Widening Participation issues among postgraduate students at the University of Bristol, 2008/09: a preliminary investigation Dr Tony Hoare and Beatriz Caicedo Velasquez, on behalf of the University’s Widening Participation Research Cluster University of Bristol December 2010 1 Contents 1. Preamble.......................................................................................................... 3 2. Practicalities .................................................................................................... 10 2.1 Source of data ............................................................................................ 10 2.2 Study population ........................................................................................ 12 2.3 Dependent Variables .................................................................................. 13 2.4 Independent variables ............................................................................... 13 3. Results 1 – simple descriptive ....................................................................... 17 3.1 Composition of undergraduates and postgraduates – demographic and educational characteristics ............................................................................. 17 3.2 Composition of undergraduates and postgraduates – widening participation students...................................................................................... 20 3.3 Motivations ................................................................................................. 23 4. Results 2 - Multivariate analysis .................................................................. 30 4.1 Our modelling approach ............................................................................ 30 4.2 Results…………………………………………………………………………..33 5. Summary and Discussion ............................................................................ 45 6. Postscript………………………………..……………………………………………52 7. References…….…………………….….……………….…………………………..53 2 Preamble Since Lord Dearing‟s Report of 1997 the focus of widening participation (WP) policy and practice in the British Higher Education (HE) sector has been on the under-representation of certain groups of young people among the nation‟s undergraduate population. The several subsequent WP policy statements and reports, from Government and otherwise, have differed in the ways they defined such under-represented groups, and, amongst universities, Bristol has done so using a wider range of criteria than any other. As a result of an institution-wide consultation the University identified eight groups to which the efforts of widening participation has being focused: 1) mature students, 2) students from low performing schools and colleges, 3) students from low participation areas, 4) students from low socio economic groups, 5) minority ethnic students, 6) disabled students, 7) state school students, and, 8) local students (a full description for each of these categories is described in section 2.4). This emphasis on undergraduates is entirely understandable. Not only are they emphatically the dominant numerical component among all university students but almost all the minority that pursue postgraduate study do so through having first taken an undergraduate degree, immediately or some time previously. Hence, other things being equal, any patterns of bias and under-representation at the undergraduate level will be replicated to some extent at the later, postgraduate, one, and any success in broadening the intake of undergraduate by social groups should find its subsequent reflection among postgraduate populations as well. However, this does not necessarily mean that the latter will be a perfect image in miniature of the former. Some of the barriers and impediments that have led to comparatively few young people from certain backgrounds undertaking a first degree (eg the lack of peer-group graduate role models at their school, or in their 3 local residential neighbourhood) may have little or no impact three or four years later. But others might still be influential (eg the lack of experience of post-school education in the immediate and extended family), or even enhanced (notably the lack of family financial support for postgraduate study, which is more dependent on such „outside‟ funding than was an undergraduate degree). This will reduce still further the likelihood of students from low social class backgrounds or mature students becoming postgraduates, even had they enrolled for a first degree. Another dimension to this potential difference between undergraduate and postgraduate experiences is geographical. While the so-called elite universities, whether defined as the Russell Group, the Sutton Trust universities or however, represent, in reality and by definition, only a minority of the national undergraduate population, they provide disproportionately more postgraduate student places (Wakeling and Kyriacou, 2010). These universities tend to perform poorly against conventional undergraduate national WP Performance Indicators (such as the proportion of students from state schools or low social classes), and recent work by one of us has underlined that this may owe more to the deterrent effect of their campus images and cultures, real or supposed, on such nationally underrepresented student groups than to their academic attainments (Hoare and Aitchison, 2009). We may therefore find these different geographies also impact on the national social composition of postgraduates. Whether the elites are recruiting postgraduates from their own undergraduate ranks or from first-degree students who choose to migrate elsewhere to pursue their postgraduate interests (a distinction we enlarge on below), the pool they are drawing upon may be more socially exclusive than is the national undergraduate population. More formally, we can define several ways in which students flow between undergraduate and postgraduate study within a system of universities, and hence of ways in which any 4 one university can compile its postgraduate student population. Figure 1 shows the simplest of cases, with just two such „universities‟, Bristol and all others combined. Some Bristol students leave Higher Education after completing their first degree (Flow G). The postgraduate composition of Bristol will include some of its own undergraduate population who progress immediately to a further qualification (Flow A) and also first graduates attracted from elsewhere (Flow D). Bristol‟s postgraduates also are supplied by a further, composite component from Flow E, containing those with an earlier first degree from Bristol or elsewhere and who return to postgraduate study some time after, along with those without any first degree but who are admitted to postgraduate study based on an equivalent or better background from „the university of life‟. (The equivalent flows for the other, combined „university‟ – respectively H,C,B and F – are also shown for completeness.) Figure 1. Hypothetical inter-university interactions between undergraduate and postgraduate populations 5 We can now specify two different Bristol postgraduate populations – the one it generates (Flows A,B and part of E and F) and the other it attracts (Flows A,D and E). These may differ from each other, and either or both may also differ from its undergraduate population, to the extent that the specific differential progression barriers to postgraduate status suggested above are relevant to the attracted and generated populations. For the attracted case, Bristol‟s postgraduate composition will also reflect both the distinctive postgraduate opportunities it offers, which may differ sharply from its undergraduate programmes (in Bristol, for example, the lion‟s share of taught postgraduate courses are provided by just one of its six faculties – Social Sciences and Law) and also, as discussed already, its campus culture - the sort of place it is (or is thought to be). This is likely to be particularly important to all of Flow D and that part of Flow E that reflects the attraction of students without prior familiarity of Bristol as a place to study from their undergraduate years. The equivalent composition of the generated case will similarly reflect the differential appeal to Bristol undergraduates of its courses and cultures against other campuses offering postgraduate study. So, again, even were the research evidence to converge on a standard, consensus view of what the national landscape of postgraduate students looks like, there is no reason to assume this will be repeated at each of over 100 campuses offering postgraduate study, either to the (potentially different) postgraduate populations each attracts or that each generates. At the aggregated national level the attraction and generation profiles will converge to a common one (pace the significant complications of international student migration!) but this national picture will merely be the amalgam of two different and varying sets of local profiles. To make this clearer still, consider the hypothetical example outlined in Figure 2, based on the same two „universities‟ as before. Each contains an undergraduate cohort of 4000 students at Stage 1 which generate two equallysized sets of 1000 postgraduates (Stage 2). For whatever reason, the UoB has less attraction to „WP‟ students than the other „university‟, but these form a 6 constant fraction of each separate university‟s undergraduate and postgraduate cohorts (20% at UoB, 40% elsewhere). Figure 2: Hypothetical widening participation profiles of local and national postgraduate cohorts . Only half of each university‟s 1000 generated postgraduates now choose to remain at their home university for their second qualification, the other half moving to the alternative campus (Stage 3), but again UoB proves the less appealing to WP students irrespective of their origins, losing proportionately more of its own former WP undergraduates than it retains and attracting relatively fewer of those who were WP undergraduates elsewhere than remain at their „home‟ university. The outcome, at Stage 4, is that the two universities have cohorts of attracted 7 postgraduates that, while still identical in overall size, are now even more unequal in their proportions of WP students (a 3:1 differential) than were their equivalent cohorts at Stages 1 and 2 (a 2:1 differential). When aggregated into a national postgraduate population we find that the overall proportion of WPs is also different to both the equivalent generated and attracted proportions of each of its constituent universities, and so doubly misrepresents their separate experiences. Admittedly, this simplified illustration tells us nothing about why the differential attraction of the two universities on WP undergraduates and postgraduates comes about, and also assumes the decision to become a postgraduate and then to seek an appropriate placement arise as separate, sequential, decisions. Nevertheless, it shows how the WP composition of postgraduate cohorts can vary both from university to university, and also from the generation and attraction stages for any single university. Clearly, examination merely of the overall national picture tells us nothing about these finer-grained geographies. So in one way or the other, and for whatever reasons, WP issues amongst postgraduates have become a seriously under-researched theme. As a reaction we have recently seen reviews which have pored over the limited and sometimes contradictory research findings to date concerning the composition of the national postgraduate population its trends, funding, barriers affecting the participation, the decisions undertaken by the potential students and the impact of postgraduate degrees on future career prospects (Smith et al, 2010, Equality Challenge Unit, 2009, House, 2010, Stuart et al., 2008, HEFCE, 2004, Wakeling and Kyriacou, 2010). They are mostly focused at the aggregated (across-universities) national scale, although, as we have shown, we might reasonably expect that the picture at 8 one university could be very different from that at another, and even more so than are their equivalent undergraduate populations. However, we know of no published research into the characteristics of postgraduate student populations at, or generated by, an individual UK university, however defined in terms of the interaction flows of Figure 1. This makes our survey investigation a „first‟, not just for Bristol but for the HE sector more widely. A summary of these published results can be found in our previous report “Widening participation at Postgraduate level: Reading report and future work for the Research Cluster group” which was shared with the Social Science and Law Faculty‟s WP Forum in July 2010 by the WP Research Cluster. Overall, these published reports have showed marked differences in the access to postgraduate by gender, ethnicity and social class. According to the different authors, females predominate among UK postgraduate students especially among taught degrees, postgraduate certificate and diploma degrees ( House, 2010, Smith et al, 2010), although they are under-represented in Research degrees (Purcell et al., 2005). Although still controversial, given the difficulty in measurement, low social class appeared as the most important deterrent to progress onto postgraduate courses. Different studies suggests that postgraduate students are considerably more likely to be from a professional or managerial social class background (Wakeling, 2005, Machin and Murphy, 2010). Finally, interesting differences between minority groups across the different postgraduate degrees were also found, especially in overrepresentation on taught higher degrees for certain minority ethnic groups specifically, Black, Chinese and Asian Other. White and South Asian groups were found to be relatively underrepresented not only among taught degrees but also across all types of postgraduate degrees (Smith et al, 2010). 9 Against this background, during the academic year 2009/10 the need to know more about the nature of the local (Bristol) postgraduate population, was considered a first step to any discussion of the University‟s need to respond to WP imperatives in its postgraduate cohorts. Knowing whether there was anything we should, or could, be doing about it was given high priority by the Management Group of the University‟s Widening Participation Research Cluster (WPRC) and also strongly supported by the WP Forum of the Faculty of Social Sciences and Law and other consultees. The report to follow is the outcome of this. It concentrates first on some simple description and then on a more sophisticated quantitative analysis of the snapshot picture provided from the latest available appropriate data source. We also draw out some main summary findings, implications and conclusions, along with some of the obvious ways in which the work could be extended. Some of what we report is predictable, but not everything. Some is reassuring, but some more worrying and puzzling. We also look forward to comments and suggestions as to where, if anywhere, we should go from here. 2. Practicalities Two main objectives oriented this first report. First of all, we wanted to explore the demographic and educational characteristics of the UoB‟s undergraduates who finished in 2008/09 and were enrolled on postgraduate degrees somewhere six months after graduation. Second, we wanted to provide some initial analysis of the factors associated with being a postgraduate, with special focus on the presence or absence of WP characteristics. 2.1 Source of data 10 To achieve our main objectives we used data related to the UK and EU undergraduate population who finished their Bristol (UoB) course between the 1st August 2008 and 31st July 2009 and who then completed the Destination of Leavers from Higher Education Survey (DLHE) carried out between April 2009 and January 2010 by the University‟s Careers Service. This identifies each postgraduate student under a number of headings: Type of postgraduate course: postgraduate diploma and certificate (eg PGCE), Masters (eg M.Sc, MA) Research programmes (eg PhD, DPhil and MPhil) Location of postgraduate course: home university (UoB) or elsewhere Motivation for postgraduate study: employment or academic reasons, interest in the field, wanting to study, to postpone job hunting and others. Means of funding: self-funding, grant/award, sponsorship, employer-based funding, others This source enables us to access two of the three largest pathways of postgraduate recruitment relevant to the UoB, as mapped in Figure 1: i) Those UoB undergraduates who then immediately enrol for postgraduate study of some sort at UoB (Flow A) ii) Those from the same UoB undergraduate cohort who enrol for postgraduate study of some sort at another university (unspecified in the DLHE dataset) in the UK or abroad (Flow D) 11 However, two other postgraduate pathways are necessarily excluded: i) Those who enrolled immediately as UoB postgraduates from the equivalent undergraduate cohorts at universities other than UoB (Flow B) ii) Those, whether undergraduates from UoB or elsewhere, who enrol on postgraduate study at UoB after a delay of at least one year from their first graduation, or who have no first degree prior to becoming postgraduates (Flow E). Neither of these latter flows is accessible from our UoB-held DLHE returns, which are confined to the destinations of our own undergraduates within six months of graduation. The recent extension of the DLHE to cover destinations three years after graduation would enable us to capture some of the „delayed postgraduates‟ of that part of Flow E of Figure 1 who were Bristol undergraduates (though the value of these data is obviously conditional on the response rate and bias), but our own records of postgraduates provide very limited background detail of non-UoB first-degree postgraduates here. The current (2010/11) move at Bristol to on-line registration will not significantly change this. 2.2 Study population The total of University of Bristol undergraduate students who finished during the academic year 2008/09 was 3,050, of which 2,588 completed the DLHE survey. Additionally, by searching the student information system of the postgraduate student body (via SITS programme); we were able to find 36 12 postgraduates who also had graduated from UoB in 2008/09 but had not completed the DLHE survey. We also included them in our analysis, making a total undergraduate population with information about their study circumstances six months after graduation of 2,624. 2.3 Dependent Variables Given the nature of our current interest we adopted three different outcome variables related to being a postgraduate six months after graduation: 1. All Postgraduates: Those DLHE respondents now engaged in any further postgraduate study - PhD, DPhil, MPhil, MA, MSc, Postgraduate Diploma, Postgraduate Certificate, other Diploma or Certificate and Professional qualification – wherever they may be studying. 2. Bristol Postgraduates; Those respondents from within category 1 now studying at the University of Bristol. 3. Elsewhere Postgraduates: Those respondents from within category 1 now studying other than at the University of Bristol. 2.4 Independent variables In this analysis we assessed the effect on the probability of studying postgraduate of the WP characteristics as specified at the start of their undergraduate course. These are: 13 1. Mature students: students aging 21 or older when starting their Undergraduate course. 2. Students from low performing schools and colleges: although this has changed since in the light of other work by the WPRC, this was defined for the cohort being analysed here as schools and colleges with a UCAS average tariff score of 265 or below for their year 13 (Upper Sixth) cohorts. 3. Students from low participation neighbourhoods: calculated using POLAR2 quintiles 1 and 2. 4. Students from low socio economic groups: students whose parents are classified on their UCAS forms as being from social classes between 4 to 8: Small employers & own account workers, Lower supervisory & technical occupations, Semi-routine and Routine occupations. 5. Minority ethnic students: defined as those students who self- declared as Black Caribbean, Black African, Black other, Indian, Pakistani, Bangladeshi, Chinese, Asian other, White/black Caribbean, White/black African, White/Asian, Mixed other and Other. 6. Disabled students: those falling into any of UCAS‟s nine categories of disability: specific learning difficulty (eg. dyslexia), blind or partially 14 sighted, deaf , wheelchair or mobility difficulties, autistic spectrum disorder or Asperger syndrome, mental health difficulties, unseen disability (eg. diabetes, epilepsy, heart condition), multiple disabilities, other disability. 7. State school students: those from all types of non-independent schools. 8. Local students: students from Bristol or Bath home postcodes when they applied to the University as potential undergraduates. In addition, we explored the effect of other potential significant variables that according to the literature may influence the probability of postgraduate study. These are: 9. Gender: as we have seen, previous research has found significant gender difference in the probability of undertaking postgraduate studies, which has being more marked by different types of degrees. The existing literature suggests that women are under-represented among Masters and Research degrees while appear to be over-represented among other postgraduate courses such as postgraduate certificates and diplomas (Purcell et al., 2005, Wakeling, 2005, Stuart et al., 2008). 10. Age at graduation: age when the graduates finished their undergraduate course. There is evidence suggesting that students younger than 25 years old on graduaton are significantly more likely to continue onto further studies than older graduates (Purcell et al., 2005). 11. Living with parents: students who are reported as living with their parents when starting their undergraduate studies. Although under15 researched at postgraduate level, it has being found that secondary school leavers who were living at the parental home were more likely to undertake higher education studies than those who were living in rented accommodation (Albert, 2000, Marino et al., 2006). 12. Student loan: students who had a student loan at some time during their undergraduate course. The literature suggests that undergraduate debt influences the decision of graduates to enrol onto postgraduate studies. Graduates with a student loan are significantly less likely to go on to postgraduate studies than graduates who did not carrying a debt from their undergraduate years (Millett, 2003). The study by Stuart and colleagues also highlights that the amount of undergraduate debt is not the real barrier to apply for postgraduate studies, but that the worry of incurring more debts is (Stuart et al., 2008). 13. Subject of study: this refers to the Bristol faculty to which the student was registered as an undergraduate: Arts, Science, Engineering, Medical & Veterinary Sciences, Medicine & Dentistry and Social Sciences & Law. According to the literature, graduates from Engineering and Health disciplines are less likely to continue onto further studies than Arts and Social Sciences graduates, and more likely to enter the job market (Stuart et al., 2008, Purcell et al., 2005). 14. Undergraduate degree classification: unsurprisingly, the literature demonstrates that having a higher degree class (eg First or Upper Second class award ) increases the likelihood of progressing to postgraduate studies than those with lower class degrees. (Purcell et al., 2005). 16 3. Results 1 – simple descriptive 3.1 Composition of undergraduates and postgraduates – demographic and educational characteristics Of the total of undergraduate population of Bristol students examined 31.2% were undertaking further studies six months after graduation and 25.7% of the total were involved in postgraduate courses of some sort. Of this latter figure 35.5% were reported as studying at UoB and the remainder, nearly two thirds, were studying elsewhere (Figure 2). Unfortunately, we have no ready way of comparing these results with those from other universities in general or our research-intensive peers within the Russell Group in particular, valuable though this would be. Figure 3. Proportion of undergraduates according to their study status six months after graduation. University of Bristol, UK. 2008/09 5.45% 35.50% No study Study but no PG 68.78% Study PG 25.77% PG UoB 64.50% PG elsewhere 17 Table 1 describes the demographic, socio-economic and educational characteristics of the undergraduate cohort captured in the DLHE survey and of those also undertaking postgraduate studies. Of the total undergraduates, whether subsequently postgraduates or not, we found that their average age was 23 years (sd = 4.7), and just over half (53.6%) were females. Unsurprisingly the overwhelming majority were white (90.7%) and over 60% belonged to the highest two socio-economic classes – higher and lower administrative, managerial and professional (36.9% and 26.84% respectively). As also expected 97.4% of the study population were UK or EU in their pre-university home residence and had belonged to the Faculties of Sciences, Arts and Social Sciences and Law. Over half had had a student loan in some of their years of undergraduate study and nearly three-quarters graduated with a First-Class or Upper Second-Class Honours degree. As can be seen, the demographic, socio-economic and educational characteristics of those who then enrolled in postgraduate studies are different in detail rather than in substance from the undergraduate populations from which they are derived. Slight increases are to be noted in the percentage that is female, nonwhite non-UK/EU based and (perhaps surprisingly) living with parents. Predictably, larger deviations are found in the percentages who are young (so less likely to have dependents, though we have no direct measure of this), with good (First and Upper Second) degrees and from the three previously leading Faculties (the almost complete absence of any postgraduates from Medicine and Dentistry is striking). Most surprising, given the widespread publicity to the burden of debt repayment on current generations of graduating students, the percentage with some experience of a loan has risen by almost 20 percentage points compared to the undergraduate benchmark, to nearly three-quarters (Table 1). Finally, on social class (defined through parental occupation) there is some bimodality in the differences we find: higher percentages of postgraduates than undergraduates 18 come from the top two categories but also from the very lowest one: it‟s the centreground that seems to be squeezed. We return to this issue in the next section. Table 1. Characteristics of the Undergraduates who finished their course in 2008/09 according to their demographic, socio-economic and educational variables. University of Bristol, UK. 2008/09 Variable Total Undergraduates Studying a 2008/09 Postgraduate degree N % N % Female 1,406 53.58 371 54.88 Male 1,218 46.42 305 45.12 2,320 90.66 576 88.21 24 0.94 7 1.07 Indian/Pakistani/Bangladeshi 64 2.50 17 2.60 Chinese/Asian Other 57 2.23 23 3.52 Other Mixed 94 3.67 30 4.59 18-21 1,564 59.63 493 72.93 22-25 926 35.30 161 23.82 26-66 133 5.07 22 3.25 940 36.94 257 39.00 683 26.84 179 27.16 Intermediate occupations 257 10.10 67 10.17 Small employers and own account workers 115 4.52 20 3.03 Lower supervisory and technical occupations 56 2.20 12 1.82 Semi-routine occupations 122 4.79 20 3.03 Routine occupations 34 1.34 7 1.06 Never worked and long-term unemployed 338 13.28 97 14.72 Gender Ethnicity White Black Caribbean/Black African/Other black background Age at graduation Parent’s social class Higher managerial, administrative and professional occupations Lower managerial, administrative and professional occupations 19 Variable Total Undergraduates Studying a 2008/09 Postgraduate degree N % N % Yes 158 6.02 44 6.51 Not 2,466 93.98 632 93.49 Yes 1,435 56.54 432 73.10 Not 1,103 43.46 159 26.90 567 21.65 188 27.81 Upper second 1,347 51.43 397 58.73 Lower second 317 12.10 70 10.36 Third class 59 2.25 9 1.33 Ordinary 329 12.56 12 1.78 Arts 597 23.07 187 27.66 Science 664 25.66 213 31.51 Engineering 250 9.66 42 6.21 Medical & Veterinary Sciences 356 13.76 83 12.28 Medicine & Dentistry 275 10.63 5 0.74 Social Sciences & Law 446 17.23 146 21.60 676 25.77 Living with parents Used a loan for Undergraduate course Degree classification First class Area of study Total 3.2 2,624 Composition of undergraduates and postgraduates – widening participation students Table 2 presents a similar comparison to Table 1, but now based on the prevalence of Widening Participation characteristics. Two of them effectively reinforce Table 1 - less than one tenth of the graduates were classified on entry as Mature students (aged 21 and over) , or belong to minority ethnic groups. A similar 20 percentage on entry falls into the widely-cast definition of „disability‟. In contrast, around two thirds were coming from state schools (though here UoB is far below the national average, of course), and one quarter come from low performing schools. Although the overall prevalence of these individual Bristol „WP‟ characteristics is clearly very different, to summarize their overall presence or absence we created a score which is based on a simple count, ranging from 0 (no Widening Participation characteristic present) to eight (all the Widening Participation characteristics are present). Table 2 also presents the distribution of this score. It can be seen that nearly 80% of undergraduates presented with at least one WP characteristics and nearly 20% with at least three. Again, the distribution of these same characteristics in the equivalent population of postgraduates shows some differences, but these are mostly minor and also in the „right‟ direction for those who feared that the deterrents of further study (and hence postponed employment and greater likely post-university debt) would be relatively greater for WP first degree students. So there are small increases in the percentages of a number of WP characteristics amongst the postgraduates – students from state schools, from low participation neighbourhoods, from ethnic minorities and from the local region. The only downwards movement, but also the largest such change in either direction, is in those from low social groups, which falls by nearly 50% from its undergraduate representation to barely one in ten postgraduates. Given that this WP measure probably best captures both the financial and cultural capital of the background of UoB students, and their capacity to fund postgraduate studies independently of grants, scholarships and other awards, this is not surprising. But it is also particularly concerning from the perspective of the potential benefits of postgraduate training for occupational, financial and social mobility of individuals and for society as a whole. 21 Table 2. Description of the Undergraduate population with widening participation characteristics who finished their course in 2008/09. University of Bristol, UK. Variable Total Undergraduates Studying a Relative 2008/09 Postgraduate degree proportion N % N % Yes 178 6.79 33 4.90 Not 2,442 93.21 641 95.10 Yes 285 13.29 67 12.27 Not 1,860 86.71 479 87.73 Yes 474 18.11 130 19.35 Not 2,143 81.89 542 80.65 Yes 327 14.82 59 10.50 Not 1,880 85.18 503 89.50 Yes 239 9.34 77 11.79 Not 2,320 90.66 576 88.21 Yes 247 9.43 54 8.00 Not 2,371 90.57 621 92.00 Yes 1,562 64.33 409 65.97 Not 866 35.67 211 34.03 Yes 277 10.58 77 11.46 Not 2,340 89.42 595 88.54 613 23.36 160 23.67 1,606 61.20 416 61.54 405 15.43 100 14.79 Mature students 0.72 Students from low performing schools 0.86 Students from low participation areas 1.06 Socio-economic groups 4-8 0.70 Minority ethnic students 1.25 Disabled students 0.85 State School students 1.02 Local Students 1.08 Number of WP characteristics None 1-2 3 or more Total 2,624 676 We also examined the finer-grained profiles of WP characteristics by type of postgraduate study for these two largest categories - those following Masters and 22 research programmes (PhD, DPhil and MPhil) – the two largest such categories – and found the distribution of WP characteristics within these two sub-groups were very different. Not surprisingly, most of the students with WP characteristics were studying Master courses, with greatest proportions observed within mature students, from minority ethnic groups and the local region. Research programmes were mainly studied by graduates from low socio-economic groups, state schools and, again, local students (Figure 4). 3.3 Motivations When analyzing the motivations for pursuing postgraduate programmes we find that four reasons were each quoted by over 50% of postgraduates: 23 to develop a broader or more specialist range of skills or knowledge (60.2%), to change or improve career options (56.9%), because they were interested in the content of the course (56.6%) and because it was requirement of their employment (55.2%). Only 4.5% indicated that they were studying because they were unable to find a suitable job. As Table 3 shows, though, the profile of motivations varies substantially between those studying in UoB and those elsewhere. However, this is largely the result of a complete void in responses from UoB postgraduates citing the „requirement of my employment’ motive. It seems surprising (to put it mildly) that no postgraduate study of any kind in Bristol is seen as career-related, but taking this datum at face means, inevitability, that the UoB responses on all other motivations are higher that of the equivalents in postgraduates elsewhere (where the void in Bristol corresponds to the highest single motive !). This aside, UoB postgraduates seem relatively more motivated by two particular considerations – the ability to continue with previously enjoyed coursework (presumably with previously enjoyed teachers too), and the more negative wish to avoid moving from „studenthood‟ to seeking a job. The first, a clear advantage UoB has in retaining such students over competing postgraduate opportunities, is presumably something to be welcomed, though the second less so, with its inference of a lack of explicit enthusiasm and commitment to postgraduate work. (It would be interesting to know whether this is reflected in a lower completion rate of UoB postgraduates who stay here than those moving away: we cannot access the latter data directly but could at least compare the profile of withdrawals of UoB and non-UoB students on the same 24 postgraduate courses we offer, since our results suggest these might be different, and the former higher.) Table 3. Motivations of pursuing postgraduate programmes of graduates who were enrolled onto postgraduate courses everywhere, at the University of Bristol and elsewhere, 2008/09 Motivations It was a requirement of my employment To develop a broader or more specialist range of skills or knowledge To change or improve my career options Because I was interested in the content of the course Because I had enjoyed my first course and wanted to continue studying I wanted to go on being a student/I wanted to postpone job hunting I had been unable to find a suitable job Studying PG PG UoB PG elsewhere 55.17 0.00 66.67 60.24 61.76 59.57 56.93 51.96 59.13 56.63 65.69 52.61 38.86 60.78 29.13 17.77 30.39 12.17 4.52 2.94 5.22 Figure 5 shows these motivations for postgraduate study against the proportion of WP characteristics, combining all postgraduates, irrespective of destination. In general, the postgraduates drawn from the WP fractions of our undergraduates either take a longer-term career view of the value of postgraduate work or are driven by more academic than employability considerations. For instance, graduates from low participation areas, disabled and local students were 25 less employment-motivated and the most highly academic-oriented. Disabled students and those from low performing schools also stated they had failed to find a job, while a quarter of the local students and those from minority groups wanted to avoid moving from the student life to seeking a job. For mature students “employment requirements”, the wish for “changing or improving career options” or “to postpone job hunting” were not important reasons for choosing postgraduate study, rather they were mainly motivated by their “interest in the content of the course” and “developing a broader or more specialist skill”. All PG 3.4 Mode of study Overall, the clear majority (nearly 90%) of the postgraduates in our dataset are studying full-time but, as Figure 6 shows, this does vary by WP characteristic. 26 Predictably, for Mature postgraduates in particular the figure is much lower, just over 50%. In most other cases the differences between WP students and the overall total are fairly small. Local students are similarly more oriented to part-time study (many will also be part of the „Mature‟ group, of course, as are those from low participation neighbourhoods). On the other hand, those from ethnic minority groups are more attracted to full-time study (not surprising, given the well documented high value places on maximizing educational opportunities in the cultural contexts from which may of these will come – see Table 1), as are those from low performing schools. 3.5 Funding sources Given that funding for postgraduate studies is far less secure and predictable than for undergraduate degrees this is a strong prima facie barrier to 27 such a transition in general and those form less affluent backgrounds in particular. In examining our postgraduates according to their main source of funding we found that 26,1% have a grant or award, 12,9% have support from their employer, only 6,2% have a scholarship and 4,0% have other source of funding. But, predictably, the largest single source is self-funding (50.7%). Important variations were also observed across WP characteristics. Figure 7 shows the sources of funding according of each of the WP groups. Positive values indicate proportions higher than the average for all postgraduates (ie the values stated above), negative values means lower proportions than the average for all postgraduates. As we can see, Mature students present the higher proportion of self-funded students, exceeding the total proportion by around 30% (doubtless linked in a chicken-andegg fashion to their predilection for part-time study), with disabled and ethnic minority students (perhaps drawing disproportionately on support from extended family networks) also having higher than normal self-funding levels. The fact that disabled students are in the same category is very surprising, and may say something about the poor levels of financial provision for extended educational support for such groups. In contrast, dependence on some sort of external grant or award to fund postgraduate study seems an essential underpinning for many of the WP groups drawn from lower income social strata – those from low participation neighbourhoods, low performing schools, lower social groups, state and low performing schools. Local students and those from low participation areas are also in this category, the former maybe able to draw upon funding sources specific to postgraduates studying at their local institution. 28 29 4. Results 2 - Multivariate analysis 4.1 Our modelling approach In this final section we return to perhaps the main interest in our report – to what extent are WP undergraduates disproportionately likely or unlikely to undertake postgraduate study whether anywhere, in Bristol or elsewhere. Considering the decision to undertake a postgraduate degree is the result of a complex interaction of multiple factors that operate at different levels - individual, family, school, university, neighbourhood and community - it is important to base our analysis on a theoretical framework that allows us to identify the appropriate potential risk and/or motivators factors. Our reading of the literature suggested a set of variables that are considered important predictors of postgraduate study, although the nature of our data precludes some of them. This includes those relating to family history, such as parental experience of higher education, having dependent family members (Stuart et al., 2008), and amount of previous debt (Millett, 2003). Nonetheless, with the available data and following the strategy of analysis proposed by Victora et al. (Victora et al., 1997), we constructed our conceptual model that describes the hierarchical relationship between the variables, as in Figure 8. As can be seen, we grouped our WP and the other available individual variables in four blocks, or levels. Level 1 contains characteristics that we assume to be independent of all other influences in the model (the „distal‟ variables in Victora and et al‟s terminology) but which may have an influence on the lower-level variabes. Level 2 and 3 includes the „intermediate‟ variables which could be influenced by those at higher or the same level. Finally, at Level 1 we have the „proximate‟ variables, with both the greatest potential to be affected by others at the same or higher levels and which also exercise the most direct influence on the decision to undertake postgraduate studies. 30 Applying this conceptual framework to our own project, we placed the demographic and socio-economic characteristics like age, gender, minority and low social class at Level 1, where they may influence directly or indirectly all the other variables in the model. Level 2 includes health conditions such as disability which may be partly determined by low socio-economic status and the demographic variables. This level also includes environmental variables related to the place of residence, such as being a local student and being from a low participation area. The third level includes school characteristics such as low performance school and state school which can be affected by those above. Finally, Level 4 comprises the student‟s personal history or conditions while an undergraduate at UoB, and her/his academic attainment, which may have been affected by all the preceding variables (Figure 8). Figure 8. Conceptual model of postgraduate study. University Bristol, UK. 2008-09 Level Demographic 1 Socio-economic Gender Minority Health 2 Parents’ social class Age 2009 Environmental Local UG student Disability Low participation area School 3 Low performance school State school Undergraduate history 4 Living with parents Mature Student loan Area of study Degree classification Undertaking postgraduate studies six months after graduation 31 As explained by Victora and colleagues, the important advantage of this hierarchical conceptual model is that it allows us to adjust the effect of each variable for those in the same level or above in the model. Consequently our analyses adjust only for confounding variables with a superior (ie higher level) hierarchical effect, and not for all variables present in the model, as is often done in similar analyses so leading to incorrect and misleading interpretations (Victora et al., 1997). For example, analysis studying the effect of social class or gender on the probability of undertaking postgraduate study may conclude that they do not have any effect after adjustment by “confounding variables” such as degree class or academic subject studied. As specified in our conceptual model, these later variables are proximal variables which may capture most if not all of the effect of the more distal variables (such as social class and gender), and consequently the effect of the latter result may be underestimated. We conducted both unadjusted and adjusted analysis. In the unadjusted analysis, we used the Chi-square test for heterogeneity to compare the probability of each outcome within the groups of the independent variables without controlling for any confounding variable(s). The adjusted analysis was based on the conceptual model in Figure 8 which determined the order in which the variables enter the analysis: the more distal variables (demographic factors and socio- economic) were the first to be included in the model, irrespective of whether they were significant in the unadjusted analysis. Those variables displaying an association with the outcome of interest of P<0.20 remained as „adjustment variables‟ for the hierarchically lower variables. Then, the variables of the next level were entered and a similar procedure repeated for all levels. Both unadjusted and adjusted analyses were carried out using Poisson regression and Wald's test which uses the „probability ratio‟ (PR) as the effect measure, enabling more accurate interpretations. We opted to use the Poisson regression with a robust variance instead of the alternative logistic regression with its associated effect measure of the „odds ratios‟, given the relatively high probability of studying a 32 postgraduate degree (over 5%, see Barros and Hirakata, 2003). However, when we repeated the analyses using logistic regression (data not shown) we found virtually no change in the significance of the associations that emerged. All our procedures were performed with STATA version 9.0. 4.2 Results For our undergraduate cohort as a whole, the estimated probability of studying as a postgraduate three months after graduating was 25.8% (95% confidence interval [CI] 24.1, 27.5). The probability of studying specifically at the University of Bristol was 9.1% (95% CI 8.1, 10.3) and for studying elsewhere 16.7% (95% CI 15.2, 19.1). Firstly, we analysed the factors related to studying anywhere. According to the results of the unadjusted analysis, the probability of studying a postgraduate degree decreased significantly as age increased (Table 4). Students belonging to families of a low social class status and those who initiated their undergraduate course at a mature age had respectively 37% and 29% less probability of becoming postgraduates (p=0.001 and p=0.03 respectively). Undergraduates from ethnic minority groups showed a significant increased probability (30%) than their White peers (p=0.01), while those with a student loan as undergraduates were twice as likely to have being engaged in postgraduate studies as those without. The probability of postgraduate study also decreased significantly with lower degree classifications. Those with first class degree were nine times more likely to engage in postgraduate studies in comparison with their (very few) peers who obtained an ordinary degree (p<0.001). These probabilities also were significantly different across faculties, being significantly lower for those graduating from Medicine & Dentistry, Engineering and Medical & Veterinary Sciences (p<0.001). 33 However, gender, disability, being a local student, coming from low participation neighbourhood, a low performing school, a state school and living with parents were not associated with the outcome of whether or not students became postgraduates. The adjusted analyses (Table 4) were then carried out according to the prescribed hierarchical levels of causality. We found that after controlling for the variables in the higher levels: The probability ratios for students from low social class families remained low when compared with the category of reference The positive effect of Ethnic Minority groups remained significant, being 26% higher in comparison with White graduates. Probability ratios for graduates from poor families remained low when compared with the category of reference The probability of postgraduate studies decreased significantly as age increased. For those older than 26 years the probability of becoming a postgraduate was 53% lower compared with those aged 18 to 21 years. After controlling for demographic and socio-economic characteristics, disabled students appeared less likely to pursue postgraduate studies: their probability is 25% lower when compared with those with no disability. This association, however, is only disclosed when the negative confounding effect of the socio-economic status 4-8 is controlled for. Another variable that become significant after adjustment was being a local student. Those who lived in Bristol or Bath prior to coming to UoB 34 were 31% more likely to be undertaking a postgraduate course than their counterparts living elsewhere. Student loan also remained significant. Undergraduates with a student loan showed 45% more likelihood of going on to postgraduate studies than those without. Subject of study still has an explanatory role. Medicine & Dentistry and Engineering had lower probabilities of undertaking postgraduate studies. The final influential factor, unsurprisingly, is class of first degree. Graduates who obtained a first or upper second degree were more than ten times more likely to pursue a postgraduate course six months after graduation than those with an Ordinary degree. On the other hand, we identified four factors as having no statistical relationship with postgraduate status - gender, residence in a low participation neighbourhood, attendance at a low performing school and state school. The lack of association between gender and the postgraduate outcome was unchangeable after adjustment for age, minority group and family social class 4-8: however, we kept it in the model as a potential confounding variable at lower levels. Living with parents also decreased significantly, and remained non-significant (p= 0.13) but was also retained as a confounding factor. The unadjusted effect of mature students decreased after adjustment being no longer significant (p=0.17), and was again kept in the analysis to allow for its potential effect on the other variables in the same level. 35 Table 4. Associated factors of being studying a postgraduate course after six months of graduation ANYWHERE. University of Bristol graduate students, 2008/09. Adjusted † Unadjusted Level Variables 1 Gender n Probability PR (IC95%) 1,406 26.39 1.05 (0.92 - 1.20) 1.06 (0.92 - 1.22) Male 1,218 25.04 1.00 (Reference) 1.00 (Reference) 0.01 0.04 Yes 239 32.22 1.30 (1.07 - 1.59) 1.26 (1.01 - 1.56) Not 2,320 24.83 1.00 (Reference) 1.00 (Reference) <0.001 <0.001 18-21 1,564 31.52 1.00 (Reference) 1.00 (Reference) 22-25 926 17.39 0.55 (0.47 - 0.65) 0.53 (0.45 - 0.63) 26-66 133 16.54 0.53 (0.36 - 0.78) 0.47 (0.22 - 0.99) Low Social class 0.001 0.001 Yes 327 18.04 0.67 (0.53 - 0.86) 0.66 (0.52 - 0.82) Not 1,880 26.76 1.00 (Reference) 1.00 (Reference) Disability 0.15 0.05 Yes 247 21.86 0.83 (0.65 - 1.07) 0.75 (0.56 – 1.00) Not 2,371 26.19 1.00 (Reference) 1.00 (Reference) Local student 0.39 0.03 Yes 277 27.8 1.09 (0.89 - 1.34) 1.31 (1.03 - 1.66) Not 2,340 25.44 1.00 (Reference) 1.00 (Reference) Low Participation area 0.33 0.85 Yes 474 27.43 1.08 (0.92 - 1.28) 1.03 (0.78- 1.36) Not 2,143 25.29 1.00 (Reference) 1.00 (Reference) Low Performance school 0.42 0.45 Yes 285 23.51 0.91 (0.73 - 1.14) 0.91 (0.71- 1.16) Not 1,860 25.75 1.00 (Reference) 1.00 (Reference) State School 0.33 P value* 0.44 Female Age 2009 3 PR (IC95%) 0.43 Minority groups 2 P value* 0.86 36 Adjusted † Unadjusted Level 4 Variables n Probability PR (IC95%) Yes 1,562 26.18 1.07 (0.93 - 1.24) 1.01 (0.86 - 1.19) Not 866 24.36 1.00 (Reference) 1.00 (Reference) Living with parents P value* PR (IC95%) 0.53 0.13 Yes 158 27.85 1.09 (0.84 - 1.40) 1.26 (0.93 - 1.69) Not 2,466 25.64 1.00 (Reference) 1.00 (Reference) Mature 0.03 0.17 Yes 178 18.54 0.71 (0.52 - 0.97) 0.56 (0.24- 1.28) Not 2,442 26.26 1.00 (Reference) 1.00 (Reference) Student Loan <0.001 <0.001 Yes 1,435 30.1 2.08 (1.77 - 2.46) 1.45 (1.18 - 1.78) Not 1,103 14.42 1.00 (Reference) 1.00 (Reference) Area of study <0.001 <0.001 Arts 603 31.01 1.00 (Reference) 1.00 (Reference) Science 680 31.32 1.01 (0.86 -1.19) 0.89 (0.73- 1.08) Engineering 250 16.8 0.54 (0.40 - 0.73) 0.40 (0.25 - 0.63) 363 22.87 0.74 (0.59 - 0.92) 0.78 (0.60 - 1.01) Medicine & Dentistry 275 1.82 0.06 (0.02 - 0.14) 0.21 (0.08 - 0.52) Soc Sci & Law 453 32.23 1.04 (0.87 - 1.24) 1.04 (0.85 - 1.28) Medical & Veterinary Sciences Degree classification First class <0.001 <0.001 567 33.16 9.09 (5.15 - 16.04) 14.63 (4.10- 52.20) Upper second 1,347 29.47 8.08 (4.61 - 14.17) 11.09 (3.12 - 39.38) Lower second 317 22.08 6.05 (3.35 - 10.95) 7.43 (2.05 - 26.92) Third class 59 15.25 4.18 (1.84 - 9.48) 3.51 (0.73 - 16.84) Ordinary 329 3.65 1.00 (Reference) 1.00 (Reference) 2,624 25.78 Total P value* PR: Probability ratio † Adjusted for all the variables in the same level or above with p-value 0.2 * Wald test for heterogeneity. 37 We also explored the variables separately associated with being a postgraduate at the University of Bristol and elsewhere (Tables 6 and 7), again in adjusted and unadjusted formats. To help in the interpretation of the results we created a comparative table which summarizes the outcomes of these three multivariate analyses (Table 5). In this table the check symbol () indicates a significant predictor at 5% level of confidence and the arrows indicate the direction of the effect for the group (↑:increased, ↓:decreased, ↕ increased / decreased depending on the category). Table 5. Summary of the findings of the unadjusted and adjusted analysis of the associated factors of being study postgraduate. University of Bristol. 2008-09 PG anywhere Unadjusted Adjusted PG UoB Unadjusted Adjusted Female Minority ↑ Older than 21 ↓ Low socio-economic groups ↑ ↓ ↑ Disabled Local From low participation areas From low performing schools From state Schools Living with parents Mature With student loan SS&L, Arts, Sciences First/second class ↑ ↕ ↑ ↓ ↑ PG elsewhere Unadjusted Adjusted ↑ ↑ ↓ ↑ ↑ ↑ ↑ ↕ ↑ ↕ ↑ As can be seen, the unadjusted analysis of continuing studying at the University of Bristol showed a higher number of significant predictors than for those studying anywhere. In common with the „studying anywhere‟ case, the following groups showed a significantly higher probability of association with 38 postgraduate study at their „home‟ university: those aged 18-21; who were mature when starting their undergraduate course; those who had had a student loan; those from Sciences and those holding a First class degree. Additionally, other variables were showed to be important in studying at the UoB: being previously resident in the Bristol or Bath postcodes; from low participation areas; from low performing and state schools and those who were living with their parents. As with the „studying anywhere‟ case, the results of the adjusted analysis showed that older graduates are less likely to enrol onto a postgraduate degree at the UoB. The same was observed for graduates from Engineering and Social Sciences and Law whose probabilities to continue studying at UoB were significantly decreased when compared to those graduates from Arts. Students who had a student loan and had obtained a high degree class also were more likely to be studying at UoB than their corresponding reference groups. But contrasting with studying anywhere, we now found higher probabilities of studying at UoB in other groups of students such as local students, those from a state school and who were living with their parents. The greatest probability ratios of all were found among locally-resident students and those who were not living in rented or university accommodation (Table 5 and Table 6). For those postgraduates studying elsewhere, the unadjusted analysis also showed significant statistical association with ethnic minority groups, age, social classes 4-8, mature students, from low performing schools, those with a loan, from the Social Sciences and Law faculty and holding a first class degree. Unlike the equivalent „anywhere‟ analysis, female postgraduates, those living in Bristol or Bath and from a low participation area showed higher probabilities of studying elsewhere, while those from low performance schools have lower probabilities. When reviewing the adjusted results, we found that studying elsewhere had an increased probability ratio with being young, from ethnic minority groups and social 39 classes 1 to 3. Similarly, graduates with a first class degree and from Social Sciences and Law showed a higher probability of being postgraduates elsewhere than graduates with a lower academic results, or from an Arts background. A differential gender effect was also evident, with females significantly more likely to be studying elsewhere as postgraduates than males. As with the „studying anywhere‟ case, no significant associations emerged after adjustment between studying elsewhere and previous residence in a low participation area, attending a low performing school, state school and living with parents (Table 6 and Table 7). 40 Table 6. Associated factors of being studying a postgraduate course after six months of graduation AT THE UNIVERSITY OF BRISTOL. University of Bristol graduate students, 2008/09. Adjusted † Unadjusted Level Variables 1 Gender n Probability PR (IC95%) 1,406 8.25 0.81 (0.64 - 1.03) 0.80 (0.63 - 1.02) Male 1,218 10.18 1.00 (Reference) 1.00 (Reference) 0.77 0.78 Yes 239 9.62 1.06 (0.71 - 1.60) 1.07 (0.67 - 1.69) Not 2,320 9.05 1.00 (Reference) 1.00 (Reference) <0.001 <0.001 18-21 1,564 11.13 1.00 (Reference) 1.00 (Reference) 22-25 926 5.62 0.50 (0.37 - 0.68) 0.50 (0.37 - 0.68) 26-66 133 10.53 0.95 (0.57 - 1.58) 0.95 (0.57 - 1.59) Low Social class 0.26 0.26 Yes 327 7.03 0.79 (0.52 - 1.20) 0.79 (0.52 - 1.20) Not 1,880 8.94 1.00 (Reference) 1.00 (Reference) Disability 0.72 0.87 Yes 247 8.50 0.92 (0.60 - 1.41) 0.96 (0.63 - 1.48) Not 2,371 9.19 1.00 (Reference) 1.00 (Reference) Local student <0.001 <0.001 Yes 277 18.05 2.22 (1.67 - 2.96) 2.62 (1.93 - 3.54) Not 2,340 8.12 1.00 (Reference) 1.00 (Reference) Low Participation areas <0.001 0.27 Yes 474 14.77 1.86 (1.44 - 2.41) 1.28 (0.86 - 1.97) Not 2,143 7.93 1.00 (Reference) 1.00 (Reference) Low Performance schools 0.16 0.78 Yes 285 11.23 1.29 (0.90 - 1.84) 1.06 (0.72 - 1.55) Not 1,860 8.71 1.00 (Reference) 1.00 (Reference) State Schools Yes 0.002 1,562 10.44 1.59 (1.19 - 2.12) P value* 0.07 Female Age 2009 3 PR (IC95%) 0.09 Minority groups 2 P value* 0.02 1.43 (1.07 - 1.91) 41 Adjusted † Unadjusted Level Variables Not 4 n 866 Probability 6.58 PR (IC95%) P value* 1.00 (Reference) Living with parents PR (IC95%) 1.00 (Reference) 0.03 0.01 Yes 158 13.92 1.58 (1.05 - 2.37) 2.13 (1.23 - 3.68) Not 2,466 8.84 1.00 (Reference) 1.00 (Reference) Mature 0.65 0.74 Yes 178 10.11 1.11 (0.71 - 1.75) 1.24 (0.35 - 4.30) Not 2,442 9.09 1.00 (Reference) 1.00 (Reference) Student Loan <0.001 <0.001 Yes 1,435 9.41 5.19 (3.27 - 8.24) 5.90 (2.74 -12.69) Not 1,103 1.81 1.00 (Reference) 1.00 (Reference) Area of study P value* 0.0001 <0.001 Arts 603 10.45 1.00 (Reference) 1.00 (Reference) Science 680 12.21 1.17 (0.86 - 1.59) 0.64 (0.42 - 0.96) Engineering 250 8.40 0.80 (0.50 - 1.29) 0.33 (0.12 - 0.91) 363 11.57 1.11 (0.86 - 1.59) 1.11 (0.72 - 1.71) 275 0.73 0.07 (0.02 - 0.28) 0.35 (0.05 - 2.59) 453 6.40 0.61 (0.40 - 0.94) 0.39 (0.22 - 0.66) Medical & Veterinary Sciences Medicine & Dentistry Soc Sci & Law Degree classification First class <0.001 0.003 567 13.40 4.90 (2.49 - 9.65) 2.17 (0.50 - 9.42) Upper second 1,347 9.06 3.31 (1.70 - 6.44) 1.21 (0.27- 5.37) Lower second 317 8.83 3.23 (1.55 - 6.73) 0.87 (0.18 - 4.14) Third class 59 8.47 3.10 (1.08 - 8.92) 0.80 (0.12 - 5.14) Ordinary 329 2.74 1.00 (Reference) 1.00 (Reference) 2,624 9.15 Total PR: Probability ratio † Adjusted for all the variables in the same level or above with p-value 0.2 * Wald test for heterogeneity. 42 Table 7. Associated factors of being studying a postgraduate course after six months of graduation ELSEWHERE. University of Bristol graduate students, 2008/09. Adjusted † Unadjusted Level Variables 1 Gender n Probability RP (IC95%) 1,406 18.14 1.22 (1.03 - 1.45) 1.22 (1.01 - 1.47) Male 1,218 14.86 1.00 (Reference) 1.00 (Reference) 0.01 0.04 Yes 239 22.59 1.43 (1.11 - 1.84) 1.35 (1.02 - 1.78) Not 2,320 15.78 1.00 (Reference) 1.00 (Reference) <0.001 <0.001 18-21 1,564 20.40 1.00 (Reference) 1.00 (Reference) 22-25 926 11.77 0.58 (0.47 - 0.71) 0.60 (0.48 - 0.74) 26-66 133 6.02 0.29 (0.15 - 0.58) 0.12 (0.02 - 0.85) Low Social class <0.001 0.002 Yes 327 11.01 0.62 (0.45 - 0.85) 0.59 (0.42 - 0.82) Not 1,880 17.82 1.00 (Reference) 1.00 (Reference) Disability 0.15 0.228 Yes 247 13.36 0.79 (0.57 - 1.09) 0.80 (0.55 - 1.15) Not 2,371 17.00 1.00 (Reference) 1.00 (Reference) Local student 0.002 0.09 Yes 277 9.75 0.56 (0.39 - 0.81) 0.68 (0.44 - 1.06) Not 2,340 17.31 1.00 (Reference) 1.00 (Reference) Low Participation area 0.02 0.63 Yes 474 12.66 0.73 (0.57 -0.94) 0.91 (0.63 - 1.33) Not 2,143 17.36 1.00 (Reference) 1.00 (Reference) Lower Performance schools 0.05 0.29 Yes 285 12.28 0.72 (0.52 - 1.00) 0.83 (0.58 - 1.18) Not 1,860 17.04 1.00 (Reference) 1.00 (Reference) State Schools 0.20 P value* 0.04 Female Age 2009 3 RP (IC95%) 0.03 Minority groups 2 P value* 0.18 43 Adjusted † Unadjusted Level 4 Variables n Probability RP (IC95%) Yes 1,562 15.75 0.89 (0.74 - 1.06) 0.88 (0.72 - 1.06) Not 866 17.78 1.00 (Reference) 1.00 (Reference) Living with parents P value* RP (IC95%) 0.36 0.63 Yes 158 13.92 0.83 (0.56 - 1.23) 0.90 (0.58 - 1.40) Not 2,466 16.79 1.00 (Reference) 1.00 (Reference) Mature 0.01 0.14 Yes 178 8.43 0.49 (0.30 - 0.80) 0.25 (0.04 - 1.59) Not 2,442 17.16 1.00 (Reference) 1.00 (Reference) Student Loan <0.001 0.47 Yes 1,435 20.70 1.64 (1.36 - 1.98) 1.09 (0.86 - 1.36) Not 1,103 12.60 1.00 (Reference) 1.00 (Reference) Area of study P value* <0.001 <0.001 Arts 603 20.56 1.00 (Reference) 1.00 (Reference) Science 680 19.12 0.93 (0.75 - 1.16) 0.94 (0.73 - 1.21) Engineering 250 8.40 0.41 (0.26 - 0.63) 0.45 (0.26 - 0.75) 363 11.29 0.55 (0.40 - 0.76) 0.63 (0.43 - 0.92) 275 1.09 0.05 (0.02 - 0.17) 0.11 (0.04 - 0.36) 453 25.83 1.26 (1.01 - 1.57) 1.42 (1.12 - 1.81) Medical & Veterinary Sciences Medicine & Dentistry Soc Sci & Law Degree classification First class <0.001 <0.001 567 19.75 21.66 (6.94 - 67.65) 8.54 (2.70 - 26.99) Upper second 1,347 20.42 22.39 (7.22 - 69.41) 7.24 (2.31 - 22.68) Lower second 317 13.25 14.53 (4.55 - 46.41) 4.81 (1.48 - 15.58) Third class 59 6.78 7.43 (1.17 - 32.38) 2.51 (0.52 - 12.21) Ordinary 329 0.91 1.00 (Reference) 1.00 (Reference) 2,624 16.62 Total PR: Probability ratio † Adjusted for all the variables in the same level or above with p-value 0.2 * Wald test for heterogeneity. 44 5. Summary and Discussion Our report has provided for the first time for Bristol, and we believe for any other specific university for that matter, a preliminary description and analysis of the composition of its postgraduate student population, with particular emphasis on how this chimes with the University‟s widening participation agenda. By combining together the relevant segments of the national DLHE survey and the University‟s own SITS database we were able to compare the most recent profiles of Bristolgenerated postgraduates against the corresponding Bristol undergraduate cohort from which they were derived. The headline conclusions that emerged from this are these: 1. Overall, a quarter of the Bristol undergraduates immediately registered for a postgraduate course, two-thirds of them at a university other than Bristol. 2. Comparing the matched undergraduate and postgraduate profiles shows them broadly similar, but with some predictable differences – certain faculties produce proportionately more postgraduates than others, and they are drawn disproportionately from the younger and better qualified (by degree class) undergraduates. 3. On the other hand, apart from undergraduates from low social class groups, those defined as „WP‟ by the University‟s existing range of criteria are not noticeably under-represented among those who subsequently become postgraduates. 4. Masters courses appear relatively more attractive to Ethnic minority, locallyresident and mature students while those from low social classes, low 45 participation neighbourhoods and state or underperforming schools gravitate relatively more towards research-based postgraduate studies. 5. Notwithstanding an apparent glitch in the DLHE dataset, it seems that WP students in aggregate are motivated towards postgraduate studies more by academic-related rationales and long-term perspectives than are postgraduate students as a whole. 6. Most postgraduates, WP and otherwise, are full-time, with the exception of mature students who are relatively attracted towards part-time enrolment. 7. Most are also self-funding, but aside from mature and disabled students, those deemed „WP‟ on one or more criteria are relatively more likely to have secured postgraduate funding of some sort than the Bristol postgraduate average. 8. More sophisticated analysis allows us to examine the interplay of various WP and other variables on the decision to enrol as a postgraduate anywhere, and also specifically at Bristol and other university destinations. Overall, our results showed the significant roles of high degree class, mature age and faculty of first degree on the decision to become a postgraduate in all the six analyses we ran (3 destinations x adjusted/unadjusted); otherwise the tendency for WP students to be more likely to be postgraduates was generally confirmed. This was particularly so for these with other local personal/family ties of some sort and attending a „WP‟ school, who all also disproportionately enrolled as postgraduates at Bristol rather than elsewhere. 9. However, undergraduates from low social classes were confirmed as the main exception to this, being significantly less likely to become postgraduates, both as a whole and at non-Bristol universities. 46 10. Finally, there was a tendency for students who had had loans (and hence incurred debt) during their undergraduate career to be more likely to become postgraduates overall, and at Bristol. Where we can make such comparisons, we frequently find echoes of the wider literatures on the national aggregate postgraduate population in these results (Equality Challenge Unit, 2009, House, 2010, Smith et al, 2010, Stuart et al., 2008, HEFCE, 2004, Wakeling and Kyriacou, 2010). This is certainly so with the under-representation of older students among postgraduates (Purcell et al., 2005) and of those with a declared disability (Equality Challenge Unit, 2009, Leacy et al., 2010), and, in contrast, the over-representation of postgraduates from first degree programmes in Arts and Social Sciences (Pollar et al., 2008, Stuart et al., 2008, Purcell et al., 2005), an ethnic minority background (Wakeling, 2009, Stuart et al., 2008, Smith et al, 2010), a higher social class (Wakeling, 2005, Machin and Murphy, 2010), a high quality first degree (Purcell et al., 2005, Wegner, 1969, Stuart et al., 2008, Wakeling, 2005). Data limitations precluded our comparing our results with all such previously-identified variables. On the other hand, for the first time we have shown the geography of pre-university residence and living with parents while taking a first degree to be associated with subsequent postgraduate enrolment. We found some tendency for females to be over-represented in those taking postgraduate courses outside Bristol, although we did no explore whether that association was constant across the different types of courses as found by Purcell et al (2005), where women were more likely to have undertaken other courses than Master or Research degrees. Finally, our finding that those with a student loan were also more likely to become undergraduates appears to run counter to the implications of Stuart et al (2008) that fear of debt is a deterrent to further study (we return to some complications in interpreting this result below). 47 Inevitably, there are limitations to what we have been able to achieve. Only one undergraduate-cum-postgraduate cohort has been examined. With more time others could be, though we know no reason why the one analyzed should not be representative of others as well. In addition to the restrictions, noted earlier, in comparing our own students via SITS with the wider literatures, the DLHE data tell us nothing in general, nor for WP students in particular, about the detailed destinations of the non-Bristol bound postgraduates (eg how many enrol at other „elite‟, Russell group, post-1992, overseas…. universities), nor on how they fare on the courses which they were only part-way when captured by DLHE (pass rates, completion rates etc). Furthermore, as we have noted previously, we have no other HEI-specific university data against which to compare our own experiences: our progression-to-postgraduate rate and postgraduate profiles may be par for the course for the HE sector as a whole, but how do we compare with our immediate competitors? With time and more such place-specific studies published we should be able to put our own into a richer context. A related shortcoming is the absence of any information on Bristol‟s importing of postgraduates from first degree training provided elsewhere (Flow D of Figure 1). Without this we are not in any position to judge whether Bristol is overall a net importer or exporter of postgraduates – the two-thirds export loss of locally-trained students may be more than compensated by an even larger influx from elsewhere – nor to tell how far this changes the WP profiling of Bristol‟s overall postgraduate population (ie we know our generating profile but not our attracting one). The first should be fairly easy to remedy from further work on SITS. The second is less tractable, and will depend on the interplay of 48 the factors discussed in Section 1. Confidentiality constraints on the release of DLHE data from „other‟ universities is also a continuing practical obstacle in filling this gap. The more obvious solution, if Bristol is at least as interested in its attracting than generating capacity for postgraduates (realistically it is probably more so), is for Bristol‟s new on-line postgraduate recruitment processes to broaden the range of data it recruits from external applicants, to match those for its undergraduate cohorts. These caveats aside, the university might take reasonable comfort in the lessons from our results. For the most part, fears that postgraduate study is a significant barrier to further study across-the-board for Bristol‟s WP students, on top of those they faced when deciding upon a first degree, have not been borne out. If the University is anxious to recruit more of the under-represented mature postgraduates then expanding part-time provision is an obvious strategy to follow. It might genuinely be puzzled by the tendency for those with undergraduate loans to be disproportionately well represented amongst postgraduates, but maybe were we to have data also on undergraduate bursaries and other financial support we would see this result differently. Maybe, too, well-heeled, financially-astute students see a student loan as a cheap way to raise money for later use, against alternative forms of borrowing, rather than as something immediately needed to pay their way through their first degree studies. The one genuinely concerning outcome of our findings is that students from low social class backgrounds are deterred from postgraduate study in general and specifically from applying for courses away from Bristol. Bristol can inevitably only offer a limited range of further academic opportunities, as the two-third export rate from our locally-produced postgraduate recruits implies, and those postgraduate opportunities most suited to the interests and career 49 plans of many Bristol undergraduates must involve applying elsewhere. It might be claimed (correctly) that researchers have raised serious doubts about the dependability (and response rate) of much of the socio-economic data on undergraduates collected by UCAS (also the ultimate source of this aspect of our SITS data) , but, as there seems no obvious reason why this would also differentially distort the later profile of those who enrol as postgraduates over those who do not, we are inclined to take our results here as „real‟ rather than simply an artefact of an imperfect database. This under-representation of low social classes amongst the DLHE postgraduates appears in all our analyses, from the simple description of Section 3 to the six multivariate cases in Section 4, including the „at Bristol‟ case, though it falls short of our statistical significance thresholds. Given that postgraduate training confers a further clear economic advantage on students in addition to that of a first degree (House, 2010) this would seems a real barrier to social mobility for those otherwise with the academic potential for further study, both locally and also sector-wide. There is a limit to which we can probe these patterns of underrepresentation much further with such quantitative data as we have. A more attractive pathway would seem to be a two-pronged qualitatively-focused investigation. First, this might examine the ways that Bristol‟s postgraduate schools promote, fund/sponsor and adjudicate postgraduate applications and, second, research the particular barriers to postgraduate progression that students from lower social class backgrounds perceive both before and, for the minority taking such a step, after the event. The Social Sciences and Law faculty, with its dominance both as an undergraduate generator of postgraduates and provider of postgraduate courses, and also its reservoir of appropriate research expertise, would seem the obvious site for any such investigation. 50 6. Postscript While not directly part of our review, with its primary focus on the widening participation characteristics of postgraduate students, one further aspect of our findings is of considerable significance in the current funding climate. Until the October 2010 Browne Review on student fees and the Government‟s Comprehensive Spending Review, the main recent thrust of the University‟s student expansion had been through postgraduates. Here it has enjoyed much greater freedom over recruited numbers and charged fees than for the undergraduate enrolments, where it has even been threatened with fines should it „over-recruit‟. But, as our analysis confirms, the main emphasis of the postgraduate cohorts generated by the University, irrespective of their second-degree destination, and of the attracted postgraduates as measured by faculty of enrolled courses, is in Arts and Social Sciences, the two faculties most threatened by the withdrawal of any Government support for undergraduate teaching, whether at Bristol or beyond. The implications for the continued heath and diversity of the University‟s postgraduate community seem serious. One option is for fees for such undergraduate courses to rise disproportionately to maintain current teaching levels. If so, then it is reasonable to expect this to impact selectively downwards on the WP component of the recruited undergraduates, and hence lead to a deterioration of the diversity among those we both generate and can attract. A second is for undergraduate courses to be maintained, but on a severely bargainbasement teaching cost basis, reducing the preparedness of the students so produced for postgraduate study is likely to be lowered. And a third is for Bristol and other elite universities to withdraw from such undergraduate programmes no longer underpinned by Government teaching funding, so damaging their ability to generate the quantity and quality of postgraduate enrolments from their undergraduate feedstock. 51 7. References ALBERT, C. 2000. Higher education demand in Spain: the influence of labour market signals and family background. Higher Education, 40, 147-162. EQUALITY CHALLENGE UNIT 2009. Equality in higher education: statistical report 2009. London: Equality Challenge Unit. HEFCE (HIGHER EDUCATION FUNDING COUNCIL FOR ENGLAND) 2004. Review of widening participation research: addressing the barriers to participation in higher education. Bristol: HEFCE. HOARE, A. and AITCHISON, R. 2009. PQA: Pretty Questionable Assumptions. Higher Education Review, v42 n1 p19-51. HOUSE, G. 2010. Postgraduate education in the United Kingdom. Higher Education Policy Institute and British Library. LEACY, A., BARNES, P. and HEPWORTH, S. 2010. What happens next? A report on the first destinations of 2005 graduates with disabilities. Sheffield: Association of Graduate Careers Advisory Services Disability Development Network. MACHIN, S. & MURPHY, R. 2010. The social composition and future earnings of postgraduates, Sutton Trust. MARINO, R., MORGAN, M., WINNING, T., THOMSON, W., MARSHALL, R., GOTJAMANOS, T. and EVANS, W. 2006. Sociodemographic backgrounds and career decisions of Australian and New Zealand dental students. Journal of Dental Education, 70, 169. MILLETT, C. 2003. How Undergraduate Loan Debt Affects Application and Enrollment in Graduate or First Professional School. Journal of Higher Education, 74, 386-428. POLLAR, E., CONNOR, H. and HUNT, W. 2008. Mapping provision and participation in postgraduate creative arts and design. Brighton: University of Sussex: Institute for Em ployment Studies. PURCELL, K., ELIAS, P., DAVIES, R. and WILTON, N. 2005. The Class of'99: A study of the early labour market experience of recent graduates. London: DfES. SMITH, A, BRADSHAW, T., BURNETT, K., DOCHERTY, D., PURCELL, W and WORTHINGTON, S. 2010. One step beyond: making the most of postgraduate [Online]. Available: www.bis.gov.uk/one-step-beyond [Accessed 1st November, 2010]. STUART, M., LIDO, C., MORGAN, S., SOLOMON, L. and AKROYD, K. 2008. Widening participation to postgraduate study: Decisions, deterrents and creating success. Higher Education, 1. VICTORA, C., HUTTLY, S., FUCHS, S. and OLINTO, M. 1997. The role of conceptual frameworks in epidemiological analysis: a hierarchical approach. International Journal of Epidemiology, 26, 224. WAKELING, P. 2005. La noblesse d'état anglaise? Social class and progression to postgraduate study. British journal of sociology of education, 26, 505-522. WAKELING, P. 2009. Are Ethnic Minorities Underrepresented in UK Postgraduate Study? Higher Education Quarterly, 63, 86-111. 52 WAKELING, P. and KYRIACOU, C. 2010. Widening participation from undergraduate to postgraduate research degrees: A research synthesis. York, United Kingdom: University of York. WEGNER, E. 1969. Some factors in obtaining postgraduate education. Sociology of Education, 42, 154-169. 53
© Copyright 2026 Paperzz