Eur J Psychol Educ (2014) 29:577–601

DOI 10.1007/s10212-014-0214-9

Are teachers’ approaches to teaching responsive

to individual student variation? A two-level structural
equation modeling

Pedro Rosário & José Carlos Núñez & Guilermo Vallejo &
Olímpia Paiva & António Valle & Sonia Fuentes &
Ricardo Pinto

Received: 30 October 2013 / Revised: 26 January 2014 / Accepted: 25 February 2014 /

Published online: 11 March 2014
# Instituto Superior de Psicologia Aplicada, Lisboa, Portugal and Springer Science+Business Media Dordrecht
2014

Abstract In the framework of teacher’s approaches to teaching, this study investigates the
relationship between student-related variables (i.e., study time, class absence, domain knowl-
edge, and homework completion), students’ approaches to learning, and teachers’ approaches
to teaching using structural equation modeling (SEM) with two independent data samples. The
participants were 61 biology teachers and their corresponding 1,518 high school students (12th
grade). The first sample was used to fit the model, and the second sample was used to analyze
the consistency of the data derived from the first sample. Using a two-level SEM analysis, we
established whether the effects found at the individual level varied significantly at class level.
The students’ approaches to learning were related to the teachers’ approaches to teaching as a
function of the hypotheses established in the model, although the effect size was smaller than
expected. However, approximately 48 % of the variance of the surface approach and 46 % of
the deep approach sat at class level. At the individual level, the results of this study suggest that
students’ approaches to learning significantly explain their teachers’ approaches to teaching
and, thus, constitute important contextual variables. At the class level, the way students learn
appears to be closely associated with class-related variables. Our data stresses the importance
of promoting educational opportunities (e.g., school-based courses) for teachers to reflect upon
the teaching methodologies used in class.

Keywords Approaches to learning . Approaches to teaching . Two-level SEM . High school

P. Rosário (*) : O. Paiva : R. Pinto

Universidade do Minho, Braga, Portugal
e-mail: prosario@psi.uminho.pt

J. C. Núñez : G. Vallejo
Universidad de Oviedo, Asturias, Spain

A. Valle
University of A Coruña, Galicia, Spain

S. Fuentes
Universidad Central de Chile, Santiago, Chile
578 P. Rosário et al.

Introduction

Prosser and Trigwell (1999) and Prosser et al. (1994) in the end of the last century initiated a
line of research focusing on teachers’ approaches to teaching. These authors investigated
whether teachers approached teaching in qualitatively different ways and found that there are
two different approaches to teaching: (a) the Information Transmission/Teacher-Focused
(ITTF) approach and (b) the Conceptual Change/Student-Focused (CCSF) approach
(Lindblom-Yl nne et al. 2006; Prosser and Trigwell 1999, 2006; Ramsden 2003; Rosário
et al. 2013a).
Teachers who generally focus their teaching on the transmission of information can be
described as teacher-centered. Their teaching practices are organized to facilitate the process of
information transmission to improve their students’ comprehension and academic results.
These teachers are frequently meticulous about the technical aspects related to the teaching
process (e.g., planning classes in detail and providing the students with notes or summaries of
the contents) and believe that the organization of the class contents and their own mastery of
teaching competences play a decisive role in their students’ learning (Trigwell et al. 1999;
Ramsden et al. 2007; Rosário et al. 2013a). Other teachers are more likely to display a student-
centered approach to teaching. This approach consists of a teacher’s preference for involving
students in an active process of knowledge building. Focused on supporting the students’
conceptual change, these teachers take into account their students’ prior knowledge and
organize their classes to facilitate the students’ knowledge building (e.g., by asking questions
that evoke complex cognitive processes, stimulating debates in class, or discussing the
conclusions of reports). The teachers who are more likely to adopt a student-centered approach
believe that learning depends significantly on the students’ active role in the interaction with
the specific learning context than on the transmission of knowledge, although these teachers
also consider knowledge transmission important (Prosser et al. 1994; Rosário et al. 2013a;
Trigwell and Prosser 2003).
Although the relationship between students’ approaches to learning and their teachers’
approaches to teaching is intuitive, only few studies have addressed this relationship directly
(Ramsden et al. 2007; Rosário et al. 2013b).
In the present study, we attempted to answer the question posed by Stes et al. (2008), Why
do teachers teach the way they do?, by analyzing how the way teachers teach (teachers’
approaches to teaching) is explained by students’ characteristics. Specifically, we examined the
degree to which teachers’ approaches to teaching were responsive in the face of high school
students’ individual variation (i.e., approaches to study, domain knowledge, homework com-
pletion, study time, and class absence) using two-level structural equation modeling (SEM).

Research about teachers’ approaches to teaching

One of the first studies under this framework analyzed the relationship between approaches to
teaching and approaches to learning (Trigwell et al. 1999) using an initial version of the
Approaches to Teaching Inventory (ATI). The data revealed a relationship between the ITTF
approach and the students’ surface approach (SA) to learning, although no relationship was
found between the teachers’ CCSF approach and the students’ deep approach (DA) to
learning. However, the latter-mentioned findings proved to be worthwhile, and directions for
future research on the relationships between approaches to teaching and approaches to learning
were strengthened. Subsequent investigations within this framework were focused on the
evaluation of the relationship between approaches to teaching, contextual variables (e.g.,
discipline of teaching and class size), and teaching (e.g., gender, teacher experience). In fact,
Teachers’ approaches to teaching 579

there is more data about how the teachers’ characteristics can influence their approaches to
teaching than about how these approaches to teaching may adapt as a function of the teaching
context (e.g., teaching discipline, class size; see Lindblom-Yl nne et al. 2006; Lueddeke 2003;
Nevgi et al. 2004; Stes et al. 2008).
Research indicates that the ways teachers teach partially predict the students’ approaches to
learning and the students’ quality of learning (Prosser et al. 2003; Rosário et al. 2013b;
Trigwell et al. 1999). However, the opposite relationship has barely been investigated, and
we believe it to be an important research topic. High school classes include students varying in
domain knowledge, skills, and motivation. Additionally, in order to promote students’ en-
gagement with school, teachers are expected to adapt their approach to teaching according to
their class.

What student-related variables are related to teachers’ approaches to teaching?

Teachers’ approaches to teaching are responsive to a large set of student and context variables.
The current research investigates student variables such as students’ approaches to learning,
study time, class absence, domain knowledge, and homework completion as contextual
variables in relationship to teachers’ approaches to teaching.
A strong body of research states that students tend to approach school tasks using either a
DA or an SA. A DA occurs when students engage themselves in understanding the meaning of
the contents, and a SA occurs when students are extrinsically motivated (Biggs et al. 2001;
Entwistle 2009; Richardson 2013; Rosário et al. 2010). Students who adopt a DA are driven
by an intention to comprehend and to maximize understanding. High-quality outcomes are
generally related to this approach (Trigwell and Prosser 1991). In contrast, students using SAs
to learning put little effort and show a low engagement in completing the tasks, as tasks are
seen by as an external imposition. An example could be using the strategy of rote learning to
study the contents assigned without understanding them. The SA is then frequently related to
low-quality outcomes (Trigwell and Prosser 1991).
To our knowledge, the analysis of how a student’s approach to learning can influence
teachers’ approaches to teaching is still quite unexplored. For example, it seems plausible that
whenever students display in class a DA to learning (by asking questions to organize mind
maps and deepen the understanding of the contents), those academic behaviors are likely to
promote a teaching approach that is student-centered, to help students achieve a comprehen-
sive learning. However, this intuitive relationship between these two variables needs to be
investigated. Our research can add the literature as a preliminary study.
Domain knowledge describes general knowledge within a certain discipline of teaching
(Alexander and Jetton 2000). Studies show that students who have extensive domain knowl-
edge are more likely to be intrinsically motivated to learn, to self-set goals, and to display
behaviors for monitoring their learning and performance (Willoughby et al. 2003). Alexander
et al. (1994) analyzed students’ comprehension of two different papers used in the physics
class, one of the papers being more technical than the other, and concluded that the variable
domain knowledge contributed significantly to task performance. Alexander et al. (1994,
1997) found that individuals investigate more about domains for which they master and master
domains in which they are more interested for.
Although study time is commonly considered a good predictor of academic success, some
empirical data do not support such a clear and direct relationship (e.g., Gortner-Lahmers and
Zulauf 2000; Plant et al. 2005). For example, in the field of homework, even though the time
spent on homework is often used as an indicator of academic success, the results of the
research are scarcely conclusive. Whereas some reviews of previous works report a positive
580 P. Rosário et al.

relation between time spent on homework and achievement (e.g., Cooper 1989; Cooper et al.
2006; Cooper and Valentine 2001; Walberg 1991), other investigators report that this relation is
either rather low or even negative (e.g., De Jong et al. 2000; Núñez et al. 2013, in press;
Trautwein 2007; Trautwein et al. 2009). Therefore, it is important to analyze the behavior of
the variable named study time within a structured equation model.
Regarding homework, Kitsantas and Zimmerman (2009) noted that the quality of college
students’ homework is significantly related to the development of better study habits.
According to the literature, completing a reasonable amount of homework on a daily basis
can help to develop study habits that facilitate learning and, ultimately, improve academic
achievement (Cooper et al. 2006; Epstein and Van Voorhis 2001; Xu and Corno 2006).
Nevertheless, the strength of the relation between the amount of homework and learning
processes is not yet clear (e.g., Dettmers et al. 2009; Trautwein and Koller 2003).
Class absence, another variable that was included in the current model, is a complex
concept that requires clarification to deepen our comprehension of its causes and consequences
(Jonanssen 2011), especially as this variable is considered a key concept in the analysis of
academic failure (Ball and Connolly 2000; Reid 2006).
The investigated four student variables may impact how teachers teach, although this
impact is expected to be indirect, through students’ approaches to learning. Results from
previous research show that these four variables are related to approaches to learning (Rosário
et al. 2010; Struyven et al. 2006), but there is little information about their role regarding
approaches to teaching (Rosário et al. 2013b).

Study aims

By means of the SEM, the present study analyzes how teachers’ approaches to teaching are
responsive to student variables (i.e., study time, class absence, domain knowledge, homework
completion, and the students’ approaches to learning). The hypothesized model (see Fig. 1)
uses students’ approaches to learning in relation to other student variables and teachers’
approaches to teaching. This model was studied using two samples of data (a calibration
sample and a validation sample). The first sample was used to fit the model, and the second
sample was used to analyze the consistency of the fit of the model in the first sample.
Multigroup analyses were performed as a cross-validation strategy. Besides, as our data related
to a hierarchical structure (students nested in classes), a two-level SEM analysis has been
conducted.
We followed this design because the effect of class level can only be considered informative
once the sample includes at least 80 units of analysis (Maas and Hox 2005). If the numbers of
classes were higher than 150, it would be better to conduct only a two-level SEM analysis.
Considering the above-mentioned data provided by the investigations, the following
predictions have been established in the model (see Fig. 1):

1. Study time, domain knowledge, and homework completion influence the DA positively
and the SA negatively. In turn, a high rate of class absence has a positive impact on the SA
and a negative impact on the DA.
2. Students’ SA affects the ITTF approach positively and the CCSF approach negatively. In
turn, the ITTF approach is negatively determined by the DA.
3. The CCSF approach is positively influenced by students’ DA to learning and negatively
influenced by students’ SA to learning.
4. Study time, class absence, domain knowledge, and homework completion have an indirect
influence on CCSF and ITTF through their direct effect on the SA and DA to learning.
Teachers’ approaches to teaching 581

Fig. 1 Structural equation model for the hypothesis of the relationship between the approach to teaching and the
contextual variables (approach to learning, homework completion, domain knowledge, study time, and class
absence)

Method

Participants

In Portugal, high school is organized into three grade levels (10th, 11th, and 12th grades) or
15-, 16-, and 17-year-old students, respectively. The ten high schools enrolled in this inves-
tigation were located in an urban school district in the north of Portugal and were randomly
chosen from the pool of 45 possible high schools. The participant classes in each school were
also randomly chosen. Overall, the families of these students can be considered lower middle
class as evidenced by the high percentage of students who received free or reduced-price lunch
(39.4 %, data collected in the students’ office of the participating schools). All of the high
school students and teachers from the participating schools were invited to participate in this
investigation. Teachers sent an e-mail expressing their willingness to participate, and students
who volunteered for the study presented their parents’ authorizations.
582 P. Rosário et al.

The study involved 1,518 high school students (12th grade) from 10 public schools; 624
(41.5 %) were male and 880 (58.5 %) were female. The student ranged between 16 and
19 years of age (M=17.2, SD=.69). The majority of the participants were Caucasian (99.6 %),
.3 % of the students were African, and .1 % were Asian. The ethnic characteristics of this
sample were highly similar to those of the entire high school.
The 61 biology teachers of the 1,518 students were also enrolled: 12 (19.7 %) male and 49
(80.3 %) female, all Caucasians. The teachers’ age varied from 26 to 61 years (M=46.9, SD=
9.2). The teachers’ teaching experience ranged from 2 to 36 years (mean of 23.5 years; SD=
9.6).
The sample was randomly split into two parts to cross-validate the model, a calibration
sample, and a validation sample. Six students were eliminated from the calibration
sample, and eight students were eliminated from the validation sample because ten of
these students presented a high number of missing data and four displayed outlier values.
Ultimately, the calibration sample comprised of 726 students and in validation sample of
778 students.

Measures and instruments

Study time was assessed using an open question about the daily number of hours that students
dedicate to their personal study time and was assessed over the course of one full week. Each
student responded to this question daily by filling out a diary-form that was delivered to the
researchers at the end of the week in a closed envelope. The mean for this variable was 7.5
weekly hours (SD=5.52).
Class absence was assessed as the number of absences from the biology class and
was collected from the school secretary at the end of the academic year (M=3.18;
SD=4.16). In Portugal, high school students who miss more than nine classes in a
specific subject during the school year are required to repeat that subject the following school
year.
Domain knowledge was assessed by the mean grade in biology over the previous 2 years
(10th–11th grade). High school grades in Portugal range between the values 0 and 20, with 10
being a passing grade. The students, according to their grades, were distributed in the
following way: 1 for grade marks between 10 and 13 points (n=686, 45.6 %), 2 for grade
marks between 14 and 16 points (n=352, 23.4 %), and 3 for grade marks between 17 and 20
points (n=466, 31.0 %).
Homework completion At the end of the school year, each biology teacher reported a 1 for
students who completed less than 80 % of the assigned homework (622, 41.4 %) and a 2 for
those who completed more than 80 % of the assigned homework (882, 58.6 %).
Teachers’ Approaches to Teaching Inventory (TATI, high school) Most studies in literature
assess teachers’ approaches to teaching using the Approaches to Teaching Inventory (ATI;
Trigwell and Prosser 1993). Previous research findings indicate that the ATI is a context-
dependent instrument designed for the university context (Prosser and Trigwell 2006; Meyer
and Eley 2006; Stes et al. 2010); thus, a questionnaire adapted to the Portuguese high school
context was built under the theoretical framework of teachers’ approaches to teaching (see
Rosário et al. 2013b for further information). The TATI also consists of 12 items, with 3 items
for each of the 4 dimensions: Information Transmission/Teacher-Focused Strategy (ITTF-S;
e.g., “I only provide texts/materials/exercises on the information students need to prepare for
their assessments”), Conceptual Change/Student-Focused Strategy (CCSF-S; e.g., “I encour-
age students to investigate and read extra material so they can construct personal responses to
the assigned tasks”), Information Transmission/Teacher-Focused Intention (ITTF-I; e.g., “I
Teachers’ approaches to teaching 583

think that the learning concepts and their connections should be explicitly transmitted by the
teachers and not acquired by the students as a result of personal discovery or investigation”),
and Conceptual Change/Student-Focused Strategy Intention (CCSF-I; e.g., “In my discipline,
it is important to offer time and opportunities for the students to interact and learn with their
classmates”). The teachers responded on a five-point Likert scale that ranged from 1 (strongly
disagree) to 5 (strongly agree). The reliability indexes (Cronbach’s alpha) were .88 for the
ITTF strategy factor, .85 for the CCSF strategy factor, .81 for the ITTF intention factor, and .83
for the CCSF intention factor. For the purposes of this research, a confirmatory factor analysis
was conducted. The data analyses revealed a very good goodness-of-fit of the model (χ2(49)=
101.924, p<.001, χ2/df=2.080), goodness-of-fit index (GFI=.973), adjusted goodness-of-fit
index (AGFI=.957), comparative fit index (CFI=.987), Tucker-Lewis index (TLI=.982), and
root mean square error of approximation (RMSEA = .042 (confidence intervals
[CI]=.030–.054)).
Students’ Approaches to Learning Inventory (SALI, high school) In view of the above-
mentioned explanations regarding approaches to teaching, a questionnaire was built to assess
high school students’ approaches to learning (SALI) (see Rosário et al. 2013b for further
information). The SALI consists of 12 items, with 3 items for each of the 4 dimensions:
Surface Motivation (SM; e.g., “I believe teachers should tell me exactly what material will be
on the exam because I am only going to study that material”), Deep Motivation (DM; e.g., “I feel
pleased with my studies when I understand the answers to the ‘why’ questions”), Surface Strategy
(SS; e.g., “I only study what I think is enough to pass”), and Deep Strategy (DS; e.g., “After class,
I reread my notes to make sure they are clear and that I understand them”). The participants
responded to the statements on a five-point Likert scale that ranged from 1 (never) to 5 (always).
The reliability indexes (Cronbach’s alpha) were .83 for the DM factor, .81 for the SM factor, .80
for the DS factor, and .81 for the SS factor. As with the TATI, a confirmatory factor analysis was
conducted. The goodness-of-fit of the model was very good (χ2(49)=116.645, p<.001, χ2/df=
2.381, GFI=.987, AGFI=.980, CFI=.990, TLI=.987, RMSEA=.030 (CI=.023–.037)).

Procedure

The data were collected by the researches from biology classes during the second term of the
academic year (between January and April). The biology teachers and their students completed
the questionnaires focusing on the context of the biology class. The questionnaires were
administered to the teachers and the students at the same time, although the teachers completed
their questionnaires outside of the classroom.

Data analyses

The analysis was conducted in two stages. First, the model was fit using SEM, and data
from the two samples (calibration and validation) were compared by means of multigroup
analysis. Second, to calculate correct standard errors and to analyze the predictors from
both the individual level and the class level in the same analysis, a between model was
specified. Then, both models were simultaneously fitted using a two-level SEM analysis
to learn whether the effects found at the individual level varied significantly considering
the class level.
When working with hierarchical data (e.g., Dettmers et al. 2009; Rosário et al. 2013b;
Trautwein et al. 2009; Xu 2010), collective analysis units rather than individual analysis units
often constitute the observational reference towards which the intervention program is direct-
ed. If clusters constitute the unit of analysis, it is reasonable to think that the groups are
584 P. Rosário et al.

independent, but subjects within a group are dependent. Thus, except when there is no
association between the subjects within the group (i.e., class), statistical analysis considering
all subjects as independent observations tends to produce incorrect conclusions regarding the
effectiveness of interventions. In fact, ignoring the presence of clustered data often results in an
increased chance of committing an atomistic fallacy, which involves thinking that what is true
at the individual level must be true at the aggregated level.
A series of statistics and fit indexes were used to analyze the postulated models. In addition
to chi-square (χ2) and its associated probability (p), the GFI, AGFI, CFI, TLI, and the
RMSEA, including the CI, were used. The data were analyzed using the Mplus 6.11 statistical
package (Muthén and Muthén 2008–2011).

Results

Initial data screening

Table 1 presents the descriptive data as well as the two Pearson correlation matrixes of the two
samples. Firstly, we examined the matrices with regard to missing data, presence of outliers,
and linearity and normality of the data.
As the maximum likelihood (ML) method can produce distorted results when the assump-
tion of normality is violated (West et al. 1995), we examined the distribution of each one of the
variables in terms of their kurtosis and skewness. The guidelines by Finney and DiStefano
(2006) for whom 2 and 7, respectively, are the maximum allowable values for skewness and
kurtosis (in which case, ML should not be used) were considered. None of the above-
mentioned variables in the samples show values approaching these criteria (see Table 1).
Therefore, it was deemed appropriate to estimate the fit of the model using ML. Finally,
another important aspect in the initial analysis of the data matrix is to verify that the variables
are significantly correlated, although such correlation should not be excessively high (r>.85).
As observed in Table 1, none of the correlations exceeded .85.

Model cross-validation

Table 2 provides the values of the estimated parameters for the hypothesized model in
calibration, validation, and total samples. Table 3 shows the variance which is explained in
each of the variables in the model (observed and latent).
Taking into account the criteria that were used to calculate statistical and goodness-of-fit
index values (except for the case of χ2 and its associated probability), the fit indexes revealed
that the hypothesized model optimally represents the relationship of the empirical data
matrixes, both for calibration sample [χ2(38)=84.46, χ2/df=2.223, p<.001, GFI=.981,
AGFI=.962, CFI=.987, TLI=.977, RMSEA=.041 (CI=.029–.053), p=.889] and for valida-
tion sample [χ2(38)=87.43, χ2/df= 2.30, p< .001, GFI =.982, AGFI =.963, CFI =.987,
TLI=.977, RMSEA=.041 (CI=.030–.052), p=.904]. All of the hypotheses postulated in the
theoretical model were confirmed in both samples, with the findings of both studies being very
consistent.
The statistical analyses of the validation sample corroborate those obtained in the calibra-
tion sample, showing that there are no substantial differences in the fit of the model between
the two samples (Δχ2 =2.97) or the remaining fit indexes (ΔAGFI=.001, ΔCFI=.000,
ΔTLI=.000, ΔRMSEA=.000). Nevertheless, taking into account the small variations between
the two studies (i.e., in the fit of model or in some parameters), a multigroup analysis was
Table 1 Descriptive data (mean, standard deviation, skewness, and kurtosis) and correlation matrix corresponding to the variables included in the model (calibration and validation
samples)

1 2 3 4 5 6 7 8 9 10 11 12

Model 1 (first study—N=726)

1 –
2 .11** –
3 .22** .15** –
Teachers’ approaches to teaching

4 −.08* .07* −.12** –

5 −.13** −.08* −.28** .18** –
6 .20** .12** .34** −.22** −.60** –
7 −.17** −.13** −.28** .18** .77** −.61** –
8 .20** .17** .34** −.21** −.62** .81** −.61** –
9 .07* .02 .13** −.08* −.37** .37** −.35** .39** –
10 .09* −.00 .11** −.11** −.33** .34** −.36** .35** .68** –
11 −.00 −.01 −.10* .04 .26** −.27** .30** −.25** −.43** −.50** –
12 −.00 .13** .08* .08* .29** −.32** .31** −.31** −.46** −.52** .70** –
M 1.59 1.86 7.50 3.16 8.17 9.46 8.10 10.02 10.99 11.48 11.83 9.26
SD .49 .86 5.55 4.16 4.04 3.79 4.03 3.74 3.72 3.29 3.09 3.75
Skewness −.35 .28 .98 1.54 .32 −.22 .41 −.46 −1.07 −1.26 −1.18 .03
Kurtosis −1.88 −1.61 .63 1.90 −1.07 −1.12 −1.09 −.87 −.03 −.95 .99 −1.04
Model 2 (second study—N=778)
1 –
2 .10** –
3 .22** .15** –
4 −.09* .06 −.11** –
5 −.13** −.07* −.28** .15** –
6 .19** .12** .34** −.20** −.60** –
585
Table 1 (continued)
586

1 2 3 4 5 6 7 8 9 10 11 12

7 −.18** −.14** −.28** .17** .77** −.60** –

8 .20** .17** .33** −.21** −.60** .80** −58** –
9 .09* .02 .14** −.08* −.36** .39** −.34** .39** –
10 .08* −.00 .11** −.09* −.33** .34** −.34** .35** .68** –
11 −.00 −.01 −.09* .05 .25** −.28** .28** −.25** −.43** −.50** –
12 .00 .11** −.08* .07* .29** −.33** .30** −.31** −.47** −.53** .68** –
M 1.59 1.85 7.45 3.21 8.17 9.40 8.04 10.01 10.96 11.47 11.86 9.26
SD .49 .86 5.50 4.17 4.06 3.77 4.03 3.73 3.73 3.30 3.06 3.75
Skewness −.35 .30 .99 1.54 .32 −.21 .43 −.47 −1.06 −1.26 −1.19 .03
Kurtosis −1.88 −1.60 .70 1.87 −1.08 −1.10 −1.06 −.85 −.01 .94 1.06 −1.05

1 homework completion, 2 domain knowledge, 3 study time, 4 class absence, 5 surface motivation, 6 deep motivation, 7 surface strategy, 8 deep strategy, 9 CCSF strategy, 10 CCSF
intention, 11 ITTF strategy, 12 ITTF intention
P. Rosário et al.
Table 2 Standardized parameter estimates for the calibration, validation, and final samples

Estimate coefficients Standard errors (S.E.) T values (estimate/S.E.) Two-tailed P value

Calibration Validation Total Calibration Validation Total Calibration Validation Total Calibration Validation Total
sample sample sample sample sample sample sample sample sample sample sample sample

Structural model
Domain knowledge → Surface approach −.077 −.073 −.075 .156 .150 .108 −2.03 −1.97 −2.83 .042 .048 .005
Class absence → Surface approach .151 .151 .160 .032 .031 .022 4.47 4.13 6.08 .000 .000 .000
Teachers’ approaches to teaching

Study time → Surface approach −.269 −.269 −.266 .025 .024 .017 −6.73 −7.10 −9.79 .000 .000 .000
Homework → Surface approach −.090 −.095 −.093 .277 .265 .191 −2.34 −3.47 −3.48 .019 .011 .000
completion
Domain knowledge → Deep approach .111 .110 .110 .142 .135 .098 3.06 3.13 4.38 .002 .011 .000
Class absence → Deep approach −.201 −.190 −.195 .029 .028 .020 −5.59 −5.42 −7.79 .000 .000 .000
Study time → Deep approach .305 .305 .305 .023 .022 .016 8.20 8.44 11.78 .000 .000 .000
Homework → Deep approach .122 .125 .124 .252 .240 .174 3.33 3.52 4.85 .040 .000 .000
completion
Surface approach → ITTF approach .237 .219 .219 .068 .064 .047 3.15 2.89 4.26 .002 .004 .000
Surface approach → CCSF approach −.282 −.240 −.259 .058 .054 .040 −3.86 −3.54 −5.23 .000 .000 .000
Deep approach → ITTF approach −.198 −.237 −.218 .071 .067 .049 −2.67 −3.41 −4.31 .008 .000 .000
Deep approach → CCSF approach .255 .302 .280 .060 .057 .041 3.54 4.50 5.70 .030 .000 .000
Measurement model
Deep approach → Deep motivation .895 .899 .897 .013 .010 .010 69.89 70.44 99.21 .000 .000 .000
Deep approach → Deep strategy .906 .893 .900 .012 .010 .010 72.52 69.17 99.99 .000 .000 .000
Surface approach → Surface .877 .878 .877 .015 .020 .010 58.92 58.42 82.95 .000 .000 .000
motivation
Surface approach → Surface strategy .879 .872 .875 .015 .020 .010 59.29 57.30 82.40 .000 .000 .000
ITTF approach → ITTF intention .860 .856 .858 .021 .020 .020 40.07 40.73 57.13 .000 .000 .000
ITTF approach → ITTF strategy .808 .799 .803 .022 .020 .020 36.49 36.74 51.77 .000 .000 .000
CCSF approach → CCSF intention .857 .852 .855 .020 .020 .010 42.12 43.09 60.26 .000 .000 .000
587
Table 2 (continued)
588

Estimate coefficients Standard errors (S.E.) T values (estimate/S.E.) Two-tailed P value

Calibration Validation Total Calibration Validation Total Calibration Validation Total Calibration Validation Total
sample sample sample sample sample sample sample sample sample sample sample sample

CCSF approach → CCSF strategy .797 .796 .797 .021 .020 .020 37.29 38.41 53.52 .000 .000 .000
Covariances among exogenous variables
Homework ↔ Domain .047 .044 .106 .016 .015 .011 2.93 2.87 4.10 .003 .004 .000
completion knowledge
Homework ↔ Study time .612 .586 .220 .104 .099 .072 5.88 5.89 8.33 .000 .000 .000
completion
Homework ↔ Class absence −.159 −.182 −.083 .076 .074 .053 −2.09 −2.46 −3.23 .037 .014 .001
completion
Domain knowledge ↔ Study time .720 .728 .152 .180 .172 .124 4.00 4.23 5.83 .000 .000 .000
Domain knowledge ↔ Class absence .262 .227 .068 .134 .129 .093 1.96 1.75 2.62 .050 .079 .009
Study time ↔ Class absence −2.710 −2.539 −.114 .862 .827 .597 −3.14 −3.07 −4.40 .002 .002 .000
P. Rosário et al.
Teachers’ approaches to teaching 589

Table 3 R-square estimates for the

observed and latent variables in the Calibration Validation Total
calibration, validation, and final sample sample sample
samples
Latent variable
Surface approach .141 .139 .140
Deep approach .205 .201 .203
ITTF approach .168 .170 .169
CCSF approach .255 .259 .257
Observed variable
Surface motivation .769 .771 .770
Deep motivation .801 .808 .805
Surface strategy .773 .760 .766
Deep strategy .821 .798 .809
CCSF strategy .636 .634 .634
CCSF intention .735 .725 .730
ITTF strategy .653 .638 .645
ITTF intention .739 .732 .736

conducted to determine whether the model is equivalent in both groups and, if not, in which parts
there are statistically significant differences (factor loadings, structural weights, variance–covari-
ances, structural residuals, or measurement residuals). The procedure consisted of comparing the
fit of the nested models progressively, until one level (its equivalence in both groups) was rejected.
By progressively comparing the nested models, we determined the following: (1) when we
tested the factor loading equivalence (those that link latent variables to their observed
indicators), no statistically significant differences were found between both groups
(Δχ2(4)=.265, p=.992, NFI=.000, IFI=.000, RFI=−.002, TLI=−.002); (2) assuming that
there were no differences in factor loading, no statistically significant differences were found
in the structural weights (relationship among the independent and dependent variables of the
model) (Δχ2(12)=.444, p=1.000, NFI=.000, IFI=.000, RFI=−.005, TLI=−.005); (3) assum-
ing the absence of differences in the structural weights, no statistically significant differences at
the level of variance–covariance were found (Δχ2(10)=.230, p=1.000, NFI=.000, IFI=.000,
RFI=−.003, TLI=−.003); (4) assuming the absence of differences at the level of variance–
covariance, no statistically significant differences were found in the structural residuals
(Δχ2(6)=.510, p=.998, NFI=.000, IFI=.000, RFI=−.002, TLI=−.002); and lastly, (5) assum-
ing the absence of differences in the structural residuals, no statistically significant differences
were found in the measurement residuals (Δχ2(8)=.915, p=.999, NFI=.000, IFI=.000, RFI=
−.002, TLI=−.002). Summing up, the multigroup analyses confirm the robustness of the
hypothesized model.
Given the evidence of model cross-validation, both samples have joined in a larger sample
(total sample) to fit the model and examine proposed assumptions. The results are displayed in
Tables 2 and 3. As proposed, the SA was positively and significantly associated with the
adoption of an ITTF approach (β=.22) but negatively and significantly associated with the use
of a CCSF approach (β=−.26). In turn, the DA had a positive and significant association with
the CCSF approach (β=.28) and a negative and significant association with the ITTF approach
(β=−.22). Also, in agreement with our predictions, domain knowledge (γ=−.07), study time
(γ=−.27), and homework completion (γ=−.09) were related negatively and significantly with
an SA, whereas the students’ class absence rate was associated positively and significantly
590 P. Rosário et al.

with the SA (γ=.16). Furthermore, the students’ DA was positively and significantly associ-
ated with homework completion (γ=.12) and domain knowledge (γ=.11), and study time
(γ=.31) was negatively and significantly related with class absence (γ=−.20). Lastly, the
fourth hypothesis was confirmed as well, since study time, class absence, homework comple-
tion, and domain knowledge had an indirect association with CCSF and ITTF approaches to
teaching through their direct relationship with SA and DA to learning (study time=−.125 with
ITTF and .155 with CCSF; class absence=.078 with ITTF and −.096 with CCSF; homework
completion=−.047 with ITTF and .059 with CCSF; domain knowledge=−.040 with ITTF and
.050 with CCSF).
Finally, the study time, class absence, domain knowledge, and homework completion were
explained with a considerable proportion of the variance in the approaches to learning (see
Table 3): 14 % of the SA and 20.3 % of the DA. The approaches to teaching were explained by
all of the student variables (study time, class absence, domain knowledge, homework com-
pletion, and the two approaches to learning) with a significant percentage such as 25.7 % of the
CCSF approach and 16.9 % of the ITTF approach.

Two-level SEM analysis

The appropriateness of fitting a two-level structure was assessed by inspecting the intra-class
correlation estimates for the variables that are potentially possessed both between- and within-
level variation (see Fig. 2). The estimates were .32 (SM), .34 (DM), .31 (SS), .36 (DS), .09
(homework completion), .92 (domain knowledge), .11 (study time), and .13 (class absence).
These estimates suggested that the within-level variability was generally far larger than
between-level variability (as expected). However, these estimates clearly indicate the need
for performing a two-level assessment (see, e.g., Muthén 1991), especially considering that
these estimates are generally attenuated (downwardly biased) due to measurement error.
A preliminary inspection of the residual variances of the measurement indicators at
a within level suggested that the residual variances were all very small (see Table 4),
which is a typical result (Heck and Thomas 2009). Furthermore, the sub-model
concerning the variables that had both within- and between-cluster variability (domain
knowledge, class absence, study time, homework completion, SA, and DA) was
constrained to be the same at both levels (i.e., the between model was identical to
the within model), which was a restriction that allowed for a common interpretation.
This model fitted very well (χ 2 (52) = 68.48, χ 2 /df = 1.31, p = .039, CFI = .993,
TLI=.986, and RMSEA=.0016). The standardized root mean squared residual for
the between and within models was .044 and .008, respectively, suggesting that the
multilevel model did an adequate job in reproducing covariance at both levels.
In comparing the part of the model that was common across both levels, the SA and DA
factors were well measured at both levels. Because the invariant factor loadings made common
factor variances directly comparable across levels (Mehta and Neale 2005), a common scale
was imposed in the analysis. The likelihood ratio statistic (−2LL) shown here as the difference
between log-likelihood statistics for the unconstrained model (i.e., without the invariant factor
loadings) and the constrained model (i.e., with invariant factor loadings) was used to evaluate
the appropriateness of the restrictions imposed by the within- and between-clusters. The
difference in the −2LL between the model with and without the invariant factor loadings
was χ2(3)=.352 (p=.940), which indicated that the hypothesis of a common scale could not be
rejected. In this case, the proportion of variance of the SA factor at the within-cluster level
explained by its random intercept or ICC for the latent SA factor was .48, whereas the
corresponding proportion of variance of the individual-level DA factor explained by its
Teachers’ approaches to teaching 591

SS SM ITTF-S ITTF-I

1 1

Domain Surface ITTF

Knowledge Approach Approach

Class
Absence

Study
Time

Deep CCSF
Homework Approach
Completion Approach

1 1

DS DM CCSF-S CCSF-I

Fig. 2 Proposed two-level model. The small circles represent disturbances. The variables with varia-
tion, solely at the cluster level, are shown outside the box, and the variables that can vary at both
levels are inside the box. SS = Surface Strategy; DS = Deep Strategy; SM = Surface Motivation; DM
= Deep Motivation; ITTF-I = Information Transmission/Teacher-Focused Intention; CCSF-I = Con-
ceptual Change/Student-Focused Intention; ITTF-S = Information Transmission/Teacher-Focused Strat-
egy; CCSF-S = Conceptual Change/Student-Focused Strategy

random intercept for the latent DA factor was .46. In other words, approximately 48 % of the
variance of the SA and 46 % of the variance of the DA lie between classes.
The results for the within level of the two-level SEM analysis, which included the data from
the SEM analysis regarding studies 1 and 2, study time, class absence, domain knowledge, and
homework completion were significantly associated with the way that students study (DA and
SA), in the directions proposed in the hypotheses. Estimation errors for this level were small
(see Table 4).
Additionally, at the within level, there were significant relationships between homework
completion and study time (r=.165, p<.001), between study time and class absence (r=−.069,
p<.01), and between homework completion and domain knowledge (r=.093, p<.001). At the
between level, however, only study time was significantly associated with the DA to study and
learning (.608 with the DA) and with the SA to learning (−.637 with the SA). Conversely to
the within level of the analysis, the estimation errors were higher at the between level. These
estimation errors may have affected the estimation of the coefficients and their statistical
significance.
In the two-level SEM to analyze the relationship between the approaches to learning and
the approaches to teaching, the class was considered a unit of analysis. Findings indicated that
592 P. Rosário et al.

Table 4 Standardized parameter estimates for the two-level structural equation model

Estimate S.E. T values Two-tailed

P values

Within level
Domain knowledge → Surface approach −.159 .027 −5.841 .000
Class absence → Surface approach .163 .027 6.025 .000
Study time → Surface approach −.278 .027 −10.438 .000
Homework completion → Surface approach −.137 .028 −4.992 .000
Domain knowledge → Deep approach .300 .025 11.977 .000
Class absence → Deep approach −.171 .026 −6.671 .000
Study time → Deep approach .302 .025 12.013 .000
Homework completion → Deep approach .097 .026 3.657 .000
Surface approach → Surface motivation .832 .014 58.559 .000
Surface approach → Surface strategy .818 .014 57.417 .000
Deep approach → Deep motivation .835 .012 67.793 .000
Deep approach → Deep strategy .854 .012 71.841 .000
Surface approach ↔ Deep approach −.607 .024 −24.964 .000
Homework completion ↔ Domain knowledge .093 .026 3.553 .000
Homework completion ↔ Study time .165 .026 6.40 .000
Homework completion ↔ Class absence −.028 .027 −1.030 .303
Domain knowledge ↔ Study time .036 .026 1.342 .179
Domain knowledge ↔ Class absence −.057 .026 −2.160 .031
Study time ↔ Class absence −.069 .026 −2.631 .009
Between level
Domain knowledge → Surface approach .018 .199 .090 .928
Class absence → Surface approach .157 .253 .622 .534
Study time → Surface approach −.637 .273 −2.336 .019
Homework completion → Surface approach .422 .285 1.482 .138
Domain knowledge → Deep approach −.136 .175 −.780 .436
Class absence → Deep approach −.004 .218 −.017 .986
Study time → Deep approach .608 .229 2.653 .008
Homework completion → Deep approach .050 .244 .0.206 .837
Surface approach → Surface motivation .979 .014 71.714 .000
Surface approach → Surface strategy .979 .015 69.698 .000
Deep approach → Deep motivation .987 .008 116.483 .000
Deep approach → Deep strategy .999 .009 124.637 .000
ITTF approach → ITTF strategy .743 .078 9.564 .000
ITTF approach → ITTF intention .803 .075 10.662 .000
CCSF approach → CCSF strategy .799 .064 12.526 .000
CCSF approach → CCSF intention .786 .065 12.186 .000
Surface approach → ITTF approach .569 .288 1.977 .048
Deep approach → ITTF approach −.088 .293 −.301 .763
Surface approach → CCSF approach −.386 .256 −1.506 .132
Deep approach → CCSF approach .386 .244 1.589 .119
Surface approach ↔ Deep approach −.941 .035 −27.188 .000
ITTF approach ↔ CCSF approach −.475 .174 −2.718 .006
Teachers’ approaches to teaching 593

Table 4 (continued)

Estimate S.E. T values Two-tailed

P values

Homework completion ↔ Domain knowledge .254 .132 1.920 .055

Homework completion ↔ Study time .733 .104 7.059 .000
Homework completion ↔ Class absence −.574 .127 −4.533 .000
Domain knowledge ↔ Study time .447 .112 4.009 .000
Domain knowledge ↔ Class absence .241 .124 1.945 .052
Study time ↔ Class absence −.478 .131 −3.645 .000

the teachers’ approaches to teaching were only partially responsive to their students’ ap-
proaches to learning (see Table 4). Only one of the four hypothesized effects was confirmed
in the two-level SEM analysis. Nevertheless, it must be considered that the sizes of the
estimation errors were high, possibly related to the sample size at the between level, and that
the t values approached statistical significance in other two scenarios (i.e., SA → CCSF
approach and SA → CCSF approach). Lastly, at the between level, it can be emphasized that

Table 5 R-square estimates for the

two-level structural equation model Estimate S.E. T values Two-tailed
P values

Within level
Observed variable
Surface motivation .693 .024 29.279 .000
Deep motivation .697 .021 33.897 .000
Surface strategy .669 .023 28.709 .000
Deep strategy .730 .020 35.920 .000
Latent variable
Surface approach .178 .021 8.521 .000
Deep approach .255 .023 11.319 .000
Between level
Observed variable
CCSF strategy .638 .102 6.263 .000
CCSF intention .618 .101 6.003 .000
ITTF strategy .552 .115 4.782 .000
ITTF intention .645 .121 5.331 .000
Surface motivation .958 .027 35.242 .000
Deep motivation .974 .017 58.242 .000
Surface strategy .958 .028 34.841 .000
Deep strategy .999 .016 62.319 .000
Latent variable
Surface approach .229 .117 1.954 .051
Deep approach .361 .100 3.591 .000
ITTF approach .422 .118 3.572 .000
CCSF approach .564 .111 5.112 .000
594 P. Rosário et al.

the DA and the CCSF approach showed a higher amount of variance explained than the SA
and the ITTF approaches (see Table 5).

Discussion

Firstly, this study aimed to analyze the relationship between student variables (i.e., study time,
class absence, domain knowledge, and homework completion), the students’ approaches to
learning, and the teachers’ approaches to teaching using a sample of 12th graders. A SEM
model was built using two equivalent samples of students to increase the robustness of the
results. The first sample was used to fit the model and the second sample was used to replicate
the results found in the first study. Secondly, the extent to which the variability of the DA and
SA was explained by variables located at the class level was analyzed.

The way students approach studying in relationship with teachers’ approaches to teaching:
an individual-level analysis

At the individual level, as postulated in the model, the data provided by the SEM analysis and
by the within level of the two-level SEM analysis indicated that the approaches to learning
were related to the approaches to teaching. The SA was positively related to the ITTF approach
and negatively related to the CCSF approach. The DA was positively related to the CCSF
approach and negatively related to the ITTF approach. Our data are in line with previous
literature, which suggests that teachers are most likely to adopt a CCSF approach when their
students report using DAs to learning and that teachers are most likely to adopt an ITTF
approach when their students report using SAs to learning (Prosser et al. 2003; Rosário et al.
2013b).
In the current study, the CCSF approaches to teaching were associated with the students’
DA to learning, although the strength of this relationship was not very strong. This finding
helps to deepen our comprehension of the approaches to teaching. Stes et al. (2008), for
example, reported no effects of the contextual and teacher variables (e.g., discipline of teaching
and teachers’ age, gender, and academic status) on the CCSF approach to teaching. In contrast
to Stes et al. (2008), the present investigation included approaches to learning as contextual
variables and, as verified, these contextual variables partially explained the variance in the
approaches to teaching. Therefore, our findings stress the importance of including student
variables and variables from the educational context when answering the question “why do
teachers approach teaching the way they do?”
The students’ SAs to learning are positively associated with ITTF approaches to teaching.
This is an important finding once high school teachers are expected to be capable of modifying
their teaching practices and discouraging students’ SA to learning (Virtannen and Lindblom-
Ylänne 2010), but, according to our findings, the participant teachers seemed unable to
accomplish this goal.
Since the way students approach studying is closely related to their perception of how their
teachers teach (Rosário et al. 2010), our data stresses the importance of promoting educational
opportunities for teachers to reflect upon the teaching methodologies used in class.
Future research should analyze the relationships between teachers’ approaches to teaching
and students’ approaches to learning in depth, for instance, by conducting qualitative studies to
understand teachers’ motives and the curricular and contextual constraints they may face when
deciding their teaching approaches to teaching. For example, teachers may have the intention
to follow a certain teaching methodology and approach to learning but not being able to carry
Teachers’ approaches to teaching 595

it out in practice (e.g., due to administrative aspects, such as number of students in class and
decisions in the assessment system demanded by the department, or personal limitations, such
as low capacity of analysis of their own behavior in class and inefficient feedback mechanisms
to control their teaching). Approaches to learning and approaches to teaching are interrelated
due to the responsive nature of both variables, and none of these approaches are universally
suited to address school tasks.
Our results could be of help in designing school-based continuing educational courses for
teachers. These courses could include topics such as the importance of being aware of their
students’ approaches to learning and of reflecting upon the teaching methods that may elicit
DAs to learning (e.g., type of questions in class, type of assessment, feedback provided,
teachers’ and students’ conceptions of teaching and learning (Postareff and Lindblom-Yl nne
2008; Virtannen and Lindblom-Ylänne 2010).
With regard to the remaining student variables, evidence was found of their direct and
significant relationship with both DA and SA to learning and of their indirect impact on the
approaches to teaching. Therefore, the current data stresses that the way teachers teach is not
directly affected by any of the four variables examined in this research (i.e., study time, class
absence, domain knowledge, and homework completion).
The coefficients of the relationships in the model also follow the hypothesized direction,
although the strength of these relationships is lower than expected, especially concerning the
coefficients for homework completion and domain knowledge. This fact clearly suggests the
need of including in the model other contextual variables (e.g., work in class, perceived social
value of the school subject, instrumentality of the school subject for the student’s personal
projects) that could increase the amount of explained variance for both the approaches to
learning and the approaches to teaching. This same model should also be tested for other high
school subjects.

The way students approach studying in relationship with teachers’ approaches to teaching:
a class-level analysis

Although no class variable has been included in the current research, the two-level analysis
indicated that 39 % of the variance of the SA and 40 % of the variance of the DA were related
to variables of the class level, beyond the effects of the four individual variables that were
evaluated (study time, class absence, domain knowledge, and homework completion).
It can be stated that teachers’ approach to teaching, which is an important variable at the
class level, may be one of the class-level variables with the highest level of influence in
determining the way their students study (either using a DA or an SA).

Limitations and suggestions for future research

Although the present study has produced interesting results at both theoretical and practical
levels, we acknowledge the following limitations. Firstly, some structural relationships showed
a low explanatory level with regard to some of the constructs used in the study. While this low
level of explanation did not compromise the validity of the model, which fit very well, we
nevertheless indicate a limitation of the explanatory capacity of the effects among the
variables. Some of the variability not explained might be determined by other variables not
included in the model, at both the individual and class levels, or due to the metric of some
variables. For example, school grades and homework completions had few categories each.
This option could have helped to explain the statistic power that was found; so, future research
could consider using more categories to measure these variables.
596 P. Rosário et al.

Future models should also include other variables related to the micro-processes of teaching
(e.g., the nature and frequency of assessment, type and frequency of the homework assigned,
the nature of feedback provided in class, type of questions asked) to increase the variance
amount for the variables.
Secondly, the followed design was cross-sectional, which does not allow for causal
inferences, even from the two-level SEM perspective. Reciprocal determinations between
approaches to learning and approaches to teaching should be included in the model.
However, the modeling of that hypothesis would need data collection using repeated measures.
The data should be collected at two or more temporal moments in a nonexperimental design
with repeated measures to analyze reciprocal relationships among the variables in the model.
The absence of repeated measures is one of the limitations of the present investigation.
Prospective research should also include a greater sample size, with randomly selected teachers
as well as students (see Maas and Hox 2005). As indicated in the “Results” section, it is
possible that observed discrepancies between the two levels may simply be due to a lack of
power. Forthcoming studies should follow multilevel designs including enough units at the
between level to assure stability and accuracy at that level.
The results derived from this investigation suggest a connection between approaches to
learning and approaches to teaching, which is an important finding for these two frameworks.
However, the magnitude of the effects found suggests the need for deeper reflection. One
important aspect is that the approaches’ constructs were assessed using self-reports, which
allowed data collection from a large number of students and teachers but did not capture real-
time responses to the demands of authentic learning environments (Rosário et al. 2013a).
These possible explanations reinforce the need to include event measures in the research
designs that capture the ongoing nature of the constructs in analysis (Zimmerman 2008).

Conclusions

The present study addressed the impact of students’ approaches to learning on their teachers’
approaches to teaching with a large sample of students and their corresponding teachers.
Firstly, the SEM was fit with two independent data samples. Secondly, we intended to examine
the extent to which the variability found in the students’ approaches to learning was related to
the variables at class level. Findings suggest that, on one hand, the way teachers’ approach
teaching is not deeply explained by the way students study but, on the other hand, the way
students study is strongly associated with variables at the class level (variables to which all
students are exposed) and less associated with variables at the student level, such as study time,
class absence, domain knowledge, or homework completion. These data suggest that the
answer to our initial question (why do teachers teach the way they do?) requires the investi-
gation not only of the way students study but also of other relevant contextual variables (e.g.,
class climate, type of assessment used, type of tasks performed in class, quality of feedback
provided, and class size) and teachers’ variables (e.g., teacher experience and pedagogical
training). Our results are aligned with SAL’s tradition, in which “what the student does is
actually more important in determining what is learned than what the teacher does” (Shuell
1986, p.429). Results stress the importance of the relationship between how students approach
their learning and how teachers approach their teaching. This finding is important and may
contribute to encourage teachers to become more aware of their students’ approaches to
learning and to reflect upon the effectiveness of their own teaching methods (e.g., type of
questions in class, type of assessment, feedback provided). Besides, reflection on data can call
upon high school principals’ attention to the importance of students’ academic behavior and
Teachers’ approaches to teaching 597

attitude towards learning. For example, schools could consider organizing educational seminars
to foster parents’ involvement, help students and parents to reflect upon students’ responsibility
towards learning, and learn self-regulated learning strategies (Rosário et al. 2010).
In short, the results of this investigation confirm the need for further research concerning
motivations behind teachers’ methodologies utilized in class and students’ motivations in their
choice of study practices.

References

Alexander, P., & Jetton, T. (2000). Learning from a text: a multidimensional and developmental perspective. In
M. L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr (Eds.), Handbook of reading research (Vol. 3, pp.
285–310). Mahwah, NJ: Erlbaum.
Alexander, P. A., Kulikowich, J. M., & Schulze, S. K. (1994). The influence of topic knowledge, domain
knowledge, and interest on the comprehension of scientific exposition. Learning and Individual Differences,
6, 379–397.
Alexander, P. A., Murphy, P. K., Woods, B. S., Duhon, K. E., & Parker, D. (1997). College instruction and
concomitant changes in students’ knowledge, interest, and strategy use: a study of domain learning.
Contemporary Educational Psychology, 22, 125–146.
Ball, C., & Connolly, J. (2000). Educationally disaffected young offenders: youth court and agency responses to
truancy and school exclusion. British Journal of Criminology, 40(4), 594–616.
Biggs, J. B., Kember, D., & Leung, D. (2001). The revised two-factor Study Process Questionnaire: R-SPQ-SF.
British Journal of Educational Psychology, 71(1), 133–149.
Cooper, H. (Ed.). (1989). Homework. New York: Longman.
Cooper, H., & Valentine, J. C. (2001). Using research to answer practical questions about homework.
Educational Psychologist, 36, 143–153.
Cooper, H., Robinson, J., & Patall, E. (2006). Does homework improve academic achievement? A synthesis of
research, 1987-2003. Review of Educational Research, 76, 1–62.
De Jong, R., Westerhof, K. J., & Creemers, B. P. M. (2000). Homework and student math achievement in junior
high schools. Educational Research and Evaluation, 6, 130–157.
Dettmers, S., Trautwein, U., & Lüdtke, O. (2009). The relationship between homework time and achievement is
not universal: evidence from multilevel analyses in 40 countries. School Effectiveness and School
Improvement, 20, 375–405.
Entwistle, N. J. (2009). Teaching for understanding at university: deep approaches and distinctive ways of
thinking. Basingstoke: Palgrave Macmillan.
Epstein, J. L., & Van Voorhis, F. L. (2001). More than minutes: teachers’ roles in designing homework.
Educational Psychologist, 36, 181–193.
Finney S. J., & DiStefano C. (2006). Non-normal and categorical data in structural equation modelling. In: G. R.
Hancock & R. O. Mueller (Eds.) Structural equation modelling. A second course (pp. 269-314). Greenwich,
CT: Information Age Publishing.
Gortner-Lahmers, A., & Zulauf, C. R. (2000). Factors associated with academic time use and academic performance
of college students: a recursive approach. Journal of College Student Development, 41, 544–556.
Heck, R. H., & Thomas, S. L. (2009). An introduction to multilevel modeling techniques. London: Routledge.
Jonanssen, C. (2011). The dynamics of absence behaviour: interrelations between absence from class and
absence in class. Educational Research, 53(1), 17–32.
Kitsantas, A., & Zimmerman, B. (2009). College students’ homework and academic achievement: the mediating
role of self-regulatory beliefs. Metacognition and Learning, 4, 97–110.
Lindblom-Yl nne, S., Trigwell, K., Nevgi, A., & Ashwin, P. (2006). How approaches to teaching are affected by
discipline and teaching context. Studies in Higher Education, 31, 285–298.
Lueddeke, G. R. (2003). Professionalizing teaching practice in higher education: a study of disciplinary variation
and ‘teaching scholarship’. Studies in Higher Education, 28, 213–228.
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling. Methodology, 1, 86–92.
Mehta, P. D., & Neale, M. C. (2005). People are variables too: multilevel structural equations modeling.
Psychological Methods, 10, 259–284.
Meyer, J. H. F., & Eley, M. G. (2006). The approaches to teaching inventory: a critique of its development and
applicability. British Journal of Educational Psychology, 76, 633–649.
Muthén, B. O. (1991). Multilevel factor analysis of class and student achievement components. Journal of
Educational Measurement, 28, 338–354.
598 P. Rosário et al.

Muthén, L. K., & Muthén, B. O. (2008–2011). Mplus user’s guide. Los Angeles: Author.
Nevgi, A., Postareff, L., & Lindblom-Yl nne, S. (2004, June). The effect of discipline on motivational and self-
efficacy beliefs and on approaches to teaching of Finnish and English university teachers. A paper presented
at the EARLI SIG Higher Education Conference.
Núñez, J. C., Suárez, N., Cerezo, R., Rosário, P., & Valle, A. (2013). Homework and academic achievement
across Spanish compulsory education. Educational Psychology. DOI: 10.1080/01443410.2013.817537
Núñez, J. C., Suárez, N., Rosário, P., Vallejo, G., Cerezo, R. & Valle, A. (in press). Teachers’ feedback on
homework, homework-related behaviors and academic achievement. The Journal of Educational Research.
Plant, E. A., Ericsson, K. A., Hill, L., & Asberg, K. (2005). Why study time does not predict grade point average
across college students: implications of deliberate practice for academic performance. Contemporary
Educational Psychology, 30, 96–116.
Postareff, L., & Lindblom-Yl nne, S. (2008). Variation in teachers’ descriptions of teaching: broadening the
understanding of teaching in higher education. Learning and Instruction, 18, 109–120.
Prosser, M., & Trigwell, K. (1993), Development of an approaches to teaching questionnaire. Research and
Development in Higher Education, 15, 468–473.
Prosser, M., & Trigwell, K. (1999). Understanding learning and teaching. The experience in higher education.
Buckingham: Open University Press.
Prosser, M., & Trigwell, K. (2006). Confirmatory factor analysis of the approaches to teaching inventory. British
Journal of Educational Psychology, 76, 405–419.
Prosser, M., Trigwell, K., & Taylor, P. (1994). A phenomenographic study of academics’ conceptions of science
learning and teaching. Learning and Instruction, 4, 217–232.
Prosser, M., Ramsden, P., Trigwell, K., & Martin, E. (2003). Dissonance in experience of teaching and its relation
to the quality of student learning. Studies in Higher Education, 28, 37–48.
Ramsden, P. (2003). Learning to teach in higher education (2nd ed.). London: Routledge.
Ramsden, P., Prosser, M., Trigwell, K., & Martin, E. (2007). University teachers’ experiences of academic
leadership and their approaches to teaching. Learning and Instruction, 17, 140–155.
Reid, K. (2006). An evaluation of the views of secondary staff towards school attendance issues. Oxford Review
of Education, 32(3), 303–324.
Richardson, J. (2013). Approaches to studying across the adult life span: evidence from distance education.
Learning and Individual Differences, 26, 74–80.
Rosário, P., Núñez, J. C., González-Pienda, J. A., Valle, A., Trigo, L., & Guimarães, C. (2010). Enhancing self-
regulation and approaches to learning in first-year college students: a narrative-based program assessed in the
Iberian Peninsula. European Journal of Psychology of Education, 25, 411–428.
Rosário, P., Núñez, J. C., Valle, A., Paiva, O., & Polydoro, S. (2013a). Approaches to teaching in high school
when considering contextual variables and teacher variables. Revista de Psicodidatica, 18, 25–45.
Rosário, P., Núñez, J. C., Ferrando, P., Paiva, O., Lourenço, A., Cerezo, R., et al. (2013b). The relationship
between approaches to teaching and approaches to studying: a two-level structural equation model for biology
achievement in high school. Metacognition and Learning, 8, 47–77. doi:10.1007/s11409-013-9095-6.
Shuell, T. J. (1986). Cognitive conceptions of learning. Review of Educational Research, 56, 411–436.
Stes, A., Gijbels, D., & Van Petegem, P. (2008). Student-focused approaches to teaching in relation to context and
teacher characteristics. Higher Education, 55, 255–267.
Stes, A., Maeyer, S., & Van Petegem, P. (2010). Approaches to teaching in higher education: validation of a
Dutch version of the approaches to teaching inventory. Learning Environment Research, 13, 59–73.
Struyven, K., Dochy, F., Janssens, S., & Gielen, S. (2006). On the dynamics of students’ approaches to learning:
the effects of the teaching/learning environment. Learning and Instruction, 16, 279–294.
Trautwein, U. (2007). The homework-achievement relation reconsidered: differentiating homework time, home-
work frequency, and homework effort. Learning and Instruction, 17, 372–388.
Trautwein, U., & Koller, O. (2003). The relationship between homework and achievement—still much of a
mystery. Educational Psychology Review, 15, 115–145.
Trautwein, U., Schnyder, I., Niggli, A., Neumann, M., & Lüdtke, O. (2009). Chameleon effects in homework
research: the homework–achievement association depends on the measures used and the level of analysis
chosen. Contemporary Educational Psychology, 34, 77–88.
Trigwell, K., & Prosser, M. (1991). Relating approaches to study and the quality of learning outcomes at the
course level. British Journal of Educational Psychology, 61, 265–275.
Trigwell, K., & Prosser, M. (2003). Qualitative difference in university teaching. In M. Tight (Ed.), Access and
exclusion (pp. 185–216). London: JAI Elsevier Science.
Trigwell, K., Prosser, M., & Waterhouse, F. (1999). Relations between teachers’ approaches to teaching and
students’ approaches to learning. Higher Education, 37, 57–70.
Virtannen, V., & Lindblom-Ylänne, S. (2010). University students’ and teachers’ conceptions of teaching and
learning in the biosciences. Instructional Science, 38, 355–370.
Teachers’ approaches to teaching 599

Walberg, H. J. (1991). Does homework help? The School Community Journal, 1, 13–15.
West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with non-normal variables: problems
and remedies. In R. Hoyle (Ed.), Structural equation modelling: concepts, issues and applications (pp. 55–
75). Newbury Park, CA: Sage.
Willoughby, T., Wood, E., & Kraftcheck, E. R. (2003). When can a lack of structure facilitate strategic processing
of information? British Journal of Educational Psychology, 73, 59–69.
Xu, J. (2010). Predicting homework time management at the secondary school level: a multilevel analysis.
Learning and Individual Differences, 20, 34–39.
Xu, J., & Corno, L. (2006). Gender, family help, and homework management reported by middle school
students. Journal of Research in Rural Education, 21, 1–13.
Zimmerman, B. J. (2008). Investigating self-regulation and motivation: historical background, methodological
developments, and future prospects. American Educational Research Journal, 45, 166–183.

Pedro Rosário. Psychology Department of University of Minho, Campus de Gualtar, 4710-057, Braga,
Portugal.E-mail: prosario@psi.uminho.pt, Web site: www.guia-psi.com

Current themes of research:

Self-regulated learning. Programs to enhance self-regulation. Study skills and approaches to learning.
Homework.

Most relevant publications in the field of Psychology of Education (max: 5):

Rosário, P., Lourenço, A., Paiva, O., Valle, A., & Tuero-Herrero, E. ( 2012). Predicción del rendimiento en
matemáticas: efecto de variables personales, socioeducativas y del contexto escolar. Psicothema, 24(2), 289-295.
Rosário, P., Lourenço, A., Paiva, O., Núnez, J., González-Pienda, J. & Valle, A. (2012). Autoeficacia y utilidad
percibida como condiciones necesarias para un aprendizaje académico autorregulado. Anales de Psicologia,
28 (1), 37-44.
Rosário, P. Mourão, R., Trigo, L., Suárez, N., Fernández, E. & Tuero-Herrero, E. (2011). Uso de diarios de tareas
para casa en el inglés como lengua extranjera: evaluación de pros y contras en el aprendizaje autorregulado y
rendimiento. Psicothema, 23(4), 881-887.
Rosário, P., Núñez, J. C., González-Pienda, J., Valle, A., Trigo, L., & Guimarães, C. (2010). Enhancing self-
regulation and approaches in first-year college students: A narrative-based program assessed in the Iberian
Peninsula. European Journal of Psychology of Education, 25, 411-428.
Rosário, P., Costa, M., Núñez, J.C., González-Pienda, J., Solano, P., & Valle, A. (2009). Academic Procrastination:
Associations with personal, school, and family variables. Spanish Journal of Psychology, 12(1), 118-127.

José Carlos Núñez. Psychology Department of University of Oviedo, Plaza Feijoo s/n, 33003, Oviedo, España.
E-mail: jcnunezperez@gmail.com;Web site: www.uniovi.es/zope/departamentos/Psicologia

Current themes of research:

Self-regulated learning. Motivation in higher education. Approaches to learning. Homework and school achieve-
ment. Assessment.

Most relevant publications in the field of Psychology of Education (max: 5):

Núnez, J. C., González-Pienda, J., Rodríguez, C., Valle, A., González-Cabanach, R. & Rosário, P. (2011).
Multiple Goals Perspective in Adolescent Students with Learning Difficulties. Learning Disability Quarterly,
34, 273-286.
Núnez, J. Cerezo, R., González-Pienda, J., Rosário, P., Valle, A., Fernández, E. & Suárez, N. (2011). Imple-
mentation of training programs in self-regulated learning strategies in Moodle format: Results of a experience
in higher education. Psicothema, 23(2), 274-281.
600 P. Rosário et al.

Rosário, P., Núñez, J. C., González-Pienda, J., Valle, A., Trigo, L., & Guimarães, C. (2010). Enhancing self-
regulation and approaches in first-year college students: A narrative-based program assessed in the Iberian
Peninsula. European Journal of Psychology of Education, 25, 411-428.
Valle, A., Cabanach, R.G., Rodríguez, S., Núñez, J.C., González-Pienda, J.A., Solano, P., & Rosário, P. (2007). A
motivational perspective on the self-regulated learning in higher education. In P. B. Richards (Ed.), Global
issues in higher education (pp. 99-125). New York: Nova Science Publishers.
Valle, A., Núñez, J. C., Cabanach, R., González-Pienda, J., Rodríguez, S., Rosário, P., Cerezo, R., & Muñoz-
Cadavid, M. (2008). Self-Regulated Profiles and Academic Achievement. Psicothema, 20(4), 724-731.

Guillermo Vallejo. Psychology Department of University of Oviedo, Plaza Feijoo s/n, 33003, Oviedo, España.
E-mail: gvallejoseco@gmail.com

Current themes of research:

Linear mixed models, with an emphasis on longitudinal data analysis. Specific areas of investigation include
statistical approaches to assessing the effects of multicolineality. Heteroscedasticity on inference, selecting, and
multilevel models.

Most relevant publications in the field of Psychology of Education (max: 5):

Núñez, J.C., Vallejo, G., Rosário, P., Tuero-Herrero, E., & Valle, A. (in press). Variables from the students, the
teachers and the school context predicting academic achievement: A multilevel perspective. Journal of
Psychodidactics.
Vallejo, G., Tuero-Herrero, E., Núñez, J. C. & Rosário, P. (in press). Performance evaluation of recent
information criteria for selecting multilevel models in behavioral and social sciences. International Journal
of Clinical and Health Psychology.
Núñez, J. C., Rosário, P., Vallejo, G. & González-Pienda, J. A. (2013). A longitudinal assessment of the
effectiveness of a school-based mentoring program in middle school. Contemporary Educational Psychology,
38, 11-21.
Livacic-Rojas, P., Vallejo, G., Fernández, M. P. & Tuero-Herrero, E. (2013). Covariance structure selection and
Type I error rates in split-plot designs. Methodology: Journal of Research Methods for the Behavioral and
Social Sciences, 9, 129-136.
Vallejo, G., Fernández, M. P., Livacic-Rojas, P. E., & Tuero-Herrero, E. (2011). Comparison of modern methods
for analyzing unbalanced repeated measures data with missing values. Multivariate Behavioral Research, 46,
900-937.

Olímpia Paiva. Psychology Department of Psychology, E-mail: olimpiapaiva0212@gmail.com; Web site:

www.guia-psi.com

Current themes of research:

Self-regulated learning. Approaches to learning. Higher education.

Most relevant publications in the field of Psychology of Education (max: 5):

Rosário, P., González-Pienda, J. A., Pinto, R., Ferreira, P., Lourenço, A. & Paiva, O. (2010). Efficacy of
the program “Testas’s (mis)adventures” to promote the deep approach to learning. Psicothema, 22 (4),
828-834.
Rosário, P., Lourenço, A., Paiva, O., Núnez, J., González-Pienda, J. & Valle, A. (2012). Autoeficacia y utilidad
percibida como condiciones necesarias para un aprendizaje académico autorregulado. Anales de Psicologia,
28 (1), 37-44.
Rosário, P., Lourenço, A., Paiva, O.,Valle, A., & Tuero-Herrero, E. ( 2012). Predicción del rendimiento en
matemáticas: efecto de variables personales, socioeducativas y del contexto escolar. Psicothema, 24(2),
289-295.
Teachers’ approaches to teaching 601

Antonio Valle. Department of Developmental and Educational Psychology, University of A Coruña, Campus de
Elviña, 15071, A Coruña, España. E-mail: vallar@udc.es;Web site:http://www.udc.es/dep/psee

Current themes of research:

Goals. Self-regulated learning. Higher education. School achievement.

Most relevant publications in the field of Psychology of Education (max: 5):

Valle, A., Rodríguez, S., J. C. Núñez, J.C., Cabanach, R., González-Pienda, J. A. & Rosário, R. (2010).
Motivación y Aprendizaje Autorregulado. Interamerican Journal of Psychology, 44, 1, 86-97.
Valle, A., Núñez, J.C., Cabanach, R., González-Pienda, J., Rodríguez, S., & Rosário, P. (2009). Academic Goals
and Learning Quality in Higher Education Students. Spanish Journal of Psychology, 12(1), 96-105.
Valle, A., Núñez, J. C., Cabanach, R., Rodríguez, S., González-Pienda, J., & Rosário, P. (2009) Perfiles
Motivacionales en Estudiantes de Secundaria: Análisis Diferencial en Estrategias Cognitivas, Estrategias de
Autorregulación y Rendimiento Académico. Revista Mexicana de Psicología, 26 (1), 113-124.
Valle, A., Cabanach, R.G., Núñez, J.C., González-Pienda, J.A., Rodríguez, S., & Piñeiro, I. (2003). Multiple
goals, motivation and academic learning. British Journal of Educational Psychology, 73, 71-87.
Núnez, J. C., González-Pienda, J., Rodríguez, C., Valle, A., González-Cabanach, R. & Rosário, P. (2011).
Multiple Goals Perspective in Adolescent Students with Learning Difficulties. Learning Disability Quarterly,
34, 273-286.

Sonia Fuentes. Universidad Central de Chile, Toesca 1783 Santiago, Chile.

Current themes of research:

Self-regulated learning. Higher education. School achievement. Elementary school.

Most relevant publications in the field of Psychology of Education (max: 5):

Rosário, P.; Núñez, J. C., Trigo, L., Guimarães, C., Fernández, E., Cerezo, R.. Fuentes, S., Orellana, M.,
Santibáñez, A., Fulano, C., Ferreira, A., Figueiredo, M. (in press). Transcultural analysis of the effectiveness
of a program to promote self-regulated learning in Mozambique, Chile, Portugal, and Spain. Higher
Education Research and Development.
Rosário, P., Pereira, A., Núñez, J.C., Cunha, J., Fuentes, S., Polydoro, S., & Gaeta, M. (2013). Non-traditional
university students at University: an explanatory model of the intention to continue studying. Psicothema.

Ricardo Pinto. Universidade do Minho. E-mail: rmnpslb@gmail.com.

Current themes of research:

Self-regulated learning. Approaches to learning. ICT.

Most relevant publications in the field of Psychology of Education (max: 5):

Rosário, P., González-Pienda, J. A., Pinto, R., Ferreira, P., Lourenço, A. & Paiva, O. (2010). Efficacy of the
program “Testas’s (mis)adventures” to promote the deep approach to learning. Psicothema, 22 (4), 828-834.