Learning and Individual Differences 61 (2018) 11–20

Contents lists available at ScienceDirect

Learning and Individual Differences

journal homepage: www.elsevier.com/locate/lindif

Reciprocal effects between self-concept of ability and performance: A T

longitudinal study of children with learning disabilities in inclusive versus
exclusive elementary education☆
⁎
Julia Gorgesa, , Phillip Neumannb, Elke Wilda, Daniela Stranghönera, &Birgit Lütje-Kloseb
a
Department of Psychology, Educational Psychology, Bielefeld University, Germany
b
Faculty of Educational Science, Inclusive Education, Bielefeld University, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Ample empirical research from regular school settings documents reciprocal effects between academic perfor-
Special educational needs mance and academic self-concept of ability (ASC), supporting what is known as a reciprocal effects model
Learning disability (REM). The present article investigates a REM in the domain of reading performance in a sample of elementary
Academic selfconcept of ability students with special educational needs in learning (SEN-L) who received special educational support in ex-
Reciprocal effects model
clusive versus inclusive settings (N = 446). In exclusive settings, SEN-L students attend special schools and are
Inclusive education
completely separated from regular students. By contrast, SEN-L students in inclusive settings attend regular
schools and are educated in classes with regular students. In both settings, SEN-L students are not graded and
taught based on individual learning goals, which may affect reciprocal effects between ASC and reading per-
formance. In addition, given that special education for SEN-L students relies heavily on individual reference
standards to evaluate performance, we tested individual performance growth of SEN-L students as a predictor of
ASC. Analyses of a longitudinal dataset across 3rd and 4th grade revealed some cross-lagged effects and an effect
of performance growth on ASC in exclusive settings in particular. The discussion focuses on the role of in-
dividualized instruction, grades, peer groups, and individual versus social reference standards for reciprocal
effects between ASC and performance as well as practical implications.

1. Introduction effect predicts a negative effect of peer-group performance—i.e., the

In educational psychology, students' academic self-concept (i.e.,
subjective perceptions) of ability is among the most researched theo-
retical constructs (cf. Marsh, 2007). Due to the undeniable importance
of academic self-concept of ability as a key predictor of academic per-
formance (Wigfield, Eccles, Schiefele, Roeser, & Davis-Kean, 2007), a
bulk of literature addresses self-concept development in general and the
role of academic performance for self-concept development in parti-
cular. The present study aims at contributing to this literature by pro-
viding empirical analyses of self-concept development in a hitherto
neglected group of students, that is, students with special educational
needs.
Three models of self-concept development stand out due to their
extensive coverage by empirical studies. First, postulating social com-
parison processes with a generalized other, the big-fish-little-pond
school or class average achievement—on his or her self-concept of
ability (Marsh, 1987; Marsh & Hau, 2003). Second, the internal/ex-
ternal-frame-of-reference model predicts that domain-specific achieve-
ment positively affects within domain self-concept but negatively af-
fects cross-domain self-concepts (Marsh, 1986; Möller, Pohlmann,
Köller, & Marsh, 2009). Third, the reciprocal effects model of causal
ordering (REM) predicts that performance affects subsequent self-con-
cept and vice versa within domains over time (cf., Marsh & Martin,
2011).
Support for the REM of academic self-concept and performance
comes from research on various cultures, age groups and educational
systems (Marsh & Martin, 2011; Valentine, DuBois, & Cooper, 2004).
Although the REM focuses on intra-individual processes, self-concept
theory suggests that performance needs be evaluated against a social,
temporal, or criterial standard to become informative (Marsh, 1986,

☆
The present study was funded by a grant to Elke Wild, Birgit Lütje-Klose, and Malte Schwinger from the German Federal Ministry of Education and Research (support code:
01JC1101, support line: Equity of opportunities and participation [Chancengerechtigkeit und Teilhabe]). The sole responsibility for the article's contents lies with the authors. Parts of the
manuscript have been presented at the 5th conference of the Society for Empirical Educational Research [Gesellschaft für empirische Bildungsforschung] in Heidelberg, 15 March, 2017.
⁎
Corresponding author at: Department of Psychology, Educational Psychology, Bielefeld University, PO Box 100131, 33501 Bielefeld, Germany.
E-mail address: julia.gorges@uni-bielefeld.de (J. Gorges).

https://doi.org/10.1016/j.lindif.2017.11.005
Received 12 December 2016; Received in revised form 25 September 2017; Accepted 3 November 2017
1041-6080/ © 2017 Elsevier Inc. All rights reserved.
J. Gorges et al.
1987). Hence, contextual factors relevant for comparison processes
such as available comparison targets within classrooms may affect
intra-individual reciprocal effects between self-concept and perfor-
mance.
This assumption has led us to investigate the REM within a specific
student population whose schooling systematically differs from their
peers, and which has rarely been studied: students with special edu-
cational needs in learning (SEN-L). Because academic performance of
SEN-L students is substantially lower than the performance of their non-
SEN-L peers (Fuchs, Fuchs, Mathes, Lipsey, & Roberts, 2001; Kocaj,
Kuhl, Kroth, Pant, & Stanat, 2014), special educational support is pro-
vided for these students in various forms. In Germany, SEN-L students
can attend exclusive special or inclusive regular schools. The exclusive
setting means they attend schools in which teachers have specialized
skills to support students' special educational needs and the whole
student population consists of students with SEN-L. In inclusive set-
tings, by contrast, SEN-L students are members of a regular classroom at
a regular school and receive part time special educational support
through special education teachers mostly during regular classes or in
small group settings; although the exact format differs between federal
states or even schools (cf., Werning & Lütje-Klose, 2016). Unlike regular
students, SEN-L students in elementary schools participating in this
study—in exclusive and inclusive settings alike—do not receive formal
grades but individualized reports on their development and are taught
according to individual learning goals.
Using the case of Germany, the present study addresses three re-
search questions. Our first question is whether empirical support for a
REM may be found among SEN-L students. Second, given the distinc-
tion between inclusive and exclusive settings, Germany's school system
provides a suitable setting to examine the role of educational setting for
processes underlying a REM. Hence, our second question is whether an
inclusive versus exclusive setting moderates reciprocal effects between
self-concept and performance. Finally, because of their individualized
learning goals, SEN-L students may draw on an individual frame-of-
reference (i.e., their own performance growth) to evaluate their per-
formances (Lüdtke, Köller, Marsh, & Trautwein, 2005). Therefore, our
third question is whether performance growth over time differentially
predicts self-concept of SEN-L students. Naturally, the former two
questions lead to the idea that educational setting might also moderate
the predictive validity of performance growth.
1.1. The reciprocal effects model in regular classrooms
Ample research documents the close relation between academic
self-concept of ability (SCA) and academic performance (for elementary
education e.g., Guay, Marsh, & Boivin, 2003; Helmke & van Aken,
1995; Weidinger, Spinath, & Steinmayr, 2015; for secondary education
e.g., Marsh, Trautwein, Lüdtke, Köller, & Baumert, 2005; Marsh &
O'Mara, 2008; Retelsdorf, Köller, & Möller, 2014; for overviews see
Marsh & Martin, 2011; Valentine et al., 2004). A meta-analysis by
Valentine et al. (2004) reveals that REMs find strong support when the
respective academic self-concept and academic performance refer to the
same skills (i.e., a particular school subject), the performance measure
has some relevance for the students (i.e., grades that determine edu-
cational progress), and students are informed about their performance
(i.e., grades or feedback communicated by teachers).
The self-concept-performance REM is a product of two models in-
itially developed and tested separately. On the one hand, the self-en-
hancement model posits that positive self-concepts promote students'
motivation to engage in effortful behavior and experience themselves as
self-efficacious, which, in turn, leads to higher academic performance
(Green et al., 2012; Marsh, Byrne, & Yeung, 1999). On the other hand,
the skill-development model posits that academic performance informs
students' self-concepts (Chapman & Tunmer, 1997; Liem, McInerney, &
Yeung, 2015; Poloczek, Karst, Praetorius, & Lipowsky, 2011; Retelsdorf
et al., 2014; Skaalvik & Valås, 1999; Weidinger et al., 2015). The self-
12
Learning and Individual Differences 61 (2018) 11–20
enhancement model primarily relies on intrapersonal processes,
whereas the skill-development model takes contextual characteristics of
students' school experiences into account (Valentine et al., 2004).
When students in elementary school start to take standardized tests
as a class, receive grades on the same scale, and have a peer-group with
roughly the same level of ability to use as a social frame of reference
(e.g., same-aged peers with a similar educational biography) they can
easily compare their own performance to the performance of other
students. Such social comparisons are an important source of in-
formation to develop their self-concepts (Marsh, 2007) and, therefore,
domain-specific performance and self-concepts of ability become more
closely related starting in primary school (cf., Dweck, 2002; Wigfield
et al., 2007; see Rosenholtz & Simpson, 1984, for a detailed account of
how classroom organization affects the formation of ability concep-
tions).
Beyond social comparisons, students' self-concepts may be based on
temporal or intra-individual comparison processes (i.e., comparing
one's current performance to one's previous performance), criterion
comparison processes (i.e., comparing one's performance to a given
standard of excellence), or by drawing inferences from feedback given
by significant others (Möller & Trautwein, 2015). Supporting the re-
levance of such alternative frames of reference, parents' evaluations
affect elementary students' SCAs beyond grades (Entwisle, Alexander,
Pallas, & Cadigan, 1987; Gniewosz, 2010). Regarding intra-individual
reference standards, Lüdtke et al.’s (2005) analyses of a large cross-
sectional dataset show that an individualized teacher frame-of-refer-
ence—that is, teachers evaluating students' performance based on the
student's performance growth rather than social comparison-
s—promotes a close association of a student's self-concept and perfor-
mance within a specific domain.
Overall, various processes underlying the skill-development model
and the self-enhancement model are spelled out in the literature. At
face value these two models specify that there are reciprocal causal
effects between self-concept and performance over time. Factors af-
fecting the skill-development model in particular draw on contextual
features of students' schooling. Given distinctive features of SEN-L
students' school and educational contexts in Germany, our next step is
to review and evaluate to what extent a REM may be conferrable to
inclusive and exclusive settings.
1.2. Characteristics of SEN-L students and provision of special education
support
Students with SEN-L represent the majority of students with special
educational needs in Germany (Bildungsberichterstattung,
Autorengruppe, 2014). However, SEN-L is not uniformly defined in
Germany (Moser, 2012; see Fuchs et al., 2001, for a similar discussion
in the US) and hardly comparable to the US-American definition (Löser
& Werning, 2011). International consensus is that the term SEN-L ty-
pically refers to students whose academic performance—particularly in
basic skills such as reading, writing and arithmetic—is one to two years
behind those of their same-aged peers (Löser & Werning, 2011; Werning
& Lütje-Klose, 2016). Although some authors consider SEN-L to be a
mild intellectual disability, a below average cognitive ability is no
constitutive element of diagnosing SEN-L. Other individual and con-
textual risk factors leading to low academic performance are taken into
account when diagnosing SEN-L (Werning & Lütje-Klose, 2016). Iso-
lated specific learning difficulties such as dyslexia or dyscalculia are not
recognized as SEN-L per se, but often accompany it.
The German school system provides two prototypical forms of
special educational support. Exclusive education via special schools for
SEN-L students only is available from first grade until the end of com-
pulsory education. The basic idea of this type of school is that “students
with similar special needs are taught in small groups by special edu-
cation teachers” (Löser & Werning, 2011, p. 91). Special education
teachers are trained to use individual frame-of-reference to support
J. Gorges et al.
students' performance growth. Hence, exclusive education is supposed
to offer better support and a protected environment for students with
SEN-L. Alternatively, inclusive education for SEN-L students as part of a
regular classroom at a regular school, in which teachers may emphasize
social comparisons underlying educational credentials (Carson, 2007;
Fend, 2006), is available from grade 1 to grade 10. In inclusive settings,
SEN-L students are supposed to be better integrated with their peers
and to reach a higher performance level due to a more stimulating and
demanding environment (Werning & Lütje-Klose, 2016). On average,
three SEN-L students get integrated in classes of about 18 regular stu-
dents (FORSA, 2015). Typically, they share all classes with regular
students but may have an additional special education teacher during
scheduled classes and/or additional classes. Recent empirical studies
suggest that low-achieving SEN-L students self-select in exclusive set-
ting, whereas SEN-L students with higher competencies tend to attend
inclusive schools (Kocaj et al., 2014; Wild et al., 2015).
In both inclusive and exclusive settings, SEN-L students are mostly
taught and evaluated based on individual learning goals that are in-
dependent of the general curriculum. Teachers are supposed to use an
individual frame-of-reference when evaluating SEN-L students, re-
fraining from grading them unless they or their parents explicitly ask
for grades; and this is only possible from fourth grade onward.
1.3. Factors that may affect reciprocal relations between self-concept and
performance
Contextual factors of special educational support provision for SEN-
L students in Germany may affect reciprocal effects between SEN-L
students' self-concept and performance. Without formal grades of their
performance in elementary school it may be more difficult for SEN-L
students to engage in social comparisons. However, empirical studies
addressing the role of grades as opposed to verbal teacher feedback in
regular classrooms do not suggest systematic differences due to grading
practices (Lipowsky, Kastens, Lotz, & Faust, 2011; Wagner & Valtin,
2003). Hence, self-concept and performance should reciprocally affect
each other regardless of grading practices in the classroom.
SEN-L students are taught based on their individual learning goals.
This may make evaluating one's performance more difficult because
social comparison processes rely on the assumption of equalized
learning goals that are attained by students to varying extents. Hence,
individualized instruction—i.e., learning goals and lessons that are
different between students—means that performance feedback may not
be comparable due to differences in task difficulty. To place their per-
formance in their peer group, students would have to assess whether
the learning goals that had to be attained were comparable prior to
comparing their performance. Consequently, students should engage in
social comparison processes only if tasks with comparable levels of
difficulty were assigned to several students.
It is not quite clear whether and how SEN-L students engage in
social comparison given individualized instruction. SEN-L students may
rely on teacher feedback, which could underlie a reciprocal causal or-
dering (Wagner & Valtin, 2003), or their self-concept development may
be separated from their performance. Assuming that teachers may try to
boost SEN-L students' self-concepts by giving overly positive feedback,
teacher feedback may contribute to such separate developmental pro-
cesses. If students still engage in social comparison, the educational
setting may act as a moderator because students may dispose of a
greater number of relevant social comparison targets, that is, compar-
able peers, in exclusive educational settings (Möller, Streblow, &
Pohlmann, 2009).
Without standardized evaluation criteria, SEN-L students may use
their individual performance growth to inform their self-concept.
Focusing on an individual frame-of-reference may help students to in-
terpret their behavior and performance in a positive way. For example,
Tabassam and Grainger (2003) found that an intervention of attribu-
tional retraining successfully increased self-concept for students with
13
Learning and Individual Differences 61 (2018) 11–20
SEN-L. Thus, SEN-L students may draw primarily on intra-individual
comparisons and use themselves as a benchmark to inform their aca-
demic self-concept (Rheinberg & Krug, 1999). Accordingly, students'
individual performance growth should predict their self-concept rather
than a (specific) previous performance.
While SEN-L students in exclusive settings are taught exclusively by
special education teachers who are trained to use primarily individual
frames-of-reference, those in inclusive settings tend to spend their time
in class with regular teachers who are supported by special education
teachers part time. Hence, students in inclusive settings may face—at
least vicarious—performance evaluations based on social reference
norms. Therefore, they may be less inclined to rely on individual de-
velopments to evaluate their performance. Using social comparison,
however, will likely result in consistently lower self-concept of students'
in inclusive settings, even if their performance improves (e.g., Sauer,
Ide, & Borchert, 2007; Tracey, Marsh, & Craven, 2003; see Lindsay,
2007, for a review). Following this line of reasoning, the educational
setting may moderate the effect of individual performance growth on
self-concept in exclusive settings in particular.
1.4. The present study
To date, reciprocal effects between self-concepts and performance of
SEN-L students have scarcely been investigated, and the role of in-
clusive versus exclusive setting has not been addressed specifically.
Focusing on elementary school students, the present study addressed
reciprocal effects between self-concept and performance in the domain
of reading, which is the most basic skill for education and social par-
ticipation (OECD, 2010). The goal of the present study was threefold.
First, we investigated whether reciprocal effects between self-concept
and performance can be found in a sample of SEN-L students. We hy-
pothesized that self-concept (performance) will predict subsequent
performance (self-concept). Second, we tested the potential moderating
effect of the educational setting, that is, whether special educational
support was provided in an inclusive versus an exclusive setting. As self-
concept should affect performance due to intrapersonal processes (e.g.,
striving to enhance one's self-image or affecting motivational media-
tors; Valentine et al., 2004), self-concept was hypothesized to predict
subsequent performance in both settings. By contrast, assuming that
SEN-L students will find more appropriate comparison targets to eval-
uate their reading performance and, thus, align their self-concept to
their performance based on social comparisons with–in terms of per-
formance–similar peers in exclusive rather than in inclusive settings, we
hypothesized stronger effects of performance on students' self-concept
in exclusive than in inclusive settings. Third, assuming that contextual
characteristics of special educational support bring intra-individual
comparisons to the fore, we examined the effect of individual perfor-
mance growth on students' self-concept at the end of primary school. An
increase in performance across the three measurement points was ex-
pected to positively predict self-concept at the final measurement point
(while controlling for previous self-concept). Educational setting was
again tested as a potential moderator and we expected stronger effects
in exclusive than in inclusive settings due to individual reference
standards emphasized by special education teachers. Analyses were
based on a longitudinal dataset with three measurement points and
subject-specific assessments of self-concept and performance.
2. Method
2.1. Sample & Procedure
Data stems from a third-party funded research project that aimed at
comparing SEN-L students' psychosocial and performance develop-
ments in inclusive versus exclusive educational settings (Wild, Lütje-
Klose, Schwinger, Gorges, & Neumann, 2017). The study was carried
out in accordance with the recommendations of the German
J. Gorges et al. Learning and Individual Differences 61 (2018) 11–20

Psychological Association's guidelines for conducting research with Table 1

human subjects. Because the students participating in the study were Internal consistency and zero-order correlations of academic self-concept and perfor-
mance measures.
underage, written consent for their participation in the study had been
obtained from all parents prior to the time of data collection. Recruit- M SD α (2) (3) (4) (5) (6)
ment was based on a list of 3779 schools attended by children with
special educational needs in Germany's federal state of North-Rhine (1) SCA t1 3.17 0.89 0.87 0.55 0.47 0.36 0.34 0.31
(2) SCA t2 3.16 0.88 0.89 – 0.71 0.30 0.41 0.43
Westphalia. At t1, the researchers contacted 441 schools that had re-
(3) SCA t3 3.26 0.80 0.87 – 0.27 0.34 0.38
ported particularly high numbers of SEN-L student. For lack of na- (4) PERF t1 26.88 17.94 – – 0.90 0.86
tionwide standards the assessment, determination and labeling of stu- (5) PERF t2 38.13 21.47 – – 0.94
dents with SEN-L differs between federal states, administrative districts (6) PERF t3 48.43 23.00 – –
or even schools. Therefore, students in this study have been identified
Note. M = mean, SD = standard deviation, SCA = reading self-concept of ability,
based on the provision of SEN-L support as reported by the schools.
PERF = reading performance, significant correlation coefficients are printed in bold
At the time of the study, SEN-L could have been officially re- (p < 0.05).
cognized based on a non-standardized diagnostic procedure leading to
eligibility of additional resources from the ‘Land’ (i.e., the federal state), time period. Hence, the test captures students' literacy based on both
or it could have been recognized within a school leading to resource speed and power. The ELFE test has been validated using a sample of
allocation from the school's budget. A total of 271 schools actually had regular students and students with SEN with high internal consistency
SEN-L students in third grade in the school year 2012/13. Of these (0.92 < α < 0.97) across subtests. In our sample, the test-retest re-
schools, 45% agreed to participate in the research project. Next, the liability ranged between rtt = 0.74 und rtt = 0.92, which corresponds
parents of the respective children (N = 740) were contacted and in- to the test-retest reliability reported by Lenhard and Schneider (2006).
vited to participate in the study. The procedure to ask SEN-L students ELFE test scores correlate substantially with teachers' evaluations of
and their parents in our target schools to participate was repeated at t2 students reading performance (r > 0.71¸ Lenhard & Schneider, 2006).
and t3 to include students that were subsequently new to the school. Analyses were based on the sum of subtest raw scores at each mea-
Every family who participated at least in one measurement wave re- surement point.
ceived a 20-Euro voucher after the third measurement wave.
On three occasions from the beginning through middle of third
2.3. Statistical analyses
grade (t1: Oct. 2012 – March 2013) and fourth grade (t2: Sep. 2013 –
Jan. 2014) as well as at the end of fourth grade (t3: March 2014 – June
2.3.1. Handling of missing data
2014), participating students (N = 461) filled in a self-report ques-
The proportion of missing values on variables used in our analyses
tionnaire in a one-on-one session with the assistance of a trained in-
ranged between 13.4% at t1 and 7.6% at t2 for reading performance
vestigator and participated in a standardized reading test administered
and between 11% at t1 and 6.3% at t3 for reading SCA. We used
to groups of not more than four children. To take school into account as
multiple imputation (Schafer & Graham, 2002) based on the mice
a clustering variable, only children who did not change schools between
package (van Buuren & Groothuis-Oudshoorn, 2011) with m = 5 to
t1 and t2 (n = 13) or t2 and t3 (n = 2) remained in the final sample
impute missing values and pool imputed datasets.
(N = 446; 45% girls; age: M = 8.74; SD = 0.66; Min = 8, Max = 11 at
the beginning) The average interval between measurements was
M = 45.51 (SD = 6.39) weeks between t1 and t2, and M = 26.28 2.3.2. Measurement models and tests of measurement invariance
(SD = 4.25) weeks between t2 and t3. The questionnaire and the tests Following the recommendation by Marsh et al. (1999) we first es-
usually took place on two consecutive days, beginning with the ques- tablished measurement models for SCA. Using confirmatory factor
tionnaire on the first day. analyses (CFA), we specified multiple-group models with a latent
Of the participating children, 199 attended exclusive schools and variable indicating SCA at each measurement point. We accounted for
247 children attended inclusive, regular elementary schools. From the correlated uniqueness (Marsh et al., 1999) by specifying residual error
total of 49 exclusive schools, the sample contained between 1 and 14 correlations of parallel items across measurement points. We inspected
children from the same school (mode = 3). From a total of 110 in- the fit of our measurement models and whether factor loadings for all
clusive schools, the sample contained between 1 and 8 children from items were above 0.40 and significant (p < 0.05).
the same school (mode = 1). The exclusive subgroup had more boys To make certain that the measures to assess SCA are equivalent
than the inclusive subgroup (61.8% boys in exclusive settings and across groups and across measurement points we next tested weak
50.2% in inclusive settings). Students' cognitive ability was within the measurement invariance required for comparisons of variable inter-
average range (Stranghöner et al., 2017). There was no indication of relations (Meredith, 1993; Steenkamp & Baumgartner, 1998): Building
systematic dropout. on the baseline model, we specified a model that had factor loadings for
parallel items constrained to be equal across measurement points and
2.2. Measures compared model fit of these models. Weak measurement invariance
may be assumed as long as model fit does not decrease significantly due
Academic self-concept of ability in the domain of reading (SCA) was to imposing these equality restrictions. Measurement invariance across
assessed using three items developed by Schöne, Dickhäuser, Spinath, groups was tested as first step of the multiple-group structural equation
and Stiensmeier-Pelster (2002), which were adapted to be used for this modeling by constraining item loadings to be equal across groups (see
particular sample based on two pilot studies (e.g., “I think reading is below).
easy.”; Schwinger et al., 2015). Responses were coded using a 4-point
Likert-type scale (1 = “applies not at all”, 2 = “hardly applies”, 2.3.3. Reciprocal effects models for SEN-L students
3 = “applies somewhat”, 4 = “applies completely”). SCA measures Next, we turned to investigating the proposed REM. According to
showed good internal consistency (see Table 1). recommendations by Marsh et al. (1999), a good way to investigate
Reading performance was assessed using standardized test results reciprocal effects between performance and SCAs is to specify a full-
from the widely-used ELFE test for reading comprehension in grade 1 to forward model. In this model, performance and SCA predict all fol-
6 (Lenhard & Schneider, 2006). The ELFE comprises three subtests re- lowing assessments of performance and SCA, respectively (stability
presenting increasing literacy levels: Students taking the ELFE test have coefficients), as well as the respective other constructs (see Fig. 1). The
to identify written words, sentences, and paragraphs within a specific model accounted for correlated uniqueness of the SCA items and

14
J. Gorges et al.
correlations between performance and SCA for each measurement
point. We then fitted the data from the full sample to the full-forward
REM. Because variances of performance measures were considerably
larger than for SCA measures due to scaling differences, we used z-
standardized performance scores.
2.3.4. Reciprocal effects models with educational setting as moderator
To include educational setting as a dichotomous moderator, we
specified an unconstrained multiple-group SEM with inclusive versus
exclusive education as the grouping variable (Model A). We established
weak measurement invariance (Meredith, 1993; Steenkamp &
Baumgartner, 1998) by constraining factor loadings (Model B), and
then determined whether the groups significantly differ by additionally
constraining regression paths to be equal across groups (Model C). At
each step, we compared model fit to detect significant changes. A sig-
nificant change in model fit of Model B compared to Model C would
support the hypothesized moderation by educational setting.
2.3.5. Latent growth curve modeling with educational setting as moderator
Latent growth curve modeling is a specific application of SEM to
evaluate developmental trends over time (for an overview see Preacher,
Wichmann, MacCallum, & Briggs, 2008). This method is based on a
regression-type curve for each participant across all measurement time
points that is described by two latent factors indicating the intercept
(i.e., the starting point of the curve) and the slope (i.e., the increase or
decrease over time). The factor loadings of the intercepts are fixed to 1
at each measurement point. Factor loadings of the latent slope factor
are set to 0 at t1, to 1 at t2 (10 months after t1), and to 1.7 at t3
(7 months after t2). Thus, factor loadings were set to represent ap-
proximately linear changes. To investigate whether changes in perfor-
mance—as opposed to the status quo of performance or the latest
available performance—affect SCA we used latent growth curve mod-
eling to evaluate increase or decrease in student performance from t1 to
t3 and then used this information (i.e., the slope) to predict SCA at t3
15
Learning and Individual Differences 61 (2018) 11–20
Fig. 1. Full-forward REM based on the full
sample of students with SEN-L.
Note. SCA = reading self-concept of ability,
PERF = reading performance, significant coeffi-
cients (p < 0.05) are printed in bold.
while controlling for SCA at t1. After fitting the data from the full
sample, the moderating effect of the educational setting was tested
using multiple-group modeling as described above.
2.3.6. Goodness of fit
All models were fit to the data using the robust Maximum
Likelihood Estimator (MLR) implemented in the lavaan package
0.5–19.848 (Rosseel, 2012) in R 3.3.0 (R Core Team, 2015). To account
for the complex data structure, we adjusted standard errors using la-
vaan.survey (Oberski, 2014) with school specified as clustering vari-
able. We used several indicators to assess goodness of fit for the models
tested in this study: The comparative fit index (CFI), the root-mean-
square error of approximation (RMSEA), and the standardized root
mean square residual (SRMR) (Hu & Bentler, 1999; Schermelleh-Engel,
Moosbrugger, & Müller, 2003). The CFI ranges between 0 and 1 with
values > 0.95 indicating good and > 0.90 indicating acceptable model
fit, respectively. Similar to the CFI, values > 0.95 indicate good fit and
values > 0.90 indicate acceptable fit. Regarding the RMSEA, va-
lues < 0.08 indicate acceptable fit and values < 0.05 indicate good fit.
The SRMR ranges between 0 and 1 with values < 0.08 indicating a
good fit. To identify significant differences in model fit we relied cutoff
criteria suggested by Chen (2007) and Cheung and Rensvold (2002):
Model fit can be considered equally fitting when the change in CFI
is < 0.01 and the change in RMSEA is < 0.015.
3. Results
3.1. Descriptive statistics and correlations
Descriptive statistics and zero-order correlations for the full sample
are shown in Table 1. As expected, SCA and performance showed sig-
nificant concurrent correlations on each measurement occasion. Table 2
depicts means and standard deviations for SEN-L students in each set-
ting. Apparently, SEN-L students in both setting improved their reading
J. Gorges et al. Learning and Individual Differences 61 (2018) 11–20

Table 2
Descriptive statistics and bivariate correlations for SEN-L students in exclusive versus inclusive settings.

Exclusive setting Inclusive setting

M SD (2) (3) (4) (5) (6) M SD (2) (3) (4) (5) (6)

(1) SCA t1 3.07 0.96 0.539 0.482 0.324 0.320 0.329 3.25 0.84 0.571 0.450 0.393 0.364 0.330
(2) SCA t2 3.14 0.97 0.713 0.342 0.534 0.556 3.18 0.80 0.711 0.339 0.434 0.399
(3) SCA t3 3.24 0.90 0.316 0.458 0.509 3.28 0.72 0.265 0.315 0.341
(4) PERF t1 16.17 14.06 0.826 0.771 34.22 16.48 0.879 0.846
(5) PERF t2 24.88 17.17 0.930 48.10 18.86 0.918
(6) PERF t3 36.19 20.10 57.93 20.42

Note. SCA = reading self-concept of ability, PERF = reading performance, M = Mean, SD = Standard Deviation.

performance over time, whereas the SCA seemed to be rather stable (see
Stranghöner et al., 2017, for a detailed analyses of performance dif-
ferences across time and educational setting). As may be expected based
on previous studies, reading performance of students with SEN-L in
exclusive settings appeared to be lower.
3.2. Measurement models and measurement invariance
The multiple-group CFA for self-concept showed significant factor
loadings (all loadings > 0.60) for each item (p < 0.05). Factor-corre-
lations of all models were significant (p < 0.05) and ranged between
r = 0.52 and r = 0.79.
Model fit for the three-factor model accounting for correlated un-
iqueness was good (χ2 = 23.826; df = 15; p = 0.068; CFI = 0.995;
RMSEA = 0.042 [90%-CI: 0.000–0.068]; SRMR = 0.017). Correlated
uniqueness coefficients were positive and significant for item 1 between
t1 and t3, and item 2 and item 3 between t2 and t3. The remaining
correlated uniqueness coefficients were both positive and negative, but
non-significant. Model fit did not change significantly when factor
loadings were fixed across measurement points (χ2 = 27.929; df = 19;
p = 0.085; CFI = 0.995; ΔCFI = 0.000; RMSEA = 0.037 [90%-CI:
0.000–0.064]; ΔRMSEA = 0.005; SRMR = 0.024) supporting the as-
sumption of weak measurement invariance.
3.3. Reciprocal effects model (REM)
The first goal of the present study was to test reciprocal effects
between SCA and performance over time for students with SEN-L.
Fitting the full-forward REM to the full sample of students with SEN-L
yielded good model fit (χ2(33) = 44.552, p = 0.086; CFI = 0.997;
RMSEA = 0.28 [90%-CI: 0.000–0.051]; SRMR = 0.020). Path coeffi-
cients and correlations between SCA and performance for each mea-
surement point are shown in Fig. 1. Stability coefficients for both SCA
and performance were high and significant (SCA: β > 0.561; perfor-
mance: β > 0.845); correlations between SCA and performance on
each measurement occasion were moderate and significant
(0.221 < r < 0.388). Regarding the cross-lagged paths that are at the
heart of the REM, performance at t1 predicted SCA at t2 (β = 0.102),
and SCA at t2 predicted performance at t3 (β = 0.062).
3.4. Reciprocal effects model with educational setting as moderator
Considering exclusive versus inclusive educational setting as a
moderator, we first compared an unconstrained model with a model
that had factor loadings constrained to be equal across groups (i.e.,
inclusive versus exclusive setting) to test the assumption of weak
measurement invariance. As can be seen from the results presented in
Table 4, weak measurement invariance may be assumed. Again, only
few significant cross-lagged effects reached significance (see Fig. 2).
Nevertheless, in line with our hypothesis, significant cross-lagged ef-
fects showed in the exclusive setting: t1 performance predicted t2 SCA
(β = 0.172; p < 0.05), t2 SCA predicted t3 performance (β = 0.092;
p < 0.05), and path coefficients were consistently larger in the ex-
clusive setting. Some other path coefficients also ranged around
β = 0.10 or more but did not reach significance. Overall, however,
these findings did not establish significant differences of the structural
model across groups, which can be seen from comparing the model with
factor loadings constrained to be equal across groups to the model with
factor loadings and regression paths constrained to be equal across
groups (Model B and Model C in Table 3).
3.5. Latent growth curve analyses and educational setting as moderator
Based on the full sample of SEN-L students, a latent growth curve
model fitted the data well (χ2(28) = 38.066; p = 0.10; CFI = 0.997;
RMSEA = 0.030 [90%-CI: 0.000–0.052]; SRMR = 0.047). SCA at t3
was significantly predicted by SCA at t1 (β = 0.546, p < 0.05) and
performance growth from t1 to t3 (β = 0.343, p < 0.05). Results from
multiple-group latent growth curve modeling showed good model fit
and no significant change in model fit with weak measurement in-
variance imposed (see Table 3). Students' past performance growth—-
that is, the slope of reading performance across the three measurement
points—significantly predicted their SCA at t3 in exclusive settings
(β = 0.546, p < 0.05) but not in inclusive setting (β = 0.316,
p = 0.065) beyond SCA at t1 (exclusive: β = 0.511, p < 0.05; in-
clusive: β = 0.576, p < 0.05). Again, constraining regressions to be
equal did not result in a significant change of model fit (see Table 3,
Model 2 versus Model 3).
4. Discussion
In the present study, reciprocal effects between academic self-con-
cept of ability and performance in the domain of reading were in-
vestigated in a sample of SEN-L elementary students; including tests for
a moderating impact of an inclusive versus exclusive educational set-
ting and of individual performance growth as a predictor of self-concept
of ability. Results documented cross-lagged effects in terms of perfor-
mance in third grade predicting self-concept of ability in early fourth
grade, and for reading self-concept of ability in early fourth grade
predicting performance at the end of fourth grade. Testing educational
setting as a moderator did not reveal differential effects of exclusive
versus inclusive school setting but suggested that the abovementioned
cross-lagged effects have been driven primarily by students in exclusive
settings. Results from latent growth curve models showed that in-
dividual performance growth (i.e., the slope of the curve) significantly
predicted self-concept of ability in the full sample, which again seemed
to be driven by students in exclusive educational settings although the
structural model have been invariant.
4.1. The role of individualized instruction and (no) grading
Despite empirical evidence of a REM for regular students from first
grade onwards even without grades (Helmke & van Aken, 1995;
Wagner & Valtin, 2003), our results indicate that a REM may not

16
J. Gorges et al. Learning and Individual Differences 61 (2018) 11–20

Fig. 2. Full-forward REM, coefficients for both

settings (exclusive/inclusive) with weak mea-
surement invariance imposed.
Note. SCA = reading self-concept of ability,
PERF = reading performance, significant coeffi-
cients (p < 0.05) are printed in bold.

Table 3
Model fit for multiple-group structural equation models and latent growth curve models.

χ2 df p CFI ΔCFI RMSEA [90%-CI] ΔRMSEA SRMR

Structural equation models

Model A 74.084 66 0.231 0.998 0.025 [0.000–0.050] 0.022
Model B 81.108 72 0.216 0.997 − 0.001 0.025 [0.000–0.049] + 0.000 0.028
Model C 90.919 84 0.284 0.998 + 0.000 0.020 [0.000–0.045] − 0.005 0.036

Latent growth curve models

Model 1 68.487 56 0.122 0.996 0.033 [0.000–0.057] 0.044
Model 2 73.365 60 0.115 0.995 − 0.001 0.033 [0.000–0.056] + 0.000 0.045
Model 3 73.673 62 0.147 0.996 + 0.001 0.030 [0.000–0.054] − 0.003 0.045

Note. Model A: full-forward model with correlated uniqueness, unconstrained across groups; Model B: Model A with factor loadings of academic self-concept measures constrained to be
equal across groups; Model C: Model B with all regression paths constrained to be equal across groups; Model 1: latent growth curve model with previous self-concept and performance
growth predicting self-concept at t3, unconstrained across groups; Model 2: Model 1 with factor loadings of academic self-concept measures constrained to be equal across groups; Model
3: Model 1 with both regression paths constrained to be equal across groups; for further description see text.

generalize to students with SEN-L in inclusive settings in particular.
Given that we did find cross-lagged effects for SEN-L students in ex-
clusive setting, alignment of self-concept and performance appears to
rely more on teacher feedback and on peers than on grading practices
or individualized learning goals, which are probably consistent in both
inclusive and exclusive setting.
In addition, using individual learning goals as a framework for
teaching is specific to special education in both inclusive and exclusive
settings. To date, it is unclear how this teaching approach may affect
students' self-concept development. In fact, little is known about how
teachers implement individual learning goals in their classrooms. With
respect to self-concept development, for example, setting an individual
learning goal for each and every students as opposed to setting the same
individual learning goal for pairs or small groups of students with si-
milar performance may have differential effects because sharing a
learning goal probably facilitates social comparison processes. Hence,
how teachers set learning goals for students may act as a moderator in
self-concept development, whereas the educational setting (i.e.,
exclusive versus inclusive) may not. Consequently, future studies need
to assess how teachers teach in the classroom to take these differential
procedural variables into account when investigating student develop-
ment.
With individual learning goals, teachers' abilities to adequately
evaluate individual learning progress—of low-achieving students in
particular—appears to be an important factor for students' self-concept
development. Teachers' abilities to set individual learning goals in
student's zones of proximal development (Hedegaard, 2005) and to
raise students' awareness with respect to their performance growth
might be the most important form of support in their learning progress
and self-concept development. Given that special education teachers
are specifically trained for such tasks, further research on regular tea-
chers' abilities to use information on student performance for feedback
and support seems desirable. However, given that we did not find
systematic differences between exclusive and inclusive educational
settings, the intensity of special education support (performed by spe-
cial education teachers) may not be as relevant as one may think.

17
J. Gorges et al.
On the one hand, decoupling self-concept of performance may be
beneficial by protecting SEN-L students from negative perceptions of
their own ability. On the other hand, however, decoupling self-concept
of performance precludes an at least approximately realistic self-eva-
luation, which is important for future achievement-related choices
(Wigfield & Eccles, 2000). Perhaps SEN-L students in inclusive settings
will show reciprocal effects only later in their educational biography, or
reciprocal effects depend on some other attribute of teaching and/or
performance feedback such as its frequency, quality or wording to take
effect. Overall, key questions to address in further research comprise
the role of grading, verbal teacher feedback and other potential influ-
encing factors in inclusive classes for students' self-concepts and per-
formance developments.
4.2. The role of social comparison targets
Although our results do not support concluding systematic differ-
ences of the REM as a function of educational setting, evidence from
regular classrooms (Huguet et al., 2009) suggest that having seemingly
similar students in a class—who, in our case, may be fellow SEN-L
students—is relevant for whether or not social comparison processes
take place. Hence, the availability of reasonable comparison targets
may depend on the educational setting after all. In addition, sharing
individual learning goals may enable social comparison, or students
may use teacher feedback—that is, compare teacher feedback so-
cially—regardless of the performance underlying the teacher's evalua-
tion. Future studies addressing performance and self-concept of full
classrooms (i.e., collecting data from all the children in a class) could
help answer these questions. In addition, assessing comparison pro-
cesses closer to the moment they actually occur appears to be an in-
teresting avenue for future research on how students (in any setting)
choose comparison targets.
Drawing on effect sizes rather than statistical significance, results
reveal path coefficients that may well be significant when using a larger
sample for SEN-L students in exclusive settings in particular. Hence,
SEN-L students in exclusive settings may align their self-concept and
performance after all, thus supporting the hypothesis regarding the
relevance of social comparison targets. In light of our findings, in-
vestigations of SEN-L students' self-concept development using larger
samples from exclusive and inclusive settings alike are needed to shed
light onto the relevance of reciprocal effects between self-concept and
performance.
4.3. The role of individual reference standards for developing ability
conceptions
Individual reference standards in evaluating one's own performance
(Lüdtke et al., 2005) appear to be used in exclusive settings in parti-
cular, which is in line with our assumption regarding special education
teacher training. Because using an individual reference standard is not
standardized, this strategy offers room for substantial variation. For
example, the time period considered for judging improvement may be
chosen by the individual or the teacher or in this case, the researchers.
Similarly, teachers may help to evaluate whether a particular perfor-
mance gain meets general expectations or whether it is above or below
what can typically be expected of a student with similar prerequisites.
Such variation may also be exploited to give SEN-L students the most
positive feedback possible to boost their self-concepts. At this point,
there is too little empirical evidence on the interrelation of individual,
social and criterion comparisons provided by teachers and used by
students. Hence, future studies should take the reference standard used
by students and teachers into account to estimate their differential in-
fluence on self-concept of ability.
In light of the present findings, individual performance growth of
SEN-L students appears to be an important source of information for
their self-concept development. Further, given that previous research
18
Learning and Individual Differences 61 (2018) 11–20
mostly used full-forward structural equation models as REMs, the im-
portance of individual performance growth may not be limited to stu-
dents with SEN-L (see Weidinger et al., 2015). Thus, future studies
applying latent growth curve modeling to regular student samples are
necessary to learn about the role of individual performance growth for
self-concept development across different student groups, older stu-
dents and students without disabilities in particular.
4.4. Practical implications
According to the ability formation theory put forward by Rosenholtz
and Simpson (1984), contextual factors in school are key determinants
of the development of ability conceptions. More specifically, a uni-
dimensional classroom organization in which all students work on the
same tasks, are evaluated publically on the same standard, and grouped
according to ability should foster the development of highly differ-
entiated patterns of ability conceptions (i.e., distinctions into low-,
medium- and high-achieving students). By contrast, classroom organi-
zations that enable selective social comparisons (e.g., comparing effort
rather than test performance) and are based on grouping students in-
dependent of their abilities leads to more students with moderately
positive ability conceptions and less differentiation. The principle of
individual learning goals focuses on establishing such a multi-
dimensional context in which every student may excel at something,
and this should be beneficial for SEN-L (and all other) students in any
type of education.
Teacher feedback is presumably an important source of information
for SEN-L students to develop their self-concepts of ability. However,
when regular teachers meet new SEN-L students, they may have little
experience with performance levels common to SEN-L students as well
as no individual reference for these students. Hence, they likely face
difficulty giving adequate feedback to these students. In such cases, it
may be helpful to receive information on the students' previous per-
formances (e.g., in their former classes or schools) or to get advice from
SEN-L teachers. This would be in line with the strong evidence of the
importance of interdisciplinary collaboration between special educators
and regular teachers in inclusive settings (Friend, Cook, Hurley-
Chamberlain, & Shamberger, 2010). The downside of this information
transfer is, however, that potentially unfounded or biased negative
evaluations may be transferred as well.
4.5. Limitations and outlook
Our study was based on the assessment of SCA and performance on
three occasions. Thus, it meets requirements to apply a REM as outlined
by Marsh et al. (1999). Measures to assess academic self-concepts of
ability have been adapted and pretested twice. Data collection was
conducted by student assistants on a one-to-one basis to ensure valid
data. Despite these efforts to generate informative REM tests some
limitations exist. In our study, the period between measurement waves
was longer between t1 and t2 (10 months) than between t2 and t3
(7 months). Furthermore, the transition from third grade to fourth
grade was between t1 and t2 introducing a potential confounding
context effect; however, teachers and peers typically do not change
between third and fourth grade, and these are the key components of
context.
Our sample of SEN-L students in exclusive and inclusive education
although broad, was not representative of all of Germany and not as
large as in some previous research on the REM (e.g., Guay et al., 2003;
Marsh et al., 2005). It was, however, quite large compared to previous
studies focusing on SEN-L students. Given the prevalence of SEN-L
students, which is about 2.64% of the student population in Germany,
of which one third attends inclusive schools (Klemm, 2015), recruiting
a sample comparable in size to sample sizes of regular student REM
studies is quite challenging. In previous studies, only Möller et al.
(2009) report results based on a sizable sample of 270 SEN-L students,
J. Gorges et al.
which were recruited from exclusive schools only. Hence, although our
study may have failed to detect significant differences due to limitations
in sample size, it nevertheless adds to the scarce literature on SEN-L
students' self-concept development based on a sufficient sample size to
use sophisticated methodology.
Results may be biased by the substantial difference in the perfor-
mance levels of children in inclusive versus exclusive settings (also see
Stranghöner et al., 2017). However, tests of measurement invariance
indicate that children in both settings interpret the items used to assess
self-concept in a comparable way. Similarly, a recent study by Heine
et al. (subm.) has demonstrated comparability of a standardized test
assessing cognitive ability in both settings. In addition, potential effects
of the weaker reading performance of SEN-L students in exclusive set-
tings would primarily be associated with weaker reciprocal effects be-
tween self-concept and performance due to a lack of self-reflection. Our
results show that reciprocal effects are stronger rather than weaker in
exclusive settings. Therefore, we conclude that the performance dif-
ference between SEN-L students from both settings does not affect re-
ciprocal effects between self-concept and performance.
Self-concept development has mainly been conceptualized as a
function of in-class (or in-school) processes (Marsh, 2007). Thus, an
objective in-survey test may not be sensitive to detect in-class
achievement processes that undergird self-concept development. Ac-
cordingly, standardized tests associate with weaker reciprocal effects
compared to teacher ratings or grades, which are more relevant to and
transparent for students (Valentine et al., 2004). Because the students in
our sample did not receive grades, however, grades were not available
as indicators of performance. In addition, as we tested a basic skill, test
scores should be reasonably related to individual learning goals and
performance feedback from teachers (Lenhard & Schneider, 2006).
Young children typically show weaker reciprocal effects (Valentine
et al., 2004). Students in our sample were relatively young and—con-
sidering their SEN-L—may not reflect about their performance as much
as regular students. Hence, one could argue that their self-concepts of
ability are less aligned to their performance because these students
focus less on evaluating their performance against one reference stan-
dard or the other. However, other researchers document reciprocal
effects between self-concepts and performance for students as early as
first grade, which suggests that even children with a competence de-
velopment lower than that of our target third grade children (due to the
age difference) are able to draw inferences from their performance to
shape their SCAs.
Although the goal of our study was to shed light on the role of social
comparison processes, these were not directly measured or observed.
Study designs that are close to the core processes under investigation,
for example, by using observations, face-to-face interviews or the day
reconstruction method (Kahneman, Krueger, Schkade, Schwarz, &
Stone, 2004), may give a more detailed account of what happens in
exclusive and inclusive classrooms with respect to academic self-con-
cept development.
Finally, our study was limited to SEN-L students of a specific age
range and to reading performance. Future studies should address stu-
dents with other disabilities in various educational settings, at various
ages and with respect to mathematical performance as well to examine
the generalizability of the results presented here. Particular attention
should be paid to analyzing the effect of contextual factors and the
mechanisms potentially underlying or replacing the skill-development
and the self-enhancement model in inclusive settings.
4.6. Conclusion
The present study is among the first comprehensive investigations of
self-concept development in children with SEN-L using a longitudinal
dataset comparing prototypes of special education support. Results
suggest that self-concept development follows theoretical assumptions
in exclusive settings, whereas self-concept and performance appear to
19
Learning and Individual Differences 61 (2018) 11–20
develop separately in inclusive setting across the 3rd and 4th year in
elementary school. These findings may be explained by students pur-
suing individualized learning goals, teachers using individual reference
standards to evaluate student performance, and differences in class-
room composition. In the light of the current modification of segregated
school systems into inclusive school systems, more research is needed to
address motivational and competence development under less-stan-
dardized conditions.
References
Bildungsberichterstattung, Autorengruppe (2014). Nationaler Bildungsbericht 2014.
[National educational report 2014.]. Bielefeld: Bertelsmann.
van Buuren, S., & Groothuis-Oudshoorn, K. (2011). Mice: Multivariate imputation by
chained equations. R. Journal of Statistical Software, 45(3), 1–67. Retrieved from
http://www.jstatsoft.org/v45/i03/.
Carson, J. (2007). The measure of merit: Talents, intelligence, and inequality in the French and
American republics, 1750–1940. Princeton, NJ: Princeton University Press.
Chapman, J. W., & Tunmer, W. E. (1997). A longitudinal study of beginning reading
performance and reading self-concept. British Journal of Educational Psychology, 67(3),
279–291. http://dx.doi.org/10.1111/j.2044-8279.1997.tb01244.x.
Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement in-
variance. Structural Equation Modeling, 14(3), 464–504. http://dx.doi.org/10.1080/
10705510701301834.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing
measurement invariance. Structural Equation Modeling, 9(2), 233–255. http://dx.doi.
org/10.1207/S15328007SEM0902_5.
Dweck, C. S. (2002). The development of ability conceptions. In A. Wigfield, & J. Eccles
(Eds.). Development of achievement motivation (pp. 57–88). San Diego, CA: Academic
Press.
Entwisle, D. R., Alexander, K. L., Pallas, A. M., & Cadigan, D. (1987). The emergent
academic self-image of first graders: Its response to social structure. Child
Development, 58(5), 1190–1206. http://dx.doi.org/10.2307/1130614.
Fend, H. (2006). Neue Theorie der Schule. Einführung in das Verstehen von Bildungssystemen
[New theory of school. An introduction to understanding educational systems].
Wiesbaden: VS.
FORSA (2015). Inklusion an Schulen aus Sicht der Lehrerinnen und Lehrer – Meinungen,
Einstellungen und Erfahrungen Ergebnisse einer repräsentativen Lehrerbefragung.
[Inclusive education from the perspective of teachers – opinions, attitudes and ex-
perience. Results from a representative survey.]. Retrieved from http://www.vbe-
berlin.de/files/Inklusion%20Ergebnisse%20-%20Bund.pdf.
Friend, M., Cook, L., Hurley-Chamberlain, D., & Shamberger, C. (2010). Co-teaching: An
illustration of the complexity of collaboration in special education. Journal of
Educational and Psychological Consultation, 20(1), 9–27. http://dx.doi.org/10.1080/
10474410903535380.
Fuchs, D., Fuchs, L. S., Mathes, P. G., Lipsey, M. W., & Roberts, P. H. (2001). Is "learning
disabilities" just a fancy term for low achievement? A meta-analysis of reading dif-
ferences between low achievers with and without the label. Paper presented at the
Learning Disabilities Summit: Building a Foundation for the Future (Washington, DC,
August 27–28, 2001). Retrieved from http://files.eric.ed.gov/fulltext/ED459544.pdf.
Gniewosz, B. (2010). Die Konstruktion des akademischen Selbstkonzeptes. Eltern und
Zensuren als Informationsquellen [the construction of academic self-concept. Parents
and grades as sources of information]. Zeitschrift für Entwicklungspsychologie und
Pädagogische Psychologie, 42(3), 133–142. http://dx.doi.org/10.1026/0049-8637/
a000014.
Green, J., Liem, G. A. D., Martin, A. J., Colmar, S., Marsh, H. W., & McInerney, D. (2012).
Academic motivation, self-concept, engagement, and performance in high school:
Key processes from a longitudinal perspective. Journal of Adolescence, 35(5),
1111–1122. http://dx.doi.org/10.1016/j.adolescence.2012.02.016.
Guay, F., Marsh, H. W., & Boivin, M. (2003). Academic self-concept and academic per-
formance: Developmental perspectives on their causal ordering. Journal of
Educational Psychology, 95(1), 124–136. http://dx.doi.org/10.1037/0022-0663.95.1.
124.
Hedegaard, M. (2005). The zone of proximal development as basis for instruction. In H.
Daniels (Ed.). An introduction to Vygotsky (pp. 227–252). London: Routeledge.
Heine, J.-H., Gebhardt, M., Schwab, S., Neumann, P., Gorges, J., & Wild, E. (subm.).
Testing Psychometric Properties of the CFT 1‑R for Students with Special Educational
Needs. Manuscript submitted for publication.
Helmke, A., & van Aken, M. A. (1995). The causal ordering of academic achievement and
self-concept of ability during elementary school: A longitudinal study. Journal of
Educational Psychology, 87(4), 624–637. http://dx.doi.org/10.1037/0022-0663.87.4.
624.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation Modeling,
6(1), 1–55. http://dx.doi.org/10.1080/10705519909540118.
Huguet, P., Dumas, F., Marsh, H., Régner, I., Wheeler, L., Suls, J., ... Nezlek, J. (2009).
Clarifying the role of social comparison in the big-fish-little-pond effect (BFLPE): An
integrative study. Journal of Personality and Social Psychology, 97(1), 156–170. http://
dx.doi.org/10.1037/a0015558.
Kahneman, D., Krueger, A. B., Schkade, D. A., Schwarz, N., & Stone, A. A. (2004). A
survey method for characterizing daily life experience: The day reconstruction
method. Science, 306(5702), 1776–1780. http://dx.doi.org/10.1126/science.
1103572.
Klemm, K. (2015). Inklusion in Deutschland. Daten und Fakten [Inclusive education in
Germany. Facts and figures]. Gütersloh: Bertelsmann Stiftung.
Kocaj, A., Kuhl, P., Kroth, A. J., Pant, H. A., & Stanat, P. (2014). Wo lernen Kinder mit
J. Gorges et al.
sonderpädagogischem Förderbedarf besser? Ein Vergleich schulischer Kompetenzen
zwischen Regel- und Förderschulen in der Primarstufe [Where do students with
special educational needs learn better? A comparison of achievement between reg-
ular primary schools and special schools]. Kölner Zeitschrift für Soziologie und
Sozialpsychologie, 66, 165–191. http://dx.doi.org/10.1007/s11577-014-0253-x.
Lenhard, W., & Schneider, W. (2006). Leseverständnistest für Erst-Sechstklässler (ELFE 1–6).
Göttingen: Hogrefe.
Liem, G. A. D., McInerney, D. M., & Yeung, A. S. (2015). Academic self-concepts in ability
streams: Considering domain specificity and same-stream peers. The Journal of
Experimental Education, 83(1), 83–109. http://dx.doi.org/10.1080/00220973.2013.
876227.
Lindsay, G. (2007). Educational psychology and the effectiveness of inclusive education/
mainstreaming. British Journal of Educational Psychology, 77(1), 1–24. http://dx.doi.
org/10.1348/000709906X156881.
Lipowsky, F., Kastens, C., Lotz, M., & Faust, G. (2011). Aufgabenbezogene Differenzierung
und Entwicklung des verbalen Selbstkonzepts im Anfangsunterricht [task-related
differentiation and the development of verbal self-concepts in early school instruc-
tion]. Zeitschrift für Pädagogik, 57(6), 868–884.
Löser, J. M., & Werning, R. (2011). Does equity exist for immigrant students in German
schools? In A. J. Artiles, E. B. Kozleski, & F. R. Waitoller (Eds.). Inclusive education
(pp. 89–100). Cambridge: Harvard Education Press.
Lüdtke, O., Köller, O., Marsh, H. W., & Trautwein, U. (2005). Teacher frame of reference
and the big-fish-little-pond effect. Contemporary Educational Psychology, 30(3),
263–285. http://dx.doi.org/10.1016/j.cedpsych.2004.10.002.
Marsh, H. W. (1986). Verbal and math self-concepts: An internal/external frame of re-
ference model. American Educational Research Journal, 23(1), 129–149. http://dx.doi.
org/10.3102/00028312023001129.
Marsh, H. W. (1987). The big-fish-little-pond effect on academic self-concept. Journal of
Educational Psychology, 79(3), 280–295. http://dx.doi.org/10.1037/0022-0663.79.3.
280.
Marsh, H. W. (2007). Self-concept theory, measurement and research into practice: The role of
self-concept in educational psychology. Vernon-Wall Lecture: British Psychological
Society.
Marsh, H. W., Byrne, B. M., & Yeung, A. S. (1999). Causal ordering of academic self-
concept and performance: Reanalysis of a pioneering study and revised re-
commendations. Educational Psychologist, 34(3), 155–167. http://dx.doi.org/10.
1207/s15326985ep3403_2.
Marsh, H. W., & Hau, K. T. (2003). Big-fish-little-pond effect on academic self-concept: A
cross-cultural (26-country) test of the negative effects of academically selective
schools. American Psychologist, 58(5), 364–376. http://dx.doi.org/10.1037/0003-
066X.58.5.364.
Marsh, H. W., & Martin, A. J. (2011). Academic self-concept and academic performance:
Relations and causal ordering. British Journal of Educational Psychology, 81(1), 59–77.
http://dx.doi.org/10.1348/000709910X503501.
Marsh, H. W., & O'Mara, A. (2008). Reciprocal effects between academic self-concept,
self-esteem, performance, and attainment over seven adolescent years:
Unidimensional and multidimensional perspectives of self-concept. Personality and
Social Psychology Bulletin, 34(4), 542–552. http://dx.doi.org/10.1177/
014616720731231.
Marsh, H. W., Trautwein, U., Lüdtke, O., Köller, O., & Baumert, J. (2005). Academic self-
concept, interest, grades, and standardized test scores: Reciprocal effects models of
causal ordering. Child Development, 76(2), 397–416. http://dx.doi.org/10.1111/j.
1467-8624.2005.00853.x.
Meredith, W. (1993). Measurement invariance, factor analysis and factorial invariance.
Psychometrika, 58(4), 525–543. http://dx.doi.org/10.1007/BF02294825.
Möller, J., Pohlmann, B., Köller, O., & Marsh, H. W. (2009). A meta-analytic path analysis
of the internal/external frame of reference model of academic achievement and
academic self-concept. Review of Educational Research, 79(3), 1129–1167. http://dx.
doi.org/10.3102/0034654309337522.
Möller, J., Streblow, L., & Pohlmann, B. (2009). Performance and self-concept of students
with learning disabilities. Social Psychology of Education, 12(1), 113–122. http://dx.
doi.org/10.1007/s11218-008-9065-z.
Möller, J., & Trautwein, U. (2015). Selbstkonzept [Self-concept]. In E. Wild, & J. Möller
(Eds.). Pädagogische Psychologie (2. Auflage) (pp. 177–199). Berlin [u.a.]: Springer.
Moser, V. (2012). Braucht die Inklusionspädagogik einen Behinderungsbegriff? [Does
inclusive education need a new understanding of disability?]. Zeitschrift für Inklusion.
3. Retrieved from http://www.inklusion-online.net/index.php/inklusion-online/
article/view/40/40.
Oberski, D. L. (2014). Lavaan.survey: An R package for complex survey analysis of
structural equation models. Journal of Statistical Software, 57(1), 1–27. Retrieved from
http://www.jstatsoft.org/v57/i01/.
Organization for Economic Cooperation and Development [OECD] (2010). PISA 2009
Assessment Framework: Key competencies in reading, mathematics and science. Paris:
OECD Publishinghttp://dx.doi.org/10.1787/9789264062658-en.
Poloczek, S., Karst, K., Praetorius, A. K., & Lipowsky, F. (2011). Generalisten oder
Spezialisten? Bereichsspezifität und leistungsbezogene Zusammenhänge des schu-
lischen Selbstkonzepts von Schulanfängern [Generalists or specialists? Domain spe-
cificity and relationships to achievement data of the academic self-concept of school
starters]. Zeitschrift für Pädagogische Psychologie, 25(3), 173–183 doi:https://doi.org/
10.1024/1010-0652/a000045.
Preacher, K. J., Wichmann, A. L., MacCallum, R. C., & Briggs, N. E. (2008). Latent growth
curve modeling. London: Sage.
R Core Team (2015). R: A language and environment for statistical computing. Retrieved
from R Foundation for Statistical Computing, Vienna, Austria: http://www.R-project.
org/.
Retelsdorf, J., Köller, O., & Möller, J. (2014). Reading achievement and reading self-
20
Learning and Individual Differences 61 (2018) 11–20
concept–testing the reciprocal effects model. Learning and Instruction, 29, 21–30.
http://dx.doi.org/10.1016/j.learninstruc.2013.07.004.
Rheinberg, F., & Krug, S. (1999). Motivationsförderung im Schulalltag [Supporting motivation
at school] (2., überarbeitete und erweiterte Auflage). Göttingen [u.a.]: Hogrefe.
Rosenholtz, S. J., & Simpson, C. (1984). The formation of ability conceptions:
Developmental trend or social construction? Review of Educational Research, 54(1),
31–63. http://dx.doi.org/10.3102/00346543054001031.
Rosseel, Y. (2012). Lavaan: An R package for structural equation modeling. Journal of
Statistical Software, 48(2), 1–36. Retrieved from http://www.jstatsoft.org/v48/i02/.
Sauer, S., Ide, S., & Borchert, J. (2007). Zum Selbstkonzept von Schülerinnen und
Schülern an Förderschulen und in integrativer Beschulung: Eine
Vergleichsuntersuchung [on the self-concept of students at special schools versus in
integrative schools: A comparative analysis]. Heilpädagogische Forschung, 33(3),
135–142. Retrieved from http://www.pedocs.de/frontdoor.php?source_opus=8786.
Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art.
Psychological Methods, 7(2), 147. http://dx.doi.org/10.1037/1082-989X.7.2.147.
Schermelleh-Engel, K., Moosbrugger, H., & Müller, H. (2003). Evaluating the fit of
structural equation models: Test of significance and descriptive goodness-of-fit
measures. Methods of Psychological Research - Online, 8(2), 23–74.
Schöne, C., Dickhäuser, O., Spinath, B., & Stiensmeier-Pelster, J. (2002). Skalen zur
Erfassung des schulischen Selbstkonzepts [scales to measure school-based self-concept]
(SESSKO). Göttingen: Hogrefe.
Schwinger, M., Wild, E., Lütje-Klose, B., Grunschel, C., Stranghöner, D., Yotyodying, S., ...
Stelling, S. (2015). Wie können motivationale und affektive Merkmale bei Kindern
mit sonderpädagogischem Förderbedarf valide erfasst werden? Erste Befunde der
Bielefelder Längsschnittstudie BiLieF [How to assess motivational and affective
characteristics in children with SEN-L: First results from the Bielefeld longitudinal
Study BiLieF]. In P. Kuhl, P. Stanat, B. Lütje-Klose, C. Gresch, H. A. Pant, & M. Prenzel
(Eds.). Inklusion von Schülerinnen und Schülern mit sonderpädagogischem Förderbedarf in
Schulleistungserhebungen (S. 273-300). Wiesbaden: Springer VS.
Skaalvik, E. M., & Valås, H. (1999). Relations among performance, self-concept, and
motivation in mathematics and language arts: A longitudinal study. The Journal of
Experimental Education, 67(2), 135–149. http://dx.doi.org/10.1080/
00220979909598349.
Stranghöner, D., Hollmann, J., Otterpohl, N., Wild, E., Lütje-Klose, B., & Schwinger, M.
(2017). Inklusion versus Exklusion: Schulsetting und Lese-Rechtschreibentwicklung
von Kindern mit Förderschwerpunkt Lernen [Inclusion vs. Exclusion: School Setting
and the Development of Reading and Writing Skills in Students with Mild Learning
Difficulties]. Zeitschrift für Pädagogische Psychologie, 31, 125–136. http://dx.doi.org/
10.1024/1010-0652/a000202.
Steenkamp, J. B. E. M., & Baumgartner, H. (1998). Assessing measurement invariance in
cross-national consumer research. Journal of Consumer Research, 25(1), 78–107.
http://dx.doi.org/10.1086/209528.
Tabassam, W., & Grainger, J. (2003). Self-concept enhancement for students with
learning difficulties with and without attention deficit hyperactivity disorder. In H.
W. Marsh, R. G. Craven, & McInerney (Eds.). International advances in self research.
Greenwich: DM Information Age Publishing.
Tracey, D. K., Marsh, H. W., & Craven, R. G. (2003). Self-concepts of preadolescent stu-
dents with mild intellectual disabilities: Issues of measurement and educational
placement. In H. W. Marsh, R. G. Craven, & McInerney (Eds.). International advances
in self research (pp. 203–230). Connecticut: DM Information Age Publishing.
Valentine, J. C., DuBois, D. L., & Cooper, H. (2004). The relation between self-beliefs and
academic achievement: A meta-analytic review. Educational Psychologist, 39(2),
111–133. http://dx.doi.org/10.1207/s15326985ep3902_3.
Wagner, C., & Valtin, R. (2003). Noten oder Verbalbeurteilungen? Die Wirkung un-
terschiedlicher Bewertungsformen auf die schulische Entwicklung von
Grundschulkindern [Marks or verbal reports? The effect of different assessment forms
on school development of elementary school children]. Zeitschrift für
Entwicklungspsychologie und Pädagogische Psychologie, 35(1), 27–36. http://dx.doi.
org/10.1026//0049-8637.35.1.27.
Weidinger, A. F., Spinath, B., & Steinmayr, R. (2015). Zur Bedeutung von
Grundschulnoten für die Veränderung von Intrinsischer Motivation und
Fähigkeitsselbstkonzept in Deutsch [The role of grades for changes in intrinsic mo-
tivation and ability self-concept in German]. Zeitschrift für Pädagogische Psychologie,
29(3–4), 193–204 doi:https://doi.org/10.1024/1010-0652/a000160.
Werning, R., & Lütje-Klose, B. (2016). Einführung in die Pädagogik bei
Lernbeeinträchtigungen [Introduction to the education of students with special educational
needs in learning] (4., überarbeitete Auflage). München [u.a.]: Reinhardt.
Wigfield, A., & Eccles, J. S. (2000). Expectancy–value theory of achievement motivation.
Contemporary Educational Psychology, 25(1), 68–81. http://dx.doi.org/10.1006/ceps.
1999.1015.
Wigfield, A., Eccles, J. S., Schiefele, U., Roeser, R. W., & Davis-Kean, P. E. (2007).
Development of achievement motivation. In W. Damon, R. M. Lerner, & N. Eisenberg
(Eds.). Handbook of child psychology. Volume three: social, emotional, and personality
development (pp. 933–1002). Wiley.
Wild, E., Schwinger, M., Lütje-Klose, B., Yotyodying, S., Gorges, J., Stranghöner, D., ...
Kurnitzki, S. (2015). Schülerinnen und Schüler mit dem Förderschwerpunkt Lernen
in inklusiven und exklusiven Förderarrangements: Erste Befunde des BiLieF-Projektes
zu Leistung, sozialer Integration, Motivation und Wohlbefinden [Students with
Learning Disabilities in Inclusive and Exclusive School Settings. First Results from the
BiLieF Project on Achievement, Social Integration, Motivation, and Well-being.].
Unterrichtswissenschaft, 43(1), 7–21.
Wild, E., Lütje-Klose, B., Schwinger, M., Gorges, J., & Neumann, P. (2017). BiLieF -
Bielefelder Längsschnittstudie zum Lernen in inklusiven und exklusiven
Förderarrangements. Version: 1. IQB. Datensatz: Institut zur Qualitätsentwicklung im
Bildungswesenhttp://dx.doi.org/10.5159/IQB_BiLieF_v1.