Student attrition at a new generation university

Transcription

1 CAO06288 Student attrition at a new generation university Zhongjun Cao & Roger Gabb Postcompulsory Education Centre Victoria University Abstract Student attrition is an issue for Australian higher educational institutions. This is particularly the case for newly established universities. In order to have a better understanding of the issue of student attrition, Victoria University recently conducted a project exploring the patterns of attrition of its bachelor level students. The students investigated were three cohorts of domestic bachelor level students who commenced in 2002, 2003 and 2004, with 4,405, 4,414, and 3,684 students in each of the cohorts respectively. Attrition rates of different groups of students were examined in each of the cohorts and significant factors contributing to attrition were identified. It is expected that the findings in this project will be both useful for the University and other similar universities in addressing the issue of student attrition effectively. Introduction Student attrition is an important issue for many Australian universities. According to the Department of Education, Science and Training (DEST) (2001; 2004a), the attrition rates for tertiary students in Australian higher education (HE) institutions ranged from 15%-5 over the past decade, with an average attrition rate of around 2. Australian universities are paying increased attention to reducing student attrition, both because it results in considerable costs to the student (e.g. fees, opportunity costs, emotional costs) and to the institution (e.g. loss of fees, recruitment costs, tuition costs). The fact that DEST now uses commencing domestic bachelor student attrition rate as one of the performance indicators in allocating its Teaching and Learning Performance Fund is another driver. Australian higher education institutions have organised themselves into several groups: the Group of Eight, the Innovative Research Universities, the Australian Technology Network, the Regional Universities and the New Generation Universities (DEST, 2004b). The New Generation Universities are all relatively new universities and, as such, face challenges in attracting and retaining students, and in attracting research funding. Victoria University is a member of this group. It is a dual-sector university that was founded in 1992 from two colleges of advanced education (Western Institute and Footscray Institute of Technology) and enlarged in 1998 by amalgamation with a large TAFE institute (Western Melbourne Institute of TAFE). It therefore provides a wide range of both TAFE and HE programs from work for the dole to PhDs. It currently has over 45,000 students and about half of these are enrolled in TAFE and half in HE. About half are part-time students. Most of the University s campuses are situated in the western Melbourne region, an area now making a transition from manufacturing to logistics and characterised by high levels of cultural and linguistic diversity and relatively low SES. The Victoria University Act states that VU has a particular responsibility for providing tertiary education for the people of the western Melbourne region and the University takes this responsibility seriously. This commitment is encapsulated in the University s mission statement: The mission of Victoria University is to: transform the lives of individuals and develop the capacities of industry and communities within the western Melbourne region and beyond through the power of vocational and higher education (VU Strategic Plan ). Like other members of the New Generation Universities group, Victoria University is working at improving student retention, especially the retention of commencing students. For this reason, the Vice-Chancellor charged the

2 University s Postcompulsory Education Centre with investigating the patterns of attrition of commencing domestic bachelor students in Victoria University. This paper reports some of the major findings of this project. Previous research on student attrition Many theories of student attrition exist. They include economic theories (e.g. Braxton, 2003; Cabrera and Nora, 1994), psychological theories (e.g. Astin, 1984), sociological theories (e.g. Berger, 2000; Kuh and Love, 2000) and interactionist theories (e.g. Tinto, 1993), with each emphasising somewhat different factors that influence student attrition. For instance, economic theories maintain that economic factors play a significant role in student departure, psychological theories claim that the psychological characteristics of individual students are the key to understanding student departure and sociological theories emphasise the importance of the influence of social forces on student departure. The interactionist theory of Tinto (1993) is one of the most studied, tested and critiqued theories and it views student departure as a process of interaction between the characteristics of individual students and the academic and social environment of the university. According to Tinto, academic integration and social integration are critical to student persistence in the institution. Academic integration refers to the extent that students are integrated into the academic system of the university (the domain of formal education activities) and social integration refers to the extent of integration into the social system of the university (the domain of interactions between students and staff outside of formal education activities). Many empirical studies based, in whole or in part, on these theories have been reported. The factors investigated can be broadly divided into socio-demographic variables, prior experience variables and institutional variables. Socio-demographic variables related to student attrition include gender, age, language, geographic location, socio-economic status and country of birth. For example, there are reports that females are more likely than males to complete a course (e.g. Martin, Maclachlan and Karmel, 2001) but a recent study of Australian first year university students (Long, Ferrier and Heagney, 2006) indicated that gender differences in first year attrition rate were tiny. A number of studies (DEST, 2004; Martin, Maclachlan and Karmel, 2001) suggest that students aged between 17 and 20 have much lower attrition rates than older students. Language background has also been studied, with a study that was part of the Longitudinal Surveys of Australian Youth (LSAY), concluding that students from a language background other than English had a much lower attrition rate than that of English background students (McMillan, 2005). With respect to geographic location, it has been reported that students from urban areas had a higher completion rate than students from isolated areas (James et al, 2004) and students from capital cities had a lower attrition rate than students from provincial cities (McMillan, 2005). James et al (2004) concluded that higher SES students had lower attrition rates than lower SES students (James et al. 2004), although another study suggested that the association between completion rate and family wealth was not consistent across different groups of students (Carpenter, Hayden, and Long, 1998). In terms of the country of birth, a UK study (Johnes and McNabb, 2004) concluded that domestic students were more likely to withdraw than international students, while Australian data suggests a different pattern (DEST 2004). Australian studies have shown that prior academic achievement, as measured by Equivalent National Tertiary Entry Ranking (ENTER) scores and other standard indexes, is negatively related to attrition. For example, the LSAY study discussed above found an attrition rate from higher education of 5% for those with ENTER scores of 90 or more compared with a rate of 23% for those with ENTER scores of less than 70 (McMillan, 2005). Similarly, a US study suggested that a one standard unit increase in high school GPA was associated with a 2% increase in student persistence (Titus, 2004). A UK study (Jones and McNabb, 2004) found that school type also contributed significantly to student attrition, although this was not the case in the LSAY study (McMillan, 2005). Variables associated with institutional experience, such as broad field of study, basis for admission, type of attendance and employment status were all found to be related to student attrition. For example, Martin et al. (2001) found that veterinary science and health students had higher completion rates than students from other fields, while adult students across fields of study had an attrition rate of 21%. 2

3 The effect of type of attendance (full-time versus part-time) has also been studied, with most reports concluding that those who study part-time are more likely to leave their courses than those who study full-time (Krause et al, 2005; Hillman, 2005). However, the LSAY study based on the 1995 Year 9 cohort found no significant difference in attrition rate between full-time and part-time students (McMillan, 2005). Academic performance has consistently been identified as a predictor of attrition. For example, a US study indicated that the probability of persistence increased by 8% with an increase in of one unit in first year GPA (Titus, 2004). Employment while studying has also been explored with mixed results. McMillan s (2005) LSAY study concluded that students who worked 21 or more hours a week were much more likely to leave than students who were not working. However, a recent study (Bradley, 2006) reported that students who did not work and students who worked over 21 hours a week were both more likely than other groups of students to continue their studies. In summary, the factors influencing student attrition have been well researched, with some findings consistent across institutions and countries but with considerable variation in other findings. Method Our task was to investigate the effects of factors such as those outlined above on the commencing domestic student attrition rate at Victoria University in order to inform decision-making on strategies for improving student retention. We therefore focused on the following research questions: 1. What are the attrition rates of different groups of commencing students at Victoria University? 2. Which factors best predict commencing student attrition at Victoria University? We selected the following variables on the basis of previous research findings, the context of the University and data availability: Three cohorts of domestic students who commenced bachelor level courses in the years 2002, 2003 and 2004 were selected. The number of students in the three cohorts was 4,405, 4,414, and 3,684 respectively. The cohorts were provided by the unit responsible for reporting student data to DEST and we then added more variables from the University s data warehouse. Data reliability was checked by drawing a random sample of 100 students from the 2004 cohort and comparing each record with individual records in the student database. This found an 18% error rate in identifying commencing students and an even higher error rate in Basis for admission, with at least half of the Other category incorrectly categorised. Table 1: Variables studied Socio-demographic Prior experience Institutional Gender Age Language background SES Country of birth Region (Western Melbourne/Other) ENTER score Broad field of study Basis for admission Type of attendance (Full-time/Part-time Double/single degree Academic progress Campus Employment status The domestic commencing bachelors student attrition rate was calculated, using methods based on the DEST definition (DEST, 2004). This rate refers to the proportion of domestic students (expressed as a percentage) who commenced a bachelor level course at the institution who neither re-enrolled nor graduated in the year following their commencing year. 3

4 As a measure of academic progress, the student progress rate was also calculated, using the DEST definition. It was calculated as the proportion of assessed load (measured in EFTSU/EFTSL and expressed as a percentage) passed by a student in their commencing year. Logistic regression analysis, one of a family of regression analysis techniques (Tabachnick and Fidell, 1996), is often used to examine the influence of independent variables on a dependent variable, when it is a categorical variable such as student attrition (Jones and McNabb, 2004; McMillan, 2005). We therefore used this form of analysis to examine the contribution made by the factors studied to commencing student attrition in each cohort. We used two approaches when conducting the analysis: entering all variables simultaneously and entering variables in a stepwise forward mode. Results Overall student attrition rate The attrition rate for these students from their commencing year to the next academic year is summarised in Figure 1 below. For the three cohorts, the attrition rate was 23.5% in 2002, 24.1% in 2003 and 19.8% in N=4,405 N=4,414 N=3,684 Figure 1: Overall student attrition rate Student attrition rate by gender Males tended to have slightly higher attrition rates than females in all three cohorts (Figure 2). The difference ranged from 0.8 to 4.6 percentage points Female Male 1 N=2,509 N=1,896 N=2,519 N=1,895 N=2,037 N=1,647 Figure 2: Student attrition rate by gender 4

5 Student attrition rate by age Students in the age range of 20 to 24 years consistently demonstrated the lowest attrition rates (values between 18% and 21%), while the attrition rates for both younger and older students were similar across the cohorts (Figure 3) & under & over 1 N=2,353 N=1,136 N=911 N=2,338 N=1,225 N=851 N=1,816 N=1,193 N=675 Figure 3: Student attrition rate by age Student attrition rate by language background The attrition rates of those with a language background other than English (LBOTE) and those with an English language background were quite close in all the three cohorts (Figure 4) LBOTE English language 1 N=1,456 N=2,816 N=1,361 N=2,108 N=1,105 N=1,686 Figure 4: Student attrition rate by language background 5

6 Student attrition rate by socio-economic status (SES) Students with low SES postcodes had the lowest attrition rates in all three cohorts (18.3% to 22.6%) (Figure 5), while students with high SES postcodes had the highest attrition rates in 2003 and 2004, with values of 26% and 24.4% respectively Low SES Medium SES High SES 1 N=1,062 N=2,108 N=1,062 N=1,096 N=2,082 N=1,096 N=885 N=1,819 N=954 Figure 5: Student attrition rate by SES Student attrition rate by country of birth The attrition rate for students born in Australia was similar to that for those born overseas across the three cohorts of students (Figure 6) Australia Overseas 1 N=3,208 N=1,089 N=3,155 N=1,089 N=2,749 N=815 Figure 6: Student attrition rate by country of birth 6

7 Student attrition rate by region Students from the western Melbourne region consistently demonstrated lower attrition rates than those from other regions (Figure 7). The difference in attrition rate between the western region and other regions was 4.9, 7.0 and 8.5 percentage points in 2002, 2003 and 2004 respectively Western Melbourne Other regions 1 N=1,891 N=2,463 N=1,905 N=2,408 N=1,743 N=1,925 Figure 7: Student attrition rate by region Student attrition rate by ENTER The pattern of attrition rate for different ENTER score groups varied somewhat between cohorts (Table 2). Students with ENTER scores of 80 and above consistently had lower attrition rates (15% to 2), but there was no consistent pattern for other groups..the high number of cases with unknown ENTER scores is notable. Table 2: Student attrition rate by ENTER score ENTER score Year 2002 Year 2003 Year 2004 N AR N AR N AR 49 & under % % % % % % , % % % % % % 80 & over % % Unknown 1, % 1, % % 7

8 Student attrition rate by broad field of study Student attrition rates varied across the different fields of study in the three cohorts (Table 3). However, the attrition rates in the field of Education were consistently low (7.7% to 13.1%) and those in the fields of Engineering and related technologies and Natural and physical sciences were consistently high (26% to 37%). Table 3: Student attrition rate by broad field of study Broad field of study Year 2002 Year 2003 Year 2004 N AR N AR N AR Agric, environ & rel studies N/A % Architecture & building % N/A N/A Creative arts % % % Education % % % Eng & related technologies % % % Health % % % Information technology % % % Management and commerce 1, % 1, % Natural and physical sciences % % % Society and culture 1, % % % Student attrition rate by degree type Student attrition rates by type of degree (single degree/double degree) did not show consistent difference across the three cohorts. For the 2002 and 2004 cohorts, the attrition rates of single degree students were slightly higher than those of double degree students. However, for the 2003 cohort, the attrition rate of single degree students was almost the same as that for double degree students Single degree Double degree 1 N=4,129 N=276 N=4,120 N=294 N=3,414 N=270 Figure 8: Student attrition rate by degree type 8

9 Student attrition rate by basis for admission Students whose basis for admission was Secondary education (i.e. mainly school leavers) consistently had higher attrition rates in all three cohorts, with values between 24% and 29%. Those admitted on the basis of TAFE qualifications had attrition rates between 2 and 22% (Figure 9). In interpreting these results, it must be remembered that many TAFE articulators were included in the Other category Secondary education TAFE Award Other 1 N=1,698 N=365 N=2,067 N=1,707 N=522 N=2,184 N=1,133 N=486 N=2,065 Figure 9: Student attrition rate by basis for admission Student attrition rate by type of attendance (full-time/part-time) Part-time students had substantially higher attrition rates than full-time students (Figure 10). The attrition rates of full-time students were between 13% and 17%, while those for part-time students were between 39% and 47% Full-time Part-time 1 N=3,193 N=1,212 N=3,317 N=1,097 N=2,810 N=874 Figure 10: Student attrition rate by type of attendance 9

10 Student attrition rate by student progress rate Student progress rates were calculated and grouped into the following six categories in Figure 11:, 1 to 24.9%, 25 to 49.9%, 50 to 74.9%, 75 to 99.9% and 10. As shown in the figure, the attrition rate increased with decreasing progress rate. Students with progress rates between 75% and 10 generally had the lowest attrition rate (9% to 14%), while those with a progress rate of 0 consistently had the highest attrition rate (57% to 67%). Students who passed all units (progress rate 10) had slightly higher attrition rates than students with progress rates in the next category % % % % 1 2,293 N=753 N=439 N=242 N=89 N=589 2,186 N=739 N=500 N=309 N=131 N=549 1, N=386 N=224 N=109 N=352 Figure 11:.Student attrition rate by progress rate Student attrition rate by campus Student attrition rates at the different campuses varied between cohorts (Table 4). However, the attrition rates for Campuses A and E were the lowest in all three cohorts. In contrast, the attrition rates for Campus I and Online were the highest in the three cohorts. Table 4: Student attrition rate by campus Campus Year 2002 Year 2003 Year 2004 N AR N AR N AR A % % % B N/A % N/A C % N/A D 2, % 2, , % E % % F N/A % G 1, % % % H % % I % % % Online % %

11 Student attrition rate by employment status Students who indicated that they had part-time employment at the time of enrolment in their commencing year tended to have the lowest attrition rates (17% to 22%) (Figure12). Those who indicated that they were employed full-time at enrolment had the highest attrition rates (24% to 26%) Not employed Employed part-time Employed full-time 1 N=1,566 N=2,116 N=443 N=1,509 N=2,129 N=489 N=1,024 N=1,873 N=399 Figure 12: Student attrition rate by employment status Factors affecting attrition The factors related to attrition were explored by means of logistic regression analysis. The results from entering all variables simultaneously are indicated in Table 6. The models obtained correctly predicted commencing student attrition in 82% to 85% of cases. Variables that were significantly associated with attrition in all three cohorts were Progress rate, Type of attendance, Region, Basis for admission and Broad field of study. The transformed coefficients (Exp (B)s in Table 6) suggest that an increase of one unit in Progress rate decreased the likelihood of attrition 0.05 to 0.13 times. For Type of attendance, the likelihood of attrition of part-time students was 2.8 to 5.0 times higher than that of fulltime students. The Region of origin of the students had an effect on attrition in favour of those from western Melbourne. The likelihood of attrition by students from the West was 0.5 to 0.6 times lower than that of students from other regions. Similarly, for Basis of admission the likelihood of attrition of students in the Other category was 0.4 to 0.8 times lower than that for the group where basis for admission was secondary education. Given the flawed nature of the data associated with this variable, it should be interpreted with caution. The effect of Broad field of study is complex, with less consistency across cohorts. The other variables were not identified as significant predictors in all or most of the three cohorts. Stepwise analysis confirmed that Progress rate, Type of attendance and Region were the three most powerful indicators of attrition. 11

12 Table 6: Logistic regression models of student attrition B Exp(B) B Exp(B) B Exp(B) Gender (ralative to Male) Female Age (relative to 25 & over) 19 & under * **.465 Language background (relative to English language background) LBOTE -.335* Unknown *** SES (relative to Low SES) High Medium Country of birth (relative to Australia) Overseas Region (relative to western Melbourne) Other region.552*** *** *** ENTER (relative to 80 & over) 49 & under Broad field of study (relative to Society and culture) Agric, environ &.rel studies Architecture & building Creative arts Education -.944* **.282 Eng & related technologies Health -.881*** * Information technology ** Management & commerce Natural & physical sciences * * Basis for admission (relative to Secondary education) TAFE Other -.305* ** ***.415 Type of attendance (relative to Full-time) Part-time 1.444*** *** *** Type of degree (relative to Single degree) Double degree Progress rate Progress rate *** *** ***.048 Campus (relative to Campus I) A B C D E F G H ** Online Employment status (relative to Not employed) Part time employed Full time employed Constant Model indicators Baseline p 80.4% 77.4% 79.9% Model N 2,779 2,882 3,497 Pseudo R square % correctly predicted 83.8% 82.1% 84.9% ***p<0.001, **p<0.01, *p<

13 Discussion In interpreting these results, it is necessary to bear in mind the limitations of the data, both in the identification of commencing students and in classifying the basis of admission. With that caveat, there are some clear implications arising from these findings. A low level of academic achievement in the commencing year was the most powerful predictor of student attrition in this study. This may reflect deficiencies in the academic integration of these students, as suggested by Tinto (1993). It certainly draws our attention to the need for the early identification of students who are not performing well in their first semester and for timely action to support them. The University s new student progress policy specifically addresses this issue. Part-time enrolment by commencing students was an emphatic predictor of attrition in all three cohorts. Other studies have shown that part-time students are more likely to drop out than full-time students (Krause et al, 2005; Hillman, 2005) but what is striking in this study is the size of the difference. From an institutional viewpoint, an attrition rate of about 15% for full-time commencing students may be acceptable but a rate three times that for part-time students is quite unacceptable. It strongly suggests that the University must do more to support its parttime students. The University is addressing this issue as a matter of urgency. Where students come from also matters. Students from other regions are more likely to discontinue their study than students from western Melbourne. This difference may be related to proximity and perhaps to local loyalty to the University. The national First Year Experience Survey showed that travel time is one factor that causes first year students at VU to consider leaving (Gabb, 2006), and this would be particularly relevant to students who live outside western Melbourne. It is also likely that some students from outside of the University s region elected to enroll at VU not because they wanted to attend the University but because they had limited options from which to choose. The University is already strengthening its recruitment from western Melbourne by building strong partnerships with the secondary schools in its region. Our findings are that low academic progress, part-time enrolment and coming from outside of Western Melbourne are the three main predictors of attrition at Victoria University. But just as notable are the variables that were not identified as good predictors of attrition. These include SES, language background, country of birth and ENTER scores. These findings are not consistent with those of some previous studies (e.g. DEST, 2004; Martin, Maclachlan and Karmel, 2001; McMillan, 2005) but, given the characteristics of the University s entering cohort, they are encouraging. They suggest that the commencing year experience of students with low SES, non-english background and lower ENTER scores was positive enough to retain many of them in their second year. The retention of these students is consistent with the University s focus on social justice. To conclude, this study has identified several significant factors that influence the attrition of commencing students in a new generation university. The institution is already acting to address the major issues identified here. In more general terms, the institutional uniqueness of some of the factors and the difference between of some of our findings and those of studies elsewhere highlights the utility of such institutional research. In making key policy decisions, an institution should not only take existing research findings into account but also supplement them with a critical examination of its own realities. 13

14 References Astin, A.W. (1984). Student involvement: a development theory for higher education. Journal of College Student Personnel, 25, Berger, J.B. (2000) Optimizing capital, social reproduction, and undergraduate persistence: a sociological perspective. In J.M. Braxton (Ed) Reworking the student departure puzzle, (pp ). Nashville: Vanderbilt University Press. Braxton, J.M. (2003) Persistence as an essential gateway to student success. In, S. Komives & D. Woodard (Eds), Student services: a handbook for the profession (4th ed), (pp ). San Francisco: Jossey-Bass. Cabrera, A.F. & Nora, A. (1994) College students perceptions of prejudice and discrimination and their feelings of alienation. Review of Education, Pedagogy and Cultural Studies, 16, Carpenter, P.G., Hayden, M. & Long, M. (1998). Social and economic influences on graduation rates from higher education in Australia. Higher Education, 35, DEST (2004a) Higher education attrition rates : a brief overview. Research note no. 1. Canberra: Department of Education Science and Training, viewed 10 November 2006, < DEST (2004b) Performance indicators and statistics used in the institution assessment framework. Canberra: Department of Education, Science and Training. DEST (2005) Student outcome indicators for the Learning and Teaching Performance Fund: Technical note no.2, Department of Education, Science and Training, Canberra, viewed 16 August 2005, < ducation/technical_note_2.htm> Gabb, R. (2006) 2004 First Year Experience Survey: findings from Victoria University respondents. Melbourne: Postcompulsory Education Centre, Victoria University. Hillman, K. (2005) The first year experience: the transition from secondary school to university and TAFE in Australia. Camberwell: Australian Council for Educational Research. James, R., Baldwin, G., Coates, H., Krause, K.L, & McInnis, C. (2004). Analysis of equity groups in higher education Canberra: Department of Education Science and Training. Johnes, G. & McNabb, R. (2004) Never give up on the good times: student attrition in the UK. Oxford Bulletin of Economics and Statistics, 66(1), Krause, K.-L., Hartley, R., James, R. & McInnis, C. (2005). The first year experience in Australian universities: findings from a decade of national studies. Canberra: Department of Education Science and Training. Kuh, G.D. & Love, P.G. (2000). A cultural perspective on student departure. In J. M. Braxton (Ed) Reworking the student departure puzzle: New theory and research on college student retention, Nashville: Vanderbilt University Press. Long, M., Ferrier, F. & Heagney, M. (2006). Stay, play or give away? Students continuing, changing, or leaving the university in the first year. Canberra: Department of Education, Science and Training, viewed 10 November 2006, < Martin, Y.M., Maclachlan, M. & Karmel, T. (2001). Undergraduate completion rates: an update, Canberra: Department of Education, Science and Training, viewed 20 October 2006, < McMillan, J. (2005) Course change and attrition from higher education. Camberwell: Australian Council for Educational Research. Milne, L., Holden, S. & Keating, S. (2006) Making articulation work: TAFE to higher education at Victoria University. Melbourne: Postcompulsory Education Centre, Victoria University. Tinto, V. (1993) Leaving college: rethinking the causes and cures of student attrition. Chicago: The University of Chicago Press. Titus, M.A. (2004) An examination of the influence of institutional context on student persistence at 4-year colleges and universities: a multilevel approach. Research in Higher Education, 45(7), Tabachnick, B. G. & Fidell, L. S. (1996) Using multivariate statistics, 3rd ed. New York NY: HarperCollins College Publishers. 14

15 Contact details Zhongjun Cao & Roger Gabb Postcompulsory Education Centre Victoria Universitry PO Box Melbourne VIC