WAGE DIFFERENTIALS BETWEEN MALES AND FEMALES IN THE INFORMATION AND TECHNOLOGY SECTOR A Thesis by Danielle Ely B.A., Wichita State University, 2008 Submitted to the Department of Sociology And the faculty of the Graduate School of Wichita State University in partial fulfillment of the requirements for the degree of Master of Arts May 2010 © Copyright 2010 by Danielle Ely, All Rights Reserved WAGE DIFFERENTIALS BETWEEN MALES AND FEMALES IN THE INFORMATION AND TECHNOLOGY SECTOR The following faculty members have examined the final copy of this thesis for form and content, and recommend that it be accepted in partial fulfillment of the requirement for the degree of Master of Arts with a major in Sociology. _____________________________ David Wright, Committee Chair _____________________________ Twyla Hill, Committee Member _____________________________ Ravi Pendse, Committee Member iii ABSTRACT This research was conducted to examine the different factors that attribute to the wage gap between men and women in the information and technology sector. Three type of factors were included in the analysis: individual level (e.g. age, education etc.), structural level (e.g. the size of business, job title, etc), and gender level (e.g. sex, occupational sex segregation). The Current Population Survey (CPS) data from years 2007 to 2009 was used for the analysis. Two sample t-tests and OLS (Ordinary Least Square) regression were used to analyze the CPS data. The findings from the analysis indicate that the wage gap out of the full sample of I.T. workers is 85.4%, meaning that women are making about 85 cents to a man’s dollar. Net of other factors, being female will decrease potential earnings by $8066 per year. Women are sorted into inferior, lower paid positions than men in the information and technology sector. The variables that had the largest impact on annual earnings for the entire sample were the individual level factors, followed by structural level factors. iv TABLE OF CONTENTS SECTIONS Page 1. INTRODUCTION ................................................................................................. 1 2. LITERATURE REVIEW ........................................................................................ 2 2.1 Individualist model .......................................................................................... 2 2.2 Structuralist model .......................................................................................... 6 2.3 Gender Model ............................................................................................... 10 2.4 Conceptual Model ......................................................................................... 13 2.5 Hypotheses ................................................................................................... 14 3. DATA AND METHODOLOGY ............................................................................ 15 3.1 Data .............................................................................................................. 15 3.2 Variables ....................................................................................................... 15 3.2.1 Individual level variables ..................................................................... 16 3.2.2 Structural level variables .................................................................... 16 3.2.3 Gender level variables ........................................................................ 18 3.3 Methodology ................................................................................................. 18 4. RESULTS ........................................................................................................... 19 5. CONCLUSION.................................................................................................... 23 5.1 Discussion .................................................................................................... 23 5.2 Limitations .................................................................................................... 27 5.3 Policy ............................................................................................................ 28 6. REFERENCES ................................................................................................... 30 7. APPENDIX ......................................................................................................... 35 v 1. Introduction The information technology (IT) sector has witnessed above average employment growth over the last 40 years (Figure 1A) (Bureau of Labor Statistics 2009). However, males and females have not equally benefitted in this growth. While males’ share of employment in IT has continued to climb, females’ share of employment began to decline in the early 1990s and has only recently begun to increase (Bureau of Labor Statistics 2009). As Figure 1B shows, this instability in females’ IT employment was due to shifts in the types of IT jobs available. In the 1980s females were heavily concentrated in computer operators and support positions (Bureau of Labor Statistics 2009). These positions became increasingly obsolete with the rise of the modern PC. However, recently, females have witnessed increased employment growth in system analyst positions, although lagging males’ growth in these IT occupations (Bureau of Labor Statistics 2009). The IT sector has also seen above average median earnings over the last 40 years and steady increasing income since the early 1990s (Figure 2) (Bureau of Labor Statistics 2009). Females’ earnings were lower than the national median during the 1970s and 1980s largely due to the occupancy of lower IT positions (computer operators and support) but have risen since the 1990s as females have made inroads into the higher paying system analyst related jobs (Bureau of Labor Statistics 2009). The positive trends in employment and earnings within the IT sector have benefited females, providing them employment opportunities that increase their odds of being economically independent. However, females still lag males in both employment and earnings within the IT sector which raises the question of why. Are these 1 differences a result of different decisions that males and females make in terms of the investment in skills and employment choice? Are males and females segregated into different occupations which cause a difference in income or employment growth possibilities? Are the differences attributable to processes of gender stratification such as discrimination and sorting? This study examines why differences in earnings among males and females in the IT sector exist. Three theoretical models have emerged to explain wage differentiation between males and females. Individualist theories, including rational choice, human capital, and comparative advantage, state that wage differences between males and females are due to the different choices in productive skills made by males and females. Structuralist theories, such as dual economy and segmented labor market, argue that the position an individual occupies dictates the income they will receive independent of individual attributes. Gender theories argue that females are systemically devalued in the labor market and sorted into inferior economic positions relative to men. From these three schools of thought a composite model is created to examine difference in earnings among males and females in the IT sector using data from the Current Population Survey (CPS) from the years 2007 to 2009. 2. Literature Review 2.1 Individualist model Rational choice theorists argue that individuals make choices about their actions based on the costs and benefits of completing the action, which are influenced by the desires and beliefs of the individual (Coleman 1986, Parsons 2005). Different choices lead to different skills, varying in marketability, which would be a factor in the difference 2 in wages among males and females. For example, a difference in wages between males and females that is relevant to rational choice theory would be that males concentrate on training or furthering their marketable skill set whereas females concentrate on household or domestic responsibilities. Regardless of the track taken, for rational choice theorists, one’s current state of being is always a consequence of past choices taken. Human capital acknowledges the role of choice but goes one step further in arguing that certain choices lead to increases in productivity that employers find attractive in increasing profits which then are compensated by increased earnings for the worker (Becker, 1992). The commonly cited skill based investment for human capitalist are on the job training and educational attainment. These investments allow the potential employee to be more attractive to an employer, and make the potential employee more likely to obtain a position with the employer. On the job training increases possible financial returns with a smaller private cost for the individual (since job training is provided by the company) than formal education which is largely financed by the individual (Mincer 1962). On the job training is an important source of large increases in earnings because workers are more valuable to the company because they have been given specific knowledge about the company (Becker 1992). This specific knowledge makes the individual more valuable to the company because they will be more skilled and productive than workers who have not had the training (Mincer 1958). The longer an individual is with a company can also increase the amount of on the job training they will receive, so the company is increasing knowledge in its workers who have more seniority (Mincer 1958). This type 3 of training has been said to be as important, or more important, as an extended education, as education is thought of as a beginning, with generalized knowledge whereas on the job training is company specific (Mincer 1962). This on the job training is less likely to be given to employees who are not likely to stay with the company, or to those who are in temporary, short-term employment (Ono and Zavodny 2005). An extended education can increase financial returns for an individual (Mincer 1958). Education, in terms of being human capital, also allows individuals to earn more because it provides knowledge, skills and a means of analyzing issues that are not available to individuals who do not further their education (Becker 1992). Females have benefited greatly from advancing their education and have become involved in traditionally “male” fields such as math, the sciences, economics, law, and medical fields (Becker 1992). Advanced education has also been seen to be beneficial in industry sectors that have changing technologies and advancement in productivity (Becker 1992). Technology is affecting the way different sectors value general skills, and those individuals without these general skills are finding themselves unable to find work (Gould 2002). In the past, these general skills would have varied across professions. For example, a factory worker would have needed to have physical strength and endurance, whereas today they would depend on computers and their ability to oversee the equipment doing the work (Gould 2002). The information and technology sector is one field for which people acquire an extended education in addition to receiving on the job training. This field, like most others, will reward people who have these types of training, as well as those who are older who have more experience and more time to receive the training (Balmaceda 2005). 4 So for the rational choice and human capital theorists, wage difference between males and females are a reflection of different investment strategies. But why do males and females have different investments in human capital? For the individualist, comparative advantage provides the answer. Comparative advantage argues that like countries, individuals will produce labor that is cheapest to the individual so they can earn the maximum amount of profit (Ricardo 1919). The tasks that an individual performs best will be the ones that they will choose to have as an occupation, because they will put forth the least amount of effort in gaining their income (Sattinger 1978). When applied to males’ and females’ wages, comparative advantage describes females’ lower wages as a feature of the amount of labor they spend on housework and childcare instead of paid labor, which causes the decrease in their productivity and their wages (Becker 1985). Regardless whether one is arguing a biological (maternal instinct) or a sex-based division of labor in the home caused by the separation of work and family in modern times, females can provide domestic labor (especially childcare) at lower cost than males. Females’ unpaid labor requires them to find paid work that is flexible in hours, sick leave, etc. and those employers pay them less because of the flexibility that is offered (Becker 1985). Females with children spend less time in the job market due to their household and childcare duties, therefore they have less experience, training, continuous employment and opportunity to advance in a company, leading to the differentiation in wages between males and females (Mincer and Polachek 1974). Experience, also referred to as loyalty when an individual stays with a particular company for an extended amount of time, is highly rewarded, so if females are 5 spending less time with a company, they will not receive the rewards given to males who do not take the time off from working (Anderson et. al. 2009). Human capitalists would make the argument that the majority of females are not choosing to go into the field of information and technology, and that the females who are in the field do not spend a majority of their efforts at work, but at home, and are spending less time in the workforce, therefore decreasing their potential income. A final note on discrimination is warranted before leaving the individualist model. While other theories argue that wage differences are based on discriminatory actions of employers, individualists argue that discrimination would be a cost to employers, ultimately forcing out of business employers who discriminate (Becker, 1971). Employers who paid males higher wages would face labor cost competition from employers who paid males and females equally. Market equilibrium would cause any discrimination that did exist to wither away. So in the absence of discrimination, wage differences must be a reflection of different investments in productivity enhancing skill among individuals. 2.2 Structuralist Model Structuralists argue that wage differences between males and females are from the structures, or company settings, in which they work. For structuralist, organizations are economic hierarchies in which each economic position has a range of income independent of individual attributes. Thus, income is first and foremost a product of the position one occupies (Piore 1973). Structuralism as related to income is based on two different but related dimensions of production—the technical relations of production (dual economy theory) and the social relations of production (segmented labor market 6 theory. The technical relations of production are rooted in the physical infrastructure of the firm and affects all workers. The social relations of production are rooted in the social organization (occupational hierarchies) of the firm and impacts economic position differentially. Dual Economy theory argues that there are two sectors that have emerged over time. One sector is called the monopoly sector (also known as the core) which is a capital intensive, technologically advanced industry with differentiated wage structures, job ladders and on the job training (Tolbert, Horan and Beck 1980; Oster 1979). The other sector is called the competitive sector (also known as the periphery) which is not capital intensive but labor intensive, with low skill jobs and instability in employment (Tolbert, Horan and Beck 1980; Oster 1979). These two sectors have significant differences in earning levels and the composition of the labor force (Beck, Horan and Tolbert 1978). Under this theory, monopoly sector industries are concerned with technical relations of production, which influences their ability to give higher wages through surplus from production (Edwards 1973). Dual economists focus on the organization of production and the industrial structure (Tolbert, Horan, and Beck 1980). These corporations have also integrated their companies to ensure they will have the needed raw materials to make their products, as well as institutionalizing research and development of products to stay ahead of technological developments outside of the firm (Edwards 1973). By having control of the materials needed for production, these monopolistic firms can have a higher profit potential, which is also attributed to by their high market share and their involvement in national and international markets (Edwards 1973). Similarly, in the IT 7 sector, large IT firms have economies of scale that make production more efficient and less costly allowing for support of higher wage structures whereas small IT firms have limited ability to expand production and lower cost making it difficult to support high wage structures. Dual economy theory discusses how large firms in the monopoly sector can influence the competitive sector through controlling the amount of competition in the market (Edwards 1973). The competitive sector is not able to compete with the monopoly sector, due to their involvement in local markets and inability to pay their workers in a higher wage structure because of a low ratio of capital to labor (Tolbert, Horan and Beck 1980; Oster 1979). An example of how this applies to the information and technology sector is if a small corporation such as Ribbit’s Computers (a small computer sales company located in Wichita, Ks) is compared to the Microsoft corporation. Ribbit’s Computers cannot offer employees the pay or benefits that Microsoft can offer due to the smaller amount of surplus from sales and the amount of labor it took to make that capital and the small, localized labor market. Microsoft can offer more to employees because they are a part of national and international markets that allow for greater capital with less labor. The segmented labor market theory argues that there are two main labor markets, primary and secondary markets. These markets are based on the social relations of production, or the occupational position hierarchy that exists within companies and the labor market. The primary market has high wages and credentials required for the positions available (Bygren and Kumlin 2005). The secondary market includes task oriented work and low wages (Bygren and Kumlin 2005). The primary and 8 secondary markets have different sets of rules, different ways of getting information, different skill requirements and different behavior requirements for their workers (Edwards 1973). The primary market mainly consists of white males with progressive careers, where they are given more responsibilities, authority, pay and an increased status the longer they are in the labor force. The primary market also has more employment stability and autonomy in decision making for the employees (Bygren and Kumlin 2005: Edwards 1973). The secondary market tends to be comprised of females, blacks, teenagers, and the urban poor who are denied opportunities of vertical movement (Edwards 1973). The secondary market does not allow the same opportunities as the primary market as there is less employment stability, lower wages and defined rules for decision making (Edwards 1973). There have been indications that the reason females are not equally paid is due to sex discrimination in the primary market (Beck, Horan and Tolbert 1978). Females are more likely to be in the periphery or secondary market, where they earn a lower income than they would in the primary market (Coverdill 1988). When females are in the same position as males in the secondary market, they are still earning less than males (Coverdill 1988). The primary sector of segmented labor markets is focused highly on internal labor markets, inside one specific firm, that reduces the possibility of vertical movement between the two sectors and the possibility of higher wages (Levitan, Magnum, and Marshall 1972; Edwards 1973). Females are less likely to be hired in the primary sector because of the possibility they will take time off for maternity leave or other family obligations, and in the areas with rapidly changing technology it will cost the employer to retrain the females when they come back into the job from leave (Estevez-Abe 2005). 9 As information and technology is a field with rapidly changing technology, it can be assumed that females will be less likely to be hired because of possible interruptions in their career and the possibility that they will leave for time off. 2.3 Gender model Individualist and structuralist theorists see gender as a variable whose possible negative effects can be modified or overcome in the labor market. Gender theorists argue that gender is a process that systemically devalues women and sorts them into inferior economic positions. Gender theorists state that minorities and females become disadvantaged through devaluation and sorting processes that occur on all levels, personal to organizational. Throughout their lives, these groups go through processes of having their skills and attributes devalued in the workforce and are sorted into lower paying jobs, which lead to the economic disadvantage for these females and minorities. The gender theories that address the pay differences among males and females include crowding theory, revolving door theory and job queuing theory. The devaluation of females occurs in many ways, for example through objectification and using females to sell products through sexualized images, the socialization of young children into gender-based roles, and the use of language that favors male pronouns. The division of household labor is a good example of how females are devalued and sorted into inferior positions. Household labor is unpaid and has historically been relegated to females. While males have increased their share of household labor in recent decades, females continue to perform the majority of this unpaid labor. The lack of labor support from males is further exasperated by only few companies providing benefits to reduce this burden, especially childcare. In addition, 10 studies show that within the work setting, females are often relegated to performing domestic labor in the workplace regardless of their occupational position (Greenstein 2000). This unpaid labor creates a disadvantage for females as they are spending a “second shift” of work at home beyond their paid labor shift (Reskin and Roos 1992, Hochschild 1989). Devaluation is often a prelude to sorting. Some of the most common sorting theories applied to wage differences between males and females are Crowing theory, Revolving Door syndrome and Job/Gender ques. Crowding theory is based on the idea that disadvantaged groups, such as females, are crowded into smaller job sectors than males, causing an overabundance of labor (Bergmann 1971). This overabundance causes the pay scale to decrease and the group is unable to obtain higher wages because of the discrimination they face (Bergmann 1971). By discrimination being present, females are crowded into the secondary labor market which influences how much they will earn, the amount of autonomy they have and whether they have a “good” job, one that they prefer to be in. Females are less likely to be in the information and technology sector because, according to crowding theory, information and technology is largely comprised of males making it a primary labor market, which leads to females being under valued in the field, facing discrimination from employers and employees, and feeling forced into another area that will pay less. The “revolving door” theory, introduced by Jerry Jacobs, asserts that events throughout a female’s life influence which career direction she will believe is appropriate for her to take (Jacobs 1989). Gender roles, or sex roles, are taught to males and females at a young age which indicate what occupations are appropriate for males and 11 females (Jacobs 1989). Due to the gendered expectations, females are more likely to choose a job that is socially acceptable, such as teaching, rather than economically useful, such as engineering. For example, the area containing work with computer expertise is dominated by males because it is technological in its nature and females are taught through role expectations that it is not an appropriate field for them to work in (Blackburn and Jarman 2006). Although today females have greater access to traditionally male dominated fields, their turnover from male traditional to gender neutral or female dominated jobs is high. Jerry Jacobs found that for every 11 women who enter a male dominated job, 10 ultimately leave for gender neutral or female dominated jobs (Jacobs 1989). This high turn-over rate for females is a product of several factors. A female being present in a male-dominated field can intimidate the male workers who will then use coercion of different sorts such as mockery or exclusion tactics to force the female out of the position (Jacobs 1989). A female will also begin to have feelings of cognitive dissonance, wondering if it is appropriate to be in a male dominated field when she is a female (Jacobs 1989). Another reason for the high turnover rate could be due to employers having second thoughts about having a female employee, leading them to question the females’ abilities in comparison to the males’ (Jacobs 1989). Both males and females attribute higher levels of information and communication technology use and expertise as a masculine trait (Selwyn 2007). Similar to crowding theory, the revolving door theory claims that the reason females are less likely to be in the information and technology sector is due to discrimination, but because females are the 12 minority in the field, they will either leave voluntarily or be forced out of the position because of their minority status. In the book “Job Queues, Gender Queues” Reskin and Roos (1990) describe the process in which employers and employees rank both the job and applicants. Employers rank potential employees from those with the most desirable individual characteristics to the least desirable individual characteristics. The highest desired quality is to have a male employee and then after all male options are exhausted an employer will take females (Reskin and Roos 1990). Employees also rank the attractiveness of the job and will attempt to get a job that is highly attractive to them, which is true for both males and females (Reskin and Roos 1990). If an employee is not able to get their first choice of job attractiveness, they will continue moving down into less attractive jobs until they become employed (Reskin and Roos 1990). For females, this means that they will not have the opportunity to make as much money as a male, due to being sorted into lower prestige, and paid, positions from the processes used by employers. This applies to the information and technology sector, as well as many other areas, because the most desired worker is a white male, and females are not the first desired choice to be chosen to fill a position. 2.4 Conceptual Model By combining the previous models, a conceptual model (Figure 3) is formed with individual, structural and gender components. Individualists argue that individuals are rational actors and make choices to invest in human capital to increase their productivity, which is expected to increase income. Examples of individual level factors would include age (a proxy for on-the-job training) and education. It is argued that as 13 education increases it would be expected that income would increase. Income is thus a reflection of choices that increases one’s productive capacity. Structuralists argue that organizations are hierarchies of economic positions, each position having a range of income that is independent of individual attributes. Thus, income is first and foremost determined by the position one occupies. Structural level factors would include hours worked, industry, company size and occupational position. These components have both direct and indirect effects on income. Factors reflective of the technical relations of production impact all workers. For example, as company size increases wage structures increase. Factors reflective of the social relations of production impact workers based on their economic position. For example, managers received greater economic return from investment in education than production workers. The gender perspective argues that gender processes systemically devalued females and sorted them into different position than males. Females are consistently sorted into inferior economic positions relative to males and even within the same position receive less economic returns on individual attributes than males. 2.5 Hypotheses: Individual model: 1. As age increases, income will increase, net of other factors. 2. As education increases, income will increase, net of other factors. Structural model: 3. As company size increases, income will increase, net of other factors. 4. As occupational skill position increases, income will increase, net of other factors. 14 Gender model: 5. Females will be in inferior positions. 6. Females will earn less income than males, net of other factors. 3. Data and methodology 3.1 Data The data for this study comes from the Current Population Survey (CPS) March Annual Socio-Economic Supplement, a national probability survey conducted by the Bureau of the Census (Bureau of Labor Statistics, 2009). CPS data are collected from approximately 60,000 households on a monthly basis to obtain national estimates on household, family and person demographics, employment and earnings. Population restrictions for this study include civilians between the age of 18 through 65 who had work experience in the previous year in IT occupations. In order to have sufficient sample size for analysis, this study uses a pooled CPS sample from the years 2007 to 2009 which yields 7,927 respondents. The study restricts the range of annual earnings to eliminate the effects from potential outliers. Annual earnings are restricted to a minimum of $9,012.50 (earnings at minimum wage for one year) and a maximum of $300,000 (only 70 respondents exceeded this earnings amount). The final sample size for this study is 7,233 respondents. 3.2 Variables The dependent variable for this study is an interval measure (in dollars) of annual earnings based on wages, salaries and self-employment earnings. CPS earnings from 2007 and 2008 have been deflated to 2009 dollars. Most scholars log earnings to compensate for their highly skewed nature, but given the sample restrictions, skew is 15 minimized eliminating the need for a logarithmic transformation. In addition, the standardized residuals from the regression models used in the study were normally distributed. For further examination, the study creates both quintiles and centiles of earnings. 3.2.1 Individual level variables Individual level factors include age, education, region and urban/rural residence. Age is an interval level variable that, in this study, was restricted to those between the ages of 18 to 65. Age is used as a proxy for experience. Education is a 5 level ordinal variable measuring whether the respondents has (1) less than a high school diploma, (2) high school diploma or equivalent, (3) some college education or associate degree, (4) bachelor’s degree or (5) advanced college degree such as a master’s, Ph.D. or professional degree. Binaries (0,1) were constructed for each education category for bivariate and multivariate analysis. Since earnings are impacted by the standard of living depending upon location of residence this study uses the region (northeast, midwest, south, west) and urban/rural variables available in the CPS. For IT employed respondents, the midwest region and rural residence have statistically the lowest earnings and binaries (0,1) for each have been created for bivariate and multivariate analysis in order to control for cost of living by location. 3.2.2 Structural-level variables Structural level factors include annual hours worked, hours worked per week, full time status, self employed, government worker, company size, industry, and occupational position. Annual hours worked, an interval level variable, indicates the total amount of hours the respondent worked for the year. Full-time and part-time status are 16 derived from the usual weekly work hours (35 or more hours equal full-time employment). Binaries (0,1) are created for self-employed respondents and respondents who work for a government agencies (federal, state, and local). Company size is an interval level variable that was used to construct binaries of small, medium and large businesses. Small business is defined as businesses with one to 99 employees, medium business is defined as businesses with 100 to 499 employees, and large business is defined as businesses with 500 or more employees. Since earnings are impacted by industrial location, a binary (0,1) is created to indicate whether the respondent works in the goods producing sector (construction, manufacturing and mining). The CPS provides data on ten IT related occupational positions: 1) chief information officer (CIO), 2) software engineer, 3) hardware engineer, 4) computer scientist and system analyst, 5) computer programmers, 6) database administrator, 7) network system administrator, 8) network system analyst, 9) computer support specialist and 10) computer operators. For presentation, these occupations are grouped into two clusters based on education, skill level and functional authority as defined by the ONET classification system (U.S. Department of Labor, 2009). The first cluster includes 1) Chief Information Officer, 2) engineers (software and hardware), 3) programmers and related (computer scientist, system analyst and programmers), 4) network related (network administrators and system analyst), and 5) computer support (support specialist and operators). The second cluster includes 1) CIO and engineers, 2) programmers and network related, and 3) support IT occupations. Binaries (0,1) are created for each of the above clusters for bivariate and multivariate analysis. 17 3.2.3 Gender variables Gender level variables include sex, occupational sex segregation, marital status, having children under 6, minority member and immigration status. A binary (0,1) was created to indicate whether the respondent was female. Occupational sex segregation is the percent of females within a 4 digit occupation divided by the percent of females in the workforce; values under 1 indicate that females are underrepresented, values over 1 indicate females are over represented and 1 indicates males and females are equally represented. Marital status, a nominal variable, contains the categories of married, ever married (divorced, separated, widowed) and never married. Binaries (0,1) were created from the marital status variable for analysis purposes. A binary (0,1) was created to indicate if the respondent has a child under the age of 6. The CPS provides information on the race and ethnicity of a respondent. Traditionally, Caucasians are classified as non-minority and African Americans, Asians, and Hispanics are classified as minority. However, as is usually the case when examining earnings, Asians have earnings that are equivalent to Whites and including them among minorities inflates earnings. So for this study Asians are classified as a non-minority group along with non-Hispanic Caucasians and only African Americans and Hispanics are classified as minority members. A minority binary (0,1) is created for bivariate and multivariate analysis. In addition to race/ethnicity, immigrant status is constructed as a binary (0,1). 3.3 Methodology This study utilizes several different statistical techniques to examine income differences between males and females in the IT sector. Univariate analysis provides 18 summaries and distributional patterns for the full sample and separately by males and females. In cases of comparing males and females, group-means-comparison tests (ttest and anova) are used to identify whether statistically significant differences exist between males and females. Differences are also evaluated based on effect size (.20 or higher). Ordinary Least Square (OLS) regression is used to examine multivariate relationships in order to identify whether factors have an independent effect net of controls and what magnitude in terms of explained variance each independent effect contributes to the estimation models. The saturated regression model is ran against the full sample and separately for males and females. A modified chow test is used to determine statistical difference comparing males’ and females’ unstandardized betas (1.96 and -1.96 or higher). 4. Results Tables 1A through 1C provide univariate and bivariate results that compare males and females in IT occupations. Table 1A shows that females earn $59,609 annually whereas males earn $70,587 annually, a statistically significant difference at the .001 level and creating a pay gap of 84.4%. This pay difference is similar for both median earnings and for full-time workers. Among individual-level factors, Table 1A shows that females are more likely than males to have a High school diploma or less (13.0% versus 8.5%) and less likely to have a bachelors degree (42.9% versus 47.2%). There are no statistically significant differences among males and females within the remaining educational categories, nor by age or region. Females are more likely to live in rural areas than males (7% versus 4.7%). 19 Among structural level factors, Table 1B shows that females worked fewer annual hours (2,053 versus 2,159), weekly hours (40.8 versus 42.6) and were less likely to work full-time (93.9% versus 97.6%). Females were more likely to work within the government sector (16.5% versus 10.7%) but less likely to be self-employed (2.4% versus 4.1%). In regards to company size, females are less likely to be in small businesses (16.8% versus 22.5%) but more likely to be in large businesses (70.1% versus 63.3%). Males and females were equally likely to be employed in medium size companies. Females are less likely to be employed in the higher-income goodsproducing sector than males (13% versus 16%). In occupational positions, females and males have the same likelihood to be a Chief Information officer. However, females are less likely to be an engineer (20.8% versus 26.7%), more likely to be programmers (38.1% versus 35%), less likely to be in network related jobs (12.9% versus 15.8%), and more likely to be employed as computer support (17.2% versus 11.2%). When the five occupational categories are collapsed into 3 categories based on the position’s skill, females are more likely to be concentrated in the lower positions relative to males lending support to hypothesis #5 which states that females will be concentrated in lower economic positions. Females are less likely to be in the high IT group (31.8% versus 37.9%) and more likely to be in the low IT group (17.2% versus 11.2%) relative to males. Among gender-level factors, Table 1C shows that females are more likely to work in occupations that have greater percentages of female workers (0.60 versus 0.55). Females are more likely to be members of a minority group (17.5% versus 13.5%). While there are no differences between males and females for being married, females 20 are more likely to have ever been married (15.4% versus 8.9%) and less likely to be never married (20.7% versus 25.2%). Females are less likely to have children under the age of 6 (18.7% versus 22.1%) but equally likely to have children under the age of 18. Males and females have the same likelihood of being an immigrant. Figure 4 shows the median annual earnings of full time IT workers based on age. Individualists argue that as one acquires age, they acquire experience which is rewarded with higher earnings. In the early ages, males and females start with equitable wages, but as they age income diverges into separate trajectories of earning ending with a pay gap of 68.8%. Table 2 provides univariate and bivariate results based on full time full year IT workers (chosen to keep a uniform sample, and based on the small percentage of the sample that was not full time full year) stratified by educational category. With the exception of the doctoral degree level, females earn less than males for each educational degree level. When collapsed into a category indicating a graduate level degree (master and doctoral), there is no statistically significant difference between males and females in educational attainment, but there is a significant difference in earnings for males and females with graduate degrees with females earning less ($75,000 versus $86,000). When collapsed into a category indicating a college degree, there is a significant difference in educational attainment between males and females and a significant difference in earnings. Table 3 shows the earnings related to occupational position for full time full years IT workers. Although males and females have the same likelihood of being a CIO, females earn significantly less than males ($80,000 versus $86,000). On the other 21 hand, there are several IT occupations where females and males statistically earn the same whether or not they are equally likely to be in the occupational category. Females and males earn the same among hardware engineers, computer system analyst, computer programmers and computer support specialist. However, in addition to CIOs, females continue to earn less than males among software engineers, database administrators, network system administrators, network system analyst, and computer operators. Table 4 shows the ordinary least squares (OLS) regression analysis, regressing income on to the individual, structural and gender level factors. In the full saturated model for the full model, the adjusted r square is .469, indicating that the model can explain almost 47% of the variation in the dependent variable annual earnings. The study hypothesizes that females will earn less income net of other factors, which was supported in Table 4 showing that females earn $8,489 less than males, net of other factors. This negative income effect could be attributed to 3 factors; 1) data error, 2) exclusion of other relevant factors, or 3) discrimination. The study hypothesizes that with increases in age income increases, which is supported by the study. For each year of increase in age there is an increase in income of $700. This holds true for both males and females. The study hypothesizes that as education level increases income will increase, net of other factors, which is supported. Using less than a college degree as a reference group, income increases for each additional educational degree attained. This pattern exists for both males and females. The study hypothesizes that as company size increases, income will increase, net of other factors, which is supported. Using small sized businesses as the reference 22 group, people working in medium business earn $3,438 more than the reference group and people working in a large business earn $7,991 more than the reference group. The relationship between company size and earnings holds true for both males and females within large businesses but only for males in medium size businesses. This study hypothesizes that as occupational position increases income will increase, net of other factors, which is supported. Using computer support workers as the reference group, people working in network related areas witness an increase of $8,607, programmers and related earn an additional $9,825, engineers earn an additional $16,195 and chief information officers earn an additional $20,764 more than the reference group. While this pattern exists for males and females among the CIOs, Programmers, Network and Support positions, for Engineers females earn considerably less than males ($9,349 versus $18,604 above the earnings of computer support positions). Figure 5 shows the partitioning of the shares of unique variance in annual earnings in order to test the 3 models in the full sample, as well as for males and females. In the full sample, structural level factors have the highest explanatory power (48.3%) followed by individual level factors (44.9%) and then gender level factors (6.8%). There is a difference in the male and female models. For males, individual-level factors provide more explanation than females (51.8% versus 46.3%) and for females, structural level factors carry more weight (49.5% versus 46%). Females also have a greater impact from gender level factors than males (4.2% versus 2.2%). 5. Conclusion 5.1 Discussion 23 The IT sector is an important employment option for both males and females as it has above average employment and earnings growth. However, females have lagged males in the IT sector and clearly have different employment and earnings trajectory than males. This study has sought to shed light on the earnings dimension using a unified model approach that incorporates individual-, structural- and gender-level components. The individualist model argues that earnings are a reflection of human capital investment choices which lead to increasing ones productivity and thereby, earnings. Often educational attainment and on the job training are used as proxies for skill development. Accordingly, this study created two hypotheses to test the individual model’s assumption, both of which were supported by findings in Figure 4 and Tables 2 and 4. Figure 4 and Table 4 showed that as age increases income increases supporting hypothesis #1. Figure 4 showed that males and females have different earnings trajectories over their lifetimes with males having a sustained earnings increase throughout the prime work years in contrast to females. One explanation for this pattern of a rising gap in wages could be the exiting from the labor market by females related to child birth. However, most studies show when females exit for childbirth related factors it is for a very small period of time ranging from less than 12 weeks to a maximum of 6 months for a majority of females that return to the work force (Hofferth and Curtin 2006, McGovern et al 2000). A more likely explanation, given the cross-sectional nature of the CPS data, is that females have fewer years of labor market experience in IT occupations than similar age males, leading to the difference in wages by age cohort. However, Table 4 showed that when controlling for other factors, the 24 earnings differences caused by age were equivalent for both males and female. It is likely that the differential earnings trajectories shown in Figure 4 are an artifact of the cross-sectional nature of the data. Educational attainment plays an important role for individualist theorists and is a requirement for employment in the IT sector. While males and females do have different concentrations within educational degrees, table 2 showed that even within similar degrees, females earned less than males except for those with doctoral degrees. However, moving from bivariate (Table 2) to multivariate (Table 4) analysis showed that earnings increases from advances in educational attainment equally benefit males and females supporting hypothesis #2. The structuralist model argues that earnings are a reflection of the economic position one occupies independent of individual attributes. The technical relations of production thesis states that firms with large economies of scale structures (such as large companies) affords the ability to increase profits which can support higher wage structures leading to increase earnings for all workers regardless of their individual human capital (hypothesis #3 which states that as company size increases, income will increase). Similarly, the social relations of production thesis states that earnings are segmented by economic position (such as occupational position) to recruit different labor markets and control risk to structural assets (hypothesis #4 which states that as occupational skill position increases, income will increase). Table 4 showed that as company size increases, earnings increase, supporting hypothesis #3. However, this increase did not hold equally true for males and females. While male earnings increased as they move from small to large companies, females witnessed increased 25 earnings only in large companies. A similar pattern emerged when examining earnings by economic position as shown in Table 4. Support for hypothesis #4, as occupational skill position increases, income will increase, net of other factors, comes from table 4, showing that with each increase in position, there is an increase in income. Supporting hypothesis #5, earnings increased as occupational skill level increased from the lowest paid computer support positions to the highest paid CIOs. Yet, this pattern of increase held true for males and females only among programmers and CIOs. Among engineers, males received a substantial pay dividend over females and among network position females received no additional increase in earnings over computer support positions. Finally, the Gender model argues that gender is a process (in contrast to an attribute) of systemic devaluation and sorting. Females will be sorted into inferior economic positions and even within the same position will receive less economic returns on individual attributes than males. Accordingly this study created two hypotheses to test the Gender model’s assumptions concerning occupational concentration (hypothesis #5) and inferior economic returns (hypothesis #6) both of which were supported by Tables 1B, 3 and 4. As shown in Table 1B, females are more likely to be concentrated in lower skill positions than males lending support to hypothesis #5. Interestingly, females have the same probability of being in the top CIO position as males, but their lower levels of participation among engineers and greater likelihood of computer support places them at a disadvantage. This disadvantage is especially prominent when one considers that this study found that female engineers earn less than male engineers net of controls, meaning they are in positions that are not 26 economically advantageous. Most damaging for females is that simply being female incurs economic cost, reflective of the devaluation thesis, shown clearly in Table 4 and lending support to hypothesis #6. Being female incurs a negative cost of $8,489 net of controls. While it is important to note, as with all multivariate coefficients, that this negative coefficient could be the result of measurement error or misspecification of the model, the negative coefficient can also be attributable to discrimination. 5.2 Limitations As with any research this study has limitations. Even though the CPS collects data monthly, it is not a longitudinal or panel study but a snapshot at one point in time, a cross-sectional survey design. This means that factor effects caused by recency are minimized. For example, if the respondent changed from full-time to part-time employment the week before the survey, the survey will not have had time to measure the negative income effects from the change in work status. While the CPS provides the most comprehensive data on employment and earnings nationally, it is missing some important data elements for examining earnings. The CPS does not provide information on labor history such as how long a respondent has been in the labor force, how many times they have exited and for how long, how many jobs they have had and how long they have worked for their present employer. Entering into IT has many various paths, and while a small majority of the individuals who are in IT went to school and entered the field, there is evidence that many people have employment or education in other areas and then move over to the IT field (Moncarz 2002). This would indicate that age may not be a valid proxy for experience or tenure in the IT field. The different earnings trajectories in Figure 4 are possibly 27 reflective of females’ later entry into IT occupations rather than exits related to child bearing but the CPS does not provide data to measure this effect, however in previous studies there has been evidence suggesting that a person will either enter into IT almost immediately after high school or some college education or the person will have several jobs in different fields before entering into IT (Messersmith et al 2008). Nor does the CPS provide data on domestic settings in which the respondent lives, such as the household division of labor or the management and care of children, which would affect the work hours a respondent is available. 5.3 Policy Without question the IT sector has provided females the opportunity to have positive employment growth and above average earnings. However, while females have the same likelihood of being Chief Information Officers as males, females are still concentrated in lower economic positions. In addition, while the pay differences between males and females are small in most of the IT occupations, females do earn substantially less than male engineers, an IT occupation that garners high earnings within this sector. These findings of representation and earnings can be changed by more aggressively pushing females into engineering fields. Along with other efforts to encourage females into the IT sector, by focusing greater effort on engineering occupations, females are able to increase their representation in higher level positions and reap the financial rewards of higher earnings such positions provide. One way females are currently being encouraged into higher positions in IT is through federal support that is giving money to colleges to increase the number of female students who go into science or engineering programs (June 2010). Another 28 way that females are being encouraged into science and technology fields is through programs such as the Society of Women Engineers (SWE) who, with alliances from other groups, create and maintain interest for young adults (more specifically females) in these fields. The SWE has made several policy implications to encourage females into the STEM (Science, Technology, Engineering and Mathematics) areas, these include providing incentives, encouraging school officials to provide STEM materials to students in K through 12 grades, supporting magnet schools that have the STEM curriculum and creating public-private alliances that will foster interest in the STEM areas (Society of Women Engineers 2006). Incentives such as financial backing make education fields with higher wages more appealing to female students and will encourage females to study and work in these previously male dominated fields. 29 REFERENCES 30 LIST OF REFERENCES Andersson, Fredrik, Matthew Freedman, John Haltiwanger, Julia Lane, and Kathryn Shaw. 2009. “Reaching for the Stars: Who Pays for Talent in Innovative Industries?” The Economic Journal 119, F308-F332. Ang, Soon and Sandra Slaughter. 2004. “Turnover of Information Technology Professionals: The Effects of Internal Labor Market Strategies.” Database for Advances in Information Systems 35 (3), 11-27. Balmaceda, Felipe. 2005. “Firm-Sponsored General Training.” Journal of Labor Economics 23 (1), 115-133. Beck, E.M., Patrick M. Horan, Charles M. Tolbert II. 1978. “Stratification in a Dual Economy: A Sectoral Model of Earnings Determination.” American Sociological Review 43 (5), 704-720. Becker, Gary. 1971. The Economics of Discrimination. Chicago, University of Chicago Press. -------. 1985. “Human Capital, Effort, and the Sexual Division of Labor.” Journal of Labor Economics 3 (1) Part 2: Trends in Women’s Work, S33-S58. -------. 1992. “Human Capital and the Economy.” Proceedings of the American Philosophical Society 136 (1), 85-92. Blackburn, Robert M. and Jennifer Jarman. 2006. “Gendered Occupations: Exploring the Relationship between Gender Segregation and Inequality.” International Sociology 21 (2), 289-315. Brown, Charles and James Medoff. 1989. “The Employer Size-Wage Effect.” The Journal of Political Economy 97 (5), 1027-1059. Bureau of Labor Statistics. 2009. Current Population Survey Annual Social and Economic (ASEC) Supplement [machine-readable data file] / conducted by the Bureau of the Census for the Bureau of Labor Statistics. Washington D.C. Bygren, Magnus and Johanna Kumlin. 2005. “Mechanisms of Organizational Sex Segregation: Organizational Characteristics and the Sex of Newly Recruited Employees.” Work and Occupations 32, 39-65. Coleman, James S. 1986. “Social Theory, Social Research, and a Theory of Action.” The American Journal of Sociology 91 (6), 1309-1335. Coverdill, James. 1988. “The Dual Economy and Sex Differences in Earnings.” Social Forces 66 (4), 970-993. 31 Edwards, Richard C. “The Social Relations of Production in the Firm and Labor Market Structure.” In Labor Market Segmentation (1973) by Richard C. Edwards, Michael Reich, and David M. Gordon Eds. Lexington: D.C. Heath and Company, 3-26. Edwards, Richard C., Michael Reich and David M. Gordon (Eds). 1973. Labor Market Segmentation. Lexington: D.C. Heath and Company. Estevez-Abe, Margarita. 2005. “Gender Bias in Skills and Social Policies: The Varieties of Capitalism Perspective on Sex Segregation.” Social Politics 12 (2), 180-215. Gould, Eric. 2002. “Rising Wage Inequality, Comparative Advantage and the Growing Importance of General Skills in the United States.” Journal of Labor Economics 20 (1), 105-147. Greenstein, Theodore N. 2000. “Economic Dependence, Gender, and the Division of Labor in the Home: A Replication and Extension.” Journal of Marriage and Family 62 (2), 322-335. Hochschild, Arlie R. and Anne Machung. 1989. The Second Shift: Working Parents and the Revolution at Home. New York: Viking. Hofferth, Sandra and Sally Curtin. 2006. “Parental Leave Statutes and Maternal Return to Work After Childbirth in the United States.” Work and Occupations 33, 73-105. Jacobs, Jerry A. 1989. Revolving Doors: Sex Segregation and Women’s Careers. Stanford: Stanford University Press. June, Audrey Williams. 2010. “Colleges look for new ways to help women in science.” The Chronicle, January 29, A1, A8. Levitan, Sar, Garth Magnum, and Ray Marshall. 1972. Human Resources and Labour Market. Labour and Manpower in the American Economy. New York, Harper and Row. McGovern, Patricia, Dowd, Bryan, Gjerdingen, Dwenda, Moscovice, Ira, Kochevar, Laura, and Sarah Murphy. 2000. “The Determinants of Time Off Work After Childbirth.” Journal of Health Politics, Policy and Law 25 (3), 527-564. Messersmith, Emily, Garrett, Jessica, Davis-Kean, Pamela, Malanchuk, Oksana, and Jacquelynne Eccles. 2008. “Career Development from Adolescence Through Emerging Adulthood: Insights from Information Technology Occupations.” Journal of Adolescent Research 23, 206-227. Mincer, Jacob. 1958. “Investment in Human Capital and Personal Income Distribution.” The Journal of Political Economy 66 (4), 281-302. 32 -------. 1962. “On-The-Job Training: Costs, Returns and Some Implications.” The Journal of Political Economy 70 (5) Part 2: Investment in Human Beings, 50-79. Mincer, Jacob and Solomon Polachek. 1974. “Family Investment in Human Capital: Earning of Women.” The Journal of Political Economy 82 (2) Part 2: Marriage, Family Human Capital and Fertility, S76-S108. Moncarz, Roger. 2002. “Training for Techies: Career Preparation in Information technology.” Occupational Outlook Quarterly 46 (3), 38-45. Ono, Hiroshi, and Madeline Zavodny. 2005. “Gender Differences in Information Technology Usage: A U.S.-Japan Comparison.” Sociological Perspectives 48 (1), 105-133. Oster, Gerry. 1979. “A Factor Analytic Test of the Theory of the Dual Economy.” The Review of Economics and Statistics, 61 (1), 33-39. Parsons, Stephen. 2005. Rational Choice and Politics: A Critical Introduction. London: Continuum. Piore, Michael J. “Notes for a Theory of Labor Market Stratification.” In Labor Market Segmentation (1973) by Richard C. Edwards, Michael Reich, and David M. Gordon Eds. Lexington: D.C. Heath and Company, 125-150. Pull, Kerstin. 2003. “Firm Size, Wages and Production Technology.” Small Business Economics 21 (3), 285-288. Reskin, Barbara F. and Patricia A. Roos. 1990. Job Queues, Gender Queues: Explaining Women’s Inroads into Male Occupations. Philadelphia: Temple University Press. Reskin, Barbara F. and Patricia A. Roos. 1992. “Occupational Desegregation in the 1970s: Integration and Economic Equity.” Sociological Perspectives 35 (1), 69-91. Ricardo, David. 1919. Principles of Political Economy and Taxation. Ed. E.C.K. Gonner. London: G. Bell and Sons, Ltd. Sattinger, Michael. 1978. “Comparative Advantage in Individuals.” The Review of Economics and Statistics 60 (2), 259-267. Selwyn, Neil. 2007. “Hi-tech = Guy-tech? An Exploration of Undergraduate Students’ Gendered Perceptions of Information and Communication Technologies.” Sex Roles 56, 525-536. 33 Society of Women Engineers. 2006. “Society of Women Engineers General Position Statement on Science, Technology, Engineering, and Mathematics (STEM) Education and the Need for a U.S. Technologically-Literate Workforce.” www.swe.org. Tolbert, Charles, Patrick M. Horan, and E. M. Beck. 1980. “The Structure of Economic Segmentation: A Dual Economy Approach.” The American Journal of Sociology, 85 (5), 1095-1116. U.S. Department of Labor. 2009. Occupational Information Network (ONET) Washington D.C. Wright, David W. 1992. “Class, Gender and Income: A Structural/ Feminist-Marxist Analysis of Income Determination and the Income Gap.” Ph.D. Dissertation. Purdue University. 34 APPENDIX 35 36 Number of Employed Civilians 0 500,000 1,000,000 1,500,000 2,000,000 2,500,000 3,000,000 3,500,000 1971 1972 1973 1974 1975 1976 f emale male Year Figure 1A Employment in IT Occupations by Sex, 1971 to 2009 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 37 Number of Employed Civilians 0 500,000 1,000,000 1,500,000 Female support Female programmer Female system analyst Male support 1981 1980 1979 1978 1977 1976 2,000,000 Male programmer Male system analyst 1982 2,500,000 Figure 1B IT Occupational Employment by Sex, 1971 to 2009 1990 1989 Year 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1988 1987 1986 1985 1984 1983 1975 1974 1973 1972 1971 38 Median Annual Earning in 2008 Constant Dollars $0 $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 1971 1972 1973 1974 all workers (incl non-IT) female male Year Figure 2 Annual Median Earnings among Full-time Full-year IT Occupations by Sex, 1971 to 2009 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Figure 3 Income Determination Model Gender Structural Individual Income (Wright, 1992) 39 Table 1A Full Sample Values for Workers by Sex Variables: Full Sample Dependent Variables Annual Earnings (mean): $67,745 Annual Earnings (median): $65,000 Annual Earnings Centile (median): 50 Males 1 2 Females $70,587 *** ^ $59,609 $70,000 $57,000 54 39 (stddev) (29889) (30111) (27682) FT Workers Annual Earnings (median): $69,000 $70,001 $60,000 39.32 38.85 40.65 (10.5) (100%) (10.6) (100%) (10.3) (100%) Individual-level Factors: Age (years) High school Dipl or less (0,1) 9.7% (0.3) Some college (0,1) 24.1% 24.4% 23.1% Bachelor degree (0,1) 46.1% (0.5) (0.5) (0.5) Master degree (0,1) 18.4% 17.9% 19.5% (0.4) (0.4) (0.4) Doctoral degree (0,1) 1.8% 2.0% 1.4% (0.1) (0.1) (0.1) % Midwest region (0,1) 22.6% 22.6% 22.6% Rural Residence (0,1) 5.3% (0.2) (0.2) (0.3) Sample N (weighted): 7,233 100.0% 5,361 74.1% 1,872 25.9% (0.4) 2 13.0% (0.3) (0.4) 1 8.5% *** (0.3) = *** p < 0.001, ** p < 0.01, * p<0.05 effect size greater than => .20 40 (0.4) 47.2% *** (0.4) 4.7% *** (0.4) 42.9% (0.4) 7.0% (pay-gap) 84.4% 81.4% 85.7% Table 1B Full Sample Values for Workers by Sex Variables: Full Sample Structural-level Factors Annual hours worked 2,132 (median) 2,080 (434.8) Weekly hours (median) 42.1 40 % Full time (0,1) 96.6% % Government Worker (0,1) 12.2% % Self-Employed (0,1) 3.7% (6.8) (0.2) (0.3) % Small Business (0,1) Females 2,159 *** ^ 2,080 2,053 2,080 (433.1) (429.9) 42.6 *** ^ 40 (6.9) 97.6% *** (0.2) 10.7% *** (0.3) 4.1% *** 40.8 40 (6.4) 93.9% (0.2) 16.5% (0.4) 2.4% (0.2) (100%) (0.2) (100%) 21.0% 22.5% *** 16.8% (0.4) (0.4) (0.4) % Medium Business (0,1) 14.0% 14.2% 13.1% % Large Business (0,1) 65.0% % Goods producing industry (0,1) 15.0% 16.0% *** 13.0% (0.4) (100%) (0.4) (100%) (0.3) (100%) 11.2% 11.3% 11.0% (0.3) (0.3) (0.3) (0.5) % CIO (0,1) % Engineers (0,1) 25.1% (0.4) % Programmers & Related (0,1) 35.8% (0.5) % Network & Related(0,1) 15.1% (0.4) % Computer Support (0,1) 12.8% (0.3) (0.3) 63.3% *** (0.5) 26.7% *** (0.4) 35.0% ** (0.5) 15.8% *** (0.4) 11.2% *** (0.3) (0.3) 70.1% (0.5) 20.8% (0.4) 38.1% (0.5) 12.9% (0.3) 17.2% (0.4) (100%) (100%) (100%) % High [CIO & Engineers] (0,1) 36.3% 37.9% *** 31.8% (0.5) (0.5) (0.5) % Medium [Prog & Network Related] (0,1) 50.9% 50.9% 51.0% % Low [Support] (0,1) 12.8% (0.3) (0.3) (0.4) Sample N (weighted): 7,233 100.0% 5,361 74.1% 1,872 25.9% (0.5) 2 1 (0.2) (100%) (0.3) 1 2 Males = *** p < 0.001, ** p < 0.01, * p<0.05 effect size greater than => .20 41 (0.5) 11.2% *** (0.5) 17.2% Table 1C Full Sample Values for Workers by Sex Variables: Full Sample Gender-level factors: Occupational Sex Segregation 0.56 (0.1) % Minority (0,1) % Married (0,1) Females 0.55 *** ^ 0.60 (0.1) (0.2) 13.5% *** 17.5% (0.4) (100%) (0.3) (100%) (0.4) (100%) 65.4% 65.9% 64.0% (0.5) (0.5) 10.6% % Never Married (0,1) 24.0% % With children under 6 (0,1) 21.2% (0.4) (0.4) (0.4) % With children under 18 (0,1) 43.7% 43.8% 43.5% (0.5) (0.5) (0.5) % Immigrant (0,1) 21.5% 21.5% 21.3% (0.4) (0.4) (0.4) Sample N (weighted): 7,233 100.0% 5,361 74.1% 1,872 25.9% = *** p < 0.001, ** p < 0.01, * p<0.05 effect size greater than => .20 42 8.9% *** (0.5) % Ever Married (0,1) (0.4) 2 1 14.5% (0.3) 1 2 Males (0.3) 25.2% *** (0.4) 22.1% *** 15.4% (0.4) 20.7% (0.4) 18.7% Annual Median Earnings 43 $0 $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $90,000 under 25 25 to 29 30 to 34 35 to 39 Age Cohorts 40 to 44 45 to 49 50 to 54 pay_gap f emale male 55 to 59 Figure 4 Median Annual Earnings of Full-time Full-Year IT Workers by Age Cohort 60 & over 0% 20% 40% 60% 80% 100% 120% Pay-gap 44 2 1 1,298 4,269 Graduate degree (master & doctoral) College degree (bachelor to doctoral) effect size greater than => .20 = *** p < 0.001, ** p < 0.01, * p<0.05 643 1,536 2,971 1,177 120 FTFY HS Diploma or less Some college (incl assoc deg) Bachelor degree Master degree Doctoral degree Education: 963 3,276 424 1,161 2,313 865 97 Male 335 993 219 375 658 312 23 19.3% 65.7% 10.4% 24.0% 46.4% 17.5% 1.8% (100%) Female FTFY 1 19.8% 67.4% *** 8.7% *** 23.9% 47.6% *** 17.8% 2.0% (100%) % of all males 2 21.1% 62.6% 13.8% 23.6% 41.5% 19.6% 1.5% (100%) % all females $85,000 $75,000 $50,000 $55,000 $70,000 $83,000 $90,000 FTFY *** *** *** *** 1 ^ ^ ^ ^ 2 $86,000 *** ^ $77,000 *** ^ $53,000 $56,000 $75,000 $85,000 $90,000 Male Table 2 Educational Attainment and Median Annual Earnings of Full-time Full-year IT Workers $75,000 $70,000 $44,000 $50,000 $65,000 $75,000 $85,001 87.2% 90.9% 83.0% 89.3% 86.7% 88.2% 94.4% Female Pay-gap 45 2,349 3,304 794 High (CIO & Engineers) Medium (Programmers & Network) Low (Support) 2 effect size greater than => .20 = *** p < 0.001, ** p < 0.01, * p<0.05 739 1,610 2,317 988 794 (1) CIO (2) Engineers (3) Programmers & Related (4) Network & Related (5) Computer Support 1 739 1,514 96 1,293 858 165 349 638 573 220 FTFY Chief Information Officer CIO (1) Software engineer (2) Hardware engineer (2) Computer scientist & system analyst (3) Computer programmers (3) Database administrator (3) Network system administrator (4) Network system analyst (4) Computer support specialist (5) Computer operators (5) Occupational Group: 1,844 2,498 519 557 1,286 1,714 784 519 557 1,210 76 939 666 109 287 497 410 108 Male 505 806 275 182 324 603 204 275 182 304 20 354 192 56 62 141 163 112 Female 36.4% 51.3% 12.3% 11.5% 25.0% 35.9% 15.3% 12.3% 11.5% 23.5% 1.5% 20.1% 13.3% 2.6% 5.4% 9.9% 8.9% 3.4% (100%) 2 *** * *** *** ^ * *** ^ ** ** ** *** 1 37.9% *** 51.4% 10.7% *** 11.5% 26.5% 35.3% 16.1% 10.7% 11.5% 24.9% 1.6% 19.3% 13.7% 2.2% 5.9% 10.2% 8.4% 2.2% (100%) % of all FTFY males 31.8% 50.8% 17.3% 11.5% 20.4% 38.0% 12.8% 17.4% 11.5% 19.2% 1.2% 22.3% 12.1% 3.6% 3.9% 8.9% 10.3% 7.1% (100%) % all females 0.48 0.58 0.75 0.60 0.42 0.64 0.44 0.75 0.60 0.43 0.32 0.66 0.57 0.85 0.34 0.49 0.62 1.09 occ sex seg index $80,000 $65,000 $45,000 $85,000 $80,000 $65,000 $60,000 $45,000 $85,000 $80,000 $79,007 $65,000 $65,000 $70,000 $60,000 $62,000 $50,000 $34,000 FTFY 2 *** *** *** *** *** ^ ^ ^ ^ *** ^ *** ^ ** ^ *** ^ * *** ^ *** ^ 1 $85,000 *** ^ $65,000 *** ^ $45,000 *** ^ $86,000 $82,000 $67,000 $65,000 $45,000 $86,000 $82,000 $79,810 $65,000 $65,000 $80,000 $60,000 $65,000 $50,000 $40,000 Male Table 3 Occupational Position and Median Annual Earnings of Full-time Full-year IT Workers $71,796 $60,000 $42,267 $80,000 $70,000 $62,929 $50,000 $42,267 84.5% 92.3% 93.9% 93.0% 85.4% 93.9% 76.9% 93.9% $80,000 93.0% $70,000 85.4% $66,833 83.7% $65,000 100.0% $65,000 100.0% $55,000 68.8% $53,747 89.6% $50,000 76.9% $50,000 100.0% $31,000 77.5% Female Pay-gap TABLE 4 OLS Regression Analysis for the Income Determination Model (Dependent variable = annual earnings) Variables: Individual-level factors: Age (years) Age centered square Doctoral degree (0,1) Master degree (0,1) Bachelor degree Less than College Degree Midwest (0,1) Rural (0,1) Full sample unstd. 1 std. Males unstd. 1 std. $700 *** 0.247 $718 *** 0.252 -$31 *** -0.128 -$34 *** -0.138 $18,804 *** 0.084 $19,726 *** 0.091 $18,062 *** 0.234 $17,331 *** 0.221 $11,707 *** 0.195 $11,156 *** 0.185 ref grp ref grp -$3,545 *** -0.050 -$3,603 *** -0.050 -$10,405 *** -0.078 -$10,488 *** -0.074 2 Females unstd. 1 std. $612 *** 0.228 -$24 *** -0.106 $14,780 *** 0.063 $19,765 *** 0.283 $12,796 *** 0.229 ref grp -$3,149 *** -0.048 -$9,677 *** -0.089 Structural-level factors: Annual Hours Worked $21 *** 0.308 $20 *** 0.292 ^ $23 *** 0.356 Government Worker -$5,878 *** -0.064 -$6,742 *** -0.069 -$3,969 ** -0.053 Self Employed -$4,847 *** -0.030 -$4,475 *** -0.029 -$4,653 -0.026 Large Business (0,1) $7,991 *** 0.127 $8,408 *** 0.135 $6,568 *** 0.109 Medium Business (0,1) $3,438 *** 0.040 $4,091 *** 0.047 $977 0.012 Small Business (0,1) ref grp ref grp ref grp Goods-producing (0,1) $1,653 * 0.020 $1,231 0.015 $2,815 * 0.034 CIO (0,1) $20,764 *** 0.219 $21,921 *** 0.230 $17,103 *** 0.193 Engineers (0,1) $16,195 *** 0.235 $18,604 *** 0.273 ^ $9,349 *** 0.137 Programmers & Related (0,1) $9,825 *** 0.158 $10,192 *** 0.161 $8,369 *** 0.147 Network & Related (0,1) $8,607 *** 0.103 $11,214 *** 0.136 $1,139 0.014 Computer Support (0,1) ref grp ref grp ref grp Gender-level factors: Female (0,1) Occ.Sex-Seg.Index Minority (0,1) Immigrant (0,1) Married (0,1) Ever-married (0,1) Never-married (0,1) % with child under age 6 (0,1) -$8,489 *** -$5,191 -$4,369 *** -$258 $4,887 *** ref grp -$3,190 *** -$617 -0.124 -0.026 -0.051 -0.004 0.078 -0.046 -0.008 (Constant): -$18,370 *** Adjusted R-sq. 0.469 *** $609 -$4,400 *** -$569 $4,957 *** ref grp -$4,171 *** -$301 -$20,332 *** 0.457 *** 1 = *** p < 0.001; ** p < 0.01; * p < 0.05 2 significant difference betw een men and w omen at the .05 level 46 0.003 -0.050 -0.008 0.078 -0.060 -0.004 -$16,547 -$4,932 $648 $4,388 ref grp -$35 -$2,587 *** *** *** -0.103 -0.068 0.010 0.076 * -0.001 -0.036 -$20,129 ** 0.461 *** Figure 5 Shares of Unique Variance Explained Full sample: 6.8% Gender 48.3% Structural 44.9% Individual Income (Wright, 1992) Males only: 2.2% Gender 46.0% Structural 51.8% Individual Income (Wright, 1992) 47