Understanding student learning in higher education is a challenging process. We are constantly looking to frameworks, tools, theories, practices that give us insight into the learning processes of students. One framework that has been used heavily in Australia is the Student Approaches to Learning (SAL) approach (Biggs, 1987b). Biggs (1993) argues that the SAL approach best accounts for the context specific motive and strategy components in students’ approaches to learning, and their relationship to students’ intentions, the teaching/learning context and the quality of the learning outcome.

SAL conceptualises student learning as a composite of motives and strategy dimensions termed surface, deep, achieving and deep achieving approaches, which categorise important differences in the way in which students approach their learning. Implicit in this theory is that the motives and strategies students bring to the instructional setting are amenable to change. The research on approaches to learning and studying gives strong support to the view that successful learning outcomes are aligned to deep and deep-achieving approaches and that students can be taught to adopt these approaches. SAL is the theory from which the Study Process Questionnaire (SPQ) is derived.

In the study reported here the SPQ was used to explore how students in psychology approach learning. It considers the relationship between approach to learning and academic success as measured by grade point average (GPA) as well as changes across time, differences between gender and the impact of entry method to university on approaches.

Participants

The participants were 297 students studying psychology in an Australian university. This sample consisted of 27 per cent males and 73% females. The average age was 24.73 (SD = 9.41) at the time of the first questionnaire. Eighty-six per cent of students were studying for a Bachelor of Arts, whilst 21 per cent were studying a Bachelor of Science and 1 per cent were studying for a Bachelor of Education. The majority (86 %) were doing their first year at University at the time of the first Questionnaire administration. Forty-nine per cent of students had been admitted to University on the basis of Year 12 marks.

A sub-sample of 64 students completed the questionnaire at the first and second administrations of the Questionnaire. This first sub-sample consisted of 84 % females and 26 % males. The average age was 21.64 (SD = 6.50). Most were studying for a Bachelor of Science (50%) followed by 45 % studying for a Bachelor of Arts and 5% for a Bachelor of Education. Ninety-two per cent of students were doing their first year at University at the time of the first Questionnaire administration. Table 1 presents the full demographics for each sub-sample.

A second sub-sample of 23 students completed the questionnaire at the second and third administrations. Seventy-one per cent of students were female and twenty-nine per cent were male. The average age was 18.58 years (SD = 0.7). Ninety-four per cent were studying for a Bachelor of Arts, three per cent for a Bachelor of Science and three per cent for a Bachelor of Education.

Three more sub-samples contained students who had completed only one SPQ. Students who completed the first and third SPQs were, by necessity, excluded from the data analysis, as they were not part of the cross-sectional or longitudinal data. The first cross-sectional sub-sample closely matched the full sample’s age and gender demographics. The biggest difference was that they were all Bachelor of Arts students. The second, smaller sub-sample was younger by an average of 4 years than the full sample. Moreover, 60% of the students were studying for a Bachelor of Science. The third sub-sample matched the sample demographics, with only slightly more Bachelor of Arts students (see Table 1).

Table 1. Comparison of demographics for each sub-sample

Instrument

The SPQ (Biggs, 1987) provides data on students' approaches to learning and studying. The SPQ is a 42-item self-report inventory designed for use with tertiary students to measure the motives and strategies that comprise their approaches to learning and studying. Items are rated on a 5-point Likert-scale in response to statements about the student's usual ways of studying. A 5 indicates that the statement is "always or almost always true of me" and a 1 indicates that the statement is "never or only rarely true of me". Nine scores are obtained from scales dealing with three motives, three strategies and three approaches (surface, deep, and achieving) and an additional deep-achieving approach score is produced by summing the deep and achieving scale scores. So, ten scores in all are available.

Procedure

The SPQ was administered to students studying psychology four times over the course of three years following the standard procedures specified in the manual. The administrations were as follows: once in first semester year 1; once in second semester year 1; once in semester 2 of their second year; and once in semester 2 of their third year. The questionnaire was administered in year 1 and 2 during tutorial periods and year 3 at the end of a lecture period. Participation was voluntary.

All analyses were conducted using the Statistical Package for the Social Sciences (SPSS) version 10. Changes in SPQ scores across time were measured by repeated-measures ANOVA and by one-way ANOVA, to maximise the available data. Test-retest correlation coefficients were used to determine stability of students’ SPQ approach scores over time. Gender differences were analysed using ANOVA. Pearson product-moment correlations were performed on SPQ scores to determine the relationship between SPQ and GPA.

The Study Processes Questionnaire (1987) has 42 items that form 6 sub-scales and 4 scales. It is normed for Bachelor of Arts, Bachelor of Science and Bachelor of Education tertiary students of both genders. Examining those categories with at least 30 subjects there were substantial differences between the norms and the current sample. Males studying for a BA in the first trial scored much higher in Surface Approach and Achieving Approach scales and only slightly higher in Deep Approach (see Table 2). Across the three trials, females studying for their BA scored consistently and substantially higher in Surface Approach. Initially, they matched the norm in Deep Approach, but their Deep scores substantially declined in subsequent trials. In contrast, females scored higher than the norm initially in Achieving Approach, but then regressed to the norm in subsequent trials (see Table 3).

Table 2. Differences in SPQ sub-scale scores from Biggs (1987) SPQ norming sample (p.30) and present sample

Table 3. Comparison of SPQ sub-scale scores from Biggs (1987) SPQ norming sample (p.31) and present sub-samples

The internal reliability of sub-scales was examined for the first trial. The SPQ Manual reported a number of studies indicating that the Surface Motive scale and sub-scales were generally the least reliable whilst the Achieving scales and sub-scales were generally higher across the different studies. These results were replicated within this study (see Table 4).

Table 4. Comparison of reliability data from past and present samples

Data taken from the SPQ manual (Biggs, 1987, p 22).

The test-retest reliability ranged from –0.052 (p = 0.812) for Achieving Strategy at the second and third trials to 0.719 (p = 0.000) for Surface Strategy at the second and third trials (see Table 5 for a complete breakdown of reliability scores).

Table 5. Test - retest reliability scores

Biggs (1982) argues that the construct validity of the SPQ as measured by grades will be low (about 0.20 – 0.30) because of other variables such as presage factors which includes IQ and home environment. Never the less, Biggs makes several predictions. Firstly, a negative correlation between Surface Approach and GPA. Secondly, a positive correlation between Achieving Approach and GPA. Thirdly, that the strongest correlation will occur between Deep Approach and the student’s grade for their favourite subject. The first two predictions are tested in this study; however, the third could not be tested, as individual ‘favourite’ subject grades were not available. Generally, the results from this study did not support Bigg’s predictions; however, there were two exceptions. Firstly, anyone continuing to use a Surface Approach in their second semester continued to receive low marks. And secondly, an Achieving Approach was moderately correlated with good grades in the first semester of a student’s second year (see Table 6).

Table 6. Construct Validity of SPQ, based on GPA

Changes across time

A series of repeated-measures ANOVA were used to test whether there were any longitudinal changes across the three SPQ administrations. Because only 14 students completed all three questionnaires a step-wise analysis was performed whereby data was analysed at time 1 and 2, then analysed at time 1 and 3 using a slightly different sub-sample of students and then analysed at Time 1 and 4 with another slightly different sub-sample. The results showed a significant decrease in Deep Approach, Achieving Approach and Deep Achieving Approach from the first to the second administrations of the questionnaire. However, there was no significant change from Time 1 to Time 3. There was no significant change at any time in the Surface Approach scale.

Surface Approach

The cross-sectional and most of the longitudinal analysis show a slight peak in Surface Approach scores at the second administration of SPQ. The exception was the ‘long’ repeated measures analysis of Times 1, 2, and 3 whereby the scores showed an overall decline. However, this is a small sub-sample at odds with all other analysis (see Figure 1 for details).

Figure 1. Surface Approach ANOVAS

Deep Approach

The cross-sectional and longitudinal data showed a decline in Deep Approach scores which was greatest between the first and second administrations.

Figure 2. Deep Achieving ANOVAS

The longitudinal data showed that the initial drop in performance was due to a decline in the Deep Strategy sub-scale, whereby scores fell from 23.54 (SD = 3.87) to 21.95 (SD = 4.00, F(1, 64) = 10.998, p = 0.002. Examining the cross-sectional changes in the Deep Approach Scale it was observed that declines occurred in both the Strategy and Motivation sub-scales. The average Deep Strategy score fell from 24.24 (SD = 3.56) to 22.19 (SD = 5.33) and then at the third administration fell to 20.84 (SD = 4.92) (F (2, 200) = 11.71, p = 0.000. Likewise the Deep Motive scores fell from 23.83 (SD = 4.57) at the first administration to 21.81 (SD = 6.09) at the second and 21.58 (SD = 5.37) (F (2,200) = 4.394, p = 0.014) (see Figure 2 for full details of changes in Deep Approach).

Achieving Approach

The longitudinal data largely shows a decline in Achieving Approach scores, which was sharpest from the first to the second administration of SPQ. The exception to this rule was for the longest longitudinal data (Time 1, 2, and 3) which failed to reach statistical significance, but still followed the same trend of a sudden drop followed by a gradual decline. The initial drop in performance was due to changes in both the Achieving Motivation and Achieving Strategy sub-scales. Achieving Motivation fell from 23.62 (SD = 4.05) to 21.87 (SD = 21.89), F(1, 63) = 14.465, p = 0.000. Achieving Strategy declined more radically from 24.50 (SD = 5.15) to 19.23 (SD = 5.19), F(1, 63) = 14.465, p = 0.000.

The cross-sectional data replicated the longitudinal data with a substantial decline in Achieving Approach scores. Both Achieving Strategy and Achieving Motivation contributed to the changes in Achieving Approach. Achieving Motivation scores fell from 23.80 (SD = 4.76) to 23.46 (SD = 4.04) and then to 21.51 (SD = 5.45), F(2, 200) = 4.520, p = 0.012. Achieving Strategy scores fell from 23.47 (SD = 5.42) to 19.39 (SD = 6.19) and then to 18.92 (SD = 6.50), F (2, 199) = 13.687, p = 0.000 (see Figure 3 for full changes in Achieving Approach scores).

Figure 3. Achieving Approach ANOVA

Deep Achieving Approach

Longitudinal and cross-sectional data showed a dramatic drop in initial Deep Achieving Approach scores, which sank still further by the third administration and levelled out by the fourth administration. The results were consistent across all data analysis, except the failure to achieve statistical significance in the longitudinal analysis from Time 1 to 4, which still showed a downward trend (see Figure 4 for full comparisons of Deep Achieving Approach analyses).

Figure 4. Deep Achieving Approach ANOVAS

SPQ Results across time

Gender differences were analysed using ANOVA. Unlike previous data analysis whereby data was examined longitudinally or cross-sectionally, all data was analysed cross-sectionally for gender differences. There were no gender differences in Surface Approach, Deep Approach, Achieving Approach or Deep Achieving Approach at the time of the first questionnaire administration. However, at the second administration there were differences between males and females on Deep Approach, Achieving Approach and Deep Achieving Approach with males outscoring females on all variables. Males scored significantly lower on the Surface Approach scale than females at third administration (see Figure 5 for cross-sectional gender differences). The available longitudinal data was examined from Time 1 to Time 2, because this was when the greatest changes occurred and we had the largest sample. There were no significant interactions between gender and time of administration.

Surface Approach

There were no gender differences in Surface Approach scores for the first two administrations of the SPQ. However, at the last administration females scored 46.38 (SD = 7.57), whilst males only scored 42.15 (SD = 7.71). This was a significant difference, F(1, 106) = 6.229, p = 0.014, which could be traced back to a significant difference in SM. For SM at the third administration females scored 24.42 (SD = 4.16), compared with 21.30 (SD = 4.69) by males. F(1, 107) = 10.988, p = 0.001.

Figure 5. One-way ANOVA for Gender * Surface Approach

Deep Approach

There were no significant differences in males and females for the first administration of the SPQ on the Deep Approach sub-scale. Males scored on average 47.18 (SD = 10.01) whilst females scored an average of 47.33 (SD = 7.21) However, at the second administration males outscored females with an average score of 46.76 (SD = 10.53), whilst females on average scored 42.33 (SD = 7.44). The result was significant. F(1, 108) = 5.602, p = 0.020. The difference in Deep Approach scores did not persist into the third administration with males scoring an average of 43.50 (SD = 8.86), whilst females scored 41.96 (SD = 8.52). F( 1, 106) = 0.630, p = 0.429 . The gender difference at the second administration may be traced to a significant difference in Deep Strategy scores with males scoring an average of 23.96 (SD = 5.51) whilst females scored an average of 21.33 (SD = 3.61). However, it should be noted that the homogeneity of variance assumption was breached for this test (L (1, 108) = 5.991, p= 0.016).

Figure 6. One-way ANOVA for Gender *Deep Approach

Achieving Approach

There was no significant gender difference in Achieving Approach

scores at the first administration. On average males scored 37.5 (SD = 10.01), whilst females scored 41.33 (SD = 9.52). By the second administration males had improved their scores to 45.32 (SD = 8.86) whilst females had maintained an average score of 40.74 (SD = 7.72). This was statistically significant, F(1, 108) = 6.349, p = 0.013. The surprise came at the third administration when the situation was totally reversed. The average score for males was 37.50 (SD = 10.2) whilst females only scored 41.36 (SD = 9.35). This difference approached statistical significance F (1, 107) = 3.208, p =0.076 and can be traced back to an almost significant difference in Achieving Motivation scores: males scored 19.70 (SD = 5.99) and females scored 21.88 (SD = 4.66). F(1, 107) = 3.818, p = 0.053.

Examination of the available longitudinal from the first to the second data collection shows that males dropped from 51.6 (SD = 8.21) to 45.20 (SD = 10.27) whilst females dropped from 47.48 (SD = 7.73) to 40.37 (SD = 7.86). There was no significant interaction between gender and time at the second stage of data collection (i.e., between Time 2 and 3). F(1, 62) = 0.089, p =0.766.

Figure 7. One-way ANOVA for Gender * Achieving Approach

Deep Achieving Approach

Gender differences appeared only at the second administrations of the Deep Achieving Approach sub-scale. Males outscored females with an average score of 92.08 (SD = 17.16) whilst females only achieved an average score of 83.07 (SD = 12.88. The difference was statistically significant, F(1, 108) = 8.066, p= 0.005. Longitudinal data indicated that there was no interaction between time and gender on DAA scores, F(1, 62) = 0.151, p = 0.699.

Figure 8. One-way ANOVA for Gender *Deep Achieving Approach

There were weak to moderate positive correlations between age and various SPQ sub-scales. Surface Approach had a consistently negative correlation with age which attained statistical significance at Time 1, r(173) = 0.212, p = 0.005 and Time 3, r(105) = -0.216, p = 0.027. Age consistently correlated with Deep Approach, r(173) = 0.347, p = 0.000 at Time 1, r(,69) = 0.295, p = 0.014 at Time 2, r(34) = 0.453, p = 0.000 at Time 3. Correlations between age and Achieving Approach were less consistent with only a moderate correlation at Time 2, r(69) = 0.347, p = 0.004. The correlation between age and Deep Achieving Approach started moderately with a correlation of 0.271 (p = 0.000) which increased to 0.369 (p = 0.002) and then decreased to 0.299 (p = 0.000) (see Table 7 for complete correlations.)

Table 7. Correlations between age and SPQ sub-scales

When the SPQ was first administered not all participants were (or would have been) in their first year of study. However, a student’s year level at the time of the first SPQ administration never affected SPQ scores at any time in any given cohort. The strongest correlation was between year level and DA (Time 3) which had a weak, non-significant correlation of 0.165 (p = 0.090).

Students can enter University straight from Year 12 or on the basis of professional qualifications, advanced standing from another degree, or marks from a bridging course or a combination of two or more entry methods. Fifty-four per cent of students entered solely on the basis of their year 12 marks. Method of entry was collapsed into those students who entered exclusively on the basis of year 12 marks and those who used one or more other entry methods. Entry method was not found to affect SPQ scores at any time on any sub-scale.

The complexity of the relationship between approach and academic performance pinpoints some of the difficulties that are encountered when using the SPQ. Interpretation of scores can be made at a relatively superficial level based on information provided in the manual. However, at this level, there is a danger of stereotyping students as either surface or deep learners and this invites undesirable value judgements (Murray-Harvey, 1994).

Academic performance is influenced by a large number of variables of which approach is one. At the institutional level, there is a growing interest in the effects on achievement of higher education selection and admission procedures and policies; teaching and assessment practices; academic curricula; and student support services. At the student level, academic achievement has been shown to be related to motivational and attitudinal factors; prior performance; cognitive variables, for example, abilities, strategies and skills; and personal characteristics, such as locus of control and self-concept. This suggests that there are other variables that also have an impact on academic achievement.

In this study, the link between Surface Approach and GPA is variable. On the one hand there was a short-lived link between high Surface Approach scores and low grades. On the other hand, there were no significant changes in Surface Approach scores over time. This suggests that either a Surface Approach is only occasionally a problem for students, depending on the syllabus, or that students become more discerning in their use of a Surface Approach. This second hypothesis is in keeping with Transfer Appropriate Processing (TAP) and also what we know from the work on profiling successful learners.

The Deep Approach to learning is based on Levels of Processes Theory (Craik and Lockhart, 1972). Consequently, the non-existent correlations between Deep Approach scores and GPA are alarming. Allowing for the fact that a Deep Approach, per se, does not necessarily guarantee high GPA, it could be expected that Deep Achieving Approach scores would correlate with academic success. However, the lack of a significant moderate correlation between Deep Achieving Approach scores and GPA raises serious concerns about SPQ and the assessment process. At the very least there is a lack of fit between the prescribed study processes and grades. The reason for this lack of fit is inexplicable assuming that the material being studied is rich enough to stimulate and sustain a deep interest in the subject, and this then raises questions about the impact of forms of assessment on approaches to study.

The unexpectedly low construct validity of Deep Approach and Achieving Approach raises questions about the syllabus and the two constructs. The fact that Zeegers (2001) also found the same trends in Deep Approach and Achieving Approach scores across time suggests that students come to University expecting to use a Deep and/or Achieving Approach, but then turn to another approach when confronted with the demands of the University syllabus. The problem with this hypothesis is the unexpected single high correlation between Achieving Approach at Time 3 and GPA at that time. However, the low sample (n= 43) for that time, makes the results less reliable than the previous non-significant results where the sample was larger. Given that there are questions that the SAL approach as measured by the SPQ can not shed light on, a complementary perspective on learning such as Transfer Appropriate Processing (TAP) may be useful to consider.

Dyne, Taylor & Boulton-Lewis (1994) introduce the idea of complementary frameworks for understanding student learning in higher education, that is, "a range of well researched and justified perspectives" (p. 370). To this end, they discuss Transfer Appropriate Processing (TAP). TAP is derived from Levels of Processing Theory (LOP) and incorporates the relationship between encoding and retrieval (Bransford, Franks, Morris & Stein, 1979; Morris, Bransford and Franks, 1977; Roediger & Blaxton, 1987; Roediger & Weldon, 1987; Stein, 1978). TAP theories suggest that test performance is dependent upon the extent to which memory permits the appropriate transfer of knowledge gained during the study phase. Consideration of the encoding and retrieval processes is inherent in this viewpoint (Dyne, Taylor & Boulton-Lewis, 1994).

According to Stein (1978) the particular processing style adopted is interactively determined by the characteristics of the presented material (e.g., stimulus structure), the task demands (e.g., levels of processing) and the learner’s ability to detect relevant information. As suggested by Stein, the value of a particular learning strategy is relative to a particular learning goal. This is in common with many SAL theories of approaches to learning (e.g., Biggs, 1989, 1990; Entwistle, 1986). Since the value of a particular learning strategy is relative to a particular learning goal, TAP theory also acknowledges the role of both the students’ learning goals and the testing procedure which seems to be a critical to understanding student learning.

TAP theory also emphasises the students’ need to be flexible in the choice of transfer appropriate strategies. TAP theory suggests that a deep approach to learning requires flexibility in the choice of learning strategies, and the ability to tailor this flexibility to the retrieval situation. This notion of flexible use is congruent with the research into describing what it is that successful learners do. This research focuses heavily on the role of cognitive strategies and transfer of these across learning contexts.

Successful learners possess an array of techniques, or cognitive strategies, for accomplishing goals; metacognitive knowledge about when and how to use these strategies; and an extensive non-strategic knowledge base (declarative or content knowledge) that can be used in conjunction with strategic and metacognitive processes (Pressley, Borkowski, & O'Sullivan, 1984; Pressley, Goodchild, Fleet, Zajchowski ,& Evans, 1989).

Not only are successful learners good strategy users but they are able to transfer strategies readily and appropriately to new settings (Pressley, 1986; Pressley, Borkowski, & Schneider, 1987). Good strategy users have many strategies, some of them specific to the domains, or subjects, in which they are an expert. Although expertise involves acquiring a lot of declarative knowledge about a subject area, it also requires mastery of domain-particular strategies.

According to Gagne, Yekovich, & Yekovich (1993) transfer depends a great deal on the amount of overlap or common elements between the learning task and the transfer task. In strategy learning, the dual role of cognitive strategy use and metacognitive strategy such as the learner's conscious evaluation of strategy effectiveness is a key factor (see also Bassok & Holyoak, 1989; Brown, Kane, & Echols, 1986; Gick & Holyoak, 1983; Holyoak & Koh, 1987; Ross, 1984, 1987).

The flexibility of transfer appropriate strategies in not presently incorporated into the approaches to learning inventories. For example, the items intended to load on the deep strategies factor of the SPQ tend to address the student’s intention to understand, integrate or organise the material, whereas items that address the student’s preference for rote learning of discrete pieces of information represent the surface strategy factor.

Thus, the structure and general nature of the SPQ does not incorporate flexibility in the choice of encoding process according to the requirements that are specific to each task. Furthermore, the inflexibility in classifying students as having a general predisposition towards ‘deep’ or ‘surface’ learning contradicts the suggestion by both TAP theory and Biggs (1993) that rote rehearsal can be an optimal learning strategy in some test situations. Perhaps then, student learning strategies are better classified as whether they are ‘task-appropriate’ or ‘task-inappropriate’ (Dyne, Taylor & Boulton-Lewis, 1994), rather than by terms such as ‘deep’ or ‘surface’ strategies.

In conclusion, in common with Dyne, Taylor & Boulton-Lewis (1994) a more encompassing view of what constitutes successful learning in higher education is essential. Complementary perspectives on learning accommodate the strengths and address the weaknesses of various theories and in doing so contribute to our understanding of student learning in higher education.

References