5: Conclusion

[next chapter] [previous chapter] [contents]

Change in government policy, or of shifts in student demand, can have a significant impact on the size and shape of higher education. It is important to assess and estimate such an impact for a number of reasons, not least for budget purposes. Completion and dropout rates can be included in a measure of output and efficiency of the higher education sector. The projection of the former can also be used to manage and develop policy on national skill formation. Despite the existence of relatively very good data, little analysis of the sort just mentioned has been done in Australia. This project was undertaken to provide a model that could be used for these purposes.

An input-output model based on the theory of absorbing Markov chains was proposed to describe the student flows through the higher education system, and also to make projections of the number of students enrolled, the number completing and the number dropping out of courses. The model was outlined in Chapter 2. Models of this sort have been considered before but have not been persisted with, probably because of the lack of sufficiently good data that is needed to estimate student flows through various stages of the system. As such, flow data are not as yet available to us either. However, the currently available stock data on higher education in Australia from DEET, allows flow statistics necessary for this model to be closely estimated. Exploratory analysis of the data is presented in Chapter 3.

5.1 Main Results

A range of summary statistics, such as the probability of completing a course, the average time in the system and the average time to complete a course are estimated for undergraduate students. It was found that these varied, and sometimes considerably, with the age at which the student commenced the course, with gender and field of study. Projections of student numbers and course completions by gender and field of study are made. Similar analysis and projections for postgraduate students by level of course, but not by field of study, are also made.

Some general results for Australian undergraduate students are:

• the probability of completing a course for an Australian student varies from 58 percent for one who commences the course at an age between 25 and 34 years, to 74 percent for one who commences it at an age of 20 years;

• a female student has a higher chance of completing a course and takes a shorter time to complete it than a male student who commences the course at the same age;

• a 20-year-old Australian student has the highest chance of completing a course, with the chances for a female being 79 percent and that for a male 69 percent;

• a student who commences a Business or Engineering course at an age of 24 years or more, or one who commences an Architecture or Science course at an age over 29 years has a 50 percent or less chance of completing it;

• a student who commences a course at an age of 21 years takes, on average, the shortest time (3.1 years) to complete it; and

• a female student takes less time, on average, to complete a course than a male student who commences at the same age, with the difference in the times as much as 0.7 years.

Some of the variation in the results may be related to the student's mode of study- that is, full-time, part-time or external. Another factor which may affect results is credit transfer from prior courses. These issues could be matters for further research.

The results for postgraduate students can be summarised as:

• the probability of completing a Research degree or an Other Postgraduate course vary quite a lot with the age at which the course is commenced, but that of completing a Master's by Coursework is almost invariant with the course commencement age;

• in general, male students have a better chance of completing a Research degree than female students who commence the degree at the same age;

• female students have a better chance of completing an Other Postgraduate course than male students who commence the course at the same age;

• the mean time to complete a Research degree varies from 4.4 to 6 years, that of completing a Master's by Coursework from 2.7 to 3.1 years and that of an Other Postgraduate course from 1.5 to 2.1 years; and

• in general, a male student takes, on average, a shorter time to complete a Research or Master's by Coursework degree than a female student commencing the course at the same age.

The same factors that may be contributing to the variation in the results for undergraduates may also be contributing to the variation in the case of postgraduates. There may also be variation due different fields of study. Since the category Research includes both doctorates and Master's by research, the results for time to completion are likely to be underestimates for doctorates and overestimates for Master's by research.

The undergraduate projections reflect:

• the 1993 to 1994 dynamics of students' progress through the system;

• the demographic changes projected to occur between 1995 and 2001;

• a constant school retention rate as calculated from the 1993 and 1994 school enrolment data;

• a constant proportion, based on the 1994 data, of Year 12 students continuing into higher education; and

• a constant proportion, based on the 1994 data, of non-school-leavers entering higher education.

Under this scenario the projections for Australian students up to the year 2001 can be summarised as follows:

• the course commencements are projected to decline from 1995 to 1998 and then slowly rise until 2001;

• the increase in the number of course commencements between 1995 and 2001 is projected to be 0.4 percent, but male numbers are projected to decline by 0.3 percent and female numbers to increase by 0.9 percent;

• course commencements in Architecture, Engineering and Science are projected to decline between 1995 and 2001;

• the number of students enrolled is projected to increase by 4.6 percent between 1995 and 2001;

• the number of students enrolled in Education (Initial training) are projected to decline by 1.4 percent, while those in Law are projected to increase by 13.2 percent between 1995 and 2001;

• the number of course completions is projected to increase by 5.2 percent between 1995 and 2001, with females making up 60 percent of all completions in 2001; and

• course completions in Education (Initial training) are projected to decline by 7.5 percent, while those in Law are projected to increase by 20 percent.

The postgraduate projections reflect:

• the 1993 to 1994 dynamics of students' progress through the system;
• the demographic changes projected to occur between 1995 and 2001; and
• a constant proportion, based on the 1994 data, of a particular age group entering post graduate courses.

Under this scenario the projections up to the year 2001 can be summarised as follows:

• total course commencements are projected to increase by 1.7 percent between 1995 and 2001, with male commencement numbers going up by 2 percent and that of females by 1.5 percent;

• the number of students enrolled for Research, Master's by Coursework and Other Postgraduate courses is projected to increase by 16.2, 14.6 and 6.7 percent, respectively;

• male Research degree completions are projected to increase by 33.7 percent and that of females by 45.4 percent between 1995 and 2001; and

• Master's by Coursework degree completions are projected to increase by 22.2 percent between 1995 and 2001, with male completions by 18.6 percent and that of females by 26.5 percent.

The model evaluation suggests that matrix Q and R are reasonably stable over at least a five year period. This is especially so when data are at a high degree of aggregation. In general, the prediction of total enrolment was more accurate than that for course completions. The mean absolute percentage error (MAPE) of predicting total enrolments of all undergraduate students was 0.81 percent and that of postgraduate students was 1.79 percent. MAPE for predicting course completions by all undergraduate students was 3.54 percent and that for postgraduates it was 9.65 percent. The exercise on model evaluation showed a structural change that has possibly occurred for the Master's by Coursework courses over the last few years.

5.2 Limitations

The results in this report are based on the movement of students who were enrolled in 1993 either to enrolment in 1994 or out of the system (as a completion or a dropout), together with estimates of commencing students for the years 1995 to 2001.

The experience of students enrolled in 1993 may not be the same as students who were enrolled in earlier years, or who enrol in later years. However, earlier analysis on data from 1989 to 1992 seem to indicate little variation, at least at the aggregate level, in the rate of progression of students through and out of the system.

The projection of new intakes is a much more tentative matter. Government policy on student places can have a major impact on this, as can factors affecting the distribution of student demand for places. In this report, the estimates of commencements for the period 1996 to 2001 are largely a reflection of the demographic changes, a constant Year 12 retention rate and a constant participation rate for non-school-leavers. The model that we have used and the assumptions underlying it is but one of a range of plausible models for projecting student intake. Projections of student intake using alternative models or underlying assumptions can easily be linked into the current input-output model to generate projections of student enrolment, completions and dropouts.

5.3 Further Work

The software used so far is based on a mixture of mainframe and desktop platforms. However, it is not inconceivable to design all procedures to run on a desktop or a mainframe platform.

With the existing data and the current model further analysis can be performed. For example, analysis can be done for each State, though some fields of study estimates for small states may not be all that reliable. Earlier work on this project (see Coleman and Burke 1993) showed there are some differences in students' rate of progress through the system between States.

Some analysis by the student's enrolment status-that is, full-time, part-time or external-can also be carried out. However, existing data does not allow the estimation of course completions and dropouts as the course completions file does not indicate enrolment status.

The estimates that are reported can be converted to obtain approximate EFTSUs using the current ratios of students to EFTSUs in the various fields of study analysed. This could be refined when analysis by enrolment status is undertaken. The projections in EFTSUs can be used to project expenditure levels.

A great deal more work can be undertaken if full cohort data were available for at least a pair of years - say 1994 and 1995 - and the student enrolment file was linked to the completions file. In particular, it may then be possible to model those students who transfer from one course to another within a university. In the current analysis these students are treated as dropouts from the first course and commencing in the other. The additional data would not much affect the capacity of the model to project total student numbers or completions. However, the picture of progression through courses and to completion could be substantially enhanced. A much clearer picture would emerge about the pathways of students and the proportion of a generation achieving a university qualification.

Projections made with the model outlined in this report will be linked with projections of immigration and of occupational demand in a joint study of Medium Term Forecasts of Supply and Demand for the Professions and some Skilled Occupations between the Centre for the Economics of Education and Training, the Centre for Policy Studies and the Centre for Population and Urban Research, at Monash University. This project is funded by a large Australian Research Council grant.

[return to top]