'[X.sub.i] = [[beta].sub.0] + [[beta].sub.1][X.sub.i1] + [[beta].sub.2][X.sub.2] + ..... + [[beta].sub.x][X.sub.jk].

More discussions on the likelihood function can be found in Neter et. al. (1996); Choudhury, Hubata and St. Louis (1999); and Strauss (1992).

RESULTS

Table 1 reports the descriptive statistics of starting age by response variable and various predictor variables. Similar average starting age (about 31 years) across gender and success factor for all education levels combined is observed (see Table 1). This suggests that there is no apparent difference due to age on the performance outcome. Relatively speaking, average starting age decreases as the level of education increases, owing to the time one needs to spend in obtaining the higher education level (see Table 1). There is a great degree of variation in percentage distributions (see Table-2) between genders when considering performance outcome for the "efficient group" (those who obtained the certification faster). More specifically, percentage of males in the efficient group is twice as much as females (32.26% vs. 15.27%). Gender difference is not quite visible when considering the "inefficient group" (i.e., those who took longer time to obtain the certification) only. Shown in Table 3 are simple pair-wise correlation coefficients among the various factors. We found that gender and shorter completion time (success) are positively associated at the 0.02 significance level. Several studies have suggested that gender differences exist in different learning environments. It is possible that gender-bound differences exert influence in the way in which males and females are inclined to learn (Choudhury, Jones, Gamage, & Ostaszewski, 2008; Gallos, 1995; Gilligan, 1982; Richardson, 2000).

Education level is found to be positively associated with shorter completion time (note that, even though simple-correlation is statistically meaningless for these binary variables, these correlations are only an indication of bi-directional association in a simple linear regression setting). This result is consistent with the perception that high-achieving individuals make greater effort in acquiring the necessary knowledge and skill; as a result they may be more competitive in their performance of human capital attainment. Only job category "executive level" seems to be positively associated with the efficient performance. This is consistent with our hypothesis that individuals at higher job level are highly motivated to invest in human capital. As a result, these individuals perform more efficiently in attaining the human capital, such as, CPCU certification in this study.

In Table 4, we report the results of the logistic regression analysis (full model). The proposed model appeared to fit well in estimating success outcomes (a binary response variable). All three reported global p2 test statistics 94.42, 91.37, 85.59 are highly statistically significant at a significance level <0.0001. Results indicate that job-level is a significant predictor of individual's performance measure in attaining human capital as measured by CPCU certification. However, with categorical variables, the effect of a particular category must be measured in comparison with other categories involved. This means that compared to job level of senior management, middle management, professionals, or administrative; executive job level is associated with the increased log odds of success in human capital attainment at significance level of 0.09. Therefore, the job-level, specifically the executive level plays an important role when efficiency is desirable in attaining human capital. Although, two other job levels, namely middle management and professionals are statistically significant (negative estimated coefficients implies odds are less than one) their odds of success in human capital attainment is opposite to those of executives. Moreover, the odds of decreasing success for these job categories stay below one even at the upper 95% Wald confidence limits (see Table-4).

We have also found gender to be a significant factor in this empirical analysis. Specifically, the result indicates that an individual's gender may contribute to the efficiency aspect of human capital attainment and is consistent with other previous studies. Severiens and Ten Dam (1997) reported similar gender effect in undirected learning. The CPCU is primarily a self-study program, nearly two-thirds of the students reported that they self-study. Although self-study could potentially be directed, it is primarily an undirected learning. Another explanation is competing time demand for different gender, which could also account for the difference. Considering that the average age of a CPCU enrollee is 31 years, competing time for family care could be a factor and time spent on family care is well-documented (McGrattan & Rogerson, 2004; Bianchi, 2000). Age is not found to be statistically significant. Therefore, there is no evidence to support that age influences performance. Although, level of education is statistically significant, the magnitude of the odds-ratio does not contribute much to the efficiency of success outcome.

We applied forward, backward, and mixed stepwise methods to select a logistic regression model through Wald statistic, the likelihood ratio statistic, and the Score statistic using significance level as a criterion to add variables into the model or delete variables from the model. All three types of stepwise methods yield the same result, as shown in Table 5. Moreover, the model resulting from the stepwise selection provides the same conclusion that gender, education level, and executive level are significant factors in increasing the likelihood of success. These variables have direct impact on the log odds of success, as indicated by the positive coefficients that resulted in odds of success greater than one. More specifically, one can assert that likelihood of an executive to perform efficiently is twice as much as others. In addition, odds of executives' success can be as high as four times (4.359 at the upper 95% confidence, see Table 5). These findings are consistent with the hypothesis that performance of top executives combined with higher education level generates a stimulus environment for an efficient accomplishment. Therefore, the result of this study suggests that higher job level (specifically executive level) induce a system for an efficient human capital attainment.

DISCUSSIONS AND CONCLUSIONS

Performance efficiency of human capital attainment due to job-level differential is examined in this research. Logistic regression analysis found the predictive power of the job-level on the performance outcome. Despite the differences among individuals education level, performance efficiency is impacted by the job-level. Thus, our estimated logistic regression model indicates that executive job level is a significant factor in increasing the likelihood of success. Although, individuals with higher level of education, as measured by their highest degree earned, has higher odds of efficiently achieving human capital; gender plays a role in the implementation phase of human capital investment. Findings from this study have important implications on the implementation of human capital investment.

This study helps to fill this gap in our knowledge about the differential performance efficiency due to job-level. We examined the determinants of performance in human capital attainment among CPCU certified individuals. Our exploration to the effect of job-level on the human capital attainment resulted in efficient performance by executives. Studies of this kind, the question of causality inevitably arises. Possible causal explanations may fall largely into two categories: a) differences in information processing and b) motivational differences. When new information is presented to learners, it is processed in a severely limited working memory. Overcoming these limitations is enabled by the use of schemas, stored in long-term memory, to process information more efficiently (Kalyuga, Ayres, Chandler, & Sweller, 2003). These schemas may be formed differently in executives. One possible explanation is that, executives are exposed to a broad array of information versus a narrower, but deeper, flow of information for lower level practitioners. Thus, executives may have the advantage of elevated level of "initial learning" because of their broader occupational exposure. This "initial learning", which occurs from their broader exposure, facilitates executives' performance efficiency via "schema", or "anchor" where new information can be placed (Bransford et. al., 1982; Mandler & Orlich, 1993; Barnett & Ceci, 2002). Executives are also more competitive in an occupational sense, more desirous of power, and more motivated to stand out from the group. This strong motivational desire could explain executives' superior performance in acquiring human capital. Therefore, if performance serves to develop motivation, then motivation may assist to cultivate efficiency. Although this research is described as a test of theory, the theory itself has considerable relevance for practice. For example, if a business organization is interested in selecting individuals with executive level potential, then the use of performance measure recommends itself. Therefore, it is imperative that business organizations include performance efficiency criteria in their descriptions for executives. This finding would be especially important in post-modern times where corporations have continued to seek efficiency.

REFERENCES

Autor, D.H., L. F. Katz, & A.B. Krueger (1998). Computing Inequality: Have Computers Changed the Labor Market? Quarterly Journal of Economics, 113(4),1169-1213.

Bagamery, B.D., J.J. Lasik & D.R. Nixon (2005). Determinants of Success on the ETS Business Major Field Exam for Students in an Undergrduate Multisite Regional University Business Program. Journal of Education for Business, 81(1), 55-63.

Barnett, S.M., & S.J. Ceci (2002). When and where do we apply what we learn? A taxonomy for far transfer. Psychological Bulletin, 128(4), 612-637.

Bassi, L. & D. McMurrer (2007). Maximizing Your Return on People. Harvard Business Review, 85(3), 115-123. Becker, G. S (1962). Investment in Human Capital: A Theoretical Analysis. The Journal of Political Economy, vol. 70(5), 9-49.

Becker, G.S, (1975). Human Capital, 2nd edn. Chicago. IL: University of Chicago Press.

Bellico, R. (1974). Student attitudes and undergraduate achievement for economics majors. Journal of Economic Education, 5(2), 67-68.

Berman, F.E. & J.B. Miner (1985). Motivation to Manage at the Top Executive Level: A Test of the Hierarchic Role-Motivation Theory. Personnel Psychology, 38(2), 377-391.

Bianchi, S.M. (2000). Maternal Employment and Time with Children: Dramatic Change or Surprising Continuity? Demography, 37(4), 401-414.

Borde, S.F. (1998). Predictors of student academic performance in the introductory marketing course. Journal of Education for Business, 73(5), 302-306.

Borde, S.F., A.K. Byrd & N.K. Modani (1996). Determinants of student performance in introductory corporate finance courses. Presented at the Southern Finance Association Annual Meeting, Key West, Florida.

Bozionelos, N. (1996). Psychology of computer use: XXXIX. Prevalence of computer anxiety in British managers and professionals. Psychological Reports, 78, 995-1002.

Brahmasrene, T., & D. Whitten (2001). Assessing Success on the Uniform CPA Exam: A Logit Approach. Journal of Education for Business, 77(1), 45-50.

Bransford, J.D., B.S. Stein, N.J. Vye, J.J. Franks, P.M. Auble, K.J. Mezynski, & G.A. Perfetto (1982). Differences in approaches to learning: An overview. Journal of Experimental Psychology: General, 111(4), 390-398.

Buddeberg-Fischer, B., M. Stamm, C. Buddeberg, & R. Klaghofer (2008). Career-Success Scale--A new instrument to assess young physicians' academic career steps. BMC Health Services Research, 8:120.

Carrera, N., S. Carmona, & I. Gutierrez (2008). Human capital, age and job stability: evidence from Spanish certified auditors (1976-1988). Accounting and Business Research, 38(4), 295-312.

Chang, W-J.A. & T.C. Huang. (2005). The distinctive effects of earnings determinants across different job levels. International Journal of Human Resource Management, 16(11), 2094-2112.

Choudhury, A., R. Hubata & R. St. Louis (1999). Understanding Time-Series Regression Estimators. The American Statistician, 53(4), 342-348.

Choudhury, A., J.R. Jones, J. Gamage, and K. Ostaszewski (2008). Structural Change in the CPCU Curriculum and its Effect on the Completion Time. Academy of Educational Leadership Journal, 12(2), 95-108.

Cohn, E. (1972). Students' characteristics and performance in economic statistics. Journal of Economic Education, 3(1), 106-111.

Dale, L.R. & J. Crawford (2000). Student Performance Factors in Economics and Economic Education. Journal of Economics and Economic Education Research, 1, 45-53.

Doms, M., T. Dunne, & K. R. Troske (1997). Workers, Wages, and Technology. Quarterly Journal of Economics, 112(1), 253-290.

Doran, B. M., M. L. Bouillon & C. G. Smith (1991). Determinants of student performance in Accounting Principles I and II. Issues in Accounting Education, 6(1), 74-84.

Eskew, R. K., & R. H. Faley (1988). Some determinants of student performance in the first college-level financial accounting course. The Accounting Review, 63(1), 137-147.

Gallos, J. V. (1995). Gender and silence: implications of women's ways of knowing. College Teaching, 43(3), 101-5.

Gilligan, C. (1982). In a Different Voice: Psychological Theory and Women's Development. Cambridge, MA: Harvard University Press.

Garcia, L. & E. Jenkins (2003). A quantitative exploration of student performance on an undergraduate accounting programme of study. Accounting Education, 12 (1), 15-32.

Gottschalk, P. (1997). Inequality, Income Growth, and Mobility: The Basic Facts. The Journal of Economic Perspectives, 11(2), 21-40.

Gropper, D.M. (2007). Does the GMAT Matter for Executive MBA Students? Some Empirical Evidence. Academy of Management Learning & Education, 6(2), 206-216.

Hannaway, J. (1985). Managerial behavior, uncertainty and hierarchy: A prelude to a synthesis. Human Relations, 38(11), 1085-1100.

Hartog, J.K. (1988). An Ordered Response Model for Allocation and Earnings. Kyklos, 41(1), 113-142.

Hartog, J.K. & N. Vriend (1990). Young Mediterraneans in the Dutch Labor Market: a Comparative Analysis of Allocation and Earnings. Oxford Economic Papers. 42(2), 379-401.

Heath, J.A. (1989). Factors affecting student learning--An econometric model of the role of gender in economic education. Journal of Economic Education, 20(2), 226-230.

Heitmueller, A. and K. Inglis (2007). The earnings of informal carers: Wage differentials and opportunity costs. Journal of Health Economics, 26(4), 821-841.

Igbaria, M. and A. Chakrabarti (1990). Computer anxiety and attitudes toward microcomputer use. Behavior and Information Technology, 9(3), 220-241.

Judge, T.A. and R.D. Bretz, Jr. (1994). Political Influence Behavior and Career Success. Journal of Management, 20(1), 43-65.

Kaenzig, R., S. H. Anderson, E. G. Lynn (2006). Gender Differences in Students' Perceptions of Group Learning Experiences. Academy of Educational Leadership Journal, 10(1), 119-127.

Kalyuga S., P. Ayres, P. Chandler, & J. Sweller (2003). The Expertise Reversal Effect. Educational Psychologist, 38(1), 23-31.

Krueger, A. B. (1993). How Computers Have Changed the Wage Structure: Evidence from Micro Data, 1984-1989. Quarterly Journal of Economics, 108(1), 33-60.

Lauermann, J. (2006). Earnings Differentials by Industry: Testing the Theory of Compensating Wage Differentials. Journal of Undergraduate Research, 9, 1-7.

Lipe, M.G. (1989). Further evidence on the performance of female versus male accounting students. Issues in Accounting Education, 4(1), 144-152.

Mandler, J. M. & F. Orlich (1993). Analogical transfer: The roles of schema abstraction and awareness. Bulletin of the Psychonomic Society, 31(5), 485-487.

McGrattan, E. R. and R. Rogerson (2004). Changes in Hours Worked, 1950-2000. Federal Reserve Bank of Minneapolis Quarterly Review, 28(1), 14-33.

Moore, W. J., R. J. Newman, and D. Terrell (2007). Academic Pay in the United Kingdom and the United States: The Differential Returns to Productivity and the Lifetime Earnings Gap. Southern Economic Journal, 73(3), 717-732.

Mutchler, J.F., J.H. Turner & D.D. Williams (1987). The performance of female versus male accounting students. Issues in Accounting Education, 2(1), 103-111.

Neal, D. (1995). Industry-Specific Human Capital: Evidence from Displaced Workers. Journal of Labor Economics, 13(4), 653-677.

Neter, J., M.H. Kutner, C.J. Nachtsheim & W. Wasserman (1996). Applied Linear Statistical Models, New York: McGraw-Hill.

Ng, T. W. H., L. T. Eby, K. L. Sorensen & D. C. Feldman (2005). Predictors of Objective and Subjective Career Success: A Meta-analysis. Personnel Psychology, 58(2), 367-408.

Petersen, T. & I. Saporta (2004). The Opportunity Structure for Discrimination. American Journal of Sociology, 109(4), 852-901.

Rice, R.E. and D.E. Shook (1990). Relationships of Job Categories and Organizational Levels to Use of Communication Channels, Including Electronic Mail: A Meta-Analysis and Extension. Journal of Management Studies, 27(2), 195-229.

Richardson, J.T.E. (2000). Researching student learning. Buckingham: SRHE and Open University Press.

Rozell, E.J., & W.L. Gardner (1999). Computer-related success and failure: A longitudinal field study of the factors influencing computer-related Performance. Computers in Human Behavior, 15(1), 1-10.

SAS/STAT User's Guide. 1993. SASInstitute, Inc, Cary, North Carolina.

Severiens, S. and G.T. Dam (1997). Gender and gender identity differences in learning styles. Educational Psychology, 17, 79-93.

Shen, J. & X. Deng (2008). Gender wage inequality in the transitional Chinese economy: A critical review of post-reform research. Journal of Organisational Transformation and Social Change, 5(2), 109-127.

Smith, B.N., & J.R. Necessary (1996). Assessing the computer literacy of undergraduate college students. Journal of Education, 117(2), 188-193.

Strauss, D. (1992). The Many Faces of Logistic Regression. The American Statistician, 46(4), 321-327.

Vermunt, J.D. (2005). Relations between Student Learning Patterns and Contextual Factors and Academic Performance. Higher Education, 49, 205-234.

Williams, S.W., S.M. Ogletree, W. Woodburn & P. Raffeld (1993). Gender Roles, Computer Attitudes, and Dyadic Computer Interaction Performance in College Students. Sex Roles, 29(7-8), 515-525.

Zook, D. R., & A. G. Bremser (1982). A correlation between the characteristics of candidates and performance on the Uniform CPA Examination. Delta Pi Epsilon Journal, 24(2), 45-52.

Askar Choudhury, Illinois State University

James Jones, Illinois State University

TABLE 1: Summary Statistics of Starting Age

By Gender and Education

EDUCATION LEVEL           SUCCESS OUTCOME

                           0

                         GENDER          ALL

                      F         M

1 Mean              38.64     30.06     37.47
Std                 9.34      5.93      9.34

2 Mean              38.05     40.55     38.59
Std                 8.85      6.51      8.39

3 Mean              30.16     30.08     30.12
Std                 7.27      6.97      7.10

4 Mean              30.83     35.07     33.40
Std                 7.24      8.13      8.03

5 Mean              33.55     40.54     37.95
Std                 3.51      7.90      7.39

6 Mean              38.21     43.02     40.62
Std                 2.83      4.42      4.11

ALL Mean            31.60     31.64     31.62
Std                 7.98      7.79      7.88

EDUCATION LEVEL               SUCCESS OUTCOME

                           1                       ALL

                         GENDER          ALL

                      F         M

1 Mean              31.06     37.98     33.83     35.99
Std                 6.39      9.09      8.07      8.92

2 Mean              35.31     34.56     34.81     37.30
Std                 7.18      6.22      6.41      7.93

3 Mean              30.12     30.23     30.19     30.15
Std                 7.18      6.85      6.96      7.04

4 Mean              32.64     34.07     33.69     33.58
Std                 7.01      7.29      7.22      7.52

5 Mean              34.00     35.98     35.43     36.40
Std                 3.91      6.13      5.63      6.43

6 Mean              30.65     42.46     40.99     40.86
Std                 0.00      11.13     11.12     9.13

ALL Mean            31.09     32.09     31.77     31.69
Std                 7.07      7.42      7.32      7.61

Note:

Success = 0 or 1; success = 0 for candidates who took longer time to
complete the program (top 25%, i.e., above Q3) and success = 1 for
candidates who took shorter time to complete the program (bottom
25%, i.e., below Q1). Education (Level of Education):

High School = 1, Associate = 2, Bachelor = 3, Masters = 4, Law = 5,
Doctorate = 6.

Table 2: Percentage Distribution of Success

By Gender and Education

EDUCATION                  Success Outcome (in %)

                   0                          1

            GENDER             ALL     GENDER             ALL

              F        M                 F        M

1            1.30     0.21     1.51     0.62     0.41     1.03
2            2.47     0.68     3.15     0.55     1.10     1.64
3           17.74    20.75    38.49    10.00    19.32    29.32
4            2.81     4.38     7.19     3.22     8.84    12.05
5            0.68     1.16     1.85     0.82     2.12     2.95
6            0.14     0.14     0.27     0.07     0.48     0.55
ALL         25.14    27.33    52.47    15.27    32.26    47.53

Table 3: Correlation Matrix

FACTORS      SUCCESS       AGE       GENDER       EDUC        EXEC

SUCCESS       1.00        0.02        0.17        0.15        0.08
AGE           0.02        1.00        0.00        0.07        0.07
GENDER        0.17        0.00        1.00        0.15        0.07
EDUC          0.15        0.07        0.15        1.00        0.08
EXEC          0.08        0.07        0.07        0.08        1.00
M_MNG         -0.02       0.05        0.03        0.06        -0.05
S_MNG         -0.01       0.08        0.03        0.01        -0.01
PROF          -0.16       -0.12       -0.05       -0.05       -0.11
ADMN          -0.04       0.03        -0.06       -0.10       -0.02

FACTORS       M_MNG       S_MNG       PROF        ADMN

SUCCESS       -0.02       -0.01       -0.16       -0.04
AGE           0.05        0.08        -0.12       0.03
GENDER        0.03        0.03        -0.05       -0.06
EDUC          0.06        0.01        -0.05       -0.10
EXEC          -0.05       -0.01       -0.11       -0.02
M_MNG         1.00        -0.03       -0.26       -0.05
S_MNG         -0.03       1.00        -0.07       -0.01
PROF          -0.26       -0.07       1.00        -0.11
ADMN          -0.05       -0.01       -0.11       1.00

Note: Job levels are introduced as indicator variables: EXEC =
Executive, M_MNG = Middle Management, S_MNG = Senior Management,
PROF = Professionals, ADMN = Administrative.

Table 4: Logistic Regression [Full Model]

LOGISTIC Procedure

Testing Global Null Hypothesis: BETA = 0

Test            Chi-Square         DF         Pr > ChiSq

Likelihood
  Ratio          94.4251           8            <.0001
Score            91.3669           8            <.0001
Wald             85.5866           8            <.0001

Maximum Likelihood Estimates

                                               Standard
Parameter           DF          Estimate        Error

Intercept           1           -1.0346         0.3389
AGE                 1           -0.00722       0.00735
GENDER              1            0.5520         0.1124
EDUCATION           1            0.3461         0.0775
EXEC                1            0.6589         0.3988
S_MNG               1           -0.5356         0.6039
M_MNG               1           -0.3835         0.1739
PROF                1           -0.5993         0.1210
ADMN                1           -0.5045         0.3966

                   Wald
Parameter       Chi-Square     Pr > ChiSq

Intercept         9.3216         0.0023
AGE               0.9654         0.3258
GENDER           24.1279         <.0001
EDUCATION        19.9179         <.0001
EXEC              2.7301         0.0985
S_MNG             0.7868         0.3751
M_MNG             4.8627         0.0274
PROF             24.5507         <.0001
ADMN              1.6187         0.2033

Odds Ratio Estimates

                  Point       95% Wald Confidence Limits
Effect           Estimate
                                  LCL            UCL

AGE               0.993          0.979          1.007
GENDER            1.737          1.393          2.165
EDUCATION         1.413          1.214          1.645
EXEC              1.933          0.885          4.223
S_MNG             0.585          0.179          1.912
M_MNG             0.681          0.485          0.958
PROF              0.549          0.433          0.696
ADMN              0.604          0.278          1.313

Table 5: Logistic Regression [Stepwise Model]

LOGISTIC Procedure

Testing Global Null Hypothesis: BETA=0

Test                Chi-Square            DF            Pr > ChiSq

Likelihood
  Ratio               91.6832              5              <.0001
Score                 88.8035              5              <.0001
Wald                  83.2802              5              <.0001

Maximum Likelihood Estimates

                                                         Standard
Parameter               DF             Estimate            Error

Intercept                1              -1.3128           0.2602
GENDER                   1              0.5610            0.1118
EDUCATION                1              0.3459            0.0767
EXEC                     1              0.6914            0.3984
M_MNG                    1              -0.3502           0.1724
PROF                     1              -0.5454           0.1177

                       Wald
Parameter           Chi-Square        Pr > ChiSq

Intercept             25.4548           <.0001
GENDER                25.1773           <.0001
EDUCATION             20.3466           <.0001
EXEC                  3.0107            0.0827
M_MNG                 4.1264            0.0422
PROF                  21.4724           <.0001

Odds Ratio Estimates

                       Point        95% Wald Confidence Limits
Effect               Estimate
                                          LCL               UCL

GENDER                 1.752             1.408             2.182
EDUCATION              1.413             1.216             1.642
EXEC                   1.996             0.914             4.359
M_MNG                  0.705             0.503             0.988
PROF                   0.580             0.460             0.730