Access policy and approaches to learning

Introduction

In most countries, higher education institutions seek to recruit the most highly qualified students leaving secondary school. However, in the United Kingdom, the provision of higher education has recently undergone rapid expansion. This is as a result of changes in educational policy and funding arrangements. To sustain this expansion, and to respond to demands to increase access to higher education for other student groups, universities and colleges in the UK are having to recruit non-traditional, older entrants from further education colleges. 42% of all individuals joining as full-time, first-year students, are now aged 21 years or over (Department for Education and Employment, 1994a). The majority of students aged 25 or over admitted to undergraduate courses in the UK are female, a trend which is more evident in those aged 35 or over (Department for Education and Employment, 1994b). Students adopting a non-traditional route to university, formed 42% of total admissions to higher educational institutions in England, in 1992 (Department for Education andEmployment, 1994a). Given the changing nature of the student population (Metcalf, 1993), calls to improve access and direct-entry procedures (Fulton and Ellwood, 1989) and that Alevels are considered only suitable for only about 20 per cent of the population (Deere, 1992); it is surprising many admissions tutors are continuing to concentrate on school-based academic achievement (Brown and Bimrose, 1993).

The National Committee of Inquiry into Higher Education (NCIHE), considering the role of higher education in the UK over the next 20 years, recognises as a principle in its terms of reference that: "There should be maximum participation in initial higher education by young and mature students and in lifetime learning by adults, having regard to the needs of individuals." (NCIHE, 1996) However, there is a general paucity of research literature on the experiences of mature and non-traditional entrants in higher education (Richardson, 1994 andWoodley, 1981). Most of the studies that consider older students have focused on their academic performance rather than their experience of higher education (Richardson, 1994). Furthermore, little research considers how students' prior learning experience may affect their approach to learning.

Approaches to Learning in Higher Education

How individuals approach learning, that is, the relative emphasis given to learning with understanding relative to the reproduction of facts, or rote learning, has been an important and long-standing interest for education researchers. A broad consensus now exists that three clearly identifiable approaches to learning exist: a deep approach, a surface approach and a strategic approach. Their characteristics are defined in Table 1 below: (1987) It is generally accepted that a deep approach is considered desirable in higher education. Considerable work has focused on how such an approach may be stimulated. Research has attempted to operationalise the distinction between the three defining approaches to learning in terms of students' responses to different scales of particular self-report inventories. Three commonly used instruments to measure students' approaches to learning are: the Approaches to Studying Inventory (Entwistle and Ramsden, 1983); the Inventory of Learning Processes (Schmeck et al, 1977) and the Study Processes Questionnaire (Biggs, 1985). Each of these instruments has undergone extensive conceptual and psychometric evaluation.

This study investigates the effect of differences in entry qualifications, gender and age in approaches to learning using a revision of the Approaches to Studying Inventory, the Revised Approaches to Studying Inventory -1995a version (Tait and Entwistle, 1995). The study uses two samples of second and third-year accounting students at a 'new' university in Scotland. Entry qualifications are categorised into two groups: those gaining entry to university at second or third-year from further education ('direct-entry' students) and those who enter in first-year. Students are categorised into two age groups: mature students (i.e. those aged 25 or over on admission to university) and non-mature students. The present study proposes a methodology to take account of identified sampling and analysis problems with previous research.

Hypotheses

The hypotheses to be tested were developed in the context of previous research considering differences in entry qualifications, gender and age in students' approaches to learning.

A literature search reveals only one published work considering the effect of different entry qualifications in students' approaches to learning. Henson and Schmeck (1993) using a sample of college and university students in the United States, and the Inventory of Learning Processes (ILP), report community college (i.e. further education) students are less likely to adopt a deep approach than their university counterparts.

However, Roberts and Higgins (1992), reporting the views and experiences of students in higher education examined differences between students with a background in further education to those with A-levels. Students entering university or college from further education felt better prepared than those holding the school-based A-level qualification in nearly every respect. In particular, students holding A-levels felt their school-based education did not encourage the development of the study skills most frequently required in higher education, such as independent study, note taking, planning and report writing. Students with a background in further education believed they were used to researching their own work, using libraries and working on real projects rather than essays. However this work does not consider the experiences of those entering from further education directly into second or third-year of their chosen course. Therefore, the first null hypothesis tested is as follows:

H 1 : There are no significant differences in approaches to learning between direct-entry students and those entering university in first-year

Previous research investigating gender differences in approaches to learning in higher education have found evidence to suggest differences exist. Investigations have typically taken the form of quantitative measurement of approaches to learning using either the Study Processes Questionnaire (SPQ) or the Approaches to Studying Inventory (ASI). The results of these studies are shown in Table 2 below:

Reviewing the gender differences in approaches to learning literature, Wilson et al (1996) report two methodological problems with prior research. A common methodological issue concerns sampling bias, with a number of studies reporting less than satisfactory response rates. Also, many studies do not provide any specification of sample characteristics in terms of gender, year or discipline of study. As approaches to learning are known to vary as a function of year and discipline of study (Biggs, 1987;Hattie, 1981, 1985) it is possible any gender related differences could be attributable to the differential distribution of male and female students across years or disciplines of study. The second methodological problem concerns the use of inappropriate statistical procedures. Many studies reporting significant gender differences made multiple use of univariate statistics, without employing a more stringent significance level to attempt to avoid family-wise Type 1 errors. None of the studies reporting statistically significant differences, consider the measurement of effect size. Therefore some studies achieve statistical significance simply as a result of sample size. In conclusion, the most consistent finding to date, is an absence of gender differences at the scale level of approach to learning. Consequently, the present study tests the following null hypothesis:

There are no significant gender differences in approaches to learning Wilson et al (1996) ASI, SPQ Not reported No significant differences The next hypothesis examines whether age differences exist in approaches to learning.

Reviewing the approaches to learning literature, Richardson (1994) observes that most research has ignored age as an important explanatory variable. He reports a consistent suggestion that mature students are likely to exhibit the more desirable deep approaches than younger students. The results of this literature search considering age and approaches to learning using self-report inventories is summarised in Table 2.

However, similar methodological problems exist with much of the literature concerning age differences in approaches to learning as with the research that considers gender differences, namely sampling problems caused by inadequate response rates, poor disclosure of sample composition, the sampling of non-homogenous populations of university students, the inappropriate use of univariate statistics and no measurement of effect size. Therefore the third null hypothesis to be tested is as follows:

H 3 : There are no significant age differences in approaches to learning.

Method

Revised Approaches to Studying Inventory (RASI)

The Revised Approaches to Studying Inventory is a more recent version of the ASI, includes a number of changes to the scales within the Deep and Surface scales. The revision includes new labels to make the terms more transparent to non-specialists and to produce comparable features between orientations. Each orientation has a distinctive form of motivation associated with it. Each of the main scales shows certain study processes associated with a distinctive intention (although this coincides with motive in the achieving orientation). The present study uses a short form of the 1995 RASI. The instrument is a 30-item, 12 scale questionnaire to establish how students tackle everyday learning tasks. Each scale consists of two to four items. Items are scored on a 5-point Likert scale. The instrument has undergone extensive psychometric investigation and demonstrates satisfactory reliability and validity (Duff, 1997). Table 3 shows the factor pattern matrix for the RASI of the sum of item scores of the 12 scales after oblique rotation and provides evidence of its internal consistency reliability. Internal consistency reliability is measured by Cronbach's alpha coefficients for the present sample. The three factor pattern matrix for the RASI for the present sample, is reported after oblimin rotation to simple structure. An examination of the pattern matrix revealed that items for the three scales loaded consistently on the four factors as expected, suggesting high construct validity. The combination of scales into 'approach' scores, created alpha coefficient values between 0.78 to 0.82 -reasonable for scales of that length (Nunnally, 1978). Some of the scales contained items which were inadequately homogenous within their scale, suggesting they should not be used as separate variables for describing differences between students. Accordingly, the present study only uses the three approach scales in its analysis.

Sample

The RASI was administered to two independent samples, being students majoring in accounting in the 1995/96 and 1996/97 academic year. The first sample consisted of 179 second and third year accounting, business information technology (accounting stream) and combined studies (majoring in accounting) students (78 males; 101 females) following BA degree programmes at a university in Scotland in 1995/96. The ages of the males ranged from 18 to 52 years, and the ages of the females from 17 to 50 years. 73 were direct-entry students (35 male; 38 female); 106 students (43 male; 63 female); had already completed 1 or 2 years study at the institution.

The second sample consisted of 137 second and third-year accounting, business information technology (accounting stream) and combined studies (majoring in accounting) students (43 males; 68 females) following BA degree programmes at the same university in Scotland in 1996/97. The ages of the males ranged from 18 to 48 years (mean = 23), and the ages of the females from 18 to 52 years (mean = 24). 27 were direct-entry students (13 male; 14 female); 84 students (30 male; 54 female); had already completed 1 years study at the institution. 26 students declined to identify themselves.

Procedure

The instrument was administered by the author to the two samples in the first week of term in two lectures to the second and third-year students respectively. The first sample was collected in October 1995 and the second in October 1996. By sampling two cohorts of second and third year students we can be more confident the results represent differences in the wider population of second and third-year accounting students. Participation was entirely voluntary and students were given an assurance that the data would not affect any assessment procedures. Administration time totalled approximately 15 minutes. For the first sample, 93.8% of the second and third-year population enrolled on the accounting programmes were present at the time of administration, and of these, 99.4% completed a useable questionnaire. For the second sample 90.2% of the second year and third-year population were present at the time of administration, and of these 100% returned a useable questionnaire. The response rates were thus, 93.2% and 90.2% respectively. In summary, we can be confident that the sampling procedures have avoided the problems inherent in previous research; including inadequate response rates and those which sample quite distinct populations of university students.

Analysis

In order to test for differences in entry qualifications, gender and age in approaches to learning, data for the RASI was analysed by means of a repeated measures analysis of variance. The causal model to test the three hypotheses was as follows: -Approach to Learning (Deep, Surface, Strategic) = a + b (Age Group) + c (Direct Entry) + d (Gender) + error Analysing age, entry and gender differences on each of the three RASI scales represents multiple testing. Accordingly, special procedures are adopted to incorporate dependencies within the experimental unit -a repeated measures design. To test the three hypotheses, the three RASI scale scores (the dependent variables) must be transformed. That is, instead of analysing the three RASI scale scores, linear combinations of their differences are analysed. These linear combinations are chosen so that they are: first, statistically independent (orthogonal) and second, so that the sum of the squared coefficients equals unity (normalised). As a result three transformed variables are formed:

T 1 = Deep scale score + Surface scale score + Strategic scale score T 2 = 2 x Deep scale score -(Surface scale score + Strategic scale score) T 3 = Surface scale score -Strategic scale score Separate analyses of variance are performed on each of the three transformed variables. Age group, entry and gender are the independent variables.

Results

H 1 :

The results of the repeated measures analysis of variance is reported in Table 4. As hypothesised, no statistically significant differences between direct-entry students and those entering university in first-year were found (F ratio = 2.17; p = 0.142). No interaction between the three independent variables is reported. The analysis of variance results for transformed variable T 2 and T 3 identify no statistically significant differences. Therefore, the overall effect is consistent across the three scales of the RASI.

H 2 : Table 4 reports statistically significant differences between gender and the three 'approach' scales of the RASI (F ratio = 11.16; p = 0.001). No interaction between the three independent variables is reported. Whilst contrast 3 reports no statistically significant relationship between gender and the three RASI scale scores, contrast 2 identifies statistically significant gender differences (F ratio = 4.94; p = 0.027). This implies the difference reported by the overall effect is not consistent across the three RASI scales. An analysis of mean scores finds females score higher than males on Surface Approach (females = 35.39; males = 32.17) and Strategic Approach (females = 40.20; males = 37.83) -see Table 5. Therefore, although the overall effect reports statistically significant differences, these are not consistent across three RASI scales, with greater differences reported across the Surface Approach and Strategic Approach scales.

H 3 : Table 4 reports statistically significant differences between age group and the three 'approach' scales of the RASI (F ratio = 7.62; p = 0.006). No interaction between the three independent variables is reported. Whilst contrast 3 reports no statistically significant relationship between age group and the three RASI scale scores, contrast 2 reports statistically significant age differences (F ratio = 9.83; p = 0.002). This implies the difference reported by the overall effect is not consistent across the three RASI scales. An analysis of mean scores finds students aged 25 years or older score higher than males on Deep Approach (aged under 25 years = 37.13; aged 25 or older = 41.34) and Strategic Approach (aged under 25 years = 38.60; aged 25 or older = 41.41) -see Table 5. Therefore, although the overall effect reports statistically significant differences, these are not consistent across three RASI scales, with greater differences reported across the Deep Approach and Strategic Approach scales.

Discussion

A general consensus exists in higher education that the encouragement of a Deep Approach is more desirable than a Surface Approach. That is, emphasising the understanding of concepts is preferable to the reproduction of facts. The results of the present study indicate there are no statistically significant differences in approaches to learning between direct-entry and non direct-entry accounting students. Therefore, an open access admissions policy is unlikely to detract from the quality of learning on a business degree course.

The statistically significant gender differences found are inconsistent with the results of previous studies using the ASI and SPQ. The study indicates female students are more likely to adopt a surface approach than males. Previous studies that report gender differences suggest females are more likely to adopt a Deep Approach and less likely to adopt a Surface Approach than their male counterparts. Possible reasons for the difference include the methodological concerns with previous studies of the inappropriate use of multiple univariate statistics and sampling problems including sample response rate, characteristics and homogeneity. Previous work has considered largely social science or mixed groups of students, the only study of gender differences to include business students (Harper and Kember, 1986) was undertaken with a multi-faculty sample of Australian students. The Harper and Kember study reported significant gender differences but failed to indicate the direction of their findings. Possible reasons for the inconsistency of these findings with earlier research include the use of a different research instrument and the use of UK students, in the second or third year of a business course.

Statistically significant differences between age groups are found in both samples. The findings indicate that mature students are more likely to adopt a Deep Approach than their younger colleagues. These results support the findings of Richardson's (1995) study of social science students and provide further evidence that older students are likely to enrich the quality of higher education by providing school-leavers with an opportunity to learn from the mature students' superior approaches to learning.

In conclusion, students with different entry qualifications do not differ in their approach to learning. Therefore, broadening institutional admissions policy is unlikely to be detrimental to the quality of learning. This is a significant finding for admissions tutors and academics who focus on narrow school-based academic achievement (Brown and Bimrose, 1993;Germon and La Gro, 1993). Furthermore, significant gender differences in approaches to study are found and older students are more likely to adopt a Deep Approach than younger students.

Finding ways of promoting deep learning is an increasingly important topic in higher education (Gibbs, 1992). Educational background, age and gender are important social variables for consideration in the design of degree programmes. The identification of consistent and substantive differences would imply differential teaching strategies for different social groups.