Cognitive style and educational performance. The case of public schools in Bogotá, Colombia

Abstract

This study analyses the relationships among educational performance, field dependence-independence cognitive style and factors traditionally associated with performance and style, to build a comprehensive model of factors associated with the levels of education performance of students in Bogotá. A total of 3003 students, of grades 8 and 10, from 62 public schools of the city of Bogotá, Colombia participated in the study. An analysis of multiple correspondences and a path analysis were carried out. A relationship between cognitive style and educational performance was found: field-independent students are more likely to obtain high-performance levels both in standardised tests and in teachers’ evaluations. The path analysis shows that there are two directions in the association: a direct path leads to a positive association: higher levels of field independence produce better performances; an indirect path leads to a negative association: higher levels of field independence produce indiscipline and, consequently, lower performances.

Full text

Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=cedp20
Download by: [Universidad Del Norte] Date: 13 October 2015, At: 08:53
Educational Psychology
An International Journal of Experimental Educational Psychology
ISSN: 0144-3410 (Print) 1469-5820 (Online) Journal homepage: http://www.tandfonline.com/loi/cedp20
Cognitive style and educational performance. The
case of public schools in Bogotá, Colombia
Christian Hederich-Martínez & Angela Camargo-Uribe
To cite this article: Christian Hederich-Martínez & Angela Camargo-Uribe (2015): Cognitive
style and educational performance. The case of public schools in Bogotá, Colombia,
Educational Psychology, DOI: 10.1080/01443410.2015.1091916
To link to this article: http://dx.doi.org/10.1080/01443410.2015.1091916
Published online: 06 Oct 2015.
Submit your article to this journal
Article views: 1
View related articles
View Crossmark data

Cognitive style and educational performance. The case of public
schools in Bogotá, Colombia
Christian Hederich-Martínez*and Angela Camargo-Uribe
Doctorado en Educación, Universidad Pedagógica Nacional, Bogotá, Colombia
(Received 25 April 2014; final version received 3 September 2015)
This study analyses the relationships among educational performance, field
dependence-independence cognitive style and factors traditionally associated
with performance and style, to build a comprehensive model of factors associ-
ated with the levels of education performance of students in Bogotá. A total of
3003 students, of grades 8 and 10, from 62 public schools of the city of Bogotá,
Colombia participated in the study. An analysis of multiple correspondences and
a path analysis were carried out. A relationship between cognitive style and
educational performance was found: field-independent students are more likely
to obtain high-performance levels both in standardised tests and in teachers’
evaluations. The path analysis shows that there are two directions in the associa-
tion: a direct path leads to a positive association: higher levels of field indepen-
dence produce better performances; an indirect path leads to a negative
association: higher levels of field independence produce indiscipline and, conse-
quently, lower performances.
Keywords: academic performance; cognitive styles; adolescent; school subject
Information concerning the levels of educational performance among students in
Bogotá and factors associated with differences in those levels has been gathered
for many years (Instituto Colombiano para la Evaluación de la Educación Supe-
rior [ICFES], 2011; Piñeros & Rodríguez, 1999). Up to now, studies on factors
associated with educational performance have not been successful in explaining
students’educational performance (Departamento Nacional de Planeación [DNP],
1999).
We think that the difficulty to come up with an explanation for differences in
levels of educational performance is the lack of a theoretical basis for the analysis
of data. We have tried to overcome the restrictions of conventional studies on fac-
tors associated with educational performance by working with a well-developed
psychological construct: the concept of ‘cognitive style’(Riding, 2000). Some
authors have found that the use of this theoretical lens has important advantages
for the study of educational performance (Tinajero, Lemos, Araujo, Ferraces, &
Páramo, 2012; Tinajero & Páramo, 1997, 1998). The study reported in this paper
follows this idea.
*Corresponding author. Email: hederich@pedagogica.edu.co
© 2015 Taylor & Francis
Educational Psychology, 2015
http://dx.doi.org/10.1080/01443410.2015.1091916
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Field dependence-independence cognitive style
Understood as ‘a distinct and consistent way for an individual to encode, store and
perform’(Atkinson, 2004, p. 663), cognitive style has clear connections with educa-
tional performance, via school learning. This study is focused on one cognitive style
dimension that has been the most frequently used in educational research: the field
dependence-independence (FDI) dimension, developed by Herman Witkin and his
colleagues. Recent formulations of FDI define it as a construct that entails individual
differences in the preference for internal or external referents; preference that mani-
fests itself in a wide range of behavioural domains including perception, cognition,
social behaviour and affection (Witkin & Goodenough, 1981).
Witkin’s formulations of the FDI dimension emphasised its neutral value; which
means that both, field-independent (FI) and field-dependent (FD) individuals, are
equally well adapted to the demands of the environment; even though the modes of
adaptation are clearly different. Nevertheless, different studies have questioned the
neutrality of cognitive style in this dimension, pointing out the high correlations
between FDI and different measures of intelligence and spatial aptitudes (McKenna,
1984). Other authors have kept the difference between these two constructs pointing
out the inconsistencies of the relationships between these two variables (Forns-
Santacana & Amador-Campos, 1990).
The debate about the relevance of FDI in the educational field has been fed by
findings on the close relationship between this cognitive style dimension and differ-
ent educational performance indicators (Tinajero & Páramo, 1998). In general, most
of the studies have found better educational achievement indicators for FI students in
almost all areas, including Language (Paramo & Tinajero, 1990), Mathematics (Van
Blerkom, 1988), Natural Sciences (Arthur & Day, 1991), Social Sciences (Bowlin,
1988), Foreign Language (Chapelle & Roberts, 1986), Music (Schmidt & Lewis,
1987), Arts (Fergusson, 1992) and Computer Programming (Coventry, 1989).
In some cases the associations stop being significant when intelligence is con-
trolled (Roberge & Flexer, 1981); even though in other studies differences remain,
always in favour of FI students, after intelligence is controlled (Tinajero & Páramo,
1997). What exactly does educational performance mean?
Educational performance
We understand educational performance as a category that includes all those things
achieved by students as the direct result of their inclusion in an educational system.
Educational performance is a variable attributed to the individual. Within a particular
system, a singular person shows some degree of performance in each of the dimen-
sions of human learning defined by the system. In general, this degree is determined
by an evaluation of the student’s performance, according to specific criteria for the
minimum performance to be achieved at each stage of his/her process within the
system.
For the purposes of this study, we will assume that educational performance is
indicated by two types of evaluation: the first one corresponds to a standardised
large-scale evaluation, carried out by authorities of the system that are not directly
involved in the everyday instructional process experienced by the student; the sec-
ond one is an individual evaluation, directly carried out by the students’teachers;
evaluation that is part of the pedagogical process occurring in the classroom.
2C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

These two types of evaluation are qualitatively different and relate to each other
in complex ways. The first type seeks to give an objective measure of educational
performance using standardised tests that provides scores that are independent from
the test designers. They are valid, reliable and consistent exams, designed in a fixed
multiple-choice question format, that determine what has been achieved by a sample
population with respect to national standards. In these types of tests, such as PISA
and TIMMS, achievement is considered a direct indicator of the quality of the
educational service. The results of these tests are not given back to the students or
their teachers. They are presented to the society through the mass media as an
indicator of how the educational system works.
The second type of evaluation is the one that is given by the teacher to his/her
students. Basically, it is applied to determine how far each student has gone with
respect to a certain learning goal, so that pedagogical actions can be taken or a cer-
tain level of competence can be certified for the student to go from one grade to
the next. It should be mentioned that it is not rare to find cases in which the grade
or score given by the teacher is an expression of how pleased he/she is with
respect to the student’s behaviour in class. This gives this type of evaluation a sub-
jective character. In this sense, it should be noted that grading is frequently used
as reinforcement for good behaviour in the classroom. Additionally, teacher’s
grades are relative to the class group. This is the reason why it is difficult to com-
pare different class groups using these grades. This type of evaluation rarely uses
multiple-choice questions; it usually takes the form of a global assessment of a
written paper or a classroom presentation, even though in some subject areas open
tests are frequent.
FDI and educational performance
It is important to note that most studies that have worked with standardised mea-
sures of educational performance seem to have found significant differences in
favour of FI subjects. In contrast, most studies that have used school grades given
by teachers find no significant differences between FD and FI, or find differences
only in some subject areas, grades or ages (Paramo & Tinajero, 1990).
The above finding could be related to the fact that the particular design of each
evaluation favours the expression of educational performance of one cognitive style
or another. In particular, multiple choice tests favour FI students and disfavours FD
ones, due to the difficulty of the latter to inhibit incorrect or salient choices. On the
other hand, it has been found that FD students respond better to pop quizzes, when
material is presented in an interactive way, using examples that require certain social
sensibility for its total apprehension (Rollock, 1992).
Thus, it is important to understand the relationships between the two types of
evaluation and the links between each one of them and the students’cognitive style.
This will also lead to a comprehensive framework for the explanations of the rela-
tionships between different socio-demographic factors that have been associated
with educational performance by means of their relationship to the students’cogni-
tive style.
Many variables have been associated with FDI. Among them, we can mention sex
(Hederich & Camargo, 1999; Hederich, Camargo, Guzmán, & Pacheco, 1995; Witkin
& Goodenough, 1981), age (Hederich & Camargo, 1999; Witkin & Goodenough,
1981), family structure (Claeys & Mandosi, 1977; Hederich & Camargo, 1999),
Educational Psychology 3
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

nurturing practices and cultural background (Hederich, 2007; Páramo & Tinajero,
1992). The present study includes some of these variables, not only because of their
association with FDI, but also because of their relationship with students’educational
performance. The idea underlying this analysis is that the explanation of the associa-
tion among different individual and family factors and educational performance is due
to the mediation of cognitive style between these factors and performance.
In sum, the main purpose of this study is to build a comprehensive model of fac-
tors associated with the level of educational performance of students in Bogotá. In
particular, the study analyses the relationships among educational performance, FDI
cognitive style and factors traditionally associated with both, performance and style.
In order to do that, we will divide this study in two parts. In the first part, a factorial
analysis will be made. In the second part, a path analysis including some of the vari-
ables will be performed.
Methodology
Design
This research can be considered a probabilistic, multivariate observation, which
gathers information about levels of educational performance, cognitive styles in the
FDI dimension and a series of variables potentially associated with both constructs.
Educational performance is operationalised by two groups of indicators: (a) the
scores obtained in the standardised tests on basic competencies which were applied
by the Project for the Evaluation of Basic Competencies of the Secretary of Educa-
tion of Bogota (1999), called SABER. SABER tests evaluate students’mathematics,
language and science literacy according to what Colombian official curricular
standards expect them to be for 7th and 9th graders. The scores were divided into
quintiles of test scores, so that the first quintile represents the lowest scores and the
last one represents the highest ones; and (b) the students’school grades given by
their Math, Language, Natural Sciences and Social Sciences’teachers. It may be
important to note that these grades were collected one year after SABER tests were
administered, so these students were 8th and 10th graders by the time data were col-
lected. Basically, we asked each teacher to assess his/her students’performance in
his/her class, based on their achievements during the school year (it was the end of
the second term, of a four term school year). Grades were given using a scale that
goes from ‘deficient’(below average), ‘acceptable’(average) to ‘superior’(above
average).
On the other hand, students’scores in the embedded figures test (EFT) indicate
their FDI cognitive style. In this study, the Sawa-Gottschadt version of the EFT test
was used. The test consists of 50 complex geometric designs divided into 10 groups.
In each group, the student has to find and mark the specific simple figure that is
embedded in each complex one. There is a time limit to solve each group of figures.
As can be inferred, this instrument measures the speed of perceptual restructuring,
an ability which characterises FI individuals. This instrument achieves levels of
reliability of .91 (Hederich, 2007; Hederich & Camargo, 1999). It may be worth-
while to mention that the Sawa-Gottschadt version of the EFT is not the original
group embedded figures test (GEFT), developed by Witkin et al. in 1971 (Witkin,
Oltman, Raskin, & Karp, 1971). The most visible difference between these two tests
is that while in Witkin’s version the simple figures are located at the end of the
4C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

booklet and the student has to use his/her working memory to keep the figure in
mind, in the Sawa-Gottschadt version the simple figures are located on the same
page as the 10 complex figures in which they have to be found and marked. The fact
that, in this version, working memory capacity is not a factor involved in the task is,
in our opinion, an important advantage to be considered (Nagata, 1989). Addition-
ally, the correlation between the two versions of the test is very high: the correlation
between the original Witkin’s EFT and Sawa-Gottschadt version was .847 (Sawa,
1966) and the correlation between Witkin’s GEFT and the Sawa-Gottschadt version
was .794 (Hederich, 2007).
A set of variables associated with both, educational performance and cognitive
style, are also included in the analysis. Table 1shows variables and indicators.
Information concerning these variables was gathered by means of a questionnaire
administered to the sample.
Population and sample
The sample of this study is composed of 3003 public school students from the city
of Bogotá, Colombia, from grades 8 and 10 of primary and secondary education
(ages 14–17 years old). We worked with a total effective sample of individuals who
were in the 8th grade (50.1%) and 10th (49.9%) during the year 2000. These stu-
dents came from 62 public schools. From a technical point of view, this sample ful-
fils the following requirements: it is a random, multistage, clustered, representative
sample for the groups and it is stratified according to: (a) level of performance in
the tests and (b) socio-economic stratum, indicated by local government stratifica-
tion scale. This sample represents a total of 21,370 students in the 8th and 10th
grades attending public schools in Bogotá and accounted for 14.05% of the
population.
Analysis
The study consists of two parts. In part A, an analysis of correspondences was car-
ried out to examine the relationships among the variables. This type of multivariate
analysis is a factor analysis performed on values of nominal and ordinal variables.
This analysis permits the selection of a group of variables, called active variables, to
be used in the construction of the factorial axes. In this case a total of 27 values
from seven indicators of educational performance were considered: 15 values corre-
spond to the quintiles of SABER tests scores in Mathematics (5), Language (5) and
Natural Sciences (5), and 12 values that correspond to the three values of the grades
given by the teachers of Math (3), Language (3), Natural Sciences (3) and Social
Sciences (3). Apart from these active variables, the analysis of correspondences
allows the inclusion of ‘illustrative’variables: those that, without being part of the
construction of the factorial axes, show some visible associations with them in the
factorial plane. This analysis was performed by means of the software Spad, v. 4.5.
The results of the analysis of correspondences lead to the second part (part B) of
the study. Part B consisted of the construction of the linear structural equations
model for the prediction of the different educational performance indicators. The
model presents the relationships among the seven indicators of educational perfor-
mance, FDI cognitive style and two classroom behaviour indicators: not having been
Educational Psychology 5
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Table 1. Variables and indicators.
Variables Values
School profile Preschool Student attended preschool Yes/No
Cognitive training Use of computers at home hours/week No use of computer
Between 1 and 3 h
Use of computers at school hours/week Between 3 and 6 h
High: 6 or more hours
School history Over age Difference between student’s age and his/her grade median age Very much ahead: 3 or
more years
Ahead: 2 years ahead
On age: −1 to 1 year
Behind: year behind 1
Very much behind: 2 or
more years
Partial dropping
out
Student reports having dropped out one or more years during his/her school
life
Yes/No
School mobility Number of schools the student has attended 1–6,
7: 7 or more
Repetition Number of grades that the student has repeated 0, 1,
2: 2 or more grades
School behaviour Absenteeism Number of times the student has not attended class, has arrived late or has left
early
Never
Seldom
Often
Frequently
Discipline Detentions Yes/No
Frequency of running away Never
Seldom
Often
Frequently
Attitudes about
subjects
Student reports his/her attitude towards some school subjects: math, language,
science and social sciences
Positive
Neutral
Negative
(Continued)
6C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Table 1. (Continued).
Variables Values
Individual variables Gender Direct self-report Masculine
Feminine
Cognitive style EFT scores divided in quintiles Very field-dependent
Field-dependent
Intermediate
Field-independent
Very field-independent
Cultural and social
variables
Socio economic
strata
Citywide economic stratification of living home Very low
Low
Middle-low
Middle
Socio cultural
level
Student’s mother’s educational level Incomplete primary
school
Complete primary school
Incomplete secondary
school
Complete secondary
school
University
Family group Family size Very small: 1–3 family
members
Small: 4–5
Medium: 6–7
Large: 8 or more family
members
Presence of the father, presence of the mother, presence of the paternal or
maternal grand-parents at home
Yes/No
Family eco-
cultural origin
Place of birth of students, their parents and grand-parents (in order to identify
Colombian subcultural origin)
Andino-Santandereano
Antioqueño
Fluvio-minero
Bogotá
Educational Psychology 7
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

suspended from school and not having run away. This analysis was performed by
means of the software package Amos, from SPSS.
Results
Part A. A view of educational performance
The results show two important and separable dimensions, represented by the first
two factors of the analysis with eigenvalues of .2934 and .2090, respectively.
The first factor explains 27.74% of variance and polarises the scores of all the
standardised test scores (Math, Language and Natural Sciences) in extreme opposites
and all school grades given by teachers (Math, Language, Natural Sciences and
Social Sciences) in the same direction. Thus, low performances in both the tests
scores and the school grades are in the extreme negative pole, and high perfor-
mances are in the extreme positive pole. The standardised tests scores highly con-
tribute to this factor and the teachers’school grades contribute to a lesser extent.
Table 2presents the description of the first and second factors according to the
values of the active nominal variables in which they are largely polarised. It should
be noted that the order of the values agrees with the order of their distance to the
central zone of the factor. The farther it stands from the central zone, the better it
Table 2. Description of the first and second factors by active variables.
First factor Second factor
‘What is common in all evaluations’‘What is inconsistent among evaluations’
V-test Variable Value nV-test Variable Value n
−18,59 Language test
score
Quintile 1 374 −19,25 Soc. Science
school grade
Superior 374
−18,37 Nat. Science test
score
Quintile 1 374 −16,82 Nat. Science
school grade
Superior 374
−15,09 Math test score Quintile 1 378 −16,39 Math test score Quintile
1
378
−14,61 Math school
grade
Deficient 399 −15,02 Language school
grade
Superior 399
−11,64 Language test
score
Quintile 2 362 −14,19 Math school
grade
Superior 362
−11,43 Soc. Science
school grade
Acceptable 941 −10,44 Nat. Science test
score
Quintile
1
941
Central zone Central zone
17,21 Soc. Science
school grade
Superior 738 14,61 Soc. Science
school grade
Deficient 176
20,07 Nat. Science
school grade
Superior 609 15,80 Language school
grade
Deficient 187
20,47 Math school
grade
Superior 584 16,22 Nat. Science
school grade
Deficient 302
21,32 Math test score Quintile 5 375 16,79 Math school
grade
Deficient 399
23,39 Nat. Science test
score
Quintile 5 375 17,19 Math test score Quintile
5
375
23,39 Language test
score
Quintile 5 370 18,22 Nat. Science test
score
Quintile
5
375
8C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

characterises the factor. Values of the proof for examining the levels of significance
of the contribution of each value to the characterisation of the factor are also
included. To obtain this proof, a statistical U is used, with significant values outside
the interval (−1.96, 1.96).
As can be seen, in the first factor all the measures of educational performance
are associated among themselves in a consistent direction. We call this first factor:
that which is common in all the evaluations. When summarising all the indicators in
a consistent direction, we can conclude that this factor, generally, represents a
unique, broad dimension of educational performance to which all the measures of
performance (types of evaluation and subject areas) contribute with different
weights.
On the other hand, the second factor explains 11.83% of the variance and it is
extremely interesting in as much as the directions of the two types of evaluations
are inverted on this factor. Certainly, students with both high performance in school
grades and low performance in all test scores are in the extreme negative pole. The
situation is completely opposite for the extreme positive pole: this extreme includes
students with low performance in school grades, but who show high performance in
all test scores. All the values of the variables contribute to this factor, especially the
teachers’school grades for Social Sciences and Natural Sciences, as well as the test
scores on Math. Table 2also presents the description of the second factor, along
with the proof values.
Undoubtedly, the most outstanding characteristic of this second factor is the
identification of a direct opposition between the measures of performance from the
teachers’perspective and the measures according to the standardised tests. For this
reason, we consider the second factor as that in which these two types of evaluations
contradict each other: that which is contradictory between the evaluations.
When considering the results as a whole there is a clear similarity among the tra-
jectory of the standardised tests, on the one hand, and among the teachers’school
grades, on the other hand. The trajectory of the standardised tests in all the areas are
quite similar, starting from the extreme negative pole in factor 1 and the extreme
negative pole in factor 2 which represent the lowest performances, and arriving at
the extreme positive poles in factor 1 and 2, which represent the highest perfor-
mances. Teachers’school grades have a similar behaviour regardless of the subject
area considered starting from the extreme negative pole in factor 1 and the extreme
positive pole in factor 2, where the lowest school grades are located, and arriving at
the extreme positive pole in factor 1 and extreme negative pole in factor 2, where
the best school grades are located. This shows a structure of associations that
favours the type of evaluation (or, one may think, the source of information) over
the specific features of a particular subject area.
Concerning the tests on competencies, these results indicate a clear association
among the three standardised tests for Math, Language and Sciences, which allows
us to presume that there is an underlying common factor in the three tests. This
would explain the similarity mentioned above. To some extent, the three tests evalu-
ate the same issues.
Now, with respect to the obvious similarity among the evaluations of the four
teachers of the subject areas concerned, the results indicate that they evaluate the
same issue in each student rather than the specific features of each subject area. In
fact, students tend to be evaluated in the same way by their four teachers: always
‘superior’in all areas, always ‘acceptable’or always ‘deficient’. Again, the specific
Educational Psychology 9
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

features of the areas and the possibilities of a student standing out in a specific
content and achieving an acceptable or poor performance in others diminish here.
What could be the common factor in the four evaluations of the teachers? Our most
immediate hypothesis is that teachers evaluate certain characteristics of school beha-
viour together with the student’s performance in an area, regardless of the specific
area involved.
Variables that characterise educational performance
The analysis of the associations between educational performance and the variables
potentially associated was done including them as illustrative variables in the
factorial plane of educational performance.
The analysis of the variables that characterise the first factor reveals a great deal
of illustrative values with meaningful associations with the factor. The negative pole
of the factor, which represents low performance in all the indicators, is characterised
by: (a) a very field-dependent or field-dependent cognitive style, (b) neutral or nega-
tive attitudes towards Math, (c) a high level of repetition (two or more grades), (d) a
high level of overage (two or more years), (e) no use of computers, (f ) absentee
behaviour (missing classes or frequently arriving late to class), (g) low socio-cultural
levels (parents who did not finish primary school), (h) low or very low socioeco-
nomic strata, (i) females and (j) a large and extended family (more than eight rela-
tives). Other variables showing a meaningful, though less strong association with
low performance are: (k ) having studied only in the present school and (l) showing
relative ‘well behaved’behaviours (not running away from class with a few rare
exceptions).
On the other hand, the following values are in the extreme positive sense of
factor 1, which is characterised by high performance in all the evaluations: (a) very
field-independent or field-independent cognitive style, (b) positive attitudes concern-
ing Math and Sciences, (c) no repetition, (d) high use of computers, especially at
home, and (e) relatively high socioeconomic levels (parents with higher education)
and (f ) relatively high socioeconomic levels (lower-middle class). Other factors also
associated with this direction are: (g) non-absentee school behaviours, (h) different
Colombian cultural origins, such as the ‘Andean-Santandereana’or the ‘Paisa’;as
defined by Gutiérrez de Pineda (1975), (i) underage (against overage, that is, two
years ahead, (j) males and (k ) living in small family groups. See Table 3.
A quick look at these results shows that cognitive style is a variable that highly
discriminates against the differences in the first factor of educational performance;
the one in which all the evaluations agree: field-independent students achieve better
results. Cognitive style is followed by positive attitudes towards math and natural
sciences, school history (low repetition and underage), high cognitive training (use
of computers), low absenteeism and, finally, students’gender and some of their
family and cultural aspects. See Table 3.
Regarding the second factor, something very interesting happens. It reverses the
two types of evaluations: the standardised test scores and the teachers’school grades
go in opposite directions.
The extreme negative pole of factor 2, which is characterised by high levels in the
school grades, but low levels in the standardised test scores, discriminates against stu-
dents who are (a) females, (b) do not repeat grades, (c) conform with the school disci-
pline (they do not run away from classes or school), (d) observe non-absentee school
10 C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Table 3. Characterisation of the factor 1 by illustrative variables.
Variable Value V-test n
Repetition Yes −6.87 692
Cognitive style Very field-dependent −5.95 150
Attitudes towards math Neutral −5.59 1014
Number of years repeated One year −5.41 527
Overage Two years behind −5.34 154
Computer time at school None −4.79 442
Cognitive Style Field-dependent −4.31 429
Computer at home No computer at home −4.24 1562
Grade Eighth −4.09 870
Missing classes Very frequently −3.43 241
Number of years repeated Two or more years −3.14 157
Arriving late to class Very frequently −3.08 249
Attitudes towards Math Negative −2.92 146
Father’s level of education Incomplete primary school −2.89 396
Mother`s level of education Incomplete primary school −2.81 359
Attitudes towards Natural Sciences Negative −2.78 86
Socioeconomic stratum Low −2.68 582
Gender Feminine −2.68 933
Overage One year behind −2.60 380
Family size Eight relatives or more −2.55 194
Missing classes Frequently −2.39 211
Computer time at school No computer at school −2.31 111
Running away Rarely −2.12 250
School mobility One school −2.08 389
Paternal grandmother Presence of paternal grand mother −2.04 146
Socioeconomic stratum Very low −2.03 126
Central zone
Paternal grand-mother Absence of paternal grand-mother 2.04 1709
Overage One year ahead 2.18 509
Cultural origin of maternal grand-mother Andean 2.21 447
Computer at school There are computers at school 2.31 1744
Cultural origin of maternal grand-father Paisa 2.32 78
Missing classes No 2.34 1027
Family size 1–3 relatives 2.35 419
Computer time at home 3–6 h per week at home 2.4 79
Overage Same as the median age 2.43 726
Attitudes towards Natural Sciences Positive 2.44 1123
Gender Masculine 2.68 922
Cultural origin of paternal grand-mother Andean 2.72 418
Arriving late Never 2.82 989
Over age Two years ahead 2.91 59
Cultural origin of paternal grand-father Andean 3.28 378
Running away Never 3.43 1281
Socioeconomic stratum Middle –low 3.47 795
Father’s educational level Father’s higher education 4.09 286
Grade Tenth 4.09 985
Computer at home There are computers at home 4.24 293
Mother’s educational level Mother’s higher education 4.24 211
Computer time at home More than 6 h per week 5.04 84
Cognitive style Field independent 6.27 414
Number of years repeated None 6.87 1171
Attitudes towards Math Positive 7.38 695
Cognitive style Very field-independent 7.71 91
Educational Psychology 11
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

behaviour, (e) have positive attitudes towards Mathematics, Language and Natural
Sciences and (f) did not attend preschool. Although, in less significant values, this
pole is also characterised by cultural aspects, such as: (g) middle levels of socio-
cultural background, (h) not very extended families (from 4 to 5 relatives) and (i) an
Andean-Santandereana cultural origin. See Table 4.
The other extreme, that is, the extreme positive pole of the factorial plane, which
is characterised by high levels in the standardised test scores, but low levels in the
school grades, discriminates against students who are (a) males, (b) repeating several
grades, (c) being against school authority (‘undisciplined’), (d) some levels of absen-
teeism, (e) middle-high socio-cultural environments, (f ) preschool, (g) having nega-
tive attitudes towards all subjects, (h) field-independent cognitive style, (i) high
levels of school mobility and (j) middle socioeconomic class. See Table 4.
A general perspective of the second factor identifies gender as the variable that
best polarises the two extremes of the factorial plane in the vertical axis. Thus, while
girls obtain low performance in the standardised tests and high performance in the
school grades, boys obtain quite the opposite: high performance in the standardised
test and low performance in the teachers’school grades. Gender is followed by vari-
ables about school history (repetition) and behaviour at school (discipline, attitudes,
absenteeism) in a particular direction: higher repetition, more disciplinary problems,
more negative attitudes towards subjects and higher absenteeism are associated with
high performance in the standardised tests and with low performance in the school
grades.
These results make it clear that there is a close relationship between cognitive
style and educational performance: highly field-independent students have greater
probabilities to attain high-performance levels both in standardised tests and in indi-
vidual evaluations made by teachers.
Our results become part of the already huge list of studies that have found out
this association in similar situations (Ku & Soulier, 2009; Tinajero & Páramo,
1997). Most of them have considered the results as evidence of unfair conditions in
the educational systems. That is, while the premise that all the individuals learn in
the same way is assumed to be valid, the system favours the performance of just
one group of students and, at the same time, creates unfavourable conditions for a
high proportion of their users, with respect to what they can learn.
Part B. Path analysis model
The main purpose of studies like this is to explore the associations between a set of
variables and educational performance. The underlying idea in carrying out these
studies is that some of these associated factors determine educational performance to
a certain extent; therefore, when affecting the levels of these factors, we are indi-
rectly affecting performance. However, this idea has not had empirical basis for any
of the studies on factors associated with performance that have been carried out in
Colombia. This is due to the fact they are all correlational studies (Gaviria &
Barrientos, 2001; Piñeros & Rodríguez, 1999). In order to deal with this level of
description, it is necessary to state a path analysis model that can establish the
specific directions of the association between variables and performance. With this,
we try to predict educational performance in terms of the variables that explain it.
As far as we know, models of this sort have not been used to predict educational
performance in Colombia.
12 C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Table 4. Characterisation of the factor 2 by illustrative variables.
Variable Value V-test n
Gender Feminine −7.23 933
Computer time at school Two hours per week −5.28 537
Repetition No −5.07 1163
Running away Never −5.03 1281
Number of years repeated None −4.81 1171
Attitudes towards Math Positive −4.37 695
Leaving school without permission No −4.36 1698
Missing classes No −4.35 1027
Detentions None −4.27 1698
Attitudes towards language Positive −3.81 833
Cultural origin of the mother Andean −3.32 563
Father’s educational level Incomplete primary school −3.26 396
Pre-school No −2.92 573
Attitudes towards Natural Sciences Positive −2.75 1123
School mobility Two schools −2.74 322
Mother’s educational level Mother’s complete primary school −2.65 314
Overage One year ahead −2.60 509
Family size 4–5 relatives −2.52 859
Cultural origin of the father Andean −2.33 570
Family size 4–5 relatives −2.32 729
Mother’s educational level Mother’s incomplete primary school −2.23 359
School mobility One school −2.09 369
Computer at home No computer at home −2.02 1562
Central zone
Overage Average age 2.01 726
Computer at home Yes 2.02 293
Father’s educational level Father’s university education 2.03 286
Number of years repeated Two years or more 2.14 157
Running away Frequently 2.16 140
Computer time at school 1 h per week 2.36 552
Socioeconomic stratum Middle 2.46 52
Mother’s cultural origin Bogotá 2.57 665
Attitudes towards Social Sciences Negative 2.59 201
Computer time at school None 2.63 442
Attitudes towards Math Neutral 2.73 1014
Arriving late Very frequently 2.74 249
School mobility Four schools 2.78 353
Attitudes towards Math Negative 2.81 146
Cognitive style Field-independent 2.82 414
Attitudes towards language Neutral 2.91 917
Pre-school Yes 2.92 1282
Mother’s educational level Mother’s complete secondary school 2.99 360
Number of years repeated One year 3.83 527
Missing classes Very frequently 3.97 241
Attitudes towards Natural Sciences Negative 4.02 86
Leaving class Frequently 4.24 184
Detentions Yes 4.27 157
Running away Yes 4.36 157
Repetition Yes 5.07 692
Gender Masculine 7.23 922
Educational Psychology 13
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

The method used to construct the contingent model is the analysis of structural
equations the result from which is presented in Figure 1. This model cannot be con-
sidered as a complete one because it only includes some of the variables that we
have considered. However, the model developed shows the technical characteristics
expected with respect to convergence, reliability and stability (the indicators of the
test of the model will be seen later on).
The first step to define the model is to state the constructs that we have consid-
ered. Those constructs are:
•Standardised test scores. The results of the standardised tests in Math, Lan-
guage and Natural Sciences are a common construct, which have been called
‘test scores’. The reliability of this construct is .58, which can be considered
quite good (the indicators of this construct are shown in Figure 1).
•Teachers’school grades. The teachers’school grades in all areas form a com-
mon construct, which has been called ‘school grades’. Again, the reliability
level of this construct can be considered quite good (.63).
•Classroom behaviour. Having not been suspended from school and having not
run away from a common construct, which has been called ‘classroom beha-
viour’. Perhaps, this is the weakest construct in the model because it is only
defined by two ‘dummy’variables. The reliability of this construct is .41,
which can be considered as an average value.
•Cognitive style. We think that cognitive style is a construct only useful for the
results in the EFT test. This construct is considered to be fixed.
As the constructs have already been defined, now it is necessary to state contin-
gent associations. As can be observed in the figure, a field-independent cognitive
Figure 1. Estructural equations model.
14 C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

style, which is indicated by the results of the EFT test, predicts the levels of the
other three constructs: test scores, school grades and discipline. On the other hand,
the construct referring to discipline can predict the construct called school grades to
a certain extent.
In terms of the equations which describe the relationships, it can be observed
that (1) greater field independence predicts higher test scores (coefficient .543); (2)
greater field-independence directly predicts higher school grades (coefficient .384);
(3) greater field-independence negatively predicts classroom behaviour (negative
coefficient −.174); that is, greater field-independence produces greater classroom
misbehaviour; (4) classroom behaviour by itself predicts higher school grades (co-
efficient .632). Thus, there are two different directions in the association between
style and school grades: the direct path leads to a positive association, that is, greater
field-independence produces higher school grades; the indirect path leads to a nega-
tive association, that is, greater field-independence produces indiscipline and, conse-
quently, lower school grades. The index values obtained for the model [χ
2
= 25.66,
p= .31, RMSEA = .017, GFI = .997, AGFI = .994, AIC = .048] indicate that it fits
the data well, since it meets the criteria: 0 < χ
2
<2, df = 56, .05 < p< 1.00,
0 < RMSEA < .05, .95 < GFI < 1.00, .90 < AGFI < 1.00.
The residual values also show certain correlations worth mentioning: especially
those that exceed the constructs. In fact, Math standard test scores and Math school
grades were correlated. This is the only case in which the two types of evaluations
show certain associations. On the other hand, it is interesting to note certain relation-
ships between the Language standard tests scores and the school grades given by
the Social Sciences’teachers. We have no clear explanation for these associations.
Discussion
The results indicate complex relationships among the different levels of educational
performance, cognitive style and the variables associated with this relationship. In
the first factor, field-independent students, especially extremely field-independent
ones, show better test scores, and less repetition, are less likely to be older than their
classmates, and have more positive attitudes towards Math and Natural Sciences. In
the opposite extreme of the same factor, highly field-dependent students show lower
test scores, lower school grades, higher levels of repetition and are more likely to be
older than their classmates. This could indicate that the educational system is biased
in the sense that it favours the performance of the individuals of only one of the
cognitive styles: FI students, who seem to be better prepared for the system
demands. Thus, our hypothesis is still valid: Bogotá’s public educational system
favours the performance of FI students in two related ways: (a) it promotes a better
learning for them, and (b) it filters those students with a FD cognitive style, keeping
them in lower grades and finally excluding them from the system.
Indeed, since field independence has been associated to psychological variables
that are predictors of performance, such as intelligence (Roberge & Flexer, 1981), or
intrinsic motivation (Goodenough, 1976), it could be the case that a model contain-
ing some of these variables might eliminate this cognitive style as a predictor. The
only way to rule out this hypothesis is by controlling these variables during the
analysis. With respect to intelligence, as has been said above, previous research has
shown that the relationship between field independence and academic achievement
keeps on, even after controlling this variable. (Tinajero & Páramo, 1997). Similar
Educational Psychology 15
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

studies should be done in the future to control other variables associated to field
independence.
There are more results. The analysis of correspondences shows that some of the
variables which characterise the second factor, also characterise the first one. How
should these two apparently contradictory results be reconciled? There should be an
answer for each variable. For example, let’s consider gender. Considering the first,
boys have higher test scores and school grades. On the other hand, considering the
second factor, girls have higher school grades and lower test scores. In short: boys
have higher test scores, but the evaluations made by teachers favours girls. In spite
of the fact that girls show lower test scores, teachers evaluate them better.
It is important to mention that, in this landscape of multiple associations, alterna-
tive interpretations should also be considered. We know that cognitive style is
associated with gender. Though in mild levels, men tend to be more field-indepen-
dent than women. Thus, if we discuss the results emphasising gender, rather than
cognitive style, we could argue that since many teachers are women, and male stu-
dents show more disruptive behaviours in class than female students, female teach-
ers tend to favour female students by giving them higher grades. The point is that
gender, disruptive behaviours and cognitive style are related in such a way that it is
very difficult to know which of these three variables has the heaviest load in the
explanation of the results.
The variable of repetition shows similar contradictory behaviour. According to the
first factor, this variable is associated with lower test scores and with lower school
grades. According to the second factor, it is associated with lower school grades, but
with higher tests scores. The result of the synthesis of these two factors could indicate
that repetition seems to be highly associated with lower school grades and, that
although it is also somewhat associated with lower test scores, this association is not
very strong. In other words, the general analysis shows that the evaluations made by
teachers determine their repetition. Other variables that show paradoxical tendencies
are the ones related to attitudes towards language and social sciences: better attitudes
show higher school grades, but lower test scores.
According to cognitive style, these apparently confused results are quite
coherent. As was observed in the second factor, students with certain levels of field-
independence are located in the positive extreme of this factor. That is, although
they continue to show higher test scores, they do not have very good school grades.
At the same time, they have repeated courses, they have been suspended from
school, they have frequently run away from school and, in general, they have been
in a higher number of schools. In contrast, field-dependent students, who show
lower test scores, are comparatively better evaluated by their teachers, they have
repeated fewer grades, they are not frequently punished with suspensions, they have
never run away from school and they have been in just a few schools.
Therefore, the analysis of these characteristics, based on the social behaviour of
each cognitive style, leads us to conclude that teachers consider certain more frequent,
autonomous behaviours in field-independent individuals as reprehensible. However,
this does not affect their scores in standardised tests. Thus, we conclude that we
should have a second hypothesis, which can apparently be paradoxical if we compare
it with the first one: the system does not favour field-independent students with respect
to their school history, although such students tend to perform better academically.
To sum up, we have found a clear contradiction inherent in Bogotá’s educational
system: it favours field-independent students with respect to their standardised tests
16 C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

achievements, but it does not favour them with respect to their school record
because of their social behaviour. Regarding those field-dependent students, it is
clear that they do not achieve the levels of achievement expected. However, these
students easily progress within the educational system because their social behaviour
allows them to easily adjust themselves to a social, hierarchical and asymmetric
system such as the school.
This reflects a reality that is not the best one: As we have already stated, any stu-
dent, regardless of his/her cognitive style should attain the basic achievements,
which are the goals of the educational system. The fact that only field-independent
students achieve them provides clear evidence to the idea that there are no develop-
ment opportunities for field-dependent students, or that standardised tests favour the
performance of field-independent students.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Christian Hederich-Martínez http://orcid.org/0000-0003-1080-9973
Angela Camargo-Uribe http://orcid.org/0000-0003-3611-810X
References
Arthur, W., & Day, D. V. (1991). Examination of the construct validity of alternative
measures of field dependence/independence. Perceptual and Motor Skills, 72, 851–859.
Atkinson, S. (2004). A comparison of pupil learning and achievement in computer aided
learning and traditionally taught situations with special reference to cognitive style and
gender issues. Educational Psychology, 24, 659–679.
Bowlin, D. A. (1988). An investigation of the relationship between field dependent/indepen-
dent cognitive style and sex, IQ, academic achievement, curriculum track selection, and
hemispheric preference in high school seniors (Doctoral dissertation). University of
Pittsburg, Dissertation Abstracts International,49, 1405-A.
Chapelle, C., & Roberts, C. (1986). Ambiguity tolerance and field independence as predictors
of proficiency in English as a second language. Language Learning, 36,27–45.
Claeys, W., & Mandosi, M. (1977). Extended family as an environmental correlate of the test
performances of eleventh grade Zairese subjects. In Y. H. Poortinga (Ed.), Basic problems
in cross-cultural psychology (pp. 12–16). Amsterdam: Swets and Zeitlinguer.
Colombia. Departamento Nacional de Planeación. (1999). Un plan educativo para la paz
[An educational plan for peaceful times]. Bogotá: Departamento Nacional de Planeación.
Colombia. Instituto Colombiano para el Fomento de la Educación Superior. (2011). Examen
de estado de la educación media. Resultados del período 2005 –2010 [Government test
for secondary education. Results from the 2005–2010 period]. Bogotá: Instituto Colom-
biano para el Fomento de la Educación Superior.
Coventry, L. (1989). Some effects of cognitive style on learning UNIX. International Journal
of Man-Machine Studies, 31, 349–365.
Fergusson, L. C. (1992). Field independence and art achievement inmeditating and nonmedi-
tating college students. Perceptual and Motor Skills, 75, 1171–1175.
Forns-Santacana, M., & Amador-Campos, J. A. (1990). Association of scores on McCarthy
scales with field dependence/independence for seven-year-old Spanish children. Percep-
tual and Motor Skills, 70, 1291–1296.
Gaviria, A., & Barrientos, J. (2001). Determinantes de la calidad de la educación en Colom-
bia [Quality of education determinants in Colombia]. Archivos de Economía (Documento
159). Bogotá: Departamento Nacional de Planeación.
Educational Psychology 17
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Goodenough, D. R. (1976). The role of individual differences in field dependence as a factor
in learning and memory. Psychological Bulletin, 83, 675–694.
Gutiérrez de Pineda, V. (1975). Familia y Cultura en Colombia [Family and culture in
Colombia]. Bogotá: Biblioteca Básica Colombiana. Instituto Colombiano de Cultura-
Colcultura.
Hederich, C. (2007). Estilo Cognitivo en la dimensión de dependencia-independencia de
campo. Influencias culturales e implicaciones educativas [Cognitive style in the field
dependence-independence dimension. Cultural influences and educational implications].
Bogotá: Universidad Pedagógica Nacional.
Hederich, C., & Camargo, A. (1999). Estilos Cognitivos en Colombia. Resultados en cinco
regiones culturales [Cognitive styles in Colombia. Results from five cultural regions].
Bogotá: Universidad Pedagógica Nacional, Colciencias.
Hederich, C., Camargo, A., Guzmán, L., & Pacheco, J. C. (1995). Regiones Cognitivas en
Colombia [Cognitive regions in Colombia]. Bogotá: Universidad Pedagógica Nacional,
Colciencias.
Ku, D. T., & Soulier, J. S. (2009). The effects of learning goals on learning performance of
field-dependent and field-independent late adolescents in a hypertext environment.
Adolescence, 44, 651–664.
McKenna, F. P. (1984). Measures of field dependence: Cognitive style or cognitive ability?
Journal of Personality and Social Psychology, 47, 593–603.
Nagata, H. (1989). Judgments of sentence grammaticality and field-dependence of subjects.
Perceptual and Motor Skills, 69, 739–747.
Paramo, M. F., & Tinajero, C. (1990). Field dependence/independence and performance in
school: An argument against neutrality of cognitive style. Perceptual and Motor Skills,
70, 1079–1087.
Páramo, M. F., & Tinajero, C. (1992). Influencia de la estructura normative familiar sobre el
estilo cognitivo dependencia-independencia de campo: un estudio prospective [A
prospective study on the influence of family environment structure on field dependence-
independence cognitive style]. Infancia y Aprendizaje, 15,89–98.
Piñeros, L. J., & Rodríguez, A. (1999). School inputs in secondary education and their
effects on academic achievement: a study in Colombia (LCSHD Paper Series No. 36).
Washington, DC: World Bank Human Development Department.
Riding, R. (2000). Cognitive style: A review. In R. Riding & S. Rayner (Eds.), International
perspectives in individual differences. Volume 1: Cognitive styles ( pp. 315–344). Stamford,
CT: Ablex Publishing Corporation.
Roberge, J. J., & Flexer, B. K. (1981). Re-examination of the covariation of field indepen-
dence, intelligence and achievement. British Journal of Educational Psychology, 51,
235–236.
Rollock, D. (1992). Field dependence/independence and learning condition: An exploratory
study of style vs. ability. Perceptual and Motor Skills, 74, 807–818.
Sawa, H. (1966). Analytic thinking and synthetic thinking. Bulletin of Faculty of Education,
Nagasaki University, 13,1–16.
Schmidt, C. P., & Lewis, B. E. (1987). Field-dependence/independence, movement-based
instruction and fourth graders’achievement in selected musical tasks. Psychology of
Music, 15,117–127.
Tinajero, C., & Páramo, M. F. (1997). Field dependence-independence and academic achieve-
ment: A re-examination of their relationship. British Journal of Educational Psychology,
67, 199–212.
Tinajero, C., & Páramo, M. F. (1998). Field dependence-independence and strategic learning.
International Journal of Educational Research, 29, 251–262.
Tinajero, C., Lemos, S., Araujo, M., Ferraces, M., & Páramo, M. (2012). Estilo cognitivo e
estrategias de aprendizagem em estudantes universitarios brasileiros: repercussoes no
rendimento academico [Cognitive style and learning strategies as factors which affect
academic achievement of brazilian university students]. Psicologia: Reflexao & Critica,
25, 105–113.
Van Blerkom, M. L. (1988). Field-dependence, sex role, self-perceptions and mathematics
achievement in college students: A closer examination. Contemporary Educational
Psychology, 13, 339–347.
18 C. Hederich-Martínez and A. Camargo-Uribe
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015

Witkin, H., & Goodenough, D. (1981). Estilos Cognitivos. Naturaleza y orígenes [Cognitive
styles. Essence and origins]. Madrid: Ediciones Pirámide.
Witkin, H., Oltman, P. K., Raskin, E., & Karp, S. S. (1971). A manual for the embedded
figures test. Palo Alto, CA: Consulting Psychologists Press.
Educational Psychology 19
Downloaded by [Universidad Del Norte] at 08:53 13 October 2015