Content uploaded by Ronald A. Ash
Author content
All content in this area was uploaded by Ronald A. Ash on Feb 01, 2018
Content may be subject to copyright.
Why Are There so Few Women in Information Technology?
Assessing the Role of Personality in Career Choices
By
Joshua L. Rosenbloom
Ronald A. Ash
Brandon Dupont
LeAnne Coder
Draft of 29 August 2006
This material is based upon work supported by the National Science Foundation under
Grant No. 0204464. Any opinions, findings, and conclusions or recommendations
expressed in this material are those of the author(s) and do not necessarily reflect the
views of the National Science Foundation.
1
I. Introduction
Despite the very substantial gains that women have made in the labor market over
the past half-century, they remain substantially under represented across a range of
technical and scientific fields. Although women make up nearly 47 percent of the labor
force today, less than 20 percent of most engineering professions are female, just 27
percent of environmental scientists, 31 percent of chemists, and 27 percent of computer
and mathematical occupations are female.
1
Given the importance of these technical
fields in our modern economy, and the rapid expansion of employment opportunities in
technical occupations, the dearth of women in these areas is puzzling from an academic
perspective. It is also troubling from a policy perspective since it suggests that the
nation’s technical workforce may be failing to fully capture the creative energies that are
potentially available.
2
The reasons why women have made such slow progress in gaining entry into
science, math, engineering and technology remain unclear and in some cases quite
controversial, a fact illustrated by the intense debate stimulated after Harvard President
Larry Summers speculated at a January 2005 conference on the possibility that
1
These figures are derived from summary statistics drawn from the 2004 Current
Population Survey downloaded from the Bureau of Labor Statistics. The number of
women among all scientists and engineers has been increasing over time. Long (2001, p.
64) reports that the number of women scientists and engineers rose from about 7 percent
of the workforce in 1973 to about 15 percent in 1995.
2
The importance of the issue of workforce diversity from a policy perspective is reflected
in the numerous programs offered by the National Science Foundation to increase female
and minority participation in technical subjects. See Xie and Shauman (2003, pp. 4-6)
for an elaboration of these points.
2
differences in the distribution of ability among men and women might play some role in
the small numbers of women at the highest levels in science.
3
The dearth of females in technical fields is part of a larger phenomenon of
occupational segregation by sex. Fuchs (1988, p. 34-35) for example noted that in 1980
the Duncan Index of occupational dissimilarity implied that differences in occupational
segregation were nearly twice as great by sex as they were by race, and had fallen much
more slowly over the previous 20 years. Although calculations of the Duncan Index
using 1990 and 2000 census data show a continued modest decline in sex segregation,
they still indicate that it would be necessary for more than 50 percent of women to
change jobs to achieve an equal distribution of men and women (Jacobsen Forthcoming,
Table 6.4).
Explanations for these occupational differences can be grouped under three broad
headings: (1) discrimination, (2) differences in ability, and (3) choice. Explanations
based on discrimination presume that women face differential barriers to entry into
technical fields that discourage their participation.
4
In this view, if these barriers could be
eliminated women and men would enter technical occupations in equal numbers.
Alternatively, the barriers that men and women face may be the same, but (as Larry
Summers suggested) the distribution of abilities differs between men and women in ways
3
A transcript of Summers’ remarks is available on the Harvard University web site
http://www.president.harvard.edu/speeches/2005/nber.html; as is a subsequent letter to
faculty responding to concerns raised following the initial remarks
http://www.president.harvard.edu/speeches/2005/facletter.html.
4
Xie and Shauman (2003, p. 2) suggest organizing explanations within the framework of
supply and demand factors. Differences in ability and preferences are both factors that
operate on the supply side to depress entry of women into technical careers, while
discrimination would be a demand-side factor reducing opportunities for women to enter
technical fields.
3
that make men more productive in technical fields. In this view, occupational differences
by gender reflect an efficient allocation of talent across different fields. The third
explanation is that women may simply place a different weight on the attractiveness of
technical occupation than do men. In this view, the relatively small number of women in
technical fields reflects a competitive market response by workers with heterogeneous
tastes to the differences in characteristics across jobs. This explanation for occupational
choice is based on the neoclassical theory that workers weigh the benefits (both expected
earnings and nonpecuniary returns) and the costs of a particular occupational choice and
will invest in changing occupations only if that cost-benefit analysis works in their favor.
5
Such an interpretation is consistent with the theoretical framework developed by Bowles,
Gintis and Osborne (2001), which explains the existence of large interpersonal earnings
differentials at any point in time to the as the consequence of disequilibrium rents caused
by the slow response of individuals to market shocks. Within this framework the
relationship between earnings and a wide range of (non-productive) personal
characteristics reflects the association of these characteristics with the willingness or
ability of individuals to respond to market shocks that produce these disequilibrium rents.
Developing an empirical strategy to disentangle these three alternatives is
extremely challenging. Our goal in this paper is to advance the discussion with a case
study of Information Technology (IT) occupations. Based on data we collected from a
sample of IT professionals and a control group of comparable non-IT professionals, we
find that much of the gender gap in IT occupations can be accounted for by differences in
5
One reason women may self-select into certain occupations is that they allow for more
interrupted work lives, which females, on average, tend to prefer. Polachek (1981), for
example, argued that women may enter certain occupations because the loss of earnings
from expected absences over time will be minimized.
4
the distribution of vocational interests between men and women. Specifically, we find
that after including a set of measures of occupational personality, gender is no longer a
large or statistically significant factor determining the choice between IT and non-IT
professions. We interpret this to mean that the under representation of women in IT
reflects their choice in response to differences in actual or perceived job characteristics.
This finding does not rule out the possibility that differences in ability or gender
discrimination are also factors, but it does suggest that more attention needs to be given
to how individuals make career choices.
In the next section we expand on the relevance and meaning of occupational
personality for a model of career choice, and set our research within the larger context of
recent research on the role of behavioral characteristics in the labor market. Next we turn
to a description of our data. The fourth section discusses our empirical approach and
presents the results of our estimates. The fifth section discusses the interpretation of our
results.
II. Occupational Personality in the Labor Market
Recent research on labor market outcomes by labor economists, sociologists and
psychologists has established the importance of a range of non-cognitive abilities as
determinants of wages even after controlling for education, experience and cognitive
skills.
6
Heckman, Stixrud and Urzua (2006) for example, use both the Rotter Locus of
Control scale and the Rosenberg Self-Esteem scale as measures of non-cognitive ability
in regressions designed to explain observed schooling, salary, and variety of other
6
See Bowles, Gintis, and Osborne (2001) for a recent survey of this literature.
5
outcomes. They find that non-cognitive skills are equally as important as cognitive skills
in explaining labor market and behavioral outcomes. Coleman and DeLeire (2003) and
Osborne Groves (2005) also find that the Rotter Locus of Control scale is a significant
determinant of earnings. Kuhn and Weinberger (2005) show that another measure of
non-cognitive ability, leadership, also has a significant effect on earnings.
Despite growing interest in the role of non-cognitive abilities as a determinant of
wages, there has so far been little research by economists into the role that such
characteristics might play in explaining patterns of occupational choice.
7
Vocational
psychologists have, however, developed a general framework for analyzing career choice
founded on the concept of occupational personality.
Vocational counselors have long relied on interest inventories (the most
prominent of which is the Strong Interest Inventory) to identify individual interests.
In an influential series of publications, Holland (1959, 1985) developed a typology of
work environments and personalities associated with each environment. According to
Holland both people and work environments can be classified by their position along six
dimensions or General Occupational Themes. The nature of each theme can be
summarized as follows:
8
o The Realistic Theme or R, refers to a person’s preference for activities that
entail the explicit, ordered, or systematic manipulation of objects, tools,
and machines. Realistic types enjoy jobs and activities that involve
mechanical manipulations or repairs and construction. They are interested
7
Jackson (2006) considers the impact of the Rotter Locus of Control scale on
occupational attainment, but she concentrates on broad differences between blue- and
white-collar employment, rather than on more narrowly defined occupational categories.
8
The descriptions are paraphrased from Harmon, et al. (1994) and Holland (1997).
6
in action rather than thought and prefer concrete problems to ambiguous,
abstract problems. Sample Realistic occupations include auto mechanic,
gardener, plumber, and engineer.
o The Investigative Theme or I, refers to a person’s preference for activities
that entail the systematic or creative investigation of physical, biological,
and cultural phenomena. Investigative types enjoy gathering information,
uncovering new facts or theories, and analyzing and interpreting data.
They prefer to rely on themselves rather than on others in a group project.
Sample Investigative occupations include college professor, physician,
psychologist, and chemist.
o The Artistic Theme or A, refers to a person’s preference for activities that
are ambiguous, free, non-systematic and that entail the manipulation of
materials to create art forms or products. Artistic types have a great need
for self-expression. They are also comfortable in academic or intellectual
environments. Sample Artistic occupations include artist, lawyer,
librarian, musician, architect, reporter and English teacher.
o The Social Theme or S, refers to a person’s preference to lead others or for
activities that entail the manipulation of others to inform, train, develop,
cure, or enlighten. Social types enjoy working with people, sharing
responsibilities, and being the center of attention. They also like to solve
problems through discussions of feelings and interactions with others.
Sample Social occupations include elementary school teacher, nurse,
social worker, and occupational therapist.
7
o The Enterprising Theme or E, refers to a person’s preference for activities
that entail the manipulation of others to attain organizational goals or
economic gain. Enterprising types seek positions of power, leadership,
and status. They like to take financial risks and participate in competitive
activities. Sample Enterprising occupations include traveling salesperson,
buyer, realtor, sales manager, and marketing executive.
o The Conventional Theme or C refers to a person’s preference for activities
that entail the explicit, ordered, systematic manipulation of data.
Conventional types often enjoy mathematics and data management
activities. These individuals work well in large organizations but do not
show a distinct preference for or against leadership positions. Sample
Conventional occupations include bookkeeper, accountant, banker,
actuary, and proofreader.
These General Occupational Themes are measured using scales derived from
responses included in interest inventories such as the Strong Interest Inventory.
Measurement of each theme is based on responses to approximately 20 to 25 questions,
and raw scores are normalized relative to a reference population assumed to have a mean
score of 50 and standard deviation of 10. These scores thus serve to locate an
individual’s occupational personality within a six-dimensional space. We make use of
this measurement of occupational personality to identify differences in preferences and to
explore how these preferences influence the choice to enter Information Technology
careers.
8
Holland’s RIASEC model of occupational personality parallels, but is distinct
from the widely used Five-Factor (or Big Five) model of general personality (see
Campbell and Borgen 1999), and is one of the leading frameworks in vocational
psychology (Tracey and Rounds 1993; Campbell and Borgen 1999). There is general
agreement that vocational interests emerge during childhood in response to the
interaction between the individual and his or her environment, and become progressively
more stable through adolescence, and stable in early adulthood. A recent meta analysis
of studies reporting longitudinal data finds that rank order correlations of interests
increase from around 0.55 for those first tested between ages 12 and 14 to over 0.8 for
those first evaluated in their mid- to late-twenties (Douglas Low et al 2005).
III. Data
Our data originate in a study of the reasons for the under representation of women
in Information Technology careers. Our methodology followed a quasi-experimental
design intended to isolate the effects of occupational personality differences from other
possible reasons for gender-based differences in career choice. Because we wished to
control for differences in career motivation, educational attainment, and cognitive ability,
our sampling scheme was designed to produce a sample of IT professionals and a control
group of non-IT professionals who were working in equally demanding careers that
required roughly comparable levels of education and skills.
Between December 2003 and September 2004 we fielded a survey intended to
gather information on work and family history, educational background, interests,
attitudes and measures of Holland’s General Occupational Themes from a sample of
9
professionals in Information Technology and other career fields. We solicited
participation in our survey through a variety of channels including several large
organizations with offices in the central United States, and lists of business and computer
science alumni of a large mid-western university. Potential participants were contacted
via e-mail, and directed to a secure web-site where they logged-in using a password
provided in the contact e-mail. After completing our survey, participants were passed to
a second web-site operated by Consulting Psychologists Press, where they completed the
Strong Interest Inventory. To encourage completion of both surveys (which we estimate
took 30-45 minutes) we offered respondents a chance to receive one of several hundred
$50 gift cards from a large electronics chain store.
A total of 567 individuals completed both parts of the survey. We classified each
respondent as either an IT or non-IT professional based on their responses to questions
asking them to indicate their current career field (one of 13 categories or “Other”) and
specific job title (open-ended). There were 415 non-IT professionals and 152 IT
professionals in our sample. The IT professionals include application developers,
programmers, software engineers, database administrators, systems analysts, web
administrators, and web developers. The non-IT professionals include accountants,
auditors, CEOs, CFOs, presidents, consultants, engineers, managers, administrators,
management analysts, scientists, technicians, nurses, teachers, etc.
Table 1 reports a number of demographic characteristics for the full sample, and
for the IT and non-IT samples separately. More than 44 percent of the respondents were
female, but the proportion of females in the IT sample, at 31.5 percent, closely
approximates the US figure of 27 percent. Women made up about half of the non-IT
10
sample, which is again close to the US figure for Management, Professional and Related
occupations (50.3 percent).
In terms of race composition our sample includes relatively few non-whites (just
8.3 percent) and Hispanics (just 2.7 percent). As we might expect given our sampling
criteria this is a highly educated sample: 45 percent have completed an advanced degree,
another 47 percent have completed a bachelors degree, and close to 80 percent have
completed college calculus. The average age for both the IT and non-IT samples is close
to 40, and both samples have had considerable work experience. Both the IT and non-IT
survey respondents report having worked in their current career fields for more than 10
years, and having been with their current employer for about 7.5 years. Almost all of the
respondents in our sample report holding full-time jobs. They report spending an average
of 48 weeks per year at their primary occupation (less than 20 percent worked fewer than
46 weeks), and working an average of 43.3 hours per week.
Consistent with the occupational composition of our sample, and the education
and experience levels, salaries reported are substantially higher than those typical for all
workers in the US. Information about income was collected in terms of relatively broad
intervals. Table 2 summarizes the distribution of income. The median income for the
group falls in the $60,000-75,000 range, and the distributions are approximately the same
for both the IT and non-IT samples.
9
9
For 2004 the Current Population Survey reports median weekly earnings for all workers
16 and over of $638. Assuming 52 weeks of paid work, this implies median annual
income of $33,176. For computer and mathematical occupations median weekly earnings
were about 50 percent higher $1,144 (equivalent to $57,928 on an annual basis); for
management, professional and related occupations they were $918 ($47,736 on an annual
basis).
11
Table 3 summarizes scores on the General Occupational Themes for our sample
broken down by gender. In addition, the table includes average scores on each Theme
from the sample of 9,500 males and 9,500 females used to establish the population norms
for each scale. In the general population it is clear that, on average, men score higher
than women on the Realistic and Investigative themes, and lower on the Artistic and
Social themes. Differences on the Enterprising and Conventional Themes are less
pronounced. For the most part the same pattern holds true in our sample, except that
women score higher than men on the Enterprising and Conventional themes.
The question, then, is whether these differences in occupational personality can
help account for differences in career choice between IT and other non-IT professional
careers?
IV. Personality and Career Choice
As the summary statistics in Table 1 make clear, women are under represented in
Information Technology Careers in our sample. As a first step to evaluating this
observation we estimate a simple probit regression in which the dependent variable
equals one if the individual is currently working in an IT career and zero otherwise. We
begin with no controls other than gender, and then add demographic variables to
condition for the effects of race, ethnicity and age. We report these regressions in the
first two columns of Table 4. Note that in this table we have expressed all coefficient
values as marginal effects. For dummy variables the magnitude of the coefficient shows
the effect on the probability of choosing IT when the variable changes from zero to one.
12
For continuous variables the coefficient is the derivative of the probability function
evaluated at the means of the independent variables.
Although the demographic controls have some relationship to the choice of IT
careers, it is apparent they don’t have an appreciable impact on the gender effect, which
is strong and statistically significant in both formulations. Based on these regressions it
appears that other things equal, women are approximately 14 percent less likely to choose
IT careers than are men.
In the third column of the table we add scores on the six General Occupational
Themes. It is clear that these occupational personality characteristics are indeed strongly
associated with career choice decisions. The model fits the data much more closely, as
indicated by the decline in the absolute value of the log likelihood and increase in the
pseudo R-squared. Both the Realisic and Enterprising scores are highly statistically and
economically significant. A high value on the Realistic GOT is positively related to the
choice of IT careers, and the coefficient implies that the difference in the mean values of
the Realistic GOT for men and women can explain 3.5 percent of the difference in career
choices (or one quarter of the total 14 percent differential). The difference in mean
values of the Enterprising GOT meanwhile can explain another 2 percent of the
differential in career choices.
Inclusion of all six GOTs reduces the effect on the female dummy variable from
14 percent to 4.5 percent, which is statistically indistinguishable from zero. Thus, while
there may be some residual gender differences in career choices between men and
women, it appears that upon controlling for differences in occupational personality the
gender differences are substantially reduced.
13
V. Discussion
One of the persistent facts of gender economics is the high degree of gender
segregation in the workforce. Men and women tend to concentrate in very different
occupations. There are many potential explanations for this pattern of segregation,
including discrimination, differences in ability, and choice. By introducing a direct
measure of individual preferences through the use of a widely accepted measure of
occupational personality and by controlling for factors like educational attainment and
attachment to the workforce, we are able to directly test the effects of preferences on
career choices in the specific context of the choice between IT and non-IT professional
careers.
We find that within a sample of full-time employed professional workers
differences in preferences can explain a large fraction of the apparent under-
representation of women in Information Technology. In other words, much of the
difference in entry into IT is the result of the fact that, on average, men and women value
different aspects of work and therefore make different career choices. Controlling for
these differences in preferences substantially reduced the differences between men and
women in the choice of IT careers.
Because our research design controls for career motivation, education, and
cognitive skills, we cannot rule out the possibility that discrimination or differences in
ability also act as filters differentially reducing the entry of women into professional
occupations more generally. Further work is needed to assess this issue, but it is worth
observing that women make up nearly half of our control group, so differences in ability
14
or discrimination cannot be overwhelming barriers to their entry into professional
occupations. Although we cannot rule out a role for discrimination or differences in
ability, our results indicate that even if these factors could be eliminated women would
still be under represented in Information Technology because of differences in their
occupational preferences relative to men.
It is possible, of course, that the differences in occupational personality that we
find are a product of reverse causation. Work experience may have altered occupational
preferences. Given the evidence cited earlier on the stability of occupational preferences
we believe that such effects are likely to be small, but given our evidence we cannot rule
out this possibility. Clearly, collecting longitudinal data that would allow for
measurement of occupational personality prior to entry into the workforce would be
preferable. Budget limitations prevented us from collecting such data but our results
suggest that further work along these lines would be highly valuable.
Having identified a potential role for occupational personality in accounting for
gender differences in labor market outcomes, further work on the factors influencing the
formation of this element of career preferences also appears to be called for.
Occupational personality is not an inherent characteristic, but is a complex product of the
interaction between environmental factors and individual characteristics. Thus parental
and family influences, as well as educational and social pressures may contribute to the
divergence in patterns of occupational personality between men and women.
Understanding, how and when these differences emerge appears to be an important topic
for future research, especially for those who may wish to increase women’s participation
in the nation’s technical workforce.
15
References
Barrick, Murray R., Michael K. Mount, and Rashmi Gupta (2003). “Meta-analysis of the
Relationship Between the Five-Factor Model of Personality and Holland’s
Occupational Types.” Personnel Psychology 56, 45-73.
Bowles, Samuel, Herbert Gintis and Melissa Osborne (2001). “The Determinants of
Earnings: A Behavioral Approach.” Journal of Economic Literature 39, no. 4,
1137-1176.
Campbell, David P. and Fred H. Borgen (1999). “Holland’s Theory and the
Development of Interest Inventories.” Journal of Vocational Behavior 5, 86-101.
Coleman, Margo and Thomas DeLeire (2003). “An Economic Model of Locus of
Control and the Human Capital Investment Decision.” Journal of Human
Resources 38, no. 3, 701-721.
Donnay, D. A., Morris, M. L., Schaubhut, N. A., & Thompson, R. C. (2005). Strong
Interest Inventory Manual. Mountain View, CA: Consulting Psychologists Press.
Douglas Low, K.S, Mijung Yoon, Brent W. Roberts and James Rounds (2005). “The
Stability of Vocational Interests from Early Adolescence to Middle Adulthood: A
Quantitative Review of Longitudinal Studies,” Psychological Bulletin 131, no. 5,
713-737.
Fuchs, Victor (1988). Women’s Quest for Economic Equality. Cambridge, MA and
London: Harvard University Press.
Groves, Melissa Osborne (2005). “How Important is your Personality? Labor Market
Returns to Personality for Women in the US and UK.” Journal of Economic
Psychology 26, 827-841.
16
Harmon L, et al. (1994). Strong Interest Inventory Applications and Technical Guide.
Stanford, CA: Stanford University Press.
Heckman, James J., Jora Stixrud, and Segio Urzua (2006). “The Effects of Cognitive and
Noncognitive Abilities on Labor Market Outcomes and Social Behavior.”
National Bureau of Economic Research Working Paper, no. 12006.
Holland, John L. (1959). “A Theory of Vocational Choice,” Journal of Counseling
Psychology 6, 35-45.
Holland, John L. (1985). Making Vocational Choices. Englewood Cliffs, NJ: Prentice
Hall.
Holland, John L. (1997). Making Vocational Choices, 3
rd
edition. Lutz, FL:
Psychological Assessment Resources.
Jackson, Michelle (2006). European Sociological Review 22, no. 2, 187-199.
Jacobsen, Joyce P. (Forthcoming). The Economics of Gender 3
rd
. ed. Malden, MA:
Blackwell.
Kuhn, Peter and Catherine Weinberger (2005). “Leadership Skills and Wages.” Journal
of Labor Economics 23, no. 3, 395-436
Long, J. Scott, Ed. (2001). From Scarcity to Visibility: Gender Differences in the Careers
of Doctoral Scientists and Engineers. National Research Council. Washington,
DC: National Academy Press.
Polachek, Solomon (1981). "Occupational Self-Selection: A Human Capital Approach to
Sex Differences in Occupational Structure." Review of Economics and Statistics,
63(1), 60-69.
17
Xie, Yu and Kimberlee A. Shauman (2003). Women in Science: Career Processes and
Outcomes. Cambridge, MA and London: Harvard University Press.
18
Table 1: Demographic Characteristics of the Sample
Variable
Number
of Obs. Mean
Std.
Deviation Min Max
Full Sample
Age 566
39.299
9.906
22
70
Non White 567
0.085
0.279
0
1
Hispanic 567
0.026
0.161
0
1
Female 567
0.444
0.497
0
1
Completed BA 567
0.466
0.499
0
1
Completed BA and Higher Degree 567
0.451
0.498
0
1
Completed College Calculus 567
0.781
0.414
0
1
Years in Current Position 566
4.470
4.758
0
35
Years with Current Employer 564
7.456
6.863
0
36
Years in Current Career Field 566
11.926
8.216
0
40
Number of Jobs in Current Career Field 557
3.285
2.242
0
15
Weeks worked in past year 548
48.131
5.514
20
52
Average Hours worked per week 564
43.252
10.133
0
82
Non-IT Professionals
Age 414
38.935
9.915
22
70
Non White 415
0.089
0.285
0
1
Hispanic 415
0.029
0.168
0
1
Female 415
0.492
0.501
0
1
Completed BA 415
0.410
0.492
0
1
Completed BA and Higher Degree 415
0.533
0.500
0
1
Completed College Calculus 415
0.793
0.406
0
1
Years in Current Position 414
4.423
4.864
0
35
Years with Current Employer 412
7.488
7.061
0
36
Years in Current Career Field 414
11.273
7.982
0
40
Number of Jobs in Current Career Field 407
3.231
2.200
0
12
Weeks worked in past year 399
47.900
5.741
20
52
Average Hours worked per week 413
43.680
10.440
0
82
IT Professionals
Age 152
40.289
9.845
24
65
Non White 152
0.072
0.260
0
1
Hispanic 152
0.020
0.140
0
1
Female 152
0.316
0.466
0
1
Completed BA 152
0.618
0.487
0
1
Completed BA and Higher Degree 152
0.230
0.422
0
1
Completed College Calculus 152
0.750
0.434
0
1
Years in Current Position 152
4.599
4.471
1
25
Years with Current Employer 152
7.368
6.317
1
26
Years in Current Career Field 152
13.704
8.598
1
38
Number of Jobs in Current Career Field 150
3.433
2.353
1
15
Weeks worked in past year 149
48.752
4.817
20
52
Average Hours worked per week 151
42.079
9.170
0
60
Source: Professional Worker Career Experience and Family Background Survey,
University of Kansas.
19
Table 2: Current Annual Pre-Tax Salary Distribution
Current Salary Full Sample Non-IT Professionals IT Professionals
More than Less than N Pct Cum Pct. N Pct Cum Pct. N Pct Cum Pct.
0
$30,000
43
7.82
7.82
35
8.68
8.68
8
5.44
5.44
$30,000
$45,000
98
17.82
25.64
77
19.11
27.79
21
14.29
19.73
$45,001
$60,000
112
20.36
46
89
22.08
49.88
23
15.65
35.37
$60,001
$75,000
102
18.55
64.55
61
15.14
65.01
41
27.89
63.27
$75,001
$90,000
79
14.36
78.91
51
12.66
77.67
28
19.05
82.31
$90,001
$105,000
39
7.09
86
31
7.69
85.36
8
5.44
87.76
$105,001
$120,000
26
4.73
90.73
17
4.22
89.58
9
6.12
93.88
$120,001
$135,000
13
2.36
93.09
9
2.23
91.81
4
2.72
96.6
$135,001
$150,000
12
2.18
95.27
9
2.23
94.04
3
2.04
98.64
$150,001
$175,000
9
1.64
96.91
8
1.99
96.03
1
0.68
99.32
$175,001
$200,000 0
0.00
96.91
0
0.00
96.03
0
0.00
99.32
Greater than $200,000
17
3.09
100
16
3.97
100
1
0.68
100
20
Table 3: Mean Scores on General Occupational Themes, by Sex
Survey Population Norms
General
Occupational
Theme Males Females Males Females
Realistic 54.6
47.0
55.0
45.0
Investigative 53.9
51.5
51.4
48.6
Artistic 47.2
50.6
48.7
51.3
Social 44.9
51.6
48.0
51.9
Enterprising 47.9
50.7
49.6
50.4
Conventional 52.4
55.2
49.4
50.6
21
Table 4: Probit Regressions of the determinants of Choice of IT Career
I II III
Female -0.1397* -0.1418* -0.0456
-(0.0363) (0.0365) (0.0468)
Non-White 0.0185 -0.0066
(0.0815) (0.0773)
Hispanic -0.0626 -0.0696
(0.1079) (0.1050)
Age 0.0135 0.0044
(0.0148) (0.0145)
Age2 -0.0001 -0.0001
(0.0002) (0.0002)
Married -0.0007 -0.0163
(0.0426) (0.0433)
General Occupational Themes
Realistic 0.0105*
(0.0028)
Investigative 0.0000
(0.0024)
Artistics 0.0029
(0.0025)
Social -0.0004
(0.0028)
Enterprising -0.0157*
(0.0024)
Conventional 0.0035
(0.0025)
Observed P 0.2686
0.2686
0.2686
Predicted P 0.2632
0.2623
0.2362
N obs 566
566
566
Pseudo R2 0.0216
0.0263
0.123
Log Likelihood -322.5
-320.66
-288.72
* Statistically significantly different from zero at the one percent level.
Notes: Standard errors are in parentheses. Coefficients are transformed from their
original values to reflect the marginal effects of a change in the relevant variable
22
evaluated at the sample mean values. For zero-one dummies the coefficient shows the
effect of changing the independent variable from zero to one.
