Undergraduate Completion Rates: An update

(internet only publication)

Yew May Martin
Maureen Maclachlan
Tom Karmel

01/F
Occasional Paper

Higher Education Group
Department of Education, Science and Training

December 2001

• Introduction

• Updating the results of the 1992 cohort from 1997 to 1999

• Revision of the final completion rate model

• Robustness of completion rate estimates

• Comparison of the logistic regression models

• Conclusion

• Appendix A

Introduction

The success of undergraduates in completing their studies is an important issue for universities and the Australian community in general. The inaugural study by the Department of Education, Training and Youth Affairs (DETYA) on this topic was the paper by Urban et al (1999). This paper found that around 60 per cent of students who commenced an undergraduate higher education award in 1992 had completed an award at the same university by 1997. It further estimated, making use of the numbers of students who return to university with an incomplete award, that the final completion rate of a commencing cohort is around 80 per cent.

The longitudinal data base on which that paper was based has been expanded. It now includes an additional two years of data and an additional cohort of students who commenced in 1993. The purpose of this paper is to update the earlier findings and to test their robustness using the new cohort. We have three specific purposes. First, we update the estimates of completion for the 1992 cohort by tracking the students for an additional two years. Second, we re-examine the model used to estimate the final completion rate and modify it to take account of students who change awards within a university¹. Third, we test our estimate for the robustness by re-estimating the models using the 1993 cohort.

The main findings are:

• In respect of the 1992 cohort of commencing students, 64 per cent had completed an award at the same university by 1999 compared to 60 per cent by 1997;

• A significant proportion of students change awards within a university (that is, recommenced at that university without completing an award). When this is taken into account the estimate of the final completion rate is reduced;

• The estimates of completion rates are similar for the two cohorts. The rate for the 1993 commencing cohort is about one percentage point lower than the corresponding rate for the 1992 cohort: the estimates of the final completion rate are 71.6 per cent and 70.8 per cent for the 1992 and 1993 cohorts, respectively;

• The earlier findings on the relationship of student characteristics are robust. Higher completion rates are associated with:

  • Women;

  • Full-time students;

  • Younger students;

  • Students with a TER;

  • Non-English speaking background students;

  • Non-Aboriginal and Torres Strait Islander students;

  • Urban students;

  • High socio-economic status students; and

  • Veterinary science and health students.

top

Updating the results of the 1992 cohort from 1997 to 1999

Based on outcomes to date, approximately 64 per cent of all undergraduate students who commenced at an institution in 1992 had completed an award at that institution by 1999. As expected, this figure is higher than that in the previous study of 60 per cent by 1997. Around 33 per cent had not completed an award and were not studying at the institution of commencement in 1999 compared to 34 per cent by 1997. Close to 3 per cent had not yet completed an award but were still studying. The corresponding figure was 6 per cent by 1997 in the previous study.

The same trends in performance for the 1999 results compared to the 1997 results in terms of student characteristics continue to prevail. Those of major interest include:

• Women are still more likely to complete an award than men, 67.8 per cent compared to 59.9 per cent;

• Students who study full-time have higher completion rates (69.6 per cent) than part-timers (53.2 per cent) while external students have the lowest completion rates (44.0 per cent);

• Completion rates are highest for younger students and generally decline as age increases;

• Students with a Tertiary Entrance Ranking (TER) score have significantly higher completion rates (68.7 per cent) than non-TER students (61.3 per cent);

• Compared to other domestic students Aboriginal and Torres Strait Islander students have significantly lower completion rates, 35.2 per cent compared to 64.8 per cent;

• Students with a non-English speaking background have completion rates (67.2 per cent) significantly higher than students with an English speaking background (64.0 per cent);

• Socio-economic status (SES) continue to have an adverse effect on completions, with students who have a SES status of most disadvantaged having the lowest completion rates;

• Urban students have higher completion rates (65.3 per cent) than students living in rural (63.2 per cent) and isolated areas (54.1 per cent); and

• Considerable variability of completion rates is still found across fields of study.

top

Revision of the final completion rate model

In their estimation of the final completion rates, Urban et al made use of the numbers of students who return to university with an incomplete award. Their model did not, however, take into account students who change awards within a university. From the available data, it is clear that a significant proportion of students change awards within a university (that is, they recommence at that university without completing an award). Subsequently, students who recommenced in subsequent years (that is, 1993 to 1999) at the same university without completing an award are excluded from our final estimation. We explain our approach using Figure 1. It should be noted that the flow chart presented in Figure 1 differs from the one in Urban et al in respect of boxes 1.1 to 1.3 and 3.1 to 3.3. This is because we exclude from the estimation students who recommenced in subsequent years at the same institution as their subsequent attempts are already captured by those who commenced with an incomplete award.^2,3

³ In addition, the expected completions by students who may be taking a break in 1999 and later return to resume study are not considered separately in our model. This number would be so small (see Urban et al) as to not materially affect the estimate of the completion rate.

top

Figure 1: Undergraduate flows and completions as of 1999 (1992 Cohort)

The final probability of completion for the 1992 cohort is, therefore, defined as:⁴

where

  and the relevant terms are defined as:

P₁ the probability of completion in the first period of attending university;

P₂ the probability of completing in subsequent periods of attending university;

q the probability of returning after leaving university without completing an award;

N the size of the cohort; and

N_R the number of returning students.

From Figure 1,

P₁ = [(1.1)+(1.2.1)]/(1) = 0.577

P₂ = [(3.1)+(3.2.1)]/(3) = 0.592

N = (1) = 87,180

N_R = (3) = 21,110

and q = 0.464

Based on the results of performance of students to date, it is estimated that 72.1 per cent of undergraduate students who are new to higher education (i.e. without a previous award) in the 1992 student cohort will complete an award at some stage. This final estimate is 9.5 percentage points lower than the final estimate of 81.6 per cent in the Urban et al (1999) paper.

top

Robustness of completion rate estimates

To test our estimates for robustness, we re-estimate the model using the 1993 cohort of commencing students.

In terms of outcomes, approximately 63 per cent of undergraduate students who commenced at an institution in 1993 had completed an award at that institution by 1999. Around 33 per cent had not completed an award and were not studying at the institution of commencement in 1999. Just over 4 per cent had not yet completed an award but were still studying (Table 1).

Table 1: Undergraduate flows and completion rates for the 1993 Cohort

Source: 1993 Student Cohort.

Figure 2 shows a generally similar pattern of distribution in the proportion of undergraduates completed for both the 1992 and 1993 cohorts. Completions peaked at the end of the third year of commencement, fell slightly in the fourth year, but dropped sharply in the fifth year and declined gradually after that. At the end of the seventh year since commencement, 63.3 per cent of undergraduates in the 1992 cohort and 62.8 per cent of undergraduates in the 1993 cohort respectively have completed an award.

top

Figure 2: Proportion completed, 1992 and 1993 Cohorts

Note: Completion includes those who had completed the same or other levels.
Source: 1992 and 1993 Cohorts

Completion performance according to major student characteristics for the 1993 cohort is almost identical in terms of trend and magnitude as those for the 1992 cohort. These are:

• Women are much more likely to complete an award than men, with completion rates of 66.5 per cent compared to 57.9 per cent for men;

• Students who study full-time have higher completion rates (67.7 per cent) than part-timers (52.8 per cent) while external students have the lowest completion rates (41.9 per cent);

• Completion rates are highest for younger students and generally decline as age increases;

• Students with a Tertiary Entrance Ranking (TER) score have significantly higher completion rates (66.8 per cent) than non-TER students (59.7 per cent);

• Compared to other domestic students, Aboriginal and Torres Strait Islander students have significantly lower completion rates, 37.7 per cent, compared to 63.3 per cent for non-ATSI students;

• Students with a non-English speaking background have completion rates slightly higher (65.3 per cent) than students with an English speaking background (62.4 per cent);

• Socio-economic status (SES) also affects completion rates, with students who have a SES status of most disadvantaged having the lowest completion rates (59.6 per cent);

• Urban students have higher completion rates (63.7 per cent) than students living in rural (61.6 per cent) and isolated areas (54.8 per cent); and

• Considerable variability of completion rates is found across fields of study, with, for example, veterinary science (91.8 per cent) and health students (79.3 per cent) having higher completion rates than agriculture (56.8 per cent) and arts students (57.9 per cent).

Figure 3 sets out the pathways to completion for students from the 1993 cohort who are new to higher education, those who commenced on the basis of a previous award and those who commenced on the basis of an incomplete prior award.

top

Figure 3: Undergraduate flows and completions as of 1999 (1993 Cohort)

From Figure 3,⁵

P₁ = [(1.1)+(1.2.1)]/(1) = 0.565

P₂ = [(3.1)+(3.2.1)]/(3) = 0.609

N = (1) = 92,507

N_R = (3) = 21,702

and q = 0.446

Applying the same final estimation as for the 1992 cohort, it is estimated that 70.8 per cent of undergraduate students who are new to higher education in the 1993 student cohort will complete an award at some time. This compared with a final estimate of 71.6 per cent for the 1992 cohort based on performance over the seven years since commencement.⁶

⁶ For purposes of comparison, these estimates are based on performance over the seven years since commencement, that is, completions by 1998 for the 1992 cohort and completions by 1999 for the 1993 cohort.

top

Comparison of the logistic regression models

When the same logistic model is applied to the 1993 cohort, little difference is found between the two cohorts. The characteristics that influence completion in the 1993 cohort are the same characteristics that influence completion in the 1992 cohort. These results are consistent with those reported earlier using cross-tabulation analysis.

As indicated in Table 2, other thing being equal, female students are again more likely to complete a course than male students. Completion rates are highest for younger students and generally decline as age increases (see Table A1 in Appendix A).

It is also evident that the greater the involvement in academic study the greater the probability of completing an award, with full-time students having the highest completion rates, followed by part-timers and then external students.

Students who enter university on the basis of previous higher education, professional qualification or work experience also have higher probabilities of completion than students who enter on some other basis. On the contrary, students who finished secondary education and those on special entry provision are less likely to complete than students who enter on other basis.

Aboriginal and Torres Strait Islander (ATSI) students are less likely to complete than non-ATSI students, while students from a Non-English speaking background are more likely to complete than their English-speaking counterparts.

Socio-economic status (SES) continues to have an adverse effect on completions, with students who have a SES status of most disadvantaged having the lowest completion rates. Tertiary entrance ranking also plays an important role in determining whether a student completes or not. Students with a TER have significantly higher completion rates than non-TER students.

While there is little difference in the completion rates among students from either urban or rural regions, students living in isolated area have significantly lower completion rates. Finally, considerable variability in completion rates is still found across fields of study, being higher for veterinary science and health and lower for arts, humanities and socio science.

top

Table 2: Predicted probability of completion, 1992 and 1993 cohorts

top

Conclusion

It is clear that there is little variation between the 1992 and 1993 cohorts in terms of the pattern of distribution in completion rates, in terms of final estimates (less than one percentage point), and in terms of the trends and magnitude of completion rates in regards to key student characteristics. This provides reassurance as to the representativeness of the findings and the robustness of the underlying methodology.

top

Appendix A

Table A1: Binomial Logistic Regression of Completion/Non-completion, 1992 and 1993 cohorts

Source: 1992 and 1993 Student Cohorts.

home  |  search  |  site map

Any comments or queries should be sent to: highered@dest.gov.au

This page was last updated on Tuesday, 04 December 2001
Department of Education, Science and Training
Copyright © Commonwealth of Australia
DEST Web Site Privacy Statement
Disclaimer