267

British Journal of Educational Psychology (2003), 73, 267–290

2003 The British Psychological Society
www.bps.org.uk

Learning styles and academic outcome: The

validity and utility of Vermunt’s Inventory of
Learning Styles in a British higher education
setting

Elizabeth A. Boyle*, Tim Duffy and Karen Dunleavy

School of Social Sciences, University of Paisley, UK

Background. Vermunt’s (1994) Inventory of Learning Styles (ILS) integrates four

components of learning: processing strategies, regulation strategies, mental models of
learning and learning orientations. Using explanatory factor analysis, Vermunt (1998)
identified four different learning styles, meaning-directed, reproduction-directed,
application-directed and undirected, which displayed characteristic patterns of factor
loadings across the four components of learning.
Aims. The aims of the current study were to test the generalisability of Vermunt’s
integrated model of learning with a sample of students from a British university and to
establish whether different learning styles were associated with different academic
outcomes.
Sample. A total of 273 students from a British university took part.
Methods. Confirmatory factor analysis was used to test Vermunt’s four-factor model
of learning styles and compare it with alternative models. Interrelationships between
components of the ILS were examined in more detail using regression analyses. The
relationship between learning style and academic outcome was also examined.
Results. Fit indices indicated that Vermunt’s four-factor model of learning styles
provided the best fit for the current sample. Path estimates associated with meaning-
directed and reproduction-directed learning styles loaded across components as
described by Vermunt but application-directed and undirected learning styles loaded
mainly on conceptions and orientations components. Undirected learning style had a
low negative association with academic performance, while the meaning-directed
learning style had a low positive association with academic performance.

* Requests for reprints should be addressed to Dr Elizabeth A. Boyle, School of Social Sciences, University of Paisley, High
Street, Paisley, PA1 2BE, UK (e-mail: liz.boyle@paisley.ac.uk).
268 Elizabeth A Boyle et al.
Conclusions. Although the ILS did identify Vermunt’s four learning styles, different
learning environments influence the precise characteristics of each learning style.

Over the past 20 years there have been important changes in our understanding of the
nature of effective learning. Traditional views characterise effective learning as
involving the passive transfer of objective knowledge from teacher to student (Biggs,
1996) with the teacher firmly in control of the student’s learning and assessment testing
the student’s retention of knowledge. However, more recent research in educational
pedagogy has challenged these traditional views, characterising effective learning as
active, deep (Entwistle & Ramsden, 1983), self-regulated (Boekaerts, 1997), situated
(Brown, Collins, & Duguid, 1989) and constructivist (Duffy & Jonassen, 1992).
Constructivism refers to a new philosophical approach to learning, which includes
many of these modern ideas about the nature of learning (Jonassen, 1991; Pressley,
Harris, & Marks, 1992).
Deep learning has been one of the most influential constructs to emerge in the
literature on effective learning in higher education. Marton and Säljö (1976) used the
distinction between deep and surface learning to describe qualitative differences in the
processes students engage in while carrying out a prose-reading task, while Entwistle
developed and clarified the distinction in terms of processing strategies and
motivational aspects of learning (Entwistle, 1997; Entwistle & Ramsden, 1983; Tait &
Entwistle, 1996). Deep learners actively try to understand meaning by working out
relationships between concepts, relating new material to previously known information
and adopting a critical attitude to information. Deep learners are intrinsically motivated
by intrinsic interest. Surface learners on the other hand focus on memory strategies
which emphasise the retention of knowledge and are extrinsically motivated by the
desire to get qualifications or a job. Several questionnaires, such as Entwistle’s
Approaches to Study Inventory (ASI) (Entwistle, Hanley, & Hounsell, 1979) and Biggs’
Study Processes Questionnaire (SPQ) (Biggs, 1987), have been developed which assess
differences between students with respect to these approaches to studying.
Another increasingly important construct in recent research on effective learning
concerns self-regulated (Boekaerts, 1997) or self-directed (Candy, 1991) learning.
Evidence is accumulating that self-regulation, that is the ability to plan, direct and select
relevant information processing activities, is as important as content knowledge in
learning effectively (Vermunt & van Rijswijk, 1988). Boekaerts (1997) found that
encouraging students to accept the primary responsibility for controlling their studying,
rather than relying on teachers to direct their learning, led to more effective learning,
while Minnaert and Janssen (1999) found that regulatory activities accounted for just as
much variance in academic performance as intelligence test scores did. Questionnaires
such as the Strategic Flexibility Questionnaire (Cantwell & Moore, 1996) and the
Leuven Executive Regulation Questionnaire (Minnaert & Janssen, 1997) specifically
address regulation of learning.
The way in which students understand the nature of learning has also been proposed
as an important determiner of effective learning. Using a phenomenographic
methodology, Säljö (1975) distinguished five different conceptions of learning that
students had, ranging from ‘increase in knowledge’ to ‘interpretative process aimed at
understanding the nature of reality’. These distinctions were confirmed by Marton,
Dall’Alba, and Beaty (1993) who included a further category ‘changing as a person’.
Entwistle, Thompson, and Tait (1992) argued that these different conceptions of
 Validity of Vermunt’s ILS with British students 269

learning can be roughly categorised as deep and surface. Lonka and Lindblom-Ylanne
(1996) examined constructivism, active epistemology and mental representations in
their study of conceptions of learning. They found that the constructivist view of
learning was a sophisticated view, which was only evident in advanced students with
detailed knowledge of the nature of learning.
Research on learning styles has developed alongside research on effective learning.
Learning styles research has addressed many different aspects of individual differences
in learning, ranging from environmental and emotional preferences in learning (Price,
Dunn, & Dunn, 1991) to differences in cognitive style (Sadler-Smith, 1996). There have
been a number of attempts to categorise learning styles, such as Curry’s onion model
(Curry, 1983) and to identify and focus on the most useful and important features of
learning style (Geisler-Brenstein, Schmeck, & Hetherington, 1996). Researchers are
increasingly recognising the complexity of learning and are interested in examining
how different aspects of learning work together. For example Van Rossum and Schenk
(1984) looked at the relationship between learning conceptions and learning strategies,
while VanZile-Tamsen and Livingston (1999) looked at links between regulation of
learning and motivation. Pintrich, Smith, Garcia, and McKeachie (1993) addressed
motivational, cognitive and metacognitive aspects of learning but did not examine links
between these separate components.
In a recent study Vermunt (1998) proposed an approach to studying learning styles,
grounded in modern constructivist views of learning, which explicitly attempted to
provide a more comprehensive and integrated account of learning by bringing together
four different aspects of learning, cognitive processing strategies, regulation strategies,
students’ views of learning and orientations to learning (similar to motives). Vermunt
developed the Inventory of Learning Styles (ILS), a questionnaire based on this
integrated model of learning. He confirmed the reliability and validity of the four ILS
components and, using factor analysis, identified four different learning styles, a
meaning-directed style, a reproduction-directed style, an application-directed style and
an undirected style. Each learning style loaded on subscales across the four ILS
components, showing ‘strong interrelations among these learning components’
(Vermunt, 1998). Vermunt also examined links between the ILS components in more
detail using multiple regression, arguing that this is a superior, but infrequently used,
method for examining directionality between different learning styles constructs.
Regression analyses supported his view that the regulation component plays a central
role in co-ordinating the other learning components, with clear links between a
constructivist conception of learning, self-regulation of learning and the adoption of
deep processing strategies and between an intake model of learning, external regulation
and surface processing strategies.
Vermunt’s integrated model of learning has been very influential within the higher
education community in Holland. The four distinct learning styles have been confirmed
in different student samples both in a traditional university environment (Busato, Prins,
Elshout, & Hamaker, 1998; Vermetten, Vermunt, & Lodewijks, 1999) and a distance
learning environment (Vermunt, 1998) in Holland, providing support for Vermunt’s
claim that his model of learning has a high degree of generalisability. However, the ILS
has not been tested across the different higher educational environments and contexts
found in different countries. The main aim of the current study therefore was to
examine the reliability and validity of the ILS within a British university setting. Vermunt
(1998) used exploratory factor analysis to identify the four distinct learning styles, but
confirmatory factor analysis allows for the testing of existing models (Thompson &
270 Elizabeth A Boyle et al.

Daniel, 1996) and is a superior method to use in this study to test the validity of
Vermunt’s four-factor model of learning styles. Confirmatory factor analysis has been
used infrequently in the literature on learning questionnaires, but a number of studies
are now emerging which suggest that it has an important role to play in testing the
models underlying these questionnaires (Kember & Leung, 1998).
There is an implicit assumption underlying many learning questionnaires that one
approach to learning is more desirable than others. Deep learning, self-regulated
learning, intrinsic motivation and a constructivist conception of learning are regarded as
preferable to surface learning, teacher-centred learning, extrinsic motivation and an
objectivist conception of learning. If these more constructivist characteristics do lead to
more effective learning, students’ scores on learning questionnaires should be good
predictors of their academic performance. However Busato et al. (1998) found no
statistically significant relationships between meaning-directed, reproduction-directed
or application-directed learning styles and academic success, although they did report a
moderate negative correlation between an undirected learning style and academic
success. Before advocating that students should adopt a meaning-directed style of
learning, it is probably wise to look for more evidence that this learning style is linked to
academic success. Consequently a further aim of the current study was to investigate
whether variations in learning style identified by Vermunt’s ILS are systematically
related to variation in academic outcome.
Different methods of assessment may promote or discourage different styles of
learning. For example, it has been suggested (Knight, 1995) that exams encourage
students to ‘mug up’ and use a reproduction-oriented style of learning, while
coursework assessments encourage students to develop a deeper understanding of
course material. Consequently a meaning-directed learning style may be a better
predictor of coursework marks than exam marks. In the current study the relationships
between learning style and academic performance were also examined as a function of
the nature of the assessment.

Method
Samples
Participants were 273 students from a department of Social Science in a British
university, comprising 107 second year students, 89 third year and 77 fourth year
students. 198 students were female and 75 were male. The mean age of the students
was 25 years and the response rate was 57%.

Inventory of Learning Styles

The instrument used was the 100-item version of Vermunt’s Inventory of Learning
Styles (ILS) (Vermunt, 1994). The inventory consists of two parts, Part A and Part B. Part
A, called Study Activities, included questions on two domains, processing strategies and
regulation strategies. Part B, called Study Motives and Views on Studying, is further
divided into B1, study motives, which addresses learning orientations and B2, study
views, which addresses mental models of learning. Each of the four components
includes five subscales containing between four and six items (see Table 1). Each item
comprises a statement for which the participant has to indicate on a 5-point scale to
 Validity of Vermunt’s ILS with British students 271
Table 1. Components, subscales (abbreviated names used in CFA shown in brackets), number of
items in each component and Cronbach’s a for subscales for Vermunt’s Inventory of Learning Styles
(ILS)

ILS Scale N Cronbach’s a

Part A: Study activities

Processing strategies .86

Relating and structuring (ssdeep1) 6 .74
Critical processing (ssdeep2) 4 .74
Memorising and rehearsing (ssstep1) 5 .74
Analysing (ssstep2) 5 .70
Concrete processing (concrete) 5 .77

Regulation strategies .78

Self-regulation of learning process and results (ssselfr1) 6 .72
Self-regulation of learning content (ssselfr2) 4 .72
External regulation of learning process (ssexter1) 5 .46
External regulation of learning results (ssexter2) 5 .61
Lack of regulation (lackreg) 5 .66

Part B: Study motives and study views

Mental models of knowledge .78

Construction of knowledge (construc) 5 .60
Intake of knowledge (intake) 5 .68
Use of knowledge (use) 5 .66
Stimulating education (stimed) 5 .64
Cooperative learning (cooper) 5 .83

Learning orientations .67

Personally interested (interest) 5 .54
Certificate-oriented (certific) 5 .49
Self-test oriented (selftest) 5 .70
Vocation oriented (vocation) 5 .73
Ambivalent (ambivale) 5 .77

what extent the statement applies to them. 1 means ‘disagree entirely’ and 5 means
‘agree entirely’ (for statements in Part A) or ‘I do this seldom or never’ to ‘I do this
almost always’ (for statements in Part B). Clear written instructions are presented at the
beginning of the inventory and at the start of each part of the inventory.

Academic performance
Each student in second, third and fourth year is required to take six core Social Science
modules including psychology, sociology and research methods. Academic perfor-
mance was measured in terms of grade point average (GPA), the student’s average mark
across these six core modules. Academic performance was also examined with respect
to method of assessment. An average coursework mark (GPACW) and an average exam
mark (GPAEXAM) were also calculated, based on the average coursework and average
272 Elizabeth A Boyle et al.

exam marks across the six modules. For most modules overall marks are weighted on
40% coursework and 60% exam performance.

Procedure
The ILS questionnaire was administered to second-, third- and fourth-year students
during scheduled lecture slots. Most students took between 20 and 25 minutes to
complete the inventory.

Design
There were four different aspects of the analysis in this study.

(1) Confirmatory factor analysis

Confirmatory factor analysis (CFA) was used to test Vermunt’s (1998) four-factor model
of learning styles. This model (shown in Figure 1) had four latent factors corresponding
to Vermunt’s meaning-directed, reproduction-directed, undirected and application-
directed learning styles. Variable indicators for the four factors were taken from the
factor loadings reported by Vermunt (1998) for the exploratory factor analysis for
regular university students. (To avoid duplication, Figure 1 also shows the path
estimates and error terms found in the current CFA.) As Vermunt described, learning
styles load across the components of the ILS. Each of the 20 ILS subscales loaded on at
least one factor, although some, such as intake of knowledge and certificate orientation,
loaded on more than one factor. Since Vermunt used oblique rotation, the CFA model
specified correlations between factors. Error terms in the processing and regulation
strategies components were adjusted to take account of the differing number of items
in these subscales. The SPSS version of AMOS v 4.0 (Arbuckle & Wothke, 1999) was
used to estimate the parameters and test the model.
MacCallum and Austin (2000) argue that CFA methodology is most useful when it
compares alternative models. Vermunt’s model of learning styles is rather complex,
making it difficult to suggest realistic alternative models, but there may be cultural
differences in the number of factors emerging as Wong, Wen-Ling, and Watkins (1996)
found using the Learning Process Questionnaire. In the simplest model tested in this
study all 20 ILS subscales were included as one factor.
Consideration of findings from the learning styles and approaches to learning
literature suggests a number of plausible alternative models. While learning styles
similar to the meaning-directed, reproduction-directed and undirected learning styles
have previously been reported in the literature, Vermunt (1998) acknowledged that
there have been few reports of a separate application-directed learning style. Vermunt
noted that the application-directed learning style emerges later than the other learning
styles and is most similar to deep learning. Certainly learning to apply knowledge
requires a thorough understanding of theory. It seems sensible to test whether a three-
factor model is a better fit to the current data. In this three-factor model variable
indicators associated with the application-directed learning style are attached primarily
to the meaning-directed learning style.
The ILS undirected learning style has been reported infrequently in the literature, but is
similar to the apathetic approach found in Entwistle’s revised Approaches to Studying
Inventory which is characterised by a lack of direction and a lack of interest (Tait &
Entwistle, 1996). However, this approach has emerged inconsistently as a distinct factor
 Validity of Vermunt’s ILS with British students 273

Figure 1. Vermunt’s four-factor subscale level model for the structure of the ILS

and Entwistle now proposes a surface apathetic approach where subscales from the
apathetic approach load alongside subscales from the surface approach (Entwistle,
McCune, & Walker, 2001). Similarly it is possible that subscales from the undirected
learning style may load alongside the reproduction-directed learning style from the ILS.
Consequently another three-factor model was tested in which variable indicators from the
undirected learning style and reproduction-directed learning style loaded on one factor.
Preliminary analysis of the current data had indicated that correlations between
meaning-directed and reproduction-directed learning styles and between application-
274 Elizabeth A Boyle et al.

directed and undirected learning styles were high. This suggested that a four-factor
model with two higher order factors may provide a better fit to the data. Consequently a
four-factor model with two higher order factors was also tested. In all five models were
tested:
Model 1: Vermunt’s original four-factor model.
Model 2: A three-factor model with variable indicators associated with application-
directed learning style merged with meaning-directed learning (concrete processing
strategies and interest orientation) and undirected learning style (vocation and
certificate orientation and use model of knowledge).
Model 3: A three-factor model with variable indicators from the undirected learning
style loading alongside reproduction directed learning style.
Model 4: A four-factor model with two higher order factors, one corresponding to
meaning-directed and reproduction-directed learning styles and one to application-
directed and undirected learning styles.
Model 5: A one factor model with all items loading on the same factor.

(2) Reliability and validity of ILS items and subscales

To further examine the reliability and item level validity of the four components of the
ILS for the current sample, internal consistency reliability estimates were calculated for
the ILS subscales. To examine whether the items load on the component subscales in
the way that Vermunt proposed, separate item-level confirmatory factor analyses were
carried out on each of the four ILS domains.

(3) Multiple regression analyses

To test Vermunt’s claims about the directional links between the different components,
with regulation of learning playing a central role in co-ordinating the other components
of learning, multiple regression analyses were carried out. Regression analysis looks at
links between subscales with no commitment to a model of learning styles. Two
separate regression models were considered. In the first, processing strategies were the
dependent variables with regulation strategies, mental models and orientations
subscales as predictor variables. In the second regression analysis, the regulation
subscales were the dependent variables with mental models and orientations subscales
as predictor variables.

(4) Learning style and academic success

Finally, links between learning style and academic success were examined using
correlation and multiple regression.

Results
Confirmatory factor analysis of Vermunt’s four-factor model of learning styles
Confirmatory factor analysis studies on learning questionnaires have reported a number
of different indices of fit. In deciding which fit indices to report in the current study, the
guidelines of Hoyle and Panter (1995) and MacCallum and Austin (2000) were
considered as well as current practice in published CFA studies of learning
questionnaires. Hoyle and Panter recommended the reporting of several fit indices
including an absolute fit index such as À2, the traditional fit index, an adjusted goodness
 Validity of Vermunt’s ILS with British students 275

of fit index such as AGFI, which takes into account the number of estimated model
parameters, and incremental fit indices such as TLI (NNFI) and CFI. MacCallum and
Austin strongly recommended the use of RMSEA as a measure of absolute fit because it
is sensitive to model misspecification and adjusts for sample size and degrees of
freedom. In a recent study of the psychometric properties of the Learning Process
Questionnaire (LPQ), Sachs and Gao (2000) reported the conventional À2 along with
degrees of freedom as a measure of absolute fit, as well as RMSEA, the TLI and CFI as
measures of incremental fit, and the AGFI. Conventionally values of .90 to 1.00 for the
TLI and CFI are considered indicative of a well-fitting model (McDonald & Marsh, 1990),
while values less than 0.08 for RMSEA indicate acceptable model fit (Sachs & Gao,
2000). In the current study we report À2 along with degrees of freedom, RMSEA, TLI
and CFI.
The fit results for the confirmatory factor analysis carried out on Vermunt’s (1998)
four-factor model are shown in Table 2. À2/df was fairly low (3.865) and the CFI (.975)
and TLI (.966) were well above the .9 regarded as indicating a well fit model, but the
RMSEA (0.102) did not quite reach the .08 recommended by MacCallum and Austin
(2000) as indicative of an acceptable fit.

Table 2. Summary of fit indices for ILS subscale-level factor models

Models À2 df À2/df Bentler’s TLI RMSEA

CFI

Model 1
Vermunt’s 610.692 158 3.865 .975 .966 .102
4-factor model

Model 2 707.639 164 4.315 .970 .961 .110

3-factor model
(meaning reproduction
undirected)

Model 3 816.150 163 5.007 .963 .953 .121

3-factor model
(meaning application
undirected)

Model 4 613.584 159 3.859 .975 .966 .103

4-factor model
with 2 higher-level
factors (inadmissable)

Model 5
One-factor model 960.256 170 5.649 .956 .945 .131

Path estimates also provide information about how well the model fits the data. Path
estimates and error terms for Vermunt’s model are shown in Figure 1. Path estimates for
the reproduction-directed and meaning-directed learning styles were similar to the
factor loadings reported by Vermunt, although the negative links from intake model of
learning (7.78) and certificate orientation (7.59) to the meaning-directed learning
276 Elizabeth A Boyle et al.

style were much lower in the current study and the path estimate for interest
orientation was low (.22).
Values of the path estimates for undirected and application-directed learning styles
were somewhat different from the factor loadings in Vermunt’s EFA. Lack of regulation
and ambivalent orientation, which are regarded as hallmarks of undirected learning, had
unexpectedly low path estimates in the current analysis, .31 and .12 respectively, while
intake and stimulating education models of learning had higher path estimates, .61 and
.60. With application-directed learning one subscale, use of knowledge, had an
extremely high path estimate (.97), with low estimates on the other subscales linked to
this learning style, vocation orientation (.25) and concrete processing (.25). There were
significant correlations between all four factors with the highest correlations between
meaning-directed and reproduction-directed (r=0.64) and between undirected and
application-directed learning styles (r=0.61). These high correlations suggested that the
higher order model (model 4) might provide a better fit for the current data. Indeed, the
fit indices for this higher order model were as good as or better than those for the four-
factor model. However, the higher order model was inadmissible due to negative
variance associated with the undirected learning style.
Despite the discrepancies between the current results and Vermunt’s study on the
values of path estimates for some subscales, the four-factor model did in fact provide a
better fit than all the other models tested. The next best fit was provided by the three-
factor model with factors corresponding to meaning-directed, reproduction-directed
and undirected learning styles. With this three-factor model, path estimates for the lack
of regulation and ambivalent learning subscales of the undirected learning style were
more like those found by Vermunt.

Reliability and validity of the subscales of the four separate components of the ILS
Although Vermunt’s four-factor model provided the best fit of the models tested in the
current study, there were some notable differences found in factor loadings between
the current study and Vermunt’s (1998) study, particularly on orientations and
conceptions subscales. Consequently the reliability and validity of the ILS items and
subscale components were examined in more detail for the current sample. Internal
consistency reliability estimates (see Table 1) were calculated for the four components
and for component subscales. Values of Cronbach’s alpha were respectable for
processing strategies (.86), regulation strategies (.78) and mental models of knowledge
(.78), but were lower for learning orientations (.67). In addition, values of Cronbach’s
alpha for domain subscales were generally reasonable either reaching or very nearly
reaching the .70 recommended by De Vaus (1995). Exceptions were the values for
external regulation of learning results (.61), construction of knowledge view of
knowledge (.60), external regulation of learning processes (.46), personally interested
(.54) and certificate oriented (.49) learning orientation subscales.
Separate item-level confirmatory factor analyses were carried out on each of the four
components of the ILS. Goodness of fit indices for these are shown in Table 3, while the
path estimates for components, similar to factor loadings in EFA, are shown in Figures 2
to 5.
The fit results for the confirmatory factor analysis on the processing strategy items
indicate a very good fit with a very low À2/df (1.723), CFI (.987) and TLI (.985) both
well above .9 and RMSEA (0.052) less than .08. Figure 2 shows that values of the path
estimates for all items were well above .30; mean values for each subscale ranged from
 Validity of Vermunt’s ILS with British students 277
Table 3. Summary of fit indices for ILS subscale-level factor models

Models À2 df À2/df Bentler’s TLI RMSEA

CFI

Processing strategies 456.588 265 1.723 .987 .985 0.052

Regulation strategies 528.300 265 1.994 .981 .977 0.06

Mental models of learning 503.250 265 1.899 .988 .986 0.057

Learning orientations 681.724 265 2.573 977 .972 0.076

.57 to .64. The two deep subscales were highly correlated (r=.80) and the concrete
subscale was highly correlated with both deep subscales, indicating that these items
were tapping similar constructs.
The fit results for the confirmatory factor analysis on the regulation strategy items
also indicate a very good fit with a very low À2/df (1.994), CFI (.981) and TLI (.977)
both well above .9 and RMSEA (0.06) less than .08. Path estimates (see Figure 3) were
all above .30 except for those associated with the external regulation of learning
processes. Path estimates for Q4 (.05) and Q18 (.23) were both very low, bringing
down the mean value for this subscale to .36. All other subscales seemed to fit well,
with mean values ranging from .50 to .62. There were high correlations between both
self-regulation subscales (r=.83) and between both external regulation subscales
(r = .75) and both self-regulation subscales correlated with external regulation of
learning results. Lack of regulation was uncorrelated with all other regulation subscales
except external regulation of learning processes.
The fit results for the confirmatory factor analysis on the mental models of learning
items indicated a very good fit with a very low À2/df (1.899), CFI (.988) and TLI (.986)
both well above .9 and a low RMSEA (0.057). Path estimates (see Figure 5) were all .30
or above, with mean values ranging from .50 to .69. Construction, use and stimulating
education subscales had high intercorrelations.
Confirmatory factor analysis on orientations items indicated that fit for these items
was also generally good with low À2/df (2.573), CFI (.977) and TLI (.972) both well
above .9 and RMSEA (0.076) less than .08. Path estimates (see Figure 4) for ambivalent,
selftest and vocation subscales were above .30, with mean values .64, .60 and .58
respectively. Values of path estimates for interest and certificate subscales were highly
variable with values ranging from 0 to .82 for certificate, and .22 to 72 for interest
subscales. Item 55, ‘I aim at achieving high levels of study achievements’, did not fit
well with the other items on the certificate subscales such as ‘The main goal I pursue in
my studies is to pass exams’. The highest correlations between subscales were between
interest and self-test (r=.45) and certificate and vocation (r=.41) with a negative
correlation between interest and ambivalent (r=7.45).

Regression analyses of the interrelationships between learning components

The regression analyses offered a different approach to examining the relationship
between the different components, which made no assumptions about learning styles.
The results of the regression analyses with processing strategies as dependent variables
are summarised in Table 4. Self-regulation of learning processes was related to all five
278 Elizabeth A Boyle et al.

Figure 2. Five-factor model for item-level CFA of processing strategies component of the ILS
showing standardised parameter estimates

processing strategies but most strongly to the use of the two deep strategies, while self-
regulation of learning content was related only to the use of the two deep processing
strategies as well as concrete processing. External regulation of learning processes was
strongly positively related to the surface-processing strategy of memorising and
rehearsal with a weaker relationship to analysing, and low negative associations to
critical and concrete processing. External regulation of learning results was most
strongly related to analysing, but was also related to the use of both deep processing
 Validity of Vermunt’s ILS with British students 279

Figure 3. Five-factor model for item-level CFA of regulation strategies component of the ILS, showing
standardised parameter estimates

strategies and concrete processing. Lack of regulation was not related to any processing
strategies.
With respect to mental models of learning, a constructivist view of knowledge was
related only to analysing, while an intake model of knowledge was negatively related to
the use of relating and structuring activities. Stimulating education was related to both
relating and structuring and memorising and rehearsal, while a use mental model was
strongly related to concrete processing strategy and weakly negatively related to
280 Elizabeth A Boyle et al.

Figure 4. Five-factor model for item-level CFA of the learning orientations component of the ILS
showing standardised parameter estimates

analysing. With respect to learning orientations, personal interest was related to

concrete processing, vocation orientation was negatively related to both critical and
concrete processing, while an ambivalent orientation was negatively related to concrete
processing and analysing.
Five further multiple regression analyses were carried out with regulation strategies
 Validity of Vermunt’s ILS with British students 281

Figure 5. Five-factor model for item-level CFA of mental models of learning component of the ILS
showing standardised parameter estimates

as the dependent variables and mental models of learning and learning orientation as
independent variables. Table 5 presents a summary of the results of these analyses.
Mental models of learning emphasising construction of knowledge were strongly
associated with both types of self-regulation. The intake of knowledge model of
learning was linked to all processing strategies, having a positive association with both
types of external regulation and lack of regulation but a negative association with both
types of self-regulation. Stimulating education was positively linked to both self-
282 Elizabeth A Boyle et al.
Table 4. Beta weights of regulation strategies, mental models of learning and learning orientations as
predictors of processing strategies based on the stepwise regression model, and significance levels of
the F-values

Processing Relating & Critical Memorising Analysing Concrete

strategies structuring processing & rehearsal

Regulation strategies
Self-regulation of
learning processes .38 .51 .18 .26 .31
learning content .12 .20 .16
External regulation
learning processes 7.17 .43 .19 7.19
learning results .23 .23 .35 .17
Lack of regulation

Mental models of learning

Construction of knowledge .17
Intake of knowledge 7.21
Use of knowledge 7.15 .31
Stimulating education .19 .15
Co-operative learning

Learning orientations
Personally interested .18
Certificate-oriented
Self-test-oriented
Vocation-oriented 7.14 7.18
Ambivalent 7.15 7.12

R2 0.49 0.56 0.29 0.44 0.50

F 40.3 56.9 30.5 30.5 32.5

regulation subscales and the external regulation of learning results. An ambivalent

learning orientation was a strong positive predictor of lack of regulation, and was
negatively associated with both external regulation strategies and self-regulation of
learning content.

Relationship between learning style and academic performance

To examine the relationship between learning style and academic performance a factor
score for each subject for each of Vermunt’s four learning styles was calculated by
weighting relevant subscale scores according to the factor loading for that subscale and
summing all the weighted subscale scores. Moderate negative correlations were found
between undirected learning style and grade point average (GPA) (r=7.26, p<.01),
undirected learning style and exam grade point average (GPAEXAM) (r=7.17, p<.01)
and undirected learning style and coursework grade point average (GPACW) (r=7.21,
p<.01). In addition statistically significant positive correlations were found between
meaning-directed learning style and GPA (r=.23, p<.01) and also for meaning-directed
learning style and exam performance GPAEXAM (r=.13, p<.05) and GPACW (r=.22,
 Validity of Vermunt’s ILS with British students 283
Table 5. Beta weights of mental models of learning and learning orientations as predictors of
regulation strategies based on the stepwise regression model, and significance levels of the F-values

Regulation Self- Self- External External Lack of

Strategies regulation regulation regulation regulation regulation
of learning of learning of learning of learning
processes content processes results
and results

Mental models
Construction .33 .25
of knowledge
Intake of knowledge 7.23 716 .32 .16 .14
Use of knowledge .13 .19
Stimulating education .16 .21 .17
Co-operative learning 7.13

Learning orientations
Personally interested
Certificate-oriented .19 .14
Self-test-oriented .15
Vocation-oriented
Ambivalent 7.19 7.12 7.26 .46

R2 0.27 0.26 0.16 0.16 0.30

F 24.4 18.5 16.3 11.0 34.1

p<.01). Neither reproduction-directed learning style nor application-directed learning

style was correlated with any performance variables.
To examine the impact of specific subscales on academic outcome, correlations
between mean ILS subscale scores and GPA, GPAEXAM and GPACW were also
calculated and are shown in Table 6. Significant positive correlations were found with
GPA for both the deep and analysing processing subscales, and both self-regulation
subscales. In addition significant negative correlations were found between GPA and
intake and cooperative models of learning and ambivalent and vocation learning
orientations. Correlations between exam performance and learning style subscales
reflected those found for overall performance but the pattern found for coursework
was slightly different. Positive correlations with coursework performance were found
for self-regulation of learning content and personal interest, while negative correlations
with coursework performance were found for ambivalent and vocation learning
orientations and for cooperative learning and intake models of learning.
A regression analysis with GPA as the dependent variable and the subscale scores as
independent variables indicated that intake of knowledge was the best negative
predictor of performance but predicted only 9% of the variance in scores. Ambivalent
learning orientation was also a negative predictor accounting for 4% of the variance
while analysing was the only positive predictor accounting for 1.5% of the variance.
Intake of knowledge and analysing were the only predictors of exam performance
while ambivalence, vocation directed, cooperation and self-regulation of learning
content subscale scores predicted performance on coursework.
284 Elizabeth A Boyle et al.
Table 6. Correlations between ILS subscale scores and Grade Point Average (GPA), GPAEXAM and
GPACW

ILS scales GPA GPAEXAM GPACW

Processing strategies
Deep processing .16* .16* .09
Relating and structuring .14* .13* .08
Critical Processing .14* .15* .08
Stepwise processing .15* .15* .11
Memorising and rehearsing .08 .06 .05
Analysing .17** .18** .12
Concrete processing .09 .05 .08

Regulation strategies
Self-regulation .16** .16** .13*
Learning process and results .12* .14* .08
Learning content .18** .15* .18**
External regulation .05 .08 .05
Learning process .01 .03 7.01
Learning results .10 .10 .11
Lack of regulation 7.11 7.06 7.12

Mental models of knowledge

Construction of knowledge .03 .04 .05
Intake of knowledge 7.23** 7.18** 7.12*
Use of knowledge 7.02 7.03 7.03
Stimulating education 7.03 7.02 7.01
Co-operative learning 7.20** 7.15* 7.21**

Learning orientations
Personally interested .09 .05 .12*
Certificate-oriented 7.02 7.04 .00
Self-test-oriented 7.04 7.05 7.01
Vocation-oriented 7.12* 7.11 7.14*
Ambivalent 7.28** 7.19** -.25**

*p<0.05 **p<0.01

Discussion and conclusions

The results of the current study of students in a British university provide some
evidence that Vermunt’s constructivist, integrated model of learning can be generalised
across the different learning environments and contexts found in higher education in
different countries and indicate that the ILS is a useful instrument for identifying distinct
learning styles. Reliability estimates for the four ILS domains and subscales were
reasonable and confirmatory factor analysis indicated that Vermunt’s (1998) four-factor
model, reflecting four distinct learning styles, provided the best fit overall compared
with the other models tested.
For the meaning-directed and reproduction-directed learning styles associated with
the four-factor model, path estimates generally matched quite closely the factor
loadings reported by Vermunt (1998) for regular university students. These learning
 Validity of Vermunt’s ILS with British students 285

styles are clearly similar to the deep and surface approaches to learning widely reported
in the literature (Entwistle et al., 1979; Biggs, Kember, & Leung, 2001). However
Vermunt’s model of learning provides a fuller characterisation of these learning styles,
identifying metacognitive aspects of learning in addition to strategies and motivation
components as important features of learning styles.
Factor three corresponded to the undirected learning style, but this factor showed
less integration across components than the first two learning styles. Apart from a link
to lack of regulation, the undirected learning style was exclusively linked to learning
conceptions (intake of knowledge, stimulating education and cooperation) and learning
orientations (certificate, self-test and ambivalent) subscales. Furthermore, items in the
lack of regulation subscale are unlike items in the other regulation subscales in that they
refer to the absence of the use of systematic and effective regulation strategies. The
undirected learning style is a negative learning style characterised by naive views about
and orientations to learning which are not linked to systematic strategy use. The
superior fit of the four-factor model over model three (the three-factor model) indicates
that undirected learning remains a distinct style, in contrast with Entwistle’s merged
surface apathetic approach.
The application-directed learning style, like the undirected learning style, was
specified largely by orientations and views of learning, not integrated across
components in the way that Vermunt described. While concrete processing strategies
were strongly linked to the application-directed learning style in Vermunt’s study, in the
current study concrete processing strategies had stronger links to the meaning-directed
learning style. These links were also indicated by the high correlations found between
concrete and deep processing strategies in the CFA on processing strategies. The
students in the current study were all from a social sciences department where there is
a strong emphasis on the application of theoretical knowledge. It appears that this
emphasis changes the focus of meaning-directed learning style, and leads to a
dissociation of the strategies components from the views and orientations components
of the application-directed learning style. Application-directed learning had closer links
to the undirected learning style (r=.62) than meaning-directed learning style (r=.27),
while meaning-directed and reproduction-directed learning styles are also highly
correlated (r=.61). This suggests that an important distinction between learning styles
in the current study concerns the extent of integration across components.
Links between conceptions and orientations subscales and learning styles varied
across the two studies, although some consistencies did emerge. The constructivist
model of learning was only associated with the meaning-directed learning style, while
the use conception was exclusively and strongly linked to the application-directed
learning style. A number of the ILS orientations and conceptions subscales, such as
stimulating education conception, do not play an important role in distinguishing
different learning styles or, like input and certificate conception, are linked to several
learning styles. In addition the reliabilities of the personal interest and certificate
orientation subscales are low. Taken together these results suggest that it may be
beneficial to simplify some of the ILS subscales.
The regression analyses showed that, apart from the link between ambivalent
learning orientation and lack of regulation, there were few strong links between
orientations and strategies. This finding is interesting as it is inconsistent with the
research on approaches to learning where different approaches to learning are
characterised in terms of related strategies and motives components (Biggs, 1987;
Entwistle & Tait, 1995). It seems likely that the inclusion of the regulation component
286 Elizabeth A Boyle et al.

in the ILS reduces or changes the impact of the orientations component compared with
the motivational subscales in approaches to learning. Certainly the main distinctions
within the ILS regulation component between lack of regulation, internal regulation
and external regulation have parallels in the traditional literature on motivation.
Recently, in research using the LPQ, Kember, Wong, and Leung (1999) reported that
links between motivation and strategy use are more diverse and variable than previous
approaches to learning studies have suggested. Comparing responses on the ILS and
LPQ might help to examine the relationships between motivation, regulation and
strategy use in more detail.
The regression analysis also suggested that links between components were more
flexible than Vermunt found, underlining Vermetten, Lodewijks, and Vermunt’s (1999)
claim that we need to consider both variability and consistency when looking at
learning styles. For example, where Vermunt found no links between external
regulation and deep processing, in the current study external regulation of learning
results was related to the selection of both the deep and concrete processing strategies.
This suggests that our students have more of a ‘partnership’ (Lonka & Ahola, 1995) or
shared regulation (Vermunt & Verloop, 1999) view of the relative responsibilities of
teachers and students than their Dutch counterparts, perceiving assignments and
exercises provided by teachers as more effective than Vermunt (1998) claimed in
guiding the selection of deep processing activities. The strong correlation (r=.82)
between the constructivist (student-centred) and stimulating education (teacher-
focused) subscales in the CFA of the mental models of learning component also
supported this view. This difference between British and Dutch university students may
arise from the variety of both student-centred and teacher-centred teaching methods
that students in the current study are exposed to, or may be attributable to the explicit
emphasis that the Dutch educational system places on self-regulated learning (Vermunt,
1995).
The current study adds to accumulating evidence of the importance of systematic
regulation strategies in learning (Minnaert & Janssen, 1999). The different types of
regulation, internal regulation, external regulation and lack of regulation, played an
important role in characterising meaning-directed, reproduction-directed and undir-
ected learning styles, providing support for Vermunt’s claim that regulation in learning
plays a mediating role in learning. Typically students adopting an undirected learning
style fail to regulate their learning and consequently fail to adopt systematic processing
strategies. However, both the current study and Minnaert and Janssen (1997) report
problems in the reliability and validity of external regulation of learning subscales.
These results add weight to Boekaert’s (1997) claim that regulation and its links to other
domains deserve further investigation.
Different learning styles, with their differing types of regulation, also had a
differential impact on academic outcome. In accordance with Busato et al. (1998), the
undirected learning style had a negative impact on academic outcome, showing that
learners who fail to adopt systematic learning strategies put themselves at a
disadvantage academically. In contrast with Busato and his colleagues, the meaning-
directed style of learning was rewarded with an increased likelihood, albeit a small one,
of academic success. These results show a clear order of desirability in learning styles,
from undirected as the least desirable to meaning-directed as the most desirable.
However, correlations between subscale scores and academic performance
suggested more specific links to academic success. Scores on the intake and
cooperative conceptions of knowledge and ambivalent and vocation learning
 Validity of Vermunt’s ILS with British students 287

orientations had negative links to academic success. These results suggest that the main
reason for poor academic performance is inappropriate motives and views of learning.
Students with undirected learning styles fail to adopt any consistent strategies when
studying. The ‘added value’ of the meaning-directed learning style was attributable to
the use of self-regulation and deep processing strategies. Although ‘analyzing’ is
categorised in the ILS as a surface-processing strategy, it was also linked to academic
success. The small improvement in academic performance associated with meaning-
directed learning style was related to exam performance rather than coursework,
indicating that examinations may be better methods of assessing deep learning than is
commonly supposed (Knight, 1995).
Factor scores accounted for a relatively small amount of variability in academic
performance, suggesting that the ILS has a rather limited role to play in predicting
academic outcome. However, the ILS may have a more useful diagnostic role to play in
higher education in detecting at an early stage learners with inappropriate views of and
orientations to learning, who fail to adopt systematic processing strategies and
consequently are in danger of failing to achieve the maximum benefit from their time in
higher education. The ILS may be more useful in this respect than similar instruments,
such as Entwistle’s ASSIST (Tait & Entwistle, 1996), since it provides a more explicit
characterisation of undirected learners. But we should bear in mind Lonka’s caution
that students’ views on the nature of learning can be extremely resistant to change
(Lonka, Joram, & Bryson, 1996). Given the low reliabilities of some ILS regulation and
orientation subscales it would be desirable to clarify and simplify the ILS if it is to be
used as a diagnostic tool. Regardless of the benefits of the ILS, Vermunt’s research has
played an important role in helping to develop a more comprehensive and more clearly
articulated model of effective learning.
The current study provides some evidence that Vermunt’s integrated model of
learning has some generality across the differing learning environments found in higher
education in different countries. There was broad support for the four distinct learning
styles identified by the ILS, although only meaning-directed and reproduction-directed
learning styles were integrated across the four components in the way Vermunt
suggested. The relative impact of specific subscales, particularly from the orientations
and conceptions components, is more variable than Vermunt proposed. However,
Vermunt’s view that regulation plays a central role in determining learning styles was
supported and there was evidence that different learning styles had a differential impact
on academic success. Concerns about the reliability of some of the ILS subscales
indicate that further consideration should be given to clarifying and probably
simplifying the constructs underlying the ILS model, focusing on the subscales which
make the most useful distinctions. The links between components and how these may
change in different educational environments also demand further investigation
(VanZile-Tamsen & Livingston, 1999). To extend our understanding of effective
learning and learning styles, it will be useful to compare different styles and approaches
instruments and their underlying models of learning.

References
Arbuckle , J. L., & Wothke, W. (1999). Amos 4.0 user’s guide. Chicago, IL: SmallWaters
Corporation.
288 Elizabeth A Boyle et al.
Biggs, J. B. (1987). The Learning Process Questionnaire (LPQ): Manual. Hawthorn, Vic.:
Australian Council for Educational Research.
Biggs, J. B. (1993). What do inventories of students’ learning processes really measure? A
theoretical review and clarification. Educational Psychology, 63, 3–19.
Biggs, J. B. (1996). Enhancing teaching through constructive alignment. Higher Education, 32,
347–364.
Biggs, J., Kember, D., & Leung, D. Y. P. (2001). The revised two-factor Study Process
Questionnaire. British Journal of Educational Psychology, 71, 133–149.
Boekaerts, M. (1997). Self-regulated learning: A new concept embraced by researchers, policy
makers, educators, teachers, and students. Learning and Instruction, 7(2), 161–186.
Brown, J. S., Collins, A., & Duguid, P. (1989) Situated cognition and the culture of learning.
Educational Researcher, 18(1), 32–43.
Busato, V. V., Prins, F. J., Elshout, J. J., & Hamaker, C. (1998). Learning styles: A cross-sectional
and longitudinal study in higher education. British Journal of Educational Psychology, 68,
427–441.
Candy, P. C. (1991). Self-direction for lifelong learning. San Franciso, CA: Jossey-Bass.
Cantwell, R. H., & Moore, P. J. (1996). The development of measures of individual differences in
self-regulatory control and their relationship to academic performance. Contemporary
Educational Psychology, 21, 500–517.
Curry, L. (1983). An organisation of learning style theory and constructs. In L. Curry (Ed.),
Learning styles in continuing education (pp. 115–131). Halifax.
De Vaus, D. A. (1995). Surveys in social research (4th ed). London: UCL.
Duffy, T. M., & Jonassen, D. H. (1992). Constructivism: New implications for instructional
technology. In T. M. Duffy & D. H. Jonassen (Eds.), Constructivism and the technology of
instruction – A conversation (pp.1–16). Hillsdale, NJ: Erlbaum.
Entwistle, N. J. (1997). Reconstituting approaches to learning: A response to Webb. Higher
Education, 33, 213–218.
Entwistle, N. J., & Ramsden, P. (1983). Understanding student learning. London: Croom Helm.
Entwistle, N. J., & Tait, H. (1995) The Revised Approaches to Study Inventory. University of
Edinburgh: Centre for Research on Learning and Instruction.
Entwistle, N. J., Hanley, M., & Hounsell, D. (1979) Identifying distinctive approaches to studying.
Higher Education, 8, 365–380.
Entwistle, N., McCune, V., & Walker, P. (2001). Conceptions, styles and approaches within higher
education: Analytic abstractions and everyday experience. In R. J. Sternberg & L. Zhang (Eds.),
Perspectives on thinking, learning and cognitive styles. London: Lawrence Erlbaum
Associates.
Entwistle, N. J., Thompson, S., & Tait, H. (1992) Guidelines for promoting effective learning in
higher education. University of Edinburgh: Centre for Teaching and Learning.
Geisler-Brenstein, E., Schmeck, R. R., & Hetherington, J. (1996). An individual difference
perspective on student diversity. Higher Education, 31, 73–96.
Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation models. In R. Hoyle (Ed.),
Structural equation modelling. Thousand Oaks, CA: Sage.
Jonassen, D. H. (1991). Objectivism versus constructivism: Do we need a new philosophical
paradigm? Educational Technology Research and Development, 40 (3), 5–14.
Kember, D., & Leung, D. Y. P. (1998). The dimensionality of approaches to learning: An
investigation with confirmatory factor analysis on the structure of the SPQ and LPQ. British
Journal of Educational Psychology, 68, 395–407.
Kember, D., Wong, A., & Leung, D. Y. P. (1999). Reconsidering the dimensions of approaches to
learning. British Journal of Educational Psychology, 69, 323–343.
Knight, P. (1995). Assessment for learning in higher education. London: Kogan Page.
Lonka, K., & Ahola, K. (1995). Activating instruction: How to foster study and thinking skills in
higher education. European Journal of Psychology of Education, 10 (4), 351–368.
Lonka, K., & Lindblom-Ylanne, S. (1996). Epistemologies, conceptions of learning and study
 Validity of Vermunt’s ILS with British students 289
practices in medicine and psychology. Higher Education, 31, 5–24,
Lonka, K., Joram, E., & Bryson, M. (1996). Conceptions of learning and knowledge: Does training
make a difference? Contemporary Educational Psychology, 21, 240–260.
MacCallum, R. C., & Austin, J. T. (2000). Applications of structural equation modelling in
psychological research. Annual Review of Psychology, 51, 201–226.
Marton, F., & Säljö, R. (1976). On qualitative differences in learning: 1: Outcome and process.
British Journal of Educational Psychology, 46, 4–11.
Marton, F., Dall’Alba, G., & Beaty, E. (1993). Conceptions of learning. International Journal of
Educational Research, 19, 277–300.
McDonald, R. P., & Marsh, H. W. (1990). Choosing a multivariate model: Noncentrality and
goodness of fit. Psychological Bulletin, 107 (2), 247–255.
Minnaert, A., & Janssen, P. J. (1997). Bias in the assessment of regulation activities in studying at
the level of higher education. European Journal of Psychological Assessment, 13 (2), 99–108.
Minnaert, A., & Janssen, P. J. (1999). The additive effect of regulatory activities on top of
intelligence in relation to academic performance in higher education. Learning and
Instruction, 9, 77–91.
Pintrich, P. R., Smith, D. A. F., Garcia, T. & McKeachie, W. J. (1993). Reliability and predictive
validity of the Motivated Strategies for Learning Questionnaire (MSLQ). Educational and
Psychological Measurement, 53, 801–813. .
Pressley, M., Harris, K. R., & Marks, M. B. (1992). But good strategy instructors are constructivists!
Educational Psychology Review, 4 (1), 3–31.
Price, G., Dunn, R., & Dunn, K. (1991). Productivity environmental preference survey (PEPS
manual). Lawrence, KS: Price Systems.
Sachs, J., & Gao, L. (2000). Item-level and subscale-level factoring of Biggs’ Learning Process
Questionnaire (LPQ) in a mainland Chinese sample. British Journal of Educational
Psychology, 70, 405–418.
Sadler-Smith, E. (1996). ‘Learning styles’ and instructional design. Innovations in Education and
Training International, 33 (4), 285–293.
Säljö, R. (1975). Qualitative differences in learning as a function of the learners’ conception of
the task. Goteburg: Acta Universitatis Gothoburgensis.
Tait, H., & Entwistle, N. (1996). Identifying students at risk through ineffective study strategies.
Higher Education, 31, 97–116.
Thompson, B., & Daniel, L. G. (1996). Factor analytic evidence for the construct validity of scores:
A historical overview and some guidelines. Educational and Psychological Measurement, 56,
197–208.
Van Rossum, E. J., & Schenk, S. M. (1984). The relationship between learning conception, study
strategy and learning outcome. British Journal of Educational Psychology, 54, 73–83.
VanZile-Tamsen, C., & Livingston, J. A. (1999). The differential impact of motivation on the self-
regulated strategy use of high- and low-achieving college students. Journal of College Student
Development, 40 (1), 54–60.
Vermetten, Y. J., Lodewijks, H. G., & Vermunt, J. D. (1999). Consistency and variability of learning
strategies in different university courses. Higher Education, 37, 1–21.
Vermetten, Y. J., Vermunt, J. D., & Lodewijks, H. G. (1999). A longitudinal perspective on
learning strategies in higher education: Different viewpoints towards development. British
Journal of Educational Psychology, 69, 221–242.
Vermunt, J. D. (1994). Inventory of Learning Styles in higher education: Scoring key. Tilburg
University, Department of Educational Psychology.
Vermunt, J. D. (1995). Process-oriented instruction in learning and thinking strategies. European
Journal of Psychology of Education, 10 (4), 325–349.
Vermunt, J. D. (1998). The regulation of constructive learning processes. British Journal of
Educational Psychology, 68, 149–171.
Vermunt, J. D., & van Rijswijk, F. A.W. M. (1988). Analysis and development of students’ skill in
self-regulated learning. Higher Education, 17, 647–682.
290 Elizabeth A Boyle et al.
Vermunt, J. D., & Verloop, N. (1999). Congruence and friction in teaching and learning. Learning
and Instruction, 9, 257–280.
Wong, N-Y, Lin, W-Y, & Watkins, D. (1996). Cross-cultural validation of models of approaches to
learning: An application of confirmatory factor analysis. Educational Psychology, 16, 317–
327.

Received 30 June, 2000; revised version received 12 December, 2001