See discussions, stats, and author profiles for this publication at: [Link]

net/publication/223109496

The construct validity of one learning styles instrument

Article in System · September 2001

DOI: 10.1016/S0346-251X(01)00027-6

CITATIONS READS

70 2,423

3 authors, including:

Ann C Wintergerst Andrea DeCapua

St. John's University Educational Consultant
10 PUBLICATIONS 395 CITATIONS 74 PUBLICATIONS 1,154 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Andrea DeCapua on 13 February 2022.

The user has requested enhancement of the downloaded file.

 System 29 (2001) 385–403
[Link]/locate/system

The construct validity of one learning styles

instrument
Ann C. Wintergersta,*, Andrea DeCapuab, Richard C. Itzenc
a
Department of Modern Foreign Languages and Classical Studies, St. John’s University, St. John Hall 434H,
8000 Utopia Parkway, Jamaica, NY 11439, USA
b
102 Phyllis Drive Old Tappan, NJ 07675, USA
c
Assessment and Testing, The School District of Lee County, 2055 Central Avenue, Ft Myers, FL 33901, USA

Received 30 October 2000; received in revised form 11 December 2000; accepted 25 February 2001

Abstract
Learning style research has become ubiquitous in second/foreign language teaching and
research. While such an approach has a great deal of intuitive appeal, difficulties arise
when researchers attempt to conceptualize actual learning styles and relate these to factors
other than individual preferences. These difficulties are further compounded when working
with second/foreign language learners and the inherent issues of learner proficiency and target
language comprehensibility. The present study examines the reliability and validity of Reid’s
(1984) Perceptual Learning Style Preference Questionnaire (PLSPQ), an instrument used to
assess the learning style of English as a Second Language (ESL) students. Exploratory factor
analysis was used to explore the dimensionality of the PLSPQ. Results showed that specific
survey items did not necessarily group into factors conceptually compatible with Reid’s
learning style model. Subsequently, an alternative learning style factor structure was explored.
# 2001 Published by Elsevier Science Ltd. All rights reserved.
Keywords: English as a Second Language; Learning styles; Learning preferences; Survey instruments;
Perceptual Learning Styles Preference Questionnaire (PLSPQ); Construct validity; Exploratory factor
analysis

* Corresponding author. Tel.: +1-718-990-5208; fax: +1-718-456-3532.

E-mail addresses: winterga@[Link] (A.C. Wintergerst); adecapua@[Link] (A. DeCapua);
richardi@lee.k12.fl.us (R.C. Itzen).

0346-251X/01/$ - see front matter # 2001 Published by Elsevier Science Ltd. All rights reserved.
PII: S0346-251X(01)00027-6
386 A.C. Wintergerst et al. / System 29 (2001) 385–403

1. Introduction

It is generally agreed that learning styles are the preferences of individuals with
respect to how they learn. Learning styles are often confused with learning strategies.
Learning strategies are specific procedures or techniques a person uses in solving a
problem or in working through an assignment. Learning strategies, unlike learning
styles, can be learned and consciously applied in different learning situations; they
are part of the process of learning. Learning styles, in contrast, are innate pref-
erences of individuals as to how they prefer to go about the process of learning.
While some individuals may prefer to hear information (auditory learners), others
may prefer to read it (visual learners). Though everyone seems to be able to define
learning styles per se, there are several issues when relating learning styles to factors
other than individual preferences in empirical research.
Learning style research has come out of the more general research on cognitive
styles, particularly from the field of psychology. The principle application of style to
language learning has been through the concept of field dependence/independence,
developed in mainstream psychology by Herman Witkin (1962). Field dependence/
independence, measured primarily through non-language-based tests, is based on the
belief that a contrast can be made between analytic and holistic individuals. Analytic
people are more inclined to focus on the parts of a whole while holistic people are
more prone to consider the whole or broader picture. Another approach to general
learning styles has been suggested by Kolb (1976) who characterizes learning style in
dynamic and static terms. According to Kolb, the natural sequence for learning is a
cycle containing four stages: concrete experience, reflection–observation, abstract
conceptualization, and active experimentation.
In applying field dependence/independence and general learning styles to language
learning styles, Chapelle and Green (1992) suggest that the analytic style could evi-
dence itself by language learners’ ability to analyze the linguistic material they are
exposed to, distinguish the components, and then, perhaps, investigate relationships
between these components. Holistic learners, on the other hand, are more likely to
perceive situations as wholes; therefore, such learners are more likely to profit from
communicative interactions that maximize chances for receiving high quality and
relevant input, and that allow these learners to engage in opportunities to use lang-
uage to express meanings. Brown (1994) points out that learners tend to have a
range of styles, so that the field dependent/independent is too rigid a dichotomy.
Different situations require people to respond differently, depending on whatever
seems appropriate to the task at hand.
Willing (1987) attempts to relate Kolb’s general learning styles to language learning.
Based on Willing’s empirical study of English as a Second Language (ESL) learners, he
proposes a two-dimensional framework encompassing four language learner types
determined by some combination of the two dimensions. These four types correspond
to the types of general learners proposed by Kolb. (See Ellis, 1994; Skehan, 1998 for
more detailed and complete discussions of cognitive styles, and language learning.)
Research has also suggested a link between learning styles and language back-
ground. Lee (1976), for instance, found that Chinese and Vietnamese speakers
 A.C. Wintergerst et al. / System 29 (2001) 385–403 387

preferred visual learning while Reid (1987) and Stebbins (1995) found that Spanish
speakers preferred kinesthetic and tactile learning styles. Yet Willing’s (1987) study
found that the Spanish speakers preferred visual and auditory styles. The relationship
between language background and learning style preference remains an open question.
Attention to learning styles has become common in language teaching and
research. Many educators are embracing learning styles as a way of addressing stu-
dents’ learning needs. While such an approach has a great deal of intuitive appeal,
difficulties arise when researchers attempt to conceptualize actual learning styles and
relate these to factors other than individual preferences.
Based on our previous research on learning styles and Russian-speaking ESL
students using Reid’s (1984) Perceptual Learning Style Preference Questionnaire
(PLSPQ), a popular assessment instrument, we found that specific survey question
items did not necessarily group into factors conceptually compatible with that model
(see Wintergerst and DeCapua, 1998a,b for specific examples). In this paper we
discuss the difficulties of conceptualizing learning style modalities and developing
assessment instruments that accurately measure those modalities.

2. Review of literature

2.1. Learning styles survey instruments

A popular learning styles survey instrument is the Dunn and Dunn learning styles
model (Dunn et al., 1975, 1989; Dunn, 1990). The Learning Style Inventory (LSI;
Dunn et al., 1989) focuses on the instructional and environmental preferences of
students and allows analysis of the conditions under which students in grades 3–12
prefer to learn. It includes 23 elements of instructional environments that learners
are affected by: their immediate environment (sound, light, temperature, seating
designs); their own emotionality (motivation, persistence, responsibility, structure);
sociological preferences (learning alone, in a pair, in a small group, as part of a
team, with an authoritative adult, and variety or routines); physiological char-
acteristics (perceptual strengths, time-of-day energy levels, and need for intake or
mobility while learning); and processing inclinations (global/analytic, right/left
hemisphere, and impulsive/reflective).
Among the most popular of the learning style assessment instruments that are
normed in the ESL/English as a Foreign Language (EFL) field are Reid’s (1984)
‘‘Perceptual Learning Style Preference Questionnaire’’, O’Brien’s (1990) ‘‘The
Learning Channel Preference Checklist’’ and Oxford’s (1993) ‘‘Style Analysis Sur-
vey’’. The Learning Channel Preference Checklist (O’Brien, 1990) examines three
sensory learning style preferences: visual, auditory, and haptic which includes
kinesthetic and tactile learning style preferences. There are a total of 36 statements
which students rate on a five-point scale. The Style Analysis Survey (Oxford, 1993)
assesses an individual’s general approach to learning and working. The style survey
is grouped into five activity types: how I use my physical senses to study or work;
how I deal with other people; how I handle possibilities; how I approach tasks; and
388 A.C. Wintergerst et al. / System 29 (2001) 385–403

how I deal with ideas. There are a total of 110 statements which students rate on a
four-point scale.
The measurement of learning styles presents the challenge of creating learning
style scales that are conceptually consistent as well as statistically reliable and
orthogonal in that they measure distinct characteristics. This latter area concerns
validity issues, especially the construct validity of survey instruments. Assessment
of validity in learning style measurement instruments has been limited with the
exception of the Kolb Learning Style Inventory (1976, 1985). Assessments have
primarily concentrated on external and predictive validity while less attention has
been placed on internal, or construct validity. In general, learning styles instruments
are questionable in terms of construct validity. For example, in a factor analytic
comparison of four learning styles instruments used with English native speakers
(Grasha-Riechmann Student Learning Style Scales in Riechmann and Grasha, 1974;
Kolb Learning Styles Inventory, 1976; Dunn et al., Learning Style Inventory, 1975;
and the Johnson Decision Making Inventory, Cascarelli, 1981), Ferrel (1983) did
not find evidence of the validity of the authors’ conceptualizations of learning styles.
After Dunn et al. (1995) synthesized the findings of 36 studies concluding that the
Dunn model was validated, Kavale et al. (1998) investigated and critiqued the Dunn
and Dunn Model of Learning-Style Preferences. Kavale et al. disagreed with this
conclusion, claiming that it had not been validated, and stressing the need for fur-
ther research before this learning style approach could be accepted. Purpura (1998)
has demonstrated the difficulties in validating instruments used in the language
learning context and the usefulness of exploratory factor analysis in assessing the
underlying factor structure of the items in these surveys.
Reid’s PLSPQ is a learning styles assessment instrument used in the ESL field
(Rossi-Le, 1995; Stebbins, 1995). The reliability and validity of the PLSPQ, with
which the present study is concerned, needs further investigation. Reid (1990)
reported difficulties in obtaining acceptable internal reliability for the six learning
style scales in the PLSPQ. In a study with Hispanic students, Itzen (1995) found
marginal internal reliability for the Auditory and Visual learning style scales. Con-
firmatory factor analyses did not confirm the six learning style scales hypothesized
by Reid, suggesting a lack of construct validity. No other investigations of the con-
struct validity of the PLSPQ have been reported.

2.2. Previous study on learning styles

Our previous study examined the learning styles of Russian-speaking ESL stu-
dents. We explored whether the learning style preferences of these students would
reflect more individual preferences or more cultural traditions, and how well student
responses would correspond with the information from the oral interviews and the
background questionnaire.
In that study we used three data collection instruments: a background ques-
tionnaire, Reid’s (1984) PLSPQ, and tape-recorded oral interviews. The background
questionnaire was a general information survey developed by the authors. Variables
such as age, gender, major, the number of years English was studied in their home
 A.C. Wintergerst et al. / System 29 (2001) 385–403 389

country and in the USA, years of schooling, parents’ educational levels and their
occupations both in their home country and in the USA were investigated.
The Perceptual Learning Style Preference Questionnaire consists of 30 randomly
ordered statements for the six learning style preferences: visual, auditory, kines-
thetic, tactile, group learning, and individual learning. Survey participants respond
on the basis of a five-point Likert scale, ranging from strongly agree to strongly
disagree. We chose to use the PLSPQ because it had been previously normed on
high, intermediate and advanced ESL classes by Reid (1987). The PLSPQ was also
one of only three known normed survey instruments in the ESL/EFL field, the
authors of which permit these instruments to be used in replication studies. Unlike
the other two survey instruments (O’Brien’s Learning Channel Preference Checklist
and Oxford’s Style Analysis Survey), the PLSPQ was neither long nor time-
consuming to complete. In addition, the PLSPQ had pre-established cutoff scores
for major, minor, and negligible learning style categories; thus, we did not have to
devise our own instrument nor determine our own cut-off scores. In the data anal-
ysis, we discovered that the results of the PLSPQ and the oral interviews contra-
dicted each other on several occasions. A number of reasons may account for this
contradiction. A lack of congruence suggests underlying problems with the test
construction, language problems on the part of the participants, test-taking prob-
lems, statement design problems, culture-specific problems, or the influence of
language proficiency on the validity and reliability of an instrument used for native
and non-native speakers of English.
The oral interviews consisted of a mix of direct and open-ended questions.
Included in these questions were some asking participants to describe how they had
been taught English in the former Soviet Union and what they thought were some of
the major similarities and differences with respect to teaching and learning there and
in the USA. Participants were also asked to share their thoughts about group work,
teacher lectures, and preferred learning styles.
Itzen (1995) conducted an in-depth confirmatory factor analysis on the norming of
Reid’s PLSPQ. Through attempts to confirm the factor structure hypothesized
by Reid, he found that regardless of the language proficiency of the participants, the
items in the survey were not clear measures of the learning styles they intended to
measure. Generally, Itzen found that students did not perceive as such items hypo-
thesized to measure particular learning styles in the PLSPQ. His results, coupled
with the outcomes of our previous study, motivated the investigation and lead to
our current study.

3. Current study

3.1. Methodology

3.1.1. Purpose of study

Specifically, the present study attempts to answer the questions, ‘‘What are
the reliability and validity of the PLSPQ when examined with ESL students?’’ And
390 A.C. Wintergerst et al. / System 29 (2001) 385–403

second, ‘‘What relationship exists between the learning styles identified in the
PLSPQ and the language background of the participants?’’ The results of these
analyses are used to examine the more general issue of the measurement difficulties
of conceptualizing learning style modalities and developing assessment instruments
that accurately measure those modalities.

3.1.2. Participants
The sample consisted of 100 ESL students, 55 females and 45 males, studying at a
major university in New York City. The students were enrolled in credit-bearing
intermediate and advanced ESL courses. The students ranged from 17 to 49 years
old with a mean age of twenty-one and a half. The students represented four lang-
uage groups (51 Chinese, 23 Korean, 11 Spanish, and 15 Russian) from 18 different
countries. Participants had studied English in their home countries for an average of
four and a half years with 20% having had no English instruction in their home
country. All but three of the students indicated that they had studied English in the
USA. The group averaged approximately three and a half years of US English
study, with 31% indicating that they had studied for one and a half years or less.
The students had a minimum score of 500 on the Test of English as a Foreign
Language (TOEFL).

3.2. Instruments

The two instruments used in this study were the background questionnaire and the
PLSPQ used in the previous study and already discussed.

3.2.1. Procedures
The background questionnaire and the PLSPQ were administered to the uni-
versity students during the eighth week of classes in the 1997 fall semester. The
teachers in the credit-bearing ESL writing/reading classes were asked to and agreed
to distribute the survey packets to all students in attendance. The packet included
the background questionnaire, instructions for the survey, and the PLSPQ. The
teachers were asked to read the instructions to the students, to allow adequate time
to complete the questionnaires, and to answer any questions. The students gave their
informed consent for the use of these data in a research project.

3.2.2. Data analysis

Initially, items in the survey were grouped according to Reid’s learning style
model. The internal consistency of these scales was assessed with the Cronbach
Alpha reliability estimate. The validity of the hypothesized factor structure of the
PLSPQ was examined through exploratory factor analysis. Results of both varimax
and oblimin rotations were reviewed.
Subsequently, an alternative learning style model was explored through a three-
factor analysis. Three new learning scales were formed and their internal reliability
estimated. These scales were utilized in examining the relationship between learning
styles and the language background of participants. Analysis of variance procedures
 A.C. Wintergerst et al. / System 29 (2001) 385–403 391

was employed to test the hypothesis that mean scores on the three learning style
measures differed across language groups. All analyses were carried out with SPSS
Version 8.0 for Windows.

4. Results

4.1. Reliability of Reid’s learning style scales

Initially, items in the PLSPQ were grouped into the six learning styles that Reid
(1984) hypothesized. Items representing these learning style scales are displayed in
Appendix A. Reliability analyses were run to examine the internal consistency of
items in each five-item scale. Item–total correlations as well as Cronbach’s alpha
were examined (Table 1).
First, the means and standard deviations were inspected in order to identify items
with little or no variability. With the exception of Q1 (4.33), means were not found
to be near the extremes of the scale. Means ranged from 3.10 to 4.33. All items
except Q1 (0.74) were found to have standard deviations between 0.80 and 1.10.
Three of the scales (Kinesthetic, Group, and Individual) yielded good reliability
estimates ranging from 0.69 to 0.87. Those of the remaining three (Visual, Auditory,
and Tactile) were found to be lower (0.48–0.59).
The item–total correlations were examined for the items in all scales. Correlations
for items in the Kinesthetic and Group scales were within acceptable ranges (0.30 or
greater). While the Individual scale had a good overall reliability (0.75), the item–
total correlation for Q27 (0.28) was found to be lower. Analyses indicated that if this
item were deleted from the scale, the scale alpha would improve to 0.78. In the
Visual scale, Q6 was found to have a low item–total correlation (0.0022) and further
analyses indicated that if this item were deleted, the scale reliability would reach
more acceptable levels (0.63). Similarly, Q11 in the Tactile scale was found to have a
low item–total correlation (0.20). Deletion of this item would increase the reliability
of this scale to 0.62. Four (Q7, Q9, Q17, and Q20) of the five items in the Auditory
scale were found to have item-total correlations lower than 0.30. Further analysis
revealed that a maximum alpha of 0.56 could be obtained by deleting Q9 and Q17.

4.2. Construct validity

The construct validity of the PLSPQ was examined through the use of exploratory
factor analysis. Initially, principal axis factoring extracted 10 factors with eigen-
values greater than 1.0 which accounted for 69% of the variance. Variable com-
munalities were all greater than 0.30 except for items Q6 (0.131) and Q1 (0.237).
Both orthogonal and oblique solutions were inspected. The interfactor correlation
matrix (Table 2) revealed only four correlations of 0.25 or greater suggesting that a
varimax rotation would appropriately represent the underlying factor structure.
Results of this analysis are displayed in Table 3. The 30 items in the survey did not
clearly load on Reid’s six hypothesized learning styles as expected.
392 A.C. Wintergerst et al. / System 29 (2001) 385–403

Table 1
Reliability analysis of Reid’s learning style scales

Scale Mean S.D. Item–total Scale alpha

correlation

Visual 0.53
Q6 3.90 0.89 0.00
Q10 3.81 0.83 0.45
Q12 4.00 0.86 0.41
Q24 3.34 1.02 0.34
Q29 3.10 0.99 0.32
Auditory 0.48
Q1 4.33 0.74 0.43
Q7 3.86 0.80 0.29
Q9 3.64 0.99 0.18
Q17 3.60 0.89 0.15
Q20 3.42 0.99 0.28
Kinesthetic 0.69
Q2 4.05 0.82 0.36
Q8 3.91 0.79 0.35
Q15 3.86 0.86 0.55
Q19 3.61 0.93 0.49
Q26 3.74 0.91 0.49
Tactile 0.59
Q11 3.91 1.02 0.20
Q14 3.62 0.94 0.44
Q16 3.43 0.95 0.35
Q22 3.68 0.82 0.34
Q25 3.18 1.04 0.42
Group 0.87
Q3 3.29 1.09 0.71
Q4 3.31 1.10 0.78
Q5 3.33 1.05 0.71
Q21 3.44 1.00 0.60
Q23 3.13 0.98 0.65
Individual 0.75
Q13 3.86 0.90 0.50
Q18 3.57 0.93 0.58
Q27 3.34 0.89 0.28
Q28 3.11 1.07 0.57
Q30 3.43 0.95 0.64

The first factor consisted of the five Group items. The second factor consisted of
four Individual items and two Visual items. Items Q12 and Q10 (Visual items refer-
ring to reading instructions) made up the third factor. Factor 4 consisted of three
Auditory items (Q7, Q1, Q20). Factors 5–9 consisted of items from different learning
style scales which were not always found to be conceptually compatible. For exam-
ple, Factor 8 consisted of Q16 (I learn better when I make drawings as I study) and
 A.C. Wintergerst et al. / System 29 (2001) 385–403 393

Table 2
Interfactor correlation matrix: Reid survey

Factor 1 2 3 4 5 6 7 8 9 10

Factor 1 1.00
Factor 2 0.21 1.00
Factor 3 0.00 0.08 1.00
Factor 4 0.20 0.00 0.20 1.00
Factor 5 0.16 0.13 0.09 0.09 1.00
Factor 6 0.22 0.13 0.10 0.19 0.01 1.00
Factor 7 0.18 0.08 0.09 0.13 0.04 0.12 1.00
Factor 8 0.26 0.11 0.07 0.25 0.01 0.13 0.19 1.00
Factor 9 0.04 0.08 0.04 0.09 0.01 0.04 0.02 0.03 1.00
Factor 10 0.30 0.07 0.10 0.21 0.08 0.33 0.13 0.10 0.05 1.00

Q17 (I learn better in class when the teacher gives a lecture). Factor 10 included only
one item, Q19 (I understand things better in class when I participate in role-playing).
Q6 did not load on any factor (with a loading of 0.35 or greater).
This factor solution did not yield the conceptual model posited by Reid and did
not appear to provide a conceptually acceptable learning style framework. The
results of a Scree Test (Cattell, 1966) indicated that a three-factor solution might
provide a more parsimonious grouping of items in the questionnaire (Appendix B).
Therefore, the authors decided to explore an alternative learning style model based
on data from this sample.

4.3. An alternative three-factor solution

A second factor analysis was run to explore an alternative learning style model
based on a three-factor solution. Items representing these learning style scales are
displayed in Appendix C. Again, both orthogonal and oblique solutions were
explored. Two inter factor correlations (Table 4) were found to be low ( 0.13 and
0.09) while the correlation between factors 1 and 3 was moderate ( 0.328) sug-
gesting that either an orthogonal or an oblique solution might be appropriate. Dif-
ferences between the solutions were minimal, so a varimax rotation was employed to
obtain orthogonal factors. The results of the three-factor solution are displayed in
Table 5. Only those loadings of 0.35 or greater are shown.
Items Q6, Q9, Q16, Q17, Q20, and Q27 were not found to have loadings of 0.35 or
greater on any factor, and are therefore not displayed. Factor one consisted of items
that focused on group activities or interaction with others (see Appendix C for item
texts). It contained 11 items: Five Group items (Q3, Q4, Q5, Q21 and Q23), two
Auditory items (Q1 and Q7) that describe learning when listening to others, and
four Kinesthetic items (Q2, Q15, Q19, and Q26) which highlight learning by ‘‘doing
experiments’’, role-playing’’ and participation in other ‘‘related activities’’. A study
by Itzen (1995) also suggested that Auditory items (e.g. listening to others)
and participating in role-playing and other group activities may be found to be an
394 A.C. Wintergerst et al. / System 29 (2001) 385–403

Table 3
Rotated factor matrix: Reid survey

Item Factors

1 2 3 4 5 6 7 8 9 10

Q4 (G)a 0.83
Q3 (G) 0.79
Q5 (G) 0.77
Q23 (G) 0.62
Q21 (G) 0.58
Q18 (I) 0.69
Q13 (I) 0.67
Q29 (V) 0.64
Q30 (I) 0.61
Q28 (I) 0.54 0.38
Q24 (V) 0.51
Q12 (V) 0.86
Q10 (V) 0.76
Q7 (A) 0.59
Q1 (A) 0.58
Q20 (A) 0.51
Q14 (T) 0.86
Q25 (T) 0.48
Q26 (K) 0.36
Q8 (K) 0.50
Q2 (K) 0.40
Q11 (T) 0.40
Q22 (T) 0.63
Q15 (K) 0.36
Q16 (T) 0.64
Q17 (A) 0.49
Q9 (A) 0.57
Q27 (I) 0.55
Q19 (K) 0.45

Only factor loadings of 0.35 and higher are displayed. Q6 did not load on any factor.
a
( ) Learning style factor the item was intended to represent: G, Group; I, Individual; V, Visual; T,
Tactile; K, Kinesthetic, A, Auditory.

integral part of a ‘‘group’’ learning style. Factor one, therefore, was termed Group
Activity Orientation and a learning style scale formed (Table 6 displays the alter-
native learning style scales and their reliability estimates). This scale was found to
have high internal reliability (Cronbach alpha=0.85) and no items displayed low
item–total correlations.
The second factor consisted of items focusing on individual learning activities. It
contained six items: Four Individual learning style items combined with two Visual
items (Q24 and Q29) to form this factor. These two visual items focus on reading,
clearly an individual learning activity. Again, the relationship between Individual
items and Visual items referring to reading was previously hinted at by Itzen
(1995). The second factor was termed Individual Activity Orientation. The internal
 A.C. Wintergerst et al. / System 29 (2001) 385–403 395

Table 4
Interfactor correlation matrix: alternative 3 — factor model

Factor 1 2 3

Factor 1 1.00
Factor 2 0.13 1.00
Factor 3 0.33 0.09 1.00

Table 5
Rotated factor matrix: alternative three — factor model

Factors

1 2 3

Q21 (G) 0.70

Q5 (G) 0.70
Q3 (G) 0.70
Q4 (G) 0.68
Q23 (G) 0.62
Q15 (K) 0.57
Q26 (K) 0.48
Q1 (A) 0.43
Q19 (K) 0.43
Q7 (A) 0.43
Q2 (K) 0.35
Q13 (I) 0.72
Q18 (I) 0.70
Q30 (I) 0.67
Q29 (V) 0.57
Q28 (I) 0.53
Q24 (V) 0.41
Q10 (V) 0.66
Q12 (V) 0.63
Q22 (T) 0.44
Q25 (T) 0.42
Q14 (T) 0.37
Q11 (T) 0.36
Q8 (K) 0.35

Items Q6, Q9, Q16, Q17, Q20, and Q27 were not found to have loadings of 0.35 or greater on any factor.

consistency of this scale (Cronbach’s alpha=0.77) as well as the item–total correla-

tions were also well within acceptable ranges (Table 6).
Factor 3 was found to contain Tactile, Visual, and Kinesthetic items. Upon close
examination of the items, it became clear that this factor focused on participating in
a class project. Two items (Q14 and Q25) refer to ‘‘making something for a class
project’’. Items Q10 and Q12 related to this concept by addressing reading instruc-
tions. Other items (Q22, Q11, for example) expressed learning by building something
and making a model. This learning style type, the third factor, was termed Project
Orientation. The Project Orientation scale consisting of eight items was found to
396 A.C. Wintergerst et al. / System 29 (2001) 385–403

Table 6
Reliability analysis of alternative learning style scales

Scale Mean S.D. Item–total Scale alpha

correlation

Group Activity Orientation 0.85

Q21 3.44 1.00 0.60
Q5 3.33 1.05 0.67
Q3 3.29 1.09 0.64
Q4 3.31 1.10 0.69
Q23 3.13 0.98 0.62
Q15 3.86 0.86 0.52
Q26 3.74 0.91 0.52
Q1 4.33 0.74 0.41
Q19 3.61 0.93 0.44
Q7 3.86 0.80 0.34
Q2 4.05 0.82 0.40
Individual Activity Orientation 0.77
Q13 3.86 0.90 0.60
Q18 3.57 0.93 0.59
Q30 3.43 0.95 0.60
Q29 3.10 0.99 0.51
Q28 3.11 1.07 0.50
Q24 3.34 1.02 0.34
Project Orientation 0.65
Q10 3.81 0.83 0.46
Q12 4.00 0.86 0.44
Q22 3.68 0.82 0.37
Q25 3.18 1.04 0.37
Q14 3.62 0.94 0.36
Q11 3.91 1.02 0.27
Q8 3.91 0.79 0.29

have an acceptable reliability of 0.65 (Table 6). Although items Q11 and Q8 had
item–total correlations of less than 0.30, further analyses indicated that removing
them would not improve the overall scale reliability.

4.4. Method effects

The data were also examined for method effects by grouping items together based
on question type (i.e. items with common phrases). Factor analysis was run on seven
different groups of items. Table 7 displays the question types and the number of
factors extracted from each group. Items of similar type but representing different
learning styles would not be expected to load together. In the event that they did,
this might indicate a method effect. Two or more factors were extracted from ques-
tion types 1, 3, 5, and 7 suggesting no method effect.
Item sets 2, 4, and 6 produced one factor solutions. In all three cases, the KMO’s
(Kaser–Meyer–Olkin Measure of Sampling Adequacy, a measure that gives an
 A.C. Wintergerst et al. / System 29 (2001) 385–403 397

Table 7
Question types and number of factors extracted

Question types Number of items Number of factors

1. I learn better (best). . . 10 4

2. I prefer. . . 4 1
3. I learn more. . . 4 2
4. I enjoy. . . 3 1
5. I remember. . .better. . . 4 2
6. I understand better. . . 3 1
7. When I work. . . 4 2

indication if there are big enough correlations among the variables to justify a factor
analysis) were sufficiently high and the percent of residuals above the diagonal
greater than 0.05 low enough (in two cases 0%) to indicate that the data were
appropriate for factor analysis. Therefore, items with the expressions ‘‘I prefer. . .’’,
‘‘I enjoy. . .’’, and ‘‘I understand better. . .’’ may show evidence of method effects (i.e.
clustering due to similar wording).

4.5. Relationship of learning styles to language groups

Based on conceptual considerations as well as the psychometric analysis presented

here, the authors believed that the three alternative learning style scales (Group
Activity Orientation, Individual Activity Orientation, and Project Orientation) pro-
vided a more realistic learning style model than the model posited by Reid in the
PLSPQ. Therefore, subsequent analyses were based on these scales.
Mean scores were calculated on the three learning style measures for four lang-
uage groups: Chinese, Korean, Spanish, and Russian (Table 8). The intention was to
examine preferences for Group Activity, Individual Activity and Project Orienta-
tions within groups as well as to compare preferences across language groups. The
Chinese and Russian students displayed a similar pattern in their learning style
preferences. Among these students, preferences for group and project activities were
found to be higher than for individual activities. Korean students expressed a

Table 8
ANOVA of learning style mean differences among language groups

Chinese Korean Spanish Russian F

(N=51) (N=23) (N=11) (N=15)

Group Activity Orientation 3.68 3.42 3.67 3.78 1.35

Individual Activity Orientation 3.28 3.66 3.77 3.13 3.90*
Project Orientation 3.75 3.62 4.01 3.61 1.81

Scheffe and Tukey HSD post-hoc analyses using Harmonic Mean Sample Sizes did not detect significant
differences among the groups.
*P <0.05.
398 A.C. Wintergerst et al. / System 29 (2001) 385–403

slightly greater preference for individual activities (3.66) and participating in class
projects (3.62) than for group learning (3.42). The Spanish students in the sample
were oriented more toward projects (4.01) and least toward group activities (3.67).
Analysis of variance procedures were performed on the learning style means in
order to detect differences among the four language groups (Table 8). No significant
differences were found between the groups on the Group Activity Orientation or the
Project Orientation. While the ANOVA test found a significant difference between
the groups on Individual Activity Orientation, post-hoc analyses failed to detect this
difference. The conservative nature of the post-hoc methods selected and adjust-
ments made for differing sample sizes may have masked a difference in the Spanish
and Russian students’ preferences for Individual learning activities. Spanish students
indicated a higher preference (3.77) for individual activities such as working alone
and reading than did the Russian speakers (3.13).

5. Discussion

5.1. Reliability and validity

In this study we examined Reid’s PLSPQ. The limited evidence provided by this
study and by a previous examination of the PLSPQ poses questions concerning the
reliability and validity of the learning style scales in this survey instrument. Reid
(1990) herself has discussed the difficulties she encountered in obtaining acceptable
internal consistency for the scales. Further research is required in order to explore the
construct validity of learning styles as measured by the PLSPQ. Itzen (1995) had uti-
lized structural equation modeling (confirmatory factor analysis) in a previous study
of the PLSPQ in an attempt to evaluate the learning style structure suggested by Reid.
Exploratory factor analysis was employed in the present study in order to explore
underlying relationships among the survey items with the hope of discovering a new
way of viewing the learning styles incorporated in the PLSPQ. Purpura (1998), among
others, has demonstrated the usefulness of exploratory factor analysis in examining
survey instruments of this type. Both exploratory and confirmatory factor analyses
will be useful in further work on the PLSPQ and other similar instruments.

5.2. Measurement difficulties

This study alludes to the difficulty of measuring learning styles. It is possible that
the items in Reid’s survey were confusing for the non-native speakers of English
in the way they were worded, and thus provided poor measures of six learning styles.
The method effects discovered for some of the items in this study indicate that stu-
dents may have responded based on similarities in wording of items rather than on
the learning styles concepts they represented. Though we do not have specific
evidence of such, we suspect that it may be more difficult to create clearly worded
survey items for non-native speakers of English that avoid this type of method
effect. Reid (1990) herself implied that the wording of the items, which make up the
 A.C. Wintergerst et al. / System 29 (2001) 385–403 399

different learning style items, may have been clearer for native speakers of English
because of their higher English language proficiency. This type of effect may be
magnified with an ESL population where there is a greater preoccupation with the
language of the survey.
Difficulties in measurement may also stem from lack of clarity in the underlying
conceptual model of learning styles. Why is it so difficult to develop a conceptually
clear model of learning styles? This may in large part be due to the difficulty of
defining certain learning styles separately. As measurement questions surfaced in
this study, we were forced to step back and examine Reid’s hypothesized model of
six distinct learning style types and wonder if in reality an individual’s learning style
preference could be divided in this way. The data were telling us this was not the
case. Learning styles for our subjects tended to be focused on group activities, indi-
vidual activities and project-related behaviors as indicated by the three-factor solu-
tion we adopted. Perhaps a new theory of learning styles will address the belief that
a preference for listening activities may be inextricably interwoven with group
activities and that a visual learning behavior such as reading cannot be separated
from the concept of individual learning.
This is not to say that the four perceptual learning style types are not valid and
cannot be measured. It is possible that in an instrument where only perceptual
learning styles are included (and not social or other types of preferences), a factor
structure distinguishing among them could be developed. However, evidence in this
study points toward a relationship between social preferences (group and individual
learning) and perceptual preferences (visual, auditory, tactile, and kinesthetic learn-
ing) which may have implications for learning style research in general.

5.3. Learning styles and language background

Previous studies (Hofstede, 1986; Willing, 1987; Hall and Hall, 1990; Oxford and
Anderson, 1995; Rossi-Le, 1995; Oxford, 1996) have suggested a relationship
between cultural and/or language background and learning styles, although our
study could find no such relationship. Our subjects were of a variety of backgrounds
and ages, and many of those that we grouped by language were in fact from different
countries. For example, the Spanish speakers came from many different Latin
American countries with differences in cultures. This may have confused the issue
and prevented us from discovering any clear relationship to learning styles. How-
ever, these results may also suggest that learning styles are in fact individual pref-
erences rather than characteristics of linguistic groups.

6. Conclusion

The value of this research lies in helping to improve a learning styles research
instrument and in alerting researchers and practitioners interested in using Reid’s
PLSPQ to the necessity of examining the underlying factor structure of the instru-
ment as well as the theory on which it is based. We have utilized exploratory factor
400 A.C. Wintergerst et al. / System 29 (2001) 385–403

analysis as an efficient tool not only for discovering measurement difficulties in the
PLSPQ, but also for highlighting the intricate relationship between theory and sur-
vey items. Furthermore, we have explored the possibility of redefining the learning
styles of ESL students as represented in the PLSPQ by proposing three alternative
learning style scales, which we have chosen to call the Learning Styles Indicator
(LSI). It is our hope that future research will lead to further evaluation and
improvement of this instrument.
In general, researchers should continue to carry out thorough assessments of the
psychometric properties of the instruments designed to measure individual learning
styles. Although a few of the learning style instruments have been in use for over two
decades, researchers are still struggling to establish valid learning style theories and
measurement instruments. The science of identifying and measuring the learning
styles of non-native speakers of English is relatively new. Exploratory factor anal-
ysis, along with other types of analyses, can play a valuable role in this field. Only
after the true factor structure of a learning style instrument has been examined, can
researchers confidently assert conclusions about individual and group differences
detected by such as instrument.

Appendix A

Learning style scales as hypothesized by Reid

Group
Q3. I get more work done when I work with others.
Q5. In class, I learn best when I work with others.
Q4. I learn more when I study with a group.
Q21. I enjoy working on an assignment with two or three classmates.
Q23. I prefer to study with others.
Individual
Q13. When I study alone, I remember things better.
Q18. When I work alone, I learn better.
Q27. In class, I work better when I work alone.
Q28. I prefer working on projects by myself.
Q30. I prefer to work by myself.
Visual
Q6. I learn better by reading what the teacher writes on the chalkboard.
Q10. When I read instructions, I remember them better.
Q12. I understand better, when I read instructions.
Q24. I learn better by reading than by listening to someone.
Q29. I learn more by reading textbooks than by listening to lectures.
 A.C. Wintergerst et al. / System 29 (2001) 385–403 401

Auditory
Q1. When the teacher tells me the instructions I understand better.
Q7. When someone tells me how to do something in class, I learn better.
Q9. I remember things I have heard in class better than things I have read.
Q17. I learn better in class when the teacher gives a lecture.
Q20. I learn better in class when I listen to someone.
Tactile
Q11. I learn more when I can make a model of something.
Q14. I learn more when I make something for a class project.
Q16.I learn better when I make drawings as I study.
Q22. When I build something, I remember what I have learned better.
Q25. I enjoy making something for a class project.
Kinesthetic
Q2. I prefer to learn by doing something in class.
Q8. When I do things in class, I learn better.
Q15. I enjoy learning in class by doing experiments.
Q19. I understand things better in class when I participate in role-playing.
Q26. I learn best in class when I can participate in related activities.
Appendix B

Scree plot of initial exploratory factor analysis

402 A.C. Wintergerst et al. / System 29 (2001) 385–403

Appendix C

Alternative learning style scales: The Learning Styles Indicator (LSI)

Learning style one: Group Activity Orientation

Q21. I enjoy working on an assignment with two or three classmates.
Q 5. In class, I learn best when I work with others.
Q 3. I get more work done when I work with others.
Q 4. I learn more when I study with a group.
Q23. I prefer to study with others.
Q15. I enjoy learning in class by doing experiments.
Q26. I learn best in class when I can participate in related activities.
Q 1. When the teacher tells me the instructions I understand better.
Q19. I understand things better in class when I participate in role-playing.
Q 7. When someone tells me how to do something in class, I learn better.
Q 2. I prefer to learn by doing something in class.
Learning style two: Individual Activity Orientation
Q13. When I study alone, I remember things better.
Q18. When I work alone, I learn better.
Q30. I prefer to work by myself.
Q29. I learn more by reading textbooks than by listening to lectures.
Q28. I prefer working on projects by myself.
Q24. I learn better by reading than by listening to someone.
Learning style three: Project Orientation
Q10. When I read instructions, I remember them better.
Q12. I understand better, when I read instructions.
Q22. When I build something, I remember what I have learned better.
Q25. I enjoy making something for a class project.
Q14. I learn more when I make something for a class project.
Q11. I learn more when I can make a model of something.
Q 8. When I do things in class, I learn better.
Statements drawn from Reid (1984).

References
Brown, H.D., 1994. Principles of Language Learning and Teaching, 3rd Edition. Prentice Hall Regents,
Englewood Cliffs.
Cascarelli, W. 1981. Personal communication.
Cattell, R.B., 1966. The meaning and strategic use of factor analysis. In: Cattell, R.B. (Ed.), Handbook of
Multivariate Experimental Psychology. Rand McNally, Chicago.
Chapelle, C., Green, P., 1992. Field independence/dependence in second language acquisition research.
Language Learning 42, 47–83.
 A.C. Wintergerst et al. / System 29 (2001) 385–403 403

Dunn, R.S., 1990. Understanding the Dunn and Dunn learning styles model and the need for individual
diagnosis and prescription. Journal of Reading, Writing, and Learning Disabilities International 6,
223–247.
Dunn, R.S., Dunn, K., Price, G.E., 1975. The Learning Style Inventory. Price Systems, Lawrence, KS.
Dunn, R.S., Dunn, K., Price, G.E., 1989. Learning Styles Inventory (LSI): An Inventory for the identifi-
cation of How Individuals in Grades 3 Through 12 Prefer to Learn. Price Systems, Lawrence, KS.
Dunn, R., Griggs, S.A., Olson, J., Beasley, M., Gorman, B.S., 1995. A meta-analytic validation of the
Dunn and Dunn model of learning style preferences. The Journal of Educational Research 88, 353–362.
Ellis, R., 1994. The Study of Second Language Acquisition. Oxford University Press, Oxford.
Ferrel, B.G., 1983. A factor analytic comparison of four learning-styles instruments. Journal of Educa-
tional Psychology 75, 33–39.
Hall, E.T., Hall, M.R., 1990. Understanding Cultural Differences. Intercultural Press, Yarmouth, ME.
Hofstede, G., 1986. Cultural differences in teaching and learning. International Journal of Intercultural
Relations 10, 301–320.
Itzen, R., 1995. The dimensionality of learning structures in the Reid perceptual learning style preference
questionnaire. Unpublished doctoral dissertation, University of Illinois at Chicago, Illinois.
Kavale, K.A., Hirshoren, A., Forness, S.R., 1998. Meta-analytic validation of the Dunn and Dunn model
of learning-style preferences: a critique of what was Dunn. Learning Disabilities Research and Practice
13 (2), 75–80.
Kolb, D., 1976. The Learning Style Inventory: Self-scoring test and interpretation. McBer and Company,
Boston.
Kolb, D., 1985. Learning Style Inventory (Revised Edition). McBer and Company, Boston.
Lee, M., 1976. Some common grammatical errors made in written English by Chinese students. CATE-
SOL Occasional Papers 3, 115–120.
O’Brien, L., 1990. Learning Channel Preference Checklist (LCPC). Specific Diagnostic Services, Rock-
ville, MD.
Oxford, R.L., 1993. Style Analysis Survey. University of Alabama, Tuscaloosa, Alabama.
Oxford, R.L., 1996. Language Learning Strategies Around the World: Cross-Cultural Perspectives. Uni-
versity of Hawaii, Honolulu, Hawaii.
Oxford, R.L., Anderson, N., 1995. A crosscultural view of learning styles. Language Teaching 28, 201–215.
Purpura, J., 1998. The development and construct validation of an instrument designed to investigate
selected cognitive background characteristics of test-takers. In: Kunnan, A.J. (Ed.), Validation in Lan-
guage Assessment. Lawrence Erlbaum Associates, Mahwah, NJ, pp. 111–139.
Reid, J., 1984. Perceptual Learning Style Preference Questionnaire. Copyrighted by Reid. Available
through Joy Reid, Department of English, University of Wyoming, Laramie, WY 82070.
Reid, J., 1987. The perceptual learning style preferences of ESL students. TESOL Quarterly 21, 87–111.
Reid, J., 1990. The dirty laundry of ESL survey research. TESOL Quarterly 24, 323–338.
Riechmann, S., Grasha, A., 1974. A rational approach to developing and assessing the construct validity
of the Student Learning Style Scales Instrument. Journal of Psychology 87, 213–223.
Rossi-Le, L., 1995. Learning styles and strategies in adult immigrant ESL students. In: Reid, J. (Ed.),
Learning Styles in the ESL/EFL Classroom. Heinle and Heinle, Boston, MA, pp. 118–125.
Skehan, P., 1998. A Cognitive Approach to Language Learning. Oxford University Press, Oxford.
Stebbins, C., 1995. Culture-specific perceptual-learning style preferences of postsecondary students of
English as a second language. In: Reid, J. (Ed.), Learning Styles in the ESL/EFL Classroom. Heinle
and Heinle, Boston, MA, pp. 108–117.
Willing, K., 1987. Learning Styles in Adult Migrant Education. Adult Migrant Education Programme,
Adelaide.
Wintergerst, A., DeCapua, A., March 1998a. Exploring learning styles of Russian ESL students. Paper
presented at the 32nd Annual TESOL Convention, Seattle, WA.
Wintergerst, A., DeCapua, A., March 1998b. Language learning styles and cultural variables. Paper pre-
sented at the American Association for Applied Linguistics Annual Conference, Seattle, WA.
Witkin, H., 1962. Psychological Differentiation. Wiley, NY.

View publication stats