Australian Educational Researcher
Vol. 19, No. 2 1992
COEDUCATION AND FACTORS AFFECTING CHOICE OF UNIVERSITY COURSES Thomas Lumley
Abstract. One of the arguments in favour of single-sex schools for girls is that they encourage the study of traditionally male-dominated subjects. To test this hypothesis, the association between coeducation and faculty choice is examined for students from private schools entering Monash University in 1990 and 1992. No significant association is found for girls at non-Catholic independent schools or boys or girls at Catholic schools. For boys at non-Catholic independent schools, coeducation is associated with higher chance of studying science-based courses at university. There is a similar, but not statistically significant, association for girls at Catholic schools.
Introduction There is still much debate over the benefits of single-sex education, especially in its effects on the study of traditionally male-dominated subjects by girls. While many surveys of the area find some connection between single-sex education and choice of, or proficiency in, these subjects, there are usually important confounding factors which make the clear interpretation of the results difficult. One of the consistent claims about single-sex schooling for girls is that it increases the likelihood that they will study science and mathematics subjects, and that they will achieve highly in these subjects. In 1985-86 the Australian Schools Commission funded a review of existing research of the effects of coeducational schooling. This survey, carried out by Judith Gill, highlighted some of the contradictions She noted (Gill, 1988: 14-15) that: The surveys that have been reported from both Australian and overseas research show that a multiplicity of interacting variables is involved in the connection between girls' school experiences and variously defined educational outcomes. and concluded
52
Lumley
At this stage it is not possible to give a clear-cut answer to the question of the relative merits of single sex and coeducation from the perspective of optimal education for girls. However, one area in which she did venture a conclusion was in the area of science and mathematics: when girls are seen to achieve highly or to enrol in large numbers in mathematics and science it seems that single sex learning experiences are frequently involved. However, a more recent study of subject choice in secondary school (Ainley et al., 1990), found very little difference between coeducational arid single-sex schools in this area. Their analysis (pp. 114-115), based on 214 schools from around Australia, found that the difference in the proportion of girls studying science or mathematics between single-sex and coeducational schools was only about 3.5%. In contrast, the difference between the genders was about 15%. These figures were based on all sectors of the education system, and when ethnic background and the differences between sectors (in socio-economic status and proportion of coeducational schools) were taken into account, the difference fell to about 2.5%. While the study by Ainley et al. shows that very little variation in subject choice at school is attributable to coeducation, it does not address the question of achievement and subsequent study. The differing views on achievement and the lack of variation observed in subject choice at school raise the question of whether any effect exists at university level. The level of attainment in science and mathematics at school may be important for improving self-esteem and altering society's perceptions of these subjects, but it will not have much impact on career choice unless it continues in higher education. In 1990, Baldwin published in this journal an analysis of faculty choice among students entering Monash University in that year (Baldwin, 1990). The study concluded that there was no significant evidence of any effect of coeducation. This paper is an extension of that analysis, based on additional data and concentrating primarily on the choice of science-related courses.
Procedure The data for this study come from a questionnaire administered every second year to first-year enrollees at Monash University. The 1992 and 1990 data sets were used in the analysis. Completion of the questionnaire is voluntary, but the tedium of the enrolment process makes students fairly receptive to the request for information. As the population being surveyed is those who enrol, even if they
Coeducation and University Courses
53
later receive an offer elsewhere, the response rate is not known accurately. However, it is probably above 95% in the first and second rounds of enrolment, thus bias due to non-response cannot be very high. The proportion of missing data is significantly higher than 5%, due to errors in filling in the form and to entries missed by the optical mark reading machine used to enter the data. Monash University has recently undergone amalgamation with the Chisholm Institute of Technology so the 1992 data, which are from the Clayton, Caulfield and Frankston campuses, cover the previous Monash University and Chisholm Institute of Technology. This gives a fairly broad sample of students entering tertiary education. The 1990 data are from before the amalgamation and so refer only to what is now the Clayton campus. Two important factors which affect any discussion of coeducation in Australian schools are the differences between private and government schools and the differences between the preferences of male and female students (irrespective of their educational situation). Table 1 shows the differences in enrolment in the various faculties at Monash University in 1992. Table 2 has the faculties divided into 'science-based' and Table 1 Gender Differences in Faculty Choice - 1992 entrants
Arts Computing & Info. Tech.* School of Business ECOM** Education Engineering Law Medicine Professional Studies Science * Computing & Information Technology ** Economics, Commerce & Management
Male
Female
Total
203 (26%) 387 (72%) 412 (54%) 255 (47%) 29 (~7%) 455 (84%) 96 (53%) 76 (50%) 172 (30%) 331 (48%)
589 (74%) 149 (28%) 347 (46%) 288 (53%) 139 (83%) 84 (16%) 86 (47%) 77 (50%) 410 (70%) 357 (52%)
792 536 759 543 168 539 182 153 582 688
54
Lumley
other courses as described below. The Faculties of Education and Professional Studies have been removed from this table for reasons explained below. Table 2 Gender Differences in Area of Study - 1992 entrants Science-Based Course No Yes Male Female Total
Total
695 (45%) 994 (63%)
863 (55%) 588 (38%)
1558
1689
1451
3140
1582
X2 = 104.87; p < 10-7 Table 3 gives the 1992 faculty enrolment patterns for the different sectors of the education system: government, Catholic and other independent schools. Students from schools other than these three types are not included in the table, but these are mainly overseas students, whose patterns of choice would be different anyway. As these tables show, there are large differences in both areas. The difference between private and government schools can be attributed almost entirely to the higher scores attained by private school students in year---12 (allowing them greater access to high-demand courses) and there is little overall difference between science and non-science courses. The gender differences do not appear to be related to differences in marks, and are predominantly divided on a science/non-science basis. Thus, to examine the effects of coeducation it is necessary to consider separately male and female students, and those from different sectors of the education system. The major problem in comparing coeducational and single-sex schools in Victoria is that nearly all the single-sex schools are in the private sector, and the small number of single-sex government schools include two selective entry schools (MacRobertson Girls High and Melbourne High). The other single-sex government schools are also atypical in other ways, as, for example, they take students from outside their local area. For this reason, it is almost impossible to give a meaningful comparison for any sample including government schools, and the rest of this analysis was restricted to students from private schools. In order to increase the power of the analysis, the faculties were divided into two groups: science-based courses (science, engineering, medicine, computer science) and other courses, loosely classifiable as humanities-related (arts, economics & commerce, business, law). Students in Education, which is hard to
Coeducation and University Courses 55 classify and also predominantly teaches graduates, and Professional Studies, which includes too wide a variety of courses, were excluded from the analysis. Combined courses (e.g. Law/Arts, Law/Science) were classified as 'science-based' if either of the component courses was science-based. This does not affect the results to any great extent, as few students combine 'science' and 'non-science' courses. As well as increasing the power and ease of interpretation of the analysis, grouping of faculties helps reduce bias due to the differing selection scores required for entrance by the various faculties. The previous analysis of the 1990 data set (Baldwin, 1990), which did not find any significant effects of coeducation, did not group the faculties in any way. Table 3 Faculty Choice by Education Sector - 1992 entrants
Arts Computing & Info. Tech.* School of Business ECOM** Education Engineering Law Medicine Professional Studies Science * **
Government
Catholic
Independent
Total
227 (34%) 217 (53%) 301 (50%) 139 (34%) 90 (60%) 234 (51%) 56 (34%) 44 (31%) 275 (53%) 227 (34%)
174 (26%) 83 (20%) 142 (23%) 101 (25%) 34 (23%) 102 (22%) 30 (18%) 25 (18%) 104 (200/0) 174 (36%)
273 (41%) 109 (27%) 162 (27%) 166 (41%) 26 (17%) 120 (26%) 78 (48%) 73 (51%) 144 (28%) 273 (41%)
674 409 605 406 150 456 164 142 523 674
Computing & Information Technology Economics, Commerce & Management
The students were asked about the type of school they attended 'for most of secondary education' and for 'the final year of secondary education'. There was very little variation in responses between these questions, .and the 'most of secondary education' response was used in the analysis. The sample used in the analysis of coeducation consisted of students entering Monash in 1990 or 1992 who attended a 'Catholic school' or 'Independent school (non-Catholic)'. The four groups (Catholic or other independent; male or female) were analysed separately giving the four 2x2 tables (Tables 4-7). As these tables
56
Lumley
are based on the private school students from both years, and the analysis of the effects of gender and school sector was done only for 1992 entrants, the totals in the tables will be different. Table 4 Boys at Catholic Schools Science-Based Course No Yes Coeducational Single Sex Total
55 (44%) 244 (49%)
71 (56%) 255 (51%)
299
326
Total 126 499
625
X2= 1.11" p = 0.29
Table 5 Girls at Catholic Schools Science-Based Course No Yes Coeducational Single Sex Total
88 (59%) 324 (67%)
60 (41°7o) 163 (33%)
412
223
X 2=2.49; p - 0 . 1 1
Total 148 487
635
Coeducation and University Courses
57
Table 6 Boys at Independent Schools Science-Based Course No Yes Coeducational Single Sex Total
Total
187 (43%) 267 (53%)
248 (57%) 240 (47%)
435
454
488
942
507
X2= 8.78" p = 0.003
Table 7 Girls at Independent Schools Science-Based Course No Yes Coeducational Single Sex Total
Total
223 (70%) 421 (67%)
97 (30%) 211 (33%)
320
644
308
951
632
X2 = 0.92; p = 0.34
Analysis In order to detect any effect of coeducation on choice, and to measure its extent, two statistics were calculated. The chi-squared statistic and its significance level are given with each table. The other measure of the extent of any effect is the odds ratio for the table. This is the odds of a student from a single-sex school being enrolled in a science course, divided by the odds of a student from a coeducational school being enrolled in a science course. If this is greater than 1 then students from single-sex schools in this group are more likely to study science than students from coeducational schools in this group. A 95% confidence interval was also calculated for the odds ratio using Cornfield's approximation, described in (Fliess, 1981: 71-75). The odds ratios and confidence intervals are given in Table 8 and a graph in Figure 1. For
58
Lumley
comparison, the odds ratio for the effect of gender on faculty choice (from Table 2) is 0.48 with a 95% confidence interval (0.41,0.55). Table 8 Effect of Coeducation on Area of Study - odds ratios
Boys Girls Boys Girls
Odds ratio
95% confidence interval
0.81 0.74 0.68 1.15
0.53-1.23 0.49-1.10 0.52-0.89 0.85-1.56
at Catholic schools at Catholic schools at other independent schools at other independent schools
Figure 1' Effect of Coeducation on Area of Study - odds ratios
8¢ienoe-based
Catholio:
Coeduoatlon favours:
ii i
boys
i
I,
Independent:
boys
ii
~ i i'l
i
i;!l ;
Catholic: girls
!
i
independent: girls
'
0.1
~ :
i
!
,~--~-+
i i ' ! J ; t
;
,
II
I
i
i
i1
~ii
i ! [-
1 Odds Ratio
Arts-based
i t
i
i
!
t t
i
1
iI
!
i
r
i
i
t
I
*
=
j
i
t !
ii
J
!
t i
10
The difference between coeducational and single-sex schools is found to be significant, in both the statistical and the usual sense, for boys in independent schools. For girls in Catholic schools the difference is also fairly large, though not statistically significant. In both cases, the odds ratio is less than 1.00,
Coeducation and University Courses
59
showing that students from coeducational schools were more likely to study science-based subjects. One problem in interpreting results about coeducation is that it may be associated with various social factors known to affect students' choices. For example, it is still true to some extent that the most prestigious independent schools are single-sex. To investigate this possibility, a stratified version of the odds ratios (Mantel, 1959) was calculated to correct for the effects of mother's & father's educational level and mother's & father's occupational level using the 1992 data. The corrected odds ratios calculated by this procedure did not differ appreciably from the crude odds ratios given above. This rules out any systematic effect of socio-economic status as responsible for the observed differences. In both cases it is quite possible that the explanation is something other than a real effect of coeducation. Anecdotal evidence suggests that some independent boys schools retain an expectation that the more able students will study humanities, which would produce an effect like that observed. For the Catholic girls, it may be the case that attitudes about suitable subjects for girls to study still persist at some Catholic girls schools. The other important result of this analysis is that single-sex education does not appear to favour the choice of science-based subjects to any significant extent in any of the groups. Even for girls at independent schools, the area where the claims of benefits are the strongest, the odds ratio is quite close to 1, showing no real evidence of an overall effect. Conclusions Examining the choice of faculty among students from private schools entering Monash University provides strong evidence that single-sex education does not have the overall effect of encouraging girls to study mathematics and science based subjects at university level. In fact, in the Catholic education sector, single-sex education may be associated with a reduced chance that girls will study these subjects. For boys in independent schools, single-sex education is also associated with a significantly lower frequency of studying science and mathematics based courses at university. All of these differences, however, are much less than the overall difference between the subject choices of boys and girls. The fact that no systematic effect of coeducation is observed for either boys or girls indicates that these differences may be due to other factors associated with coeducation. A school which decides to become coeducational is likely to differ from one that remains single-sex in other ways as well. Whether this is an important problem depends on the reason for one's interest in coeducation. From the point of view of education policy, these figures do not indicate any clear difference which is likely to result from a change to coeducation. On the other hand, from the point of view of someone choosing a school, the data show
60
Lumley
considerable differences between coeducational and single sex schools in some sectors. References
Ainley J., Jones W., and Navaratam K.K. (1990), Subject Choice in Senior Secondary School Australian Council for Education Research. Baldwin G. (1990), Single-Sex Schooling and subject choice: Patterns of enrolment at Monash University. The Australian Educational Researcher. 17(3) pp 47-64 Gill J. (1988), Which Way to School? A Review of the Evidence on the Single
Sex Versus Coeducation Debate and an annotated bibliography of the research. Canberra, Commonwealth Schools Commission. Fliess J.L..(1981), Statistical methods for rates and proportions New York: John Wiley and Sons. Mantel N., Haenszel W. (1959), Statistical aspects of the analysis of data from retrospective studies of disease J Natl Cancer Inst 22: 719-48.