Female Retention in Undergraduate Computing: Comparisons with a Women’s College J. McGrath Cohoon Research Assistant Professor Curry School of Education University of Virginia Katherine Cook Graduate Student Curry School of Education University of Virginia 2/17/2016 ABSTRACT We consider evidence of environmental effects on female retention in science, mathematics, and engineering disciplines, with a particular focus on computer science. We then compare the relevant features of one women’s college with features of the average coed institution located in one state. These observations contribute to the debate about female representation in the predominantly male discipline of computer science and the role women’s colleges can play in increasing female participation. INTRODUCTION Men and women are often segregated by academic discipline in college. For example, education and nursing are predominantly female, whereas engineering and technology are predominantly male. The degree and nature of this segregation changes over time (Jacobs, 1995). Some historically maledominated fields such as mathematics and biology/life sciences (BLS) now graduate approximately equal numbers of men and women with baccalaureate degrees. In contrast, women’s representation in the predominantly male discipline of computer science (CS) decreased in recent years. Data from the National Council on Education Statistics show that the percentage of bachelor’s degrees awarded to women in CS decreased from 37.1% in the mid-1980s to just 26.7% in 1997-98. This decline came at a time when other science, mathematics, and engineering (SME) disciplines were increasing their female representation, as shown in Figure 1. 1 Concern about female under-representation in collegiate computing majors has motivated investigations into how the situation can be affected by academic institutions and departments. This research has identified several factors in the educational environment that are associated with the equality of outcomes for male and female students. The current investigation asks whether the environment in women’s colleges helps retain women in SME disciplines. To explore this issue, we relate prior research on women’s colleges to the results from a recent inquiry into the disproportionate loss of women from the computer science major. The results we discuss were obtained from a study of CS and BLS departments at 24 institutions in the state of Virginia.1 Our interest here is whether the departmental characteristics and practices identified in the statewide study are likely to be prevalent in women’s colleges, and whether this prevalence promotes retention of computing majors at women’s colleges. The limitations of our data prevent us from reaching any conclusions, but they raise interesting issues worthy of further discussion. ATTRITION FROM THE MAJOR Female under-representation in certain SME disciplines is exacerbated by the disproportionate loss of those relatively few women who express an initial interest. Not only do fewer women than men declare SME majors, but in some disciplines, women switch out at higher rates than men. For example, 20 percent of Virginia’s undergraduate women who declared CS majors eventually switched 1 These 24 institutions were all the institutions in Virginia that awarded CS baccalaureate degrees for at least three of the years between 1992 and 1997. 2 to some other discipline between 1992 and 1997. In comparison, just 10 percent of male CS majors switched out. Numerous explanations for the disproportionate loss of women have been put forth. Some of these explanations rest upon gender differences, and some rest upon environmental factors. We stipulate gender differences but focus primarily on environmental factors. Our reason for this choice is that inherent or deeply ingrained gender differences are not very tractable, but the environment can affect behaviors that may stem from gender differences. The influence of environmental factors on female attrition Women’s low self-confidence when it comes to computing is frequently noted, and it is suspected of contributing to the disproportionate loss of women from the CS major (Astin and Astin, 1993 Strenta et.al., 1994; Fisher, Margolis, Miller, 1997; Brainard, Metz, and Gillmore, 1999). Yet, women are not inherently less confident than men. Their self-ratings are higher than men’s for traditionally female tasks and lower for tasks they believe are “male” (Beyer, 1990). Furthermore, women have been shown to internalize the low expectations of their instructors (Ancis and Phillips, 1996). Thus, environmental factors such as low faculty expectations or the gendered image of computing may lead female students to quit the CS major. At least two additional environmental factors contribute to the loss of initially interested students of both sexes from SME. They are pedagogy and peers (Seymour and Hewitt, 1997; Astin and Astin, 1993). The environment in SME has been described as a competitive academic culture that discourages or 3 forbids students from working together; assigns grades on a curve; and offers little access to help and tutoring (Seymour and Hewitt, 1997). Both male and female SME majors at multiple institutions criticized teaching in the sciences and said that it contributed to the decision to switch. These students reported dealing with faculty who were dismissive or openly insulting, advisors who gave incorrect information, curricula that had no room for one bad semester, and “weed-out” courses. They described an environment in which both faculty and students assumed attrition to be not only a given, but desirable. According to Seymour and Hewitt’s findings, the sciences harbor a culture in which it is assumed that those who drop out are deficient in some way and that those who persist and succeed do so because they are the fittest and most able students. Furthermore, Astin and Astin (1992) found that female and minority faculty were likely to use retentionpromoting pedagogy, but the dominant culture in SME encouraged student attrition. Peers can have a critical influence on the retention of students in SME. A massive empirical study of American undergraduates in the latter 1980s found “Student interest in pursuing science majors and careers can be affected both by the characteristics of the peer group as well as by the type of pedagogy the institution employs. The clearest and most consistent pattern of environmental effects on student choice outcomes is associated with the concentration of student peers in various fields of study” (Astin and Astin, 1992). In other words, SME students with many socially similar comrades in their program are more likely to persist. Conversely, those students who are in a minority are less likely to persist. 4 These pedagogical practices and demographic characteristics can have differential effects that lead to the disproportionate loss of women. In support of this finding, female students in an all-female computer science course reported higher self-confidence in the subject area, higher levels of intrinsic interest, higher future academic ambition in the subject, and more favorable perceptions of their teachers’ support than their female counterparts in a mixed-gender class (Crombie and Armstrong, 1999). The students in the all-female class also reported having fewer gender-stereotyped attitudes towards computing. A likely explanation for this finding is that when women are in the majority, they have access to same-sex peers who can support and encourage them. The reason that pedagogical practices could have differential effects is less apparent. Seymour and Hewitt (1997) speculated that early socialization caused women to be more adversely affected by the negative aspects of their learning environment than were white men. However, we offer an alternative explanation. We believe that the reason women may be more vulnerable than men to the negative effects of an inhospitable environment is that women frequently have access to fewer resources for overcoming them. Evidence in support of our proposition comes from a statewide study of undergraduate programs in two SME disciplines: CS and BLS. This research showed that women were retained at equivalent rates to men in the average BLS department, but not in the average CS department. In CS departments, women were retained comparably to men in those programs that had higher proportions of female students, a focus on teaching, female faculty members, faculty with 5 favorable attitudes toward women’s abilities, and other supportive features (Xxxx, 2001). Thus, not all SME disciplines and not all CS programs lost women at higher rates than they lost men. Some disciplines and some departments retained men and women at equal rates. The results of this statewide study and previous research indicate the importance of environmental conditions in the retention of women, particularly in computer science. We now consider whether women’s colleges provide the kind of environment that helps retain female CS majors. FAVORABLE ENVIRONMENTS AT WOMEN’S COLLEGES? Are women’s colleges more likely than coed colleges to provide the type of environment that retains women in computing majors? Some well-known, but often disputed research about women’s colleges and their education in SME disciplines suggests that they do. Descriptions of the environment in women’s colleges contrast sharply with descriptions of coed SME programs. Instead of the unhelpful, sink-or-swim type approach, descriptions of women’s colleges report supportive, encouraging environments. For example, Bryn Mawr College and Bennet College are described as follows. “During the first two years at Bryn Mawr, it appears that faculty and administrators concentrate on convincing women that they are capable students and encouraging them to pursue topics that interest them, particularly in nontraditional fields. The focus for upper-class women is on preparing them to handle graduate-level work and encouraging them to see themselves as scholars” (Tidball et al, 1999) There appears to be both an assumption and a commitment to the belief that women in general, and individual students in particular, are capable of academic success in any field of their choosing. Students at women’s colleges are given the 6 clear message that they “matter” (Tidball et al, 1999). Faculty actively mentor their students, encouraging them to fulfill their potential and even reach beyond what the students themselves thought they were capable of doing. Students report that their professors are very accessible and willing to advise them on academic, professional and personal issues while they are students and after they have graduated. (Tidball et al, 1999) A professor from Bennet College offers an example of this attitude: “To be ‘Benettized’ is really to believe in more than the potential of students…You believe that these students can learn, can be the best, and are expected to be the best, and you do whatever it takes to get them where they need to be.” (Tidball et al., 1999) For the Bennet professor, teaching undergraduates is more than a matter of presenting the students with material and waiting to see which students rise to the top. The professor begins with the assumption that all of the students are capable of learning the material. Furthermore, the professor feels a responsibility to do “whatever it takes” to assure that the students learn the material presented. The pedagogical environment at women’s colleges may be one reason their students highly rate their own academic ability. The academic selfconfidence of women’s college students was significantly higher than the academic self-confidence of women at coed colleges, and it was related to having faculty take a personal interest in them (Kim and Alvarez, 1995). These findings suggest that if low self-confidence is related to dropping a computing major, women may be retained at higher rates in women’s colleges. If the favorable environment in women’s colleges extends to the SME disciplines, the scenarios presented by Tidball et al, suggest that it should promote female retention in the CS major. The student-centered pedagogical approach and 7 the explicitly supportive attitude toward the female students in women’s colleges are two features that the research indicates would help retain women in computing majors. In addition, women’s colleges meet a third criterion for retaining women: demographic composition. By definition, there are numerous same-sex peers for women in the CS major at a women’s college. Furthermore, there are likely to be many more female faculty than are common in coed CS programs.i These descriptions of the environment depict favorable conditions for female retention in women’s colleges, but the outcomes they report offer mixed support for the hypothesis that the environment in women’s colleges could be conducive to female retention in computing majors. For example, there is much research that shows the graduates of women’s colleges have more selfconfidence, achieve more, and are more likely to major in non-traditional fields and less likely to work in female-dominated professions (Wolf-Wendel, 1999). However, there is also research that shows no differences in outcomes for women’s and coed institutions (see Levit, 1999 for an extensive discussion). Tidball’s work is criticized for not taking into consideration the selectivity of women’s colleges and the relatively privileged socio-economic status of the students who attend (or attended) them. When factors such as family socioeconomic status, pre-college aspirations and achievement, institutional size, prestige, selectivity and gender, college major and experiences and postcollege achievement are taken into account, there is “little to indicate that attending a women’s college has more than a trivial net influence on women’s postcollege educational, occupational and economic attainments” (Stoecker and Pascarella, 8 1991). Thus, rather than being a product of the environment in women’s colleges, the achievement of their graduates may be due to recruitment factors. The findings on selection and retention in non-traditional majors are also mixed. One comparison of coed and women’s colleges showed that changes in academic majors among students in women’s colleges resulted in less graduates in female-dominated disciplines and more graduates in male-dominated disciplines than did coed colleges (Solnick, 1995). However, this effect was not brought about by positive influences on women in non-traditional disciplines. Initially, women at both types of institutions were about equally unlikely to intend a male-dominated major. Only 4 to 5% of the more than 2000 students studied in the class of 1992 had planned to major in Mathematics, Engineering, or Computer Science. As this group progressed through their undergraduate studies, retention in male-dominated disciplines was actually slightly lower among women’s college students than among students at coeducational institutions. Women’s college students who declared a Mathematics, Engineering, or Computer Science major, were no more likely than those who attended coed colleges to graduate in that major. Instead, Solnick’s finding that the distribution of women at graduation was greater in male-dominated fields at women’s colleges was attributable to the number of students who switched out of traditionally female majors. Returning to the question that began this section, “Are women’s colleges more likely than coed colleges to provide the kind of environment that retains women in computing majors?” The answer is, “Perhaps.” Levit’s (1999) review of the literature on single-sex education stresses the inconclusiveness of findings 9 to date. Further research is needed before we can give a definitive answer. The next section reports on a small contribution toward this end. COMPARISON OF VIRGINIA’S INSTITUTIONS One of the institutions participating in the Virginia study of gendered attrition was a women’s college. We will call it VAW for “Virginia Women’s.” The data collected from this private, liberal arts institution provides us with an opportunity to compare its characteristics, practices, and outcomes with those of Virginia’s coed institutions. This comparison is not intended as a generalization regarding women’s colleges and their relative ability to recruit or retain women in particular SME majors. Rather, the data from this single institution are offered as one set of observations on an issue that begs further explanation. Our first observation is that between 1992 and 1997, CS was a more popular major among the women at VAW than it was among women at Virginia’s coed institutions. All three institutional groupings shown in Table 1 indicate that VAW educated a higher proportion of its female students in CS than did coed departments in Virginia. The proportion of female graduates who had a CS major from VAW was small, only 1.7%, but it was significantly higher than at the average coed institution (1.2%), the average coed Liberal Arts (LA) institution (1.0%), or the average moderately selective coed LA institution (0.7%). We also observe that VAW retained a higher proportion of its upper-level female CS majors than did the average coed institution.2 As the second row in Table 1 shows, the attrition from VAW’s CS program was lower than at 10 Virginia’s coed institutions. It was significantly lower (at the .10 level) when compared with all coed departments, but the difference between VAW and the two groups of LA institutions was not significant. The non-significant result was likely due to extreme variation within the LA group. At one institution, there was only one upper-level woman in CS during only one of the study years, and she switched to a different major before completing her degree. This action resulted in a 100% attrition rate for that department. At another LA institution, none of the women left during the two years the department had any upper-level women. These two extreme situations within the small groups of LA departments made the apparent difference with VAW’s attrition rate insignificant. These observations of choice and persistence at VAW and the coed institutions in Virginia appear to contradict Solnick’s results. Solnick, (1995) found that women in single-sex colleges were not more likely to intend a mathematics, engineering, or computer science major than were women at a coed college; they switched out of these male dominated majors at slightly higher rates than did women in coed colleges. The difference between our results and Solnick’s may stem from a variety of factors: differences in the disciplines considered, differences in the status of students considered, atypicality of the single women’s institution being described here, or a greater tendency among women’s college students to switch majors regardless of their discipline. 2 Most switching of majors occurs during the first two years of college. However, VAW, like many other institutions, has its students formally declare their major in their third year. To make our comparisons fair, we only show attrition rates for upper-level students. 11 Departmental Factors The Virginia study of gendered attrition found that CS departments that supported their female students through peers, faculty, the institution, and community resources retained women at comparable rates to men. The single factor with the strongest impact was the availability of same-sex peer support as measured by the relative numbers of women enrolled in the major. Based on these results and the observation that VAW retained a higher proportion of its female CS majors than did the coed institutions, we expect that VAW would differ from coed institutions on the support features. Of course, the first feature, gender composition of the CS program, differs by definition. The quantitative differences in gender composition between the CS program at VAW and the other Virginia programs are shown in the first row of Table 2 where you can see that the average coed CS department in Virginia was 28.3% female. The female proportion was lower still in LA departments. Students explained why same sex peers are so important during interviews conducted at five Virginia institutions. Both male and female students emphasized the necessity of having friends they could go to for help in their CS classes. In departments where women had few, if any other women to work with, they were at a disadvantage. The interview data suggest that only those women who could tolerate the teasing and the embarrassment of appearing to fit gender stereotypes were able to get the help they needed from their male peers. In this way, the gender composition of a department seriously impacted women’s access to an essential resource for success in the CS major. The comparison of VAW with 12 coed institutions indicates that VAW had a considerable advantage on this characteristic. Faculty were the next significant source of support identified by the Virginia gendered attrition study. Analyses showed that departments’ gender gaps in attrition rates were affected by their rate of faculty turnover, the satisfaction faculty derived from teaching undergraduates, the responsibility faculty took onto themselves for their students’ success, the scope of mentoring, and the presence of female faculty in the department. Comparison data for VAW were available for only 4 of these factors – satisfaction from teaching, responsibility for students’ success, hours devoted to mentoring, and presence of female faculty. Departments where faculty reported that they derived personal satisfaction from teaching undergraduates were likely to have small gender gaps in their attrition rates. This intrinsic motivation for teaching was measured on a 5-point scale of self-reported satisfaction. There was no significant difference between VAW and the average coed institution. Generally, CS faculty at Virginia institutions highly rated the personal satisfaction they felt from teaching or preparing to teach undergraduates. VAW faculty were no exception to this sentiment. Departments where faculty felt that they contributed to students’ success were more likely to retain women at comparable rates to men. The fewer faculty who agreed with the statement, “Student success required innate ability,” the smaller the gender gap in attrition rates. On a 5-point scale of agreement in our survey of faculty, VAW had lower average agreement than did the coed 13 institutions, significantly lower compared with both the average coed institution and the moderately selective coed LA institutions. These results indicated that CS faculty at VAW were more likely to emphasize the their own teaching as an important factor in students’ success. Departments where the average faculty member devoted more time to mentoring retained female students at comparable rates to men. This faculty mentoring appeared to be particularly effective during the early stage of a student’s college career. Comparing VAW with the coed institutions on this variable showed no significant difference in the number of hours per week faculty typically spent mentoring students. Thus, the significant difference between faculty behaviors at VAW and the coed institutions appears to be with respect to teaching as opposed to mentoring. The importance of same-sex faculty role models for female students has long been debated. Contrary to Tidball, but confirming Kim and Alvarez (1995), the Virginia gendered attrition study found that although the female proportion of faculty was not significant, the simple presence of women on the faculty was significant. Comparing this feature of VAW with coed institutions shows that unlike VAW, there were coed departments in each group with no women on their CS faculty. (VAW’s CS faculty was 40% female. The range in coed departments was from 0 to 100 % with an average of 27% female faculty.) Thus, the coed departments were more likely than VAW to offer no same-sex faculty role models to female students. 14 Perceived support from the institution and community were the final factors linked to female retention. The statewide study found that where chairpersons reported above average institutional support relative to other departments, the department retained women at comparable rates to men. This factor may be a matter of having the necessary resources in support of undergraduate education. Communities affected gendered attrition by supporting computing careers. Departments located in geographic areas that offered many computing career opportunities had small gender gaps in attrition rates. Unfortunately, data measuring these variables for VAW were not available for comparison with the coed institutions. DISCUSSION Of the factors we were able to compare between VAW and the coed study institutions in Virginia, the female proportion of enrollment, faculty attitudes toward their role in student success, and the presence of female faculty could be important differences. The first of these factors is clearly an advantage that women’s colleges in general have over coed institutions. Computing classes in women’s colleges offer female peers, a feature that coed colleges do not always offer. Thus, CS majors at women’s colleges are spared the type of experience reported by students in coed colleges, as described by one woman, “Everyone looks at you strangely because you’re the girl in CS.” In women’s colleges, no one is alone as “the girl” in CS. They have the support that comes from the simple presence of other women in class. 15 In addition to the benefit of not seeming strange, female CS majors at women’s colleges have access to women classmates for academic help. And, as students explained over and over, this form of peer support is crucial. One woman spoke for most CS students when she said, “If I didn't have people here that I could come to and say, ‘What does this mean?’ I don't think I would have stayed at all.” Students depend on each other to make it through the rigors of an undergraduate computing program. The help of classmates is essential, but male classmates can give this help; it does not have to come from other women. In fact, most women in coed institutions go to their male classmates for help. Many women explained this as a simple matter of numbers, “[In] the majority of the Computer Science classes, there's mostly males.” Rather than search out one of the very few women in their program these women simply ask whoever is at hand, and this is likely to be a man. Some women also noted that they have always had a preference for male friends, “all through high school, all through college, guys are my friends.” Female CS majors’ reliance on men for their academic peer support might be a necessary skill for completing the program in coed departments. However, this cross-sex behavior is not the norm; homophily is the norm (although less so among young adults). Similar people, people of the same sex, race, etc., tend to form social networks at higher rates than dissimilar people (McPherson, SmithLovin, and Cook, 2001; Lazarsfeld and Merton, 1954). Women who attend CS programs in women’s colleges have the opportunity to keep within the norm 16 because there are other female students at hand. In this way, the women’s colleges open their CS program to a broader range of female students. The second important difference between VAW and the average coed institution was a single item measuring emphasis on teaching – faculty investment in student success. Although the evidence was thin that VAW differed considerably from coed institutions when it came to teaching, on this particular teaching item, VAW’s result for its computing program was in keeping with reports from other research on teaching in women’s colleges. These published reports contrast sharply with reports of SME teaching in coed institutions. Further research is needed to determine if the difference observed here is representative of a general difference between women’s and coed computing programs. The third important difference between VAW and the average coed institution was the presence of female faculty. The value of role models is well accepted, but the particular requirements are still disputed. Tidball (1973) found that the ratio of women faculty to women students affected women’s achievement. However, Kim and Alvarez (1995) found that the percent of female faculty did “not account for the extraordinary achievement of graduates from women-only colleges.” Xxxx (2001) found no significant relationship between either the proportion of female faculty or the ratio of women faculty to women students and a CS department’s gendered attrition rate, but she did find that the simple presence or absence of female faculty was a significant factor. This latter finding may explain the difference between the results obtained by Tidball and Kim-Alvarez. Tidball’s original research focused on women who 17 achieved doctoral degrees between 1910 and 1970- before female faculty were common in academic departments. The number of coed departments with at least one woman on the faculty has increased during the years between the Tidball and the Kim-Alvarez studies. Tidball’s data may have represented a time when it was the exceptional coed SME department that had a woman on the faculty. Thus, a measure of female proportion may have been equivalent to an indication of the presence of female faculty. By the time Kim and Alvarez did their research, women were more commonly members of SME faculty, making the two measures more distinct. There is also a possibility that being an effective role model requires more than just being a woman computer scientist. Women who follow a “male professional model” may not have the positive effect expected of role models for female students. For example, physics, chemistry, electrical engineering and computer science departments lost women graduate students in part from a lack of suitable role models (Etzkowitz et al., 1994). According to this research, female students prefer to model themselves only after those women who balance their work and personal lives. Further research is needed on the influence of role models on retention in undergraduate computing. This research should address the issue of which measure of role models is appropriate – presence, proportion of the faculty, ratio to female students, or each of these considering “suitability.” It should also distinguish between effects on recruitment and retention (Canes and Rosen (1995) 18 showed that role models did not increase female enrollment in a variety of majors at 3 institutions.) VAW’s relative success with retaining women in the CS major was likely due to the presence of same-sex peers and faculty, and to faculty who were personally invested in their students’ success. These features could also be reasons why CS programs in women’s colleges might have more success than their coed counterparts in graduating female students with a computing baccalaureate. CONCLUSION The observations described here are not to be interpreted as advocacy of sex-segregated education. Certainly, the evidence presented is insufficient to support such a position. Furthermore, the departmental characteristics and practices associated with gendered attrition were linked with equality of outcomes, not necessarily high retention of women. We offered no evidence that excluding men would improve female retention in computing majors, only that including more women has a favorable effect on retention. However, this paper contributes to the debate by drawing attention to the potential for women’s colleges to increase female retention in a particular maledominated discipline. The evidence presented suggests the possibility that at the present time, the environment in women’s colleges might support women’s retention in computing majors. This retentive environment need not be an exclusive feature of women’s colleges. Coed institutions could recruit more female students and faculty, and 19 they could adopt more student-centered pedagogical practices. The decision is theirs. Yet, considering the need for qualified computer professionals and demographic trends toward high proportions of women in higher education and a more diverse workforce, the choice to increase female participation seems like a wise one. 20 APPENDIX Virginia Statewide Study Methodology The Virginia gendered attrition study employed multiple methodologies to form a consensus of findings. The methods employed were: qualitative analysis of interview data; discipline comparisons of quantitative data; and statistical analyses of CS survey data and official institutional data on student outcomes. Conclusions were based on results produced with at least two types of analyses that related these data to the gendered attrition rates we calculated from official statistics on declared majors and their eventual outcomes. Qualitative data was collected through 34 face-to-face interviews at five of the CS study departments. The five interview departments were of varying size, geographic location, institutional selectivity, and gendered attrition rates. Interviews were conducted with the chairperson, all faculty in small departments and up to five faculty in larger departments, and with an average of eleven students per department. Most students were interviewed in gender-segregated groups. The interviews lasted between 45 minutes and two hours during which the realities of undergraduate computer science education and potential explanations for disproportionate female attrition were explored. Quantitative data were obtained from a mail survey and from a state agency that collects institutional data from all colleges and universities in Virginia. The survey was constructed based on the interview results and on hypotheses from the literature. Up to 10 faculty members plus the chairperson at CS and BLS departments in 24 institutions were sent this written survey. A total 21 of 255 completed questionnaires were returned. The overall response rate for all questionnaires was 68%, with a breakdown of 90% of chairpersons responding and 65% of faculty responding. The quantitative data were statistically analyzed to determine which factors had the strongest, most consistent effects on gendered attrition. First, descriptive data from computer science and biology/life sciences were compared. Second, measures of CS department characteristics and practices were correlated with the departments’ gendered attrition rates. Third, the characteristics and practices that were significantly correlated with gendered attrition were entered into multiple regressions that compensated for differences in the size of study departments. The results of these analyses identified the factors we will discuss below in relation to the one women’s college included in the study. WORKS CITED Ancis, Julie R. Phillips, Susan D. “Academic Gender Bias and Women's Behavioral Agency Self-Efficacy.” Journal of Counseling & Development. v75 n2 p131-37 Nov-Dec 1996. Astin, Alexander W. Astin, Helen S. Undergraduate Science Education: The Impact of Different College Environments on the Educational Pipeline in the Sciences. Final Report. University of California, Los Angeles. Higher Education Research Institute November 1992 22 Astin, Helen S, Milem, Jeffrey F. “The Changing Composition of the Faculty: What Does It Really Mean for Diversity?” in: Change. v25 n2 p21-27 Mar-Apr 1993. Beyer, Sylvia. “Gender Differences in the Accuracy of Self-evaluations of Performance.” Journal of Personality and Social Psychology. v99 (5). p960-970. 1990. Brainard, Suzanne G, Susan Staffin Metz, Gerald M Gillmore. “WEPAN Pilot Climate Survey: Exploring the Environment for Undergraduate Engineering Students” Women in Engineering Programs & Advocates Network. 1999. Bureau of Labor Statistics. News release. “Table 1: Labor force status of 2000 high school graduates and 1999-2000 high school dropouts 16 to 24 years old by school enrollment, sex, race, and Hispanic origin, October 2000” Labor Force Statistics from the Current Population Survey. April 13, 2001. Canes, Brandice and Harvey Rosen. “Following in her footsteps? Faculty Gender Composition and Women’s Choices of College Major” Industrial and Labor Relations Review. Vol. 48, No. 3, April, 1995. Xxxx, Xxxx. “Toward Improving Female Retention in the Computer Science Major” Communications of the ACM. May, 2001. 23 Crombie, Gail and Patrick Ian Armstrong. “Effects of Classroom Gender composition on Adolescents’ Computer-related Attitudes and Future Intentions.” Journal of Educational Computing Research. 20(4). 1999. Etzkowitz, Henry, Carol Kemelgor, Michael Neuschatz, Brian Uzzi. “Barriers to Women in Academic Science and Engineering” Who Will Do Science: Educating the Next Generation. Edited by Willie Pearson Jr. and Irwin Fechter. Baltimore: Johns Hopkins University Press. Fisher, Alan, Jane Margolis, Faye Miller. “Undergraduate Women in Computer Science: Experience, Motivation and Culture.” SIGCSE ’97 The Proceedings of the Twenty-eighth SIGCSE Technical Symposium on Computer Science Education. ACM. 1997 Jacobs, Jerry A. “Gender and Academic Specialties: Trends among Recipients of College Degrees in the 1980s.” Sociology of Education. v68 (April): p8198. 1995. Kim, Mikyong, Rodolfo Alvarez. “Women-Only Colleges: Some Unanticipated Consequences.” Journal of Higher Education. V66 (6). p641-668. 1995 Levit, Nancy. “Separating Equals: Educational Research and the Long-Term Consequences of Sex Segregation.” The George Washington Law Review. v67 (3) p451-526. 1999. McPherson, Miller, Lynn Smith-Lovin, and James M Cook. “Birds of a Feather: Homophily in Social Networks” Annual Review of Sociology. 2001. 27:415-44. 24 Riordan, Cornelius. “Single and Mixed Gender Colleges for Women: Educational, Attitudinal, and Occupational Outcomes.” In: The Review of Higher Education. Spring 1992, Volume 15, Number 3. 327-346. Seymour, Elaine and Nancy M. Hewitt. Talking About Leaving: Why Undergraduates Leave the Sciences. Boulder, CO: Westview Press, 1997. Solnick, Sara J. “Changes in Women’s Majors from Entrance to Graduation at Women’s and Coeducational Colleges.” In: Industrial and Labor Relations Review, v48, Number 3 (April 1995) 505-514. Stoecker, Judith L. and Ernest T. Pascarella. “Women’s Colleges and Women’s Career Attainments Revisited.” In: Journal of Higher Education, v62, Issue 4(Jul-Aug., 1991) 394-406. Strenta, A Christopher. “Choosing and Leaving Science in Highly Selective Institutions.” Research in Higher Education. v35 n5 p513-47 Oct 1994. Tidball, Elizabeth M. “Comment on ‘Women’s Colleges and Women’s Career Attainments Revisited.’” In: Journal of Higher Education, v62, Issue 4(Jul-Aug., 1991) 406-409. Tidball, Elizabeth, M. “Baccalaureate Origins of Recent Natural Science Doctorates.” In: Journal of Higher Education. v57, Issue 6 (Nov.-Dec., 1986), 606-620 Tidball, Elizabeth, M. and Vera Kistiakowsky. “Baccalaureate Origins of American Scientists and Scholars. In: Science, New Series, v193, Issue 4254 (Aug. 20, 1976), 646-652 25 Tidball, Elizabeth, M., Daryl G. Smith, Charles S. Tidball, and Lisa E. WolfWendel. Taking Women Seriously: Lessons and Legacies for Education the Majority. Phoenix, AZ: The American Council on Education and the Oryx Press, 1999. Wolf-Wendel, Lisa E. “Research Issues on Women’s Colleges.” A Closer Look at Women’s Colleges. July 1999. Available on the Internet at http://ed.gov/pubs/WomensColleges/chap3fin.html. National Institute on Postsecondary Education, Libraries, and Lifelong Learning. Office of Educational Research and Improvement. U.S. Department of Education. Irene B. Harwarth (editor.) 26 27 Table 1 Undergraduate Women in CS, 1992 - 1997 All VA Coed Insitutions Liberal Arts VA Coed Institutions Moderately Selective LA Coed Institutions VAW n=23 n=8 n=4 n=1 female representation 1.2%** 1.0%** 0.7%** 1.7% upper-level female attrition 14.4%* 26.7% 34.4% 6.3% * sig at .10 level ** sig at .05 level 28 Table 2 Departmental Factors, 1992 - 1997 All VA Coed Insitutions % female Moderately Liberal Arts VA Selective LA Coed Coed Institutions Institutions VAW 28.3%** 23.6%** 22.0%** 100.0% satisfaction from teaching 4.6% 4.7% 4.9% 4.7% responsibility for success 3.3%** 3.3% 4.2%** 3.0% mentoring hrs/wk 4.2 3.5 4.7 3.3 dpts w/o female faculty 3/23 2/8 1/4 0/1 * sig at .10 level ** sig at .05 level 29