Introduction

Over the past two decades, governments, industry, and policy-makers have consistently called for higher enrolments in science, technology, engineering, and mathematics in Australia (Australian Government, 2020a), the UK (Cm8980, 2014), European Union (EU Skills Panorama, 2012), and the USA (National Academy of Sciences, 2010). Such calls underscore the current and potential future value of science graduates to national economies (Australian Academy of Science, 2015, 2016). The timeliness of science undergraduates’ degree (e.g. Bachelor of Science or BSc) completion is crucial for graduates themselves in terms of paid employment (Baum et al., 2013), remediation of tertiary debt (Williams, 2014), and further study opportunities that enhance their employability and likelihood of higher paid employment (Palmer et al., 2018). For universities, enhanced timely degree completion (TDC) is likely to generate greater productivity and efficiency. Finally, there is a monetary windfall associated with more rapid degree completion, which for Australia is estimated to be $A2.7 billion by 2030 (Ernst & Young, 2019).

Despite the recognised importance of TDC, research over the past two decades has shown that in broad terms, tertiary students are taking considerably longer than the minimum specified time to complete their degree (Bound et al., 2012; Contini et al., 2018; Edwards & McMillan, 2015). In comparing the timeliness of students’ completion of their degree across various disciplines, Tentsho et al. (2019) reported that science and technology students took significantly longer than both humanities and social science students and education students to complete their degree. This is consistent with the findings of Lassibille and Gómez (2011), among their cohorts of business, arts, engineering, and humanities students.

Factors shown to impact students’ degree completion time include their socio-economic status (SES) (Bound et al., 2007; Cabrera et al., 2005; Contini et al., 2018; Zarifa et al., 2018), age (Taniguchi & Kaufman, 2005), gender (Corbett et al., 2008; Lin et al., 2020; Tentsho et al., 2019), and measures of academic achievement (Allen & Robbins, 2010; DesJardins et al., 2003; Lin et al., 2020; Tentsho et al., 2019). Other factors include institutional resources (Bound et al., 2007), student motivation (Allen & Robbins, 2010), financial aid (DesJardins et al., 2002), prior attainment (Yue & Fu, 2017), family background (Taniguchi & Kaufman, 2005; Tentsho et al., 2019), and academic major (DesJardins et al., 2003; Lin et al., 2020; Yue & Fu, 2017).

Students’ TDC has been shown to be predicted by their academic success in first year (Allen & Robbins, 2010; DesJardins et al., 2003; Dowd, 2004; Lin et al., 2020; Tentsho et al., 2019) as well their overall academic achievement (Lassibille, 2011; Yue & Fu, 2017). Despite a burgeoning body of literature on the relationship between academic achievement and TDC, in particular the predictive nature of first year academic achievement on TDC, there is an apparent lack of research into the relationship between students’ final year academic achievement and their TDC.

In considering undergraduate students’ TDC, key time-independent variables (TIVs) include SES, Australian Tertiary Admission Rank (ATAR—aka university entrance score), gender, and age of enrolment (DesJardins et al., 2002; Tentsho et al., 2019; Yue & Fu, 2017). Previous research has shown that gender (being female) predicts both academic achievement (Anderton, 2017; Anderton et al., 2016; Li & Dockery, 2015; Messinis & Sheehan, 2015) and TDC (Lassibille, 2011; Lin et al., 2020; Tentsho et al., 2019; Yue & Fu, 2017). Of the TIVs described above, some have been shown to influence certain time-dependent variables (TDVs), such as academic achievement, academic major(s), and the number of majors (Yue & Fu, 2017). Consequently, TIVs and TDVs may interact to determine students’ TDC. Research into a broader range of TIVs and TDVs and their potential role and interactions in shaping TDC appears to be somewhat lacking, in particular for tertiary science education.

This study sets out to investigate specific TIVs and TDVs that predict students’ TDC, for a cohort of BSc undergraduates at an Australian research-intensive university. We thus sought, on this basis, to examine Yue and Fu’s (2017) conceptual model for TDC, using students enrolled in a BSc degree over 2015–2018. Specifically, we aimed to investigate the following questions about the TDC of this cohort of BSc students:

  1. 1.

    Among a range of time-independent and time-dependent variables, which ones predict timely Bachelor of Science degree completion and to what extent?

  2. 2.

    Which time-independent variables influence time-dependent variables and, therefore, timely Bachelor of Science degree completion?

  3. 3.

    What interrelationships exist among time-independent and time-dependent variables, and how does this inform a conceptual model for timely Bachelor of Science degree completion?

Methods

Participants

Study participants were BSc students who commenced their degree at Monash University (Clayton campus) in 2015 (ninitial = 1067). Participants were excluded if they did not possess an ATAR, or an Australian residential postcode, were not full-time students, or if they had discontinued their degree before March 2018 (nfinal = 511). Discontinued students were excluded on the basis that poor academic progress (i.e. achievement) may have been a confounding factor. Study participants maintained continuous enrolment (no intermission or ‘stopping out’) from commencement to completion of the study, as per Yue and Fu (2017).

Of the cohort, 45% and 55% were males and females, respectively, with 91.4% of students aged 17–21 years old and 8.6% aged > 21 years old at commencement of the study. The mean age of females (19.1 ± 0.2) was not significantly different to that of males (18.9 ± 0.1). Participants’ SES was determined by aligning their Australian residential postcode with Australian Bureau of Statistics (ABS) data. The ABS classifies postcodes in decile increments from the lowest SES (score of 1) to the highest SES (score of 10) on a national basis. This scale is based on an index encompassing economic resources, education and occupation of householders identified in the 2016 ABS census, and corresponding to a set of both metropolitan and rural postcodes, per Dollman et al. (2007). A very high proportion (> 85%) of the overall cohort was domiciled in moderate to high SES locations, balanced by gender across most of the ten SES deciles (Fig. 1).

Fig. 1
figure 1

Proportions of study participants’ SES status stratified by gender; open columns represent females, and shaded columns represent males. The aggregate for each series is 100%

Full size image

Data Analysis

Analyses of differences between means were carried out using unpaired Student’s t-test to examine differences between the input values (e.g. ATAR) for each of TDC and non-timely degree completed (NTDC) students. Students’ age at the commencement of their BSc, gender, ATAR, number of academic major(s), academic achievement for each unit, overall WAM, and degree completion status were retrieved from university records, under Monash University Human Research ethics approval number 21007. Student WAMs for years 1, 2, and 3 were derived using university data, as described by Rayner and Papakonstantinou (2020). The mean ATAR of study participants was 81.7. The mean ATAR of females (82.0 ± 0.8) was not significantly different to that of males (81.3 ± 0.8).

Logistic, multiple, and linear regression analyses were carried out using SPSS® version 27.0 for Windows, according to the vendor’s standard procedure and instructions (IBM, 2020). These analyses tested the strength and significance of potential predictive interrelationships among the TIVs and TDVs. Of the independent variables, gender was coded as ‘1’ and ‘0’ for females and males, respectively. Students completing one academic major in their degree were coded as ‘1’ and those completing two majors as ‘2’, respectively. Degree completion status was coded as ‘1’ for BSc completion within 3 years (the minimum timeframe for completion) or ‘0’ if not completed in that timeframe. All analyses used p < 0.05 as the threshold of significance for comparisons and regressions, with regression coefficients (β), Wald, F, and T statistics, as indicated.

Tests of association among categorical variables (e.g. gender) and TDC were conducted using chi-square analysis incorporating Yates’ correction, which is more conservative and less likely to generate a type 1 error (Preacher, 2001).

Results

Interrelationships Among TIVs and TDVs

In terms of students’ academic achievement in their first year of study, ATAR (T = 7.44, β = 0.24, p < 0.0001) and gender (T = 2.84, β = 0.80, p = 0.005) predicted year 1 WAM. Similarly, year 1 WAM (T = 23.99, β = 0.72, p < 0.0001), ATAR (T = 2.63, β = 0.08, p = 0.009), and gender (T = 2.48, β = 0.07, p = 0.014) predicted year 2 WAM. In terms of gender, females had a significantly higher WAM than males for each of years 1 (T = 1.52, p = 0.002), 2 (T = 1.97, p < 0.0001), and 3 (T = 1.63, p = 0.001).

TIVs, TDVs, and Their Interactions

Comparisons were made between TDC and NTDC students (Table 1). There was a significantly higher proportion of females compared to males (χ2 = 3.93, p = 0.047) who completed their BSc in the minimum timeframe.

Table 1 Study participant degree status, overall and stratified by gender. ‘Completed’ indicates those who completed their BSc in a timely (within the minimum 3-year timeframe) manner. ‘Not completed’ represents students who were still enrolled, or who were on intermission, within this timeframe. Data represents raw number and as a percentage of the total in each column
Full size table

Timely BSc completion was predicted by gender, year 2 and year 3 WAM, and the number of majors, but not by age at enrolment, ATAR, SES, or year 1 WAM (Table 2).

Table 2 Logistic regression of TIVs and TDVs for timely BSc completion. St Error refers to the standard error of the regression, and df refers to degrees of freedom. The Wald statistic tests the significance of individual coefficients in the regression model. NS stands for not significant
Full size table

Year 3 WAM, the strongest predictor of TDC, was only predicted by year 2 WAM (T = 17.81, β = 0.73, p < 0.0001).

ATAR and WAM

For the overall cohort, TDC students had significantly higher mean achievement at enrolment (ATAR), and at each BSc year level, compared to NTDC students. For the ATAR: T = 2.13, p < 0.0001; year 1 WAM: T = 4.69, p < 0.0001; year 2 WAM: T = 5.94, p < 0.0001; year 3 WAM: T = 6.05, p < 0.0001. While the mean WAM of TDC students increased significantly over years 1–3 (T = 2.06, p < 0.0001), there was no such increase for NTDC students.

The patterns described above for aggregated TDC and NTDC students were consistent when each group was segregated by gender (compare columns partitioned by significance bars in Fig. 2). However, there were no significant differences in the ATAR or WAM for years 1, 2, and 3 of female and male TDC students (compare lightly shaded columns and intermediate shaded columns in Fig. 2). While there was no difference in ATAR between female and male NTDC students, such females had higher academic achievement at each year level compared to males (compare open columns and heavily shaded columns in Fig. 2). For year 1 WAM, T = 1.28, p = 0.006; year 2 WAM: T = 1.85, p = 0.0002; year 3 WAM: T = 1.06, p = 0.018. While the mean WAM of TDC students increased significantly over years 1–3 for both females (T = 1.70, p = 0.0004) and males (T = 1.13, p = 0.013), there was no corresponding increase for NTDC students (Fig. 2).

Fig. 2
figure 2

Time course of students’ academic achievement stratified by gender and degree completion status. TDC females (lightly shaded columns), NTDC females (open columns), TDC males (intermediate shaded columns), and NTDC males (heavily shaded columns) are indicated. ** and *** denote a significant difference between mean ATARs at p < 0.01 and p < 0.001; **** denotes a significant difference between mean year WAMs at p < 0.0001. Individual cohort numbers for TDC and NTDC students are as per Table 1

Full size image

Age at Enrolment

We found no significant difference in age at enrolment between TDC (19.1 ± 0.2) and NTDC students (19.0 ± 0.1). When stratified by gender, there were no significant differences in age between TDC (19.3 ± 0.3) and NTDC (18.9 ± 0.2) females and the corresponding TDC (18.8 ± 0.2) and NTDC (19.0 ± 0.2) males.

SES

There were no significant differences between the expected proportions of TDC and NTDC students across each SES decile and also when these proportions were further stratified by gender.

Comparisons of the proportions of TDC and NTDC males and females in each SES decile (Fig. 3) found significant differences between the genders and completion status for deciles 1 (χ2 = 29.43, p < 0.0001), 2 (χ2 = 5.87, p = 0.015), and 5 (χ2 = 7.88, p = 0.005). No significant differences were determined for any of the other SES deciles.

Fig. 3
figure 3

Relative proportions of female and male students (per TDC and NTDC) for each SES decile. TDC females (lightly shaded columns), NTDC females (open columns), TDC males (intermediate shaded columns), and NTDC males (heavily shaded columns) are indicated. The number in brackets for SES decile indicates the total students for that decile

Full size image

Number of Majors

The proportion of TDC students studying two majors (49/70) was significantly higher than expected (χ2 = 5.41, p = 0.020) compared to NTDC students (21/70) (Table 3). This result was a product of each of the analyses for females (29/38 vs. 9/38: χ2 = 5.17, p = 0.023) and for males (20/32 vs. 12/32: χ2 = 5.82, p = 0.016) (Table 3).

Table 3 Timely degree completed (TDC) and non-timely degree completed (NTDC) study participant academic major(s), stratified by number of majors, degree completion status, and gender. Raw numbers are indicated in each column
Full size table

Discussion

Overall Degree Completion

While only 53.2% of study participants completed their BSc in the minimum timeframe, after an additional 20 months (beyond 3 years), the completion rate had increased to 86% (data not shown). This aligns with the overall bachelor degree completion rate of 81% for Monash University, for students who enrolled in 2014 and completed their studies by the end of 2019 (Australian Government, 2020b). Our degree completion rates are similar to those reported by Tentsho et al. (2019) for Thailand (65.1% within 4 years, 89.5% after 5 years) and Bradburn et al. (2003) for the USA (64% in the minimum time, 81% after an additional 12–23 months). Collectively, these results demonstrate the importance of students’ persistence for degree completion, regardless of the additional time taken (Dowd, 2004).

The increased time for students to complete their university degree may be due to macro-level structural changes in the broader higher education sector over the past two decades. These changes include the massification of higher education (Mok & Neubauer, 2016) and concomitant increases in student/educator ratios (Bound et al., 2007). Additionally, the increased cost of tertiary study in Australia (Min & Falvey, 2018) and elsewhere (Calitz & Fourie, 2016) is considered to be associated with more time away from study in paid work (Devlin et al., 2008). However, actual evidence of causality of these variables to bachelor degree completion is sparse, thus providing opportunities for further investigation.

Gender and Timely Degree Completion

That a higher than expected proportion of females compared to males completed their degree in a timely manner is consistent with previous research (Bradburn et al., 2003; Lassibille, 2011; Lin et al., 2020; Tentsho et al., 2019; Yue & Fu, 2017). Possible explanations are that females take more responsibility for their academic success (Young-Jones et al., 2013), may be more disciplined than males in their studies (Tessema et al., 2012), and/or place greater value on the benefits of their degree (Ilias et al., 2008). It is also possible that females have a more positive perception of their learning environment than males (Nair & Fisher, 2000) and/or are more ready to engage in supplemental instruction (Angrist et al., 2009, 2014). Such forms of instruction, including academic advising (Rodriguez-Planas, 2012) and peer-assisted learning (Bowles et al., 2008), can enhance students’ academic achievement and their TDC (Guillen, 2010).

ATAR, WAM, and Timely Degree Completion

The limited published literature on the potential relationship between ATAR and TDC lacks broader consistency. Our finding of a lack of causality between these variables aligns with that of Herbert and Herbert (2018), for ICT undergraduates. However, a predictive relationship between ATAR and TDC has been reported by Shulruf et al. (2018), for New Zealand and Australian medical students. While the nature of this relationship may be discipline-specific, further research is required at the institutional and interdisciplinary levels.

While our participants’ first-year academic achievement did not predict TDC, previous research conflicts with this (Allen & Robbins, 2010; DesJardins et al., 2003; Lin et al., 2020; Tentsho et al., 2019). Our finding regarding that relationship is discipline- and institution-specific, and the cited literature encompasses multiple disciplines and/or institutions. Further investigation of the relationship between first-year academic achievement and TDC, perhaps across multiple institutions, would provide greater understanding in that regard.

Our finding that second and final (third) year WAM predicted timely BSc completion is consistent with Yue and Fu (2017), who found that a higher cumulative GPA towards the end of the bachelor degree was associated with TDC. While there is little specific research on the relationship between these TDVs and timely bachelor degree completion, it is worth noting that in our study, final year WAM was predicted by that of second year, and that in turn by first year WAM (per Fig. 4). These findings have important implications in designing programs that will enhance student engagement and achievement through, for example, opportunities for students’ active learning and the resourcing of advisory and learning skills support (Young-Jones et al., 2013).

Fig. 4
figure 4

A conceptual model regarding timely BSc degree completion. * denotes p < 0.05 level of significance, all other levels of significance are as per Fig. 2

Full size image

Age at Enrolment and Timely Degree Completion

That age at commencement of the BSc did not predict students’ TDC is consistent with DesJardins et al. (2002). However, Lassibille and Gómez (2011) reported that this TIV is positively associated with slow progress towards completing a science degree. Additionally, Lassibille (2011) documented the lack of consistency in the relationship between age and TDC in that review of the published literature. Given this, any possible relationship between these variables is likely to be a consequence of discipline, degree length, and demographics, among other potential factors.

SES and Timely Degree Completion

The lack of a predictive relationship between SES and timely BSc completion aligns with the studies by Knipe (2013) for Australian education undergraduates, DesJardins et al. (2002) for US undergraduates, and Tumen et al. (2008) for New Zealand bachelor degree students. Conversely, a significant relationship between SES and TDC has been reported at the discipline (Contini et al., 2018), degree (Bound et al., 2007; Zarifa et al., 2018), and national levels (Australian Government, 2020b). The apparent lack of a clear relationship between SES and TDC demands further research, notwithstanding the potential value generated through better resourcing of low SES schools (Cooper et al., 2020). That our low SES (deciles 1 and 2) females completed at higher than expected rates compared to males may contribute to the generalised effect of gender on academic achievement (Zarifa et al., 2018) and TDC. Alternatively, this outcome may simply be a statistical artefact, due to the small sample size of such students in the study.

Given the inconsistencies in the literature about the connection between SES and TDC (e.g. Cabrera et al., 2005), our findings may be cohort- or discipline-specific (e.g. Elbulok-Charcape et al., 2021). Alternatively, our results may reflect the positive impact of the Australian Higher Education Participation and Partnerships Program and/or institution-specific programs on the engagement and success of low SES students. As an example of the value of such programs, Reed and Hurd (2016) reported on the substantial positive effect of provision of scholarships to disadvantaged students on their participation at university.

Number of Majors and Timely Degree Completion

Our finding that students with two majors were more likely to complete their BSc in a timely manner than those with only one major is consistent with recent literature (Lin et al., 2020; Yue & Fu, 2017). It is likely that degrees incorporating undergraduate research projects, comprising two or more majors, enhance student learning, and through that academic achievement and TDC (Olivares-Donoso & González, 2019). Further, students who engage in a range of science-related majors are more likely to persist in science careers after graduation, compared to those with a more limited range of majors (Habig et al., 2020). Foraker (2012) and Feldman et al. (1999) found that students who completed one major may have had a lower interest-environment congruence—the fit between their personal and career interests—and thus be more likely to take longer to complete their degree. Our findings point to potential value of students’ engagement in a broader range of science-related studies to enhance their TDC.

A Conceptual Model of Timely BSc Completion

Based on these results, we have devised a conceptual model (Fig. 4) for TDC based on the interrelationships among TIVs and TDVs in this study. This model of TDC, which is specific to this BSc cohort, refines the model of Yue and Fu (2017). For our BSc cohort, gender, the number of academic majors, and academic achievement at each of years 2 and 3 all predict TDC. Variable that did not predict TDC was year 1 academic achievement, students’ age at enrolment, and SES (Fig. 4).

Limitations

A limitation of this study is that it does not consider other variables (TIVs or TDVs) that may have determined or contributed to timely BSc completion. Consequently, our findings may be the result of omitted-variable bias and/or endogeneity. If correct, this would attribute the effect of missing variables to the estimated effect of gender, major number, and years 2 and 3 WAM in our study. As ATAR is a ranking, we acknowledge the limitations of testing associations between ATAR and other variables (e.g. WAM); however, such methods have been previously validated (e.g. Wright, 2015). The inclusion of variables such as students’ motivation (Allen & Robbins, 2010), ethnicity (Tumen et al., 2008), financial circumstances (DesJardins et al., 2002), and parental education (Bradburn et al., 2003; Tentsho et al., 2019) might have provided a more robust basis for analysis. However, such data were not available for this study.

Conclusions

While students’ SES and age at enrolment have been previously identified as potential predictors of TDC, for our study cohort, these factors did not predict their TDC. Importantly, this study reinforces the crucial role of gender (being female) in directly predicting TDC, and as a predictor of students’ academic achievement, which later predicts their TDC. Further, the greater the number of majors that students undertake in their BSc, the more likely they are to complete in the minimum timeframe, which reflects the importance of judicious unit choice and thus choices of majors in course planning. Our conceptual model for timely BSc completion may be useful in providing guidance and resourcing to enhance students’ engagement, retention and academic achievement, and consequently their TDC: to benefit themselves, the university and society more broadly.