Project Description

Data

Defining IT

Literature

Reports

IT Workforce: Retention of Women and Minorities

Supported by Grant # EIA 0089995 from the National Science Foundation
Principal Investigators: Paula Stephan and Sharon Levin

IT Workforce > Reports > Retention and Recruitment...

Retention and Recruitment of Women and Minorities in the IT Workforce

Paula E. Stephan
Sharon G. Levin
September 2001

Goals:

The goal of our study is to use the 1999 SESTAT data to analyze issues of retention and recruitment of women and minorities in the IT workforce.  SESTAT integrates three separate surveys:  the Survey of Doctorate Recipients (SDR), the National Survey of College Graduates (NSCG) and the National Survey of Recent College Graduates (NSRCG).  Although it is the best available database for the research we propose, SESTAT is not without problems. [1]

Retention

We will study the deployment of those trained in IT in the U.S. over the period 1993-1999.  Individuals are by definition retained in the IT workforce if they are trained in an IT field and subsequently work in an IT occupation.  The analysis will have three components:  (1) cross-tabs, examining retention differentials by gender and by minority status; (2) estimation of a logit equation where the dependent variable is equal to one if the individual remains in IT and zero otherwise.  Independent variables to control for include age, children, type of degree, gender, race, number of years since receipt of degree; (3) shift-share analysis, using the categories in IT, not in IT but still in S&E, employed out of S&E and other, which includes those retired, unemployed, or otherwise out of the labor force.  An alternative definition of retention will be explored where retention is said to occur if an individual is observed in an IT occupation in 1993 and in subsequent years found working in an IT occupation. 

Recruitment

We define recruitment to be the entry of non-IT trained individuals into IT occupations.  Recruitment is an important route to the IT workforce.  Freeman and Aspray, for example, report that in 1992-93 only about one-third of the people in computer science or programming jobs had graduated with computer and information science degrees.  The majority of the other two-thirds held degrees in business management, engineering, or mathematics.  The analysis of recruitment will use two primary methodologies:  cross-tabs and estimation of logit equations, using the types of variables discussed above.

Main Accomplishments to Date

To date, we have focused our attention on four issues:

• Defining IT fields of training and occupation that can be identified in SESTAT.
• Expanding the literature review on underrepresented minorities, retention issues in the labor market and specific circumstances affecting the IT industry.
• Obtaining access to the latest SESTAT database.
• Preparing preliminary cross-tabs

Here we focus on definitions and a presentation of preliminary results.  These results draw on the 1997 release of SESTAT given that the 1999 version has yet to be released. 

Definitions

Table 1 summaries the populations covered by the SESTAT data.

Table 1:  Summary of SESTAT Data

Note:  SESTAT is discussed in several places:  http//srsstats.sbe.nsf.gov/techinfo.html has a document called “Design and Methodology.”  Also, NSF puts out a booklet entitled SESTAT, A tool of Studying Scientists and Engineers in the U.S.”  April 1999.  Authors are Nirmala Kannankutty and R. Keith Wilkinson.

Defining IT occupations:

There have been two reports in the past five years that draw on data from SESTAT to analyze the IT workforce:  Building a Workforce for the Information Economy (National Research Council) and the IT Data Project (Richard Ellis and Lindsay Lowell). We have drawn heavily on these reports in deciding which fields in SESTAT to include as IT occupations.  Fields that have been chosen include:

• Computer analysts
• Computer scientists, except system analysts
• Information system scientists and analysts
• Other computer and information science occupations
• Computer engineers—software
• Post-secondary teachers in computer and mathematical sciences
• Computer engineers—hardware
• Computer programmers [2]

Based on these fields, SESTAT estimates that there were 1,192,897 IT workers in 1997.

Defining IT Training

Drawing on the work of the IT Data Project, we have defined the following fields as IT trained in SESTAT:

• Computer/information sciences
• Computer science
• Computer system analysts
• Information ser vice & systems
• Other computer & information sciences
• Computer and systems engineering
• Electrical, electronics & communications engineering if the recipient also reported minor or second major areas of study in computer or information sciences or in computer engineering or if they went on to earn a higher degree in an IT field

Based on this definition, there were 752,222 individuals in 1997 who had one or more degree in an IT field. [3]

Preliminary Results

For illustrative purposes, we use these definitions and 1997 data to analyze how retention varies by gender and race.  The results are presented in Table 2 and Table 3:

Table 2
Retention of IT trained by Gender
(column percent)

Chi-square statistic:  3922.  Significant at <.0001 level.

Table 3
Retention of IT Trained by Minority Status
(Column percent)

Chi-Square statistic: 6534, significant at the <.0001 level.

The tables indicate that, relative to men, women are significantly more likely to drop out of the IT workforce.  In a similar vein, we find that relative to whites (and to Asians and Pacific Islanders) African Americans, native Americans and “other” underrepresented minorities trained in an IT field are more likely to work outside their IT field of training.

In order to examine issues of recruitment, Tables 4 and 5 present cross-tabs for individuals working in an IT occupation in 1997 by training. 

Table 4
Training of IT Workers by Gender
(Row percent)

Chi-square statistic 211.90, significant at <.0001.

Table 5
Training of It Workers by Race
(Row Percent)

Chi square equals 11161.84, significant <.0001

The Chi-square statistics indicate that, in both the instance of gender and in the instance of race, the distributions are significantly different than what one would expect if the individuals had been randomly assigned in proportion to their prevalence in the population.  Of particular note is the substantial difference between African Americans and Native Americans and whites.  Only approximately 50% of the former working in IT occupations were trained in IT.  By contrast, more than 60% of white workers working in IT were ot trained in IT.

Next Steps

Our next steps are to acquire the 1999 release of SESTAT and to reexamine these distributions based on subcategories of occupations.  In particular, given the way in which programmers were treated in the data collection, we will perform certain analyses excluding programmers to see how this affects the results.  We will than proceed to estimate the logit equation.