Content uploaded by Jiesi Guo
Author content
All content in this area was uploaded by Jiesi Guo on Feb 14, 2021
Content may be subject to copyright.
REVIEW ARTICLE
Moderation of the Big-Fish-Little-Pond Effect:
Juxtaposition of Evolutionary (Darwinian-Economic)
and Achievement Motivation Theory Predictions Based
on a Delphi Approach
Herbert W. Marsh, et al. [full author details at the end of the article]
Accepted: 2 November 2020/
#Springer Science+Business Media, LLC, part of Springer Nature 2021
Abstract
The big-fish-little-pond effect (BFLPE), the negative effect of school-/class-average
achievement on academic self-concept, is one of educational psychology’smostuniversal
findings. However, critiques of this research have proposed moderators based on achieve-
ment motivation theories. Nevertheless, because these motivational theories are not
sufficiently well-developed to provide unambiguous predictions concerning moderation
of the BFLPE and underlying social comparison processes, we developed a Theory-
Integrating Approach; bringing together a panel of experts, independently making theo-
retical predictions, revising the predictions over several rounds based on independent
feedback from the other experts, and a summary of results. We pit a priori hypotheses
derived from achievement motivation theories against the more parsimonious a priori
prediction that there is no moderation based on previous BFLPE empirical research and
Darwinian-economic theory (N= 1,925 Hong Kong students, 47 classes, Mage = 12
years). Consistent with both BFLPE research and Darwinian perspectives, but in contrast
to achievement motivation theory predictions, the highly significant BFLPE was not
moderated by any of the following: prior achievement, expectancy-value theory variables,
achievement goals, implicit theories of ability, self-regulated learning strategies, and
social interdependence theory measures. Although we cannot “prove”that there are no
student-level moderators of the BFLPE, our synthesis of social comparison posited in the
BFLPE theory and an evolutionary perspective support BFLPE’sgeneralizability.We
propose further integration of our Theory-Integrating Approach with traditional Delphi
methods, combining quantitative and qualitative approaches to develop a priori theoret-
ical predictions and identify limitations in existing theory as an alternative form of
systematic review.
Keywords Big-fish-little-pond effect .Social comparison processes .Academic self-concept .
Achievement motivation theory .Darwinian economics .Theory-Integrating Delphi Method
The self-concept construct has a long history in social science research, but particularly in
educational studies (Marsh 2007; Marsh and Craven 2006). Marsh and O’Mara (2008;Morin
et al. 2015; Guo et al. 2015a,b; Eccles 2009) showed that academic self-concept (ASC)
Educational Psychology Review
https://doi.org/10.1007/s10648-020-09583-5
formed in high school contributes to the prediction of key academic outcomes and long-term
educational attainment, even after controlling the effects of school grades, standardized
achievement tests, IQ, and socioeconomic status.
An Educational Psychology Perspective on the Big-Fish-Little-Pond
Effect
ASC, one’s academic self-beliefs and perceptions of competence in different academic
domains, is based in part on social comparison processes. Thus, ASC depends not only on
one’s own academic accomplishments but also on how these compare with the accomplish-
ments of one’s classmates. These processes explain many seemingly paradoxical findings that
have important implications for theory, research, and policy/practice. In particular, one of the
most widely studied phenomena in ASC research is the seemingly paradoxical and contro-
versial big-fish-little-pond effect (BFLPE). Marsh (1984,2007) developed the theoretical
model underpinning the BFLPE that integrates diverse theoretical perspectives from many
disciplines, based in part on social comparison theory (Festinger 1954). According to the
BFLPE, students who attend schools and classes where the average ability level is high will
have lower ASCs than do equally able students who attend mixed- or low-ability classes and
schools; a negative effect of class-/school-average achievement on ASC. Similarly, academ-
ically disadvantaged students who move from special classes for disadvantaged students to
main-stream classes with mixed-ability students will suffer diminished ASCs—a negative
effect of class-average ability (Tracey et al. 2003).
The BFLPE
Originally described as “paradoxical”in relation to popular beliefs about selective schools and
classes, Marsh and Seaton’s(2015, also see Fang et al. 2018; Marsh et al. 2014,2017)
extensive review of BFLPE research based on many individual and cross-national studies
across many countries led them to conclude that the BFLPE is a universal phenomenon, one of
psychology’s most cross-culturally robust findings. In particular, there is excellent support for
this premise from data collected by the Organization for Economic Cooperation and Devel-
opment (OECD) Programme for International Student Assessment (PISA). PISA data consist
of nationally representative samples of 15-year-olds. Based on four cycles of PISA data
(2000–2012), the effect of school-average achievement on ASC was negative in all but one
of the 191 samples, and significantly so in 181 samples (Marsh and Hau 2003: 103,558
students from 26 countries; Seaton et al. 2009,2010: 265,180 students from 41 countries;
Nagengast and Marsh 2012: 397,500 students from 57 countries; Marsh et al. 2018: 485,490
students from 68 countries).
Moderators of the BFLPE
A critically important approach for extending BFLPE research, theory, and policy/practice
implications is to test whether any student-level motivation variables moderate the BFLPE (see
Dai and Rinn 2008). Seaton et al. (2010) noted that moderation is a double-edged sword.
Finding strong moderators of the BFLPE would help understand the underlying processes of
the BFLPE and allow the development of personalized interventions that could lessen its
Educational Psychology Review
negative consequences. However, if the BFLPE generalizes across diverse student character-
istics, then such evidence would strengthen support for the BFLPE’s theoretical basis,
robustness, and claims of universality.
One of the earliest proposed moderators of the BFLPE was individual achievement. In what
Marsh et al. (2018) referred to as the bright student hypothesis, some researchers argued that
BFLPEs should be substantially smaller, eliminated, or even reversed for the brightest students
in each class. According to this hypothesis, being the brightest student in high-ability classes
should enhance—not diminish—ASC (e.g., Coleman and Fults 1985; also see Davis 1966;
Huguet et al. 2009). However, according to the theoretical model underpinning the BFLPE
(Marsh 1984,2007), the size of the BFLPE should be similar for the best and worst students
within each class. According to the BFLPE theory, the frame of reference is established by the
class-/school-average achievement. However, the class-/school-average is necessarily the same
for all students within a given class or school. Thus, if the class-/school-average achievement
increases, then the ASCs of all students will decrease. Conversely, if the class-/school-average
achievement goes down, the ASCs of all students will increase. Hence, according to BFLPE
theory, the size of the BFLPE should be similar for all students within the same class or school.
A growing body of empirical research (Marsh 1984; Marsh et al. 2008,2014,2017) supports
these predictions in that interactions between school-average and individual student achieve-
ment are consistently small or non-significant, and not even consistent in direction.
Of course, it is not possible for any one study, or even a finite set of studies, to prove that
there are no moderators of the BFLPE. However, in one of the most extensive studies of the
generalizability of the BFLPE, Seaton et al. (2010; also see review by Marsh and Seaton 2015)
tested the moderation of the BFLPE for math self-concept in relation to individual student
differences on 16 potential moderators based on PISA 2003 data (41 countries, 10,221 schools,
265,180 students). Potential moderators in their study included socioeconomic status (parental
occupation, parental education, home educational resources, and cultural possessions), indi-
vidual ability, intrinsic and extrinsic motivation, self-efficacy, study methods (elaboration,
memorization, and control strategies), anxiety, competitive and cooperative learning orienta-
tions, sense of school belonging, and student-teacher relationships.
The BFLPE for math self-concept was not substantially moderated by any of these
variables. Although interactions with some of these variables were statistically significant
(due in part to the huge sample size), none were substantively significant in terms of nullifying
or changing the direction of the BFLPE. For example, the BFLPE was significantly larger for
students with high levels of anxiety, but students with low levels of anxiety suffered from the
BFLPE as well, just to a slightly lesser extent than students with high levels of anxiety. For
present purposes, we operationalize this logic by evaluating the relative size of the BFLPE and
its moderation. Thus, if the size of the moderation is less than half the size of the BFLPE, then
the direction of the BFLPE remains consistent even for students who are extreme in terms of
the moderator (i.e., two standard deviations above or below the mean of the moderator).
Based on the same logic and tests of simple slopes, Seaton et al. (2010) concluded that
none of the potential moderators in their study substantially moderated the BFLPE,
attesting to its broad generalizability. This research, along with the consistency of BFLPE
across countries in extensive cross-national studies, led Seaton, Marsh, and colleagues to
posit the BFLPE as one of psychology’s most universal phenomena (Seaton et al. 2009,
2010; also see review by Marsh and Seaton 2015). Nevertheless, this database did not
contain the achievement motivation variables that are the focus of our study—particularly
achievement goal theory constructs.
Educational Psychology Review
Jonkmann et al. (2012) took up the challenge to find moderators of the BFLPE for ASC.
They posited Big-Five personality characteristics (extraversion, agreeableness, openness,
conscientiousness, and neuroticism) as potential moderators of the BFLPE. However, they
also considered narcissism, with the rationale that students high on narcissism (i.e., those with
exaggerated feelings of superiority, self-importance, and grandiosity) might be immune to the
BFLPE. Neuroticism and narcissism were statistically significant moderators of the BFLPE,
whereas interactions with the remaining personality traits were non-significant. However, the
sizes of the two significant interactions were small relative to the size of the BFLPE (less than
one-quarter the size of the BFLPE). Thus, even students high on narcissism and low on
neuroticism experienced the BFLPE, although to a slightly lesser degree than students low in
narcissism or high in neuroticism. Nevertheless, Jonkmann et al. (2012) suggested that the
results support the construct validity of the theoretical model underlying the BFLPE, in that
students high on narcissism, who were predicted to be less affected by the relative perfor-
mances of their classmates, experienced significantly smaller BFLPEs. However, they also
noted that the results do not translate into intervention programs to counter the BFLPE through
fostering a counterproductive construct such as narcissism. In summary, the Jonkmann et al.
(2012) study demonstrated that traditional personality characteristics are not substantial mod-
erators of the BFLPE, and that moderation effects of narcissism remain small relative to the
size of the BFLPE.
Achievement Motivation Theories and Potential Moderators
of the BFLPE
Although there is growing evidence in support of the robustness of the BFLPE, there
continue to be calls for further consideration of potential moderators of the BFLPE
and its integration with motivation theories in educational science and educational
psychology. In one of the most influential critiques of the BFLPE, Dai and Rinn
(2008) argued for the need for a broader conceptualization of the BFLPE that focuses
on students having a more active role in regulating their social cognition and
motivation. They underlined the need to consider individual student characteristics
to better understand the social comparison process underlying the BFLPE, noting that
ASC is far removed from the social comparison and motivational process posited to
drive it. Knowing the moderators of the BFLPE, they argued, would facilitate the
identification of students most vulnerable to the adverse effects of the BFLPE, better
placement decisions, and more appropriate interventions. In particular, they argued for
better integration of the broad body of “motivation models”—an intentionally broad
term referring to diverse achievement motivational theories, processes, and constructs
known to be important in educational research that are likely to moderate the BFLPE.
Following from Dai and Rinn’s(2008) critique of the BFLPE, good candidates for
potential moderators of the BFLPE are motivation constructs (see Table 1)basedon
key theories of achievement motivation, such as expectancy-value theory measures of
task value (importance, interest, usefulness), achievement goal theory measures (mas-
tery, performance-approach, performance-avoidance), implicit theories of ability (fixed-
ability beliefs), self-regulated learning strategies (elaboration, rehearsal, control, effort/
persistence), and social interdependence theory measures of learning environment
preferences (cooperative, competitive). For example, two subsequent studies (Cheng
Educational Psychology Review
et al. 2014; Wouters et al. 2013) evaluated the extent to which the BFLPE varies as a
function of constructs from achievement goal theory (e.g., mastery, performance-
Table 1 Moderators of the BFLPE: a priori hypotheses about the nature of the interaction between the moderator
and class-average achievement in predicting math self-concept (MSC) based on achievement motivation theories
and BFLPE research (and Darwinian perspectives)
Moderator Achievement motivation theory hypotheses: predicted effect
(nature, direction, and rationale) when both moderator and class-
average achievement are high
BFLPE
research
(Darwinian)
Mastery goals Small decrease in BFLPE when class-average achievement and
mastery goals are both high (positive interaction); intrapersonal
and task-based comparison is salient and social comparison is
less important even though the focus on achievement could
strengthen the BFLPE.
Little or no
effect
Performance-approach
goals
Moderate increase in BFLPE (negative interaction); due to focus on
achievement and social comparison.
Little or no
effect
Performance-avoidance
goals
Moderate increase in BFLPE (negative interaction); due to focus on
achievement and social comparison.
Little or no
effect
Fixed mindset ??? Could increase BFLPE because belief in a fixed ability makes
ability relative to others more salient and damaging to MSC, but
could also decrease the BFLPE because social comparison is less
salient due to focus on own ability.
Little or no
effect
Learning preference:
competitive
??? Could result in a small increase in BFLPE due to focus on social
comparison, but enjoyment of competition and doing better than
others might offset the BFLPE.
Little or no
effect
Learning preference:
cooperative
Small decrease in BFLPE (positive interaction), due to focus on
group-based standards of collective achievement rather than so-
cial comparison even though there is a focus on achievements of
others.
Little or no
effect
Learning strategy:
rehearsal/-
memorization
??? Could result in a small decrease in BFLPE (positive interaction)
because use of learning strategies reduces salience of social
comparison, but could increase BFLPE due to a shallow learning
strategy, low achievement motivation, and low emotional
regulation (negative interaction).
Little or no
effect
Learning strategy:
metacognitive
control
Small decrease in BFLPE (positive interaction), because ability to
use control strategies (a) makes social comparison less important,
(b) makes it possible to buffer negative effects of social com-
parison.
Little or no
effect
Learning strategy:
elaboration
Small decrease in BFLPE (positive interaction), because use of
elaboration and deep processing involves mastery achievement
motivation, reduces salience of social comparison.
Little or no
effect
Learning strategy:
effort/persistence
Small increase in BFLPE (negative interaction); because trying hard
is threatening to MSC and promotes attribution of failure to lack
of ability, which makes low ability relative to others salient.
Little or no
effect
Task value: importance Moderate increase in BFLPE (negative interaction), because
importance reflects attainment value and makes achievement
salient.
Little or no
effect
Task value: utility
value
Small increase in BFLPE (negative interaction), because
instrumental value of task makes it important to achieve.
Little or no
effect
Task value: intrinsic
value/interest
Moderate increase in BFLPE (negative interaction), because
intrinsic value and interest focus attention on mastery, and reduce
importance of social comparison.
Little or no
effect
Note. Because the BFLPE is negative, negative interactions mean that the size of the negative effect of the
BFLPE is increased, whereas positive interaction mean that the size of the BFLPE is decreased (possibly
including even a reversal in direction).
??? Indicates that there was no clear consensus or alternative, unreconciled perspectives.
Educational Psychology Review
approach, and performance-avoidance goals). Although both studies found substantial
BFLPEs, moderation effects were small, non-significant, or not consistent with a
priori predictions in terms of direction. Thus, the BFLPE was slightly larger—not
smaller, as predicted—for students who had stronger mastery goals (Cheng et al.
2014; Wouters et al. 2013). Indeed, both studies suggested that endorsement of any
motivation goal tended to increase the size of the BFLPE. However, the interaction
effects were consistently small (ESs = 0 to −.10) relative to the size of the BFLPE.
Despite the plausibility of motivational constructs moderating the BFLPE, the avail-
able literature thus far has found little evidence of individual-student-level moderators
(see review by Marsh et al. 2017; Marsh and Seaton 2015). As such, in the present
research, we aim to undertake a comprehensive investigation of a broad range of
motivation constructs based on major motivational theories as potential moderators of
the BFLPE.
Students’Approaches to Learning
Consistent with Dai and Rinn’s focus on “motivation models”in a generic sense, we
felt it was important to consider key constructs from various motivation theories.
Hence, a critical first step in our research was to select the motivational constructs to
consider. Fortunately, the OECD has already developed the Students’Approaches to
Learning (SAL) instrument (Marsh et al. 2006). SAL was derived from a rigorous
process of selecting educational psychology’s most useful constructs and motivation
theories, and empirical testing of psychometric properties based on responses to SAL
in extensive cross-national pilot studies. These strong psychometric properties were
subsequently validated in the use of SAL in nationally representative samples from 26
countries as part of the PISA2000 data collection (Marsh et al. 2006).
SAL is a brief survey that measures 14 factors from a variety of theoretical perspec-
tives that assess self-regulated learning strategies, self-beliefs (ASC and self-efficacy),
expectancies, values, implicit theories of intelligence, goals, and learning preferences.
Marsh et al. (2006) described how this OECD-SAL instrument provides a standard set of
educational measures that have been selected by an OECD expert panel of substantive
researchers and that have been validated across the world in extensive pilot studies
designed by an OECD expert panel of methodological researchers (see summary by
Marsh et al. 2006). On this basis, Marsh et al. (2006) contended that SAL should be a
useful focus in diverse educational research settings, providing the longitude and latitude
against which to map new and existing educational constructs and test theoretical
predictions—the starting point for the present investigation. Starting with these con-
structs, we sought to test moderation of the BFLPE in relation to the two most widely
used self-belief constructs; ASC (Marsh 2007) and academic self-efficacy (Bandura
1986). Following from the SAL instrument, we consider the following achievement
motivation constructs as our key moderator variables.
Task Values Task value measures (importance, interest, usefulness) were based on Eccles’
expectancy-value theory (Wigfield and Eccles 2000), and the items were adapted from the
OECD-SAL instrument (Marsh et al. 2006). Types of values include attainment value
(importance), intrinsic value (e.g., interest, enjoyment), and extrinsic value (e.g., utility,
instrumental value).
Educational Psychology Review
Achievement Goals A goal is a mental representation of future possibilities that directs
proactive behavior (Elliot and Fryer 2008). In achievement settings, achievement goals
influence competence-relevant behaviors. Achievement goal theory measures (mastery, per-
formance-approach, performance-avoidance) were based on the trichotomous achievement
goal model and items adapted from Elliot and Church (1997; Elliot 1999).
The Implicit Theory of Intelligence An implicit theory of intelligence or ability is the belief
that students hold about the stability of their ability. Dweck (2000) proposed that the students
who hold an entity theory (fixed belief) believe that ability is unchangeable, whereas those
who hold an incremental theory believe that ability is malleable. The measure of fixed-ability
beliefs was based on the theoretical model and items adapted from Dweck (2000).
Self-Regulated Learning Strategy Learning strategies are the strategies that a student adopts
in order to acquire knowledge (Zimmerman 2000). Effort and persistence represent volitional
aspects of students’learning. Measures of learning strategies used here (elaboration, rehearsal,
control, effort/persistence) were adapted from the OECD-SAL instrument (Marsh et al. 2006).
Competitive or Cooperative Environment Preference Social interdependence theory proposes
that the completion of an individual’s goals is dependent on the action of others (Johnson and
Johnson 1999). The perceived learning environment is believed to be relevant to students’prefer-
ences in learning in a competitive or cooperative environment. In a cooperative learning environ-
ment, students work together in teams, whereas in a competitive environment, students’
performances are evaluated against each other. Social interdependence theory measures used here
(competitive and cooperative) were adapted from the PISA SAL instrument (Marsh et al. 2006).
A Darwinian-Economic Perspective on the BFLPE
In 1985, economist Robert E. Frank published Choosing the Right Pond: Human Behavior
and the Quest for Status (Frank 1985) just a year after Marsh and Parker (1984) published the
initial BFLPE study. Frank (2012) has since updated his thinking on the relative position and
contextual effects in The Darwin Economy: Liberty, Competition, and the Common Good.
Based on this evolutionary perspective, Frank argues that social comparison within local
contexts—the driving force behind the BFLPE—is a fundamental endowment of human
evolution. Frank states, “to survive and prosper, an individual need not be the strongest,
fastest, or smartest animal in the universe. He may be weak, slow, and stupid. What matters is
that he be able to compete successfully against members of his own species vying for the same
resources”(p. 24). In this economic research literature, there are many examples of social
comparison and frame-of-reference effects like those in BFLPE studies. The most widely
studied effects are those of income and unemployment on subjective well-being. Thus, for
example, subjective well-being is positively affected by one’s own income but negatively
affected by the income of one’s reference group due to social comparison (Clark 2018).
Nevertheless, there has been surprisingly little cross-citation between these economic studies
based on the evolutionary perspective and educational studies of the BFLPE.
Frank’s(2012) position is that the tendency to compare ourselves to immediate others is a
fundamental and largely unalterable aspect of our human nature. Frank’s perspective is similar
to Festinger’s(1954) perspective that social comparison is a universal human drive with
Educational Psychology Review
critical survival advantages. Thus, Festinger (1954,p.117)notesthat“there exists, in the
human organism, a drive to evaluate his opinions and his abilities.”For Frank, social
comparison is a universal process that is a means to achieve the survival of the fittest, not a
trait per se. As an economist, Frank does not talk explicitly about individual-level moderation,
but he does claim that social comparison is universal. For Frank, social comparison is not
something that can be eliminated, reduced, or controlled. Instead, society needs to build
environments that restrain the more destructive aspects of social comparison processes. Because
of the ubiquity and survival advantages of social comparison, moderation of social comparison
processes (the basis of the BFLPE) by individual differences is likely to be small and
inconsequential. From this Darwinian perspective, we would also expect individual-level
moderators of the BFLPE to be small and practically insignificant. Instead, our expectation is
that moderating effects will primarily be at the level of the school and school system, which can
have major consequences for the reference group that children experience on a daily basis (e.g.,
how academically selective a school system is; Parker et al. 2018,2020).
Importantly, we note that the focus of Frank’s research on the universality of social
comparison processes is relevant to the social comparison processes that underpin the BFLPE.
Specifically, his theoretical perspective is consistent with claims for the universality of the
BFLPE based on empirical research (e.g., Marsh and Seaton 2015; Marsh et al. 2017) and the
proposal that individual student variables are unlikely to moderate the BFLPE substantially.
Hence, the theoretical underpinning of Frank’s economic research and BFLPE research in
educational psychology are very similar. Thus, it is relevant to align these two areas of
research, particularly as thus far, there has been almost no cross-fertilization between them.
The Present Investigation
The Genesis of the Present Investigation
The present investigation originated in a question-answer session at the International Congress of
Applied Psychology. In a keynote presentation on goal theory by one of the main architects of goal
theory, the first author of the present investigation asked him to comment on a prediction that goal
theory constructs would moderate the BFLPE. Although not resolved at the conference, they agreed
to collaborate in pursuit of the resolution of this issue. Because they came from different theoretical
perspectives, they decided to select an intentionally diverse group of colleagues to work with—a
total of nine co-authors from Australia, the USA, Europe, and Asia. In this sense, the co-authors
were chosen explicitly as a panel of experts representing a diverse range of interests, theoretical
perspectives, and expertise in educational science, psychology, and motivation science (representing
seven universities from four continents) who agreed to collaborate on this project.
Our first task was to select an appropriate database to test our predictions. An ideal database
would include particularly the key goal theory constructs, as well as other achievement
motivation constructs, such as those in the SAL instrument (although the SAL instrument
was the basis of the PISA2000 data collection, these data did not contain measures of goal
theory that were of particular relevance). However, the appropriate database also had to be
suitable for testing the BFLPE (see Marsh and Seaton 2015; Marsh et al. 2017,for
requirements to test the BFLPE). Although the collective team was not able to locate an ideal
database, we collectively judged the one used here to be the most suitable (see subsequent
discussion in the “Methods”section).
Educational Psychology Review
Our next task was to generate a priori hypotheses in relation to moderation of the BFLPE
based on motivation theories. However, it quickly became apparent that there was disagree-
ment among the authors and that the motivational theories were not yet sufficiently developed
in relation to the specific issues considered here to generate unambiguous predictions. For this
reason, we developed a systematic process for integrating, revising, and seeking consensus
among our diverse group of co-authors—an approach that we subsequently referred to as a
“Theory-Integrating Approach.”
A Theory-Integrating Approach: Developing Predictions Based on Motivation Theory
Development of a Theory-Integrating Method We specially selected co-authors of the
present investigation as “experts”representing a diverse range of interests, theoretical per-
spectives, and expertise in educational science, psychology, and motivation science. For
present purposes, we refer to the co-authors as an “expert panel”to emphasize that the co-
authors were specifically selected to represent diversity, rather than uniformity, in relation to
the key issues.
The individual co-authors began with alternative perspectives on whether key motivation
constructs would moderate the social comparison processes underpinning the BFLPE. Using
an iterative approach, each author offered independent predictions of which of the constructs
would moderate the BFLPE in terms of direction, size, and the rationale. The open-ended
responses about the rationale for decisions provided a better basis for revising responses in
subsequent rounds, but also provided insight into underlying perspectives based on motiva-
tional theory. In this sense, our Theory-Integrating Approach is a hybrid, mixed-method
approach. It combines both quantitative and qualitative responses to juxtaposing competing
perspectives. Our Theory-Integrating Approach is also an alternative approach to a systematic
review in relation to the development of theoretical predictions in areas where existing theory
is not sufficiently well-developed to provide sufficiently unambiguous predictions when
applied to a specific issue. In this sense, the approach is important in identifying limitations
and ambiguities in existing theory and research as well as generating consensus predictions.
Because the co-authors lived on four continents, we sent responses from each co-author to
all the other co-authors via email. In this sense, the process was iterative. As part of our
approach, we summarized the initial quantitative and qualitative responses and returned them
to the co-authors for further comment. In each ensuing round, co-authors independently
revised their responses based on responses by others as appropriate, offered a rationale for
their responses, and commented on their rationale and the rationales offered by others. We
continued this iterative process of structured interaction among co-authors until a consensus
was reached, or in a few cases, there were alternative perspectives, and a consensus was not
achieved. We contend that this Theory-Integrating Approach is a healthy approach to devel-
oping theoretical predictions when there is an initial disagreement. It is also an excellent way to
advance theory and research in educational psychology where research is often conducted in
silos that do not provide a robust critique of competing perspectives.
Contrasting Sets of Predictions
In the present investigation, following from Dai and Rinn’s(2008) critique of the BFLPE, we
test predictions based on achievement motivation theories in relation to the moderation of the
Educational Psychology Review
BFLPE. More specifically, we juxtaposed two perspectives: (1) the highly parsimonious
prediction based on previous empirical BFLPE research (e.g., Marsh et al. 2017;Marshand
Seaton 2015) and consistent with a Darwinian-economic theoretical perspective (Frank 2012)
that social comparison processes underpinning the BELPE are universal, and (2) the more
nuanced theoretical predictions based on the results of our Theory-Integrating Delphi method,
as summarized in Table 1.
Tests of these competing sets of predictions are based on a set of models shown in Fig. 1.In
the basic BFLPE model (Fig. 1a), math self-concept (MSC, the dependent variable) is
regressed on individual student (L1) and class-average (L2) achievement. The BFLPE is the
direct effect of L2 achievement after controlling L1 achievement. In Fig. 1b, the L1xL2
achievement interaction is added to test the bright student hypothesis (that the individual
student achievement moderates the BFLPE such that the BFLPE is less negative for brighter
students). In Fig. 1c, achievement motivation measures and their interaction with class-average
achievement are added to test predictions in Table 1.
Methods
Sample and Measures
Sample
The data were collected from a sample of Hong Kong secondary school students. Data
collection was approved by the Research Panel, Faculty of Education, the Chinese University
of Hong Kong; school and student consents were obtained. Students (N= 1925; 47.3% boys,
52.7% girls; 11–16 years old, mean age = 12 years) from the end of the school year in Grade 7
(47 intact classes, 12 schools) were queried regarding their motivation for school learning. The
schools were sampled from various districts and broadly differentiated in terms of academic
strength. They were broadly representative of Hong Kong such that four schools were selected
from each of the above-average, average, and below-average school ability bands.
Measures
As described earlier, self-belief and achievement motivation measures were based in part on
the OECD-SAL instrument (also see Supplemental Materials for the wording of the items and
the a priori factors to which they are associated). Preliminary factor analyses of responses are
summarized below. Achievement was based on a standardized achievement test that was taken
by all students in Hong Kong (in July) before the entry of the first year of secondary schooling
(Grade 7, in September) and used as one basis for tracking students at the start of secondary
school.
Statistical Analyses
Preliminary Factor Analysis
Factor analyses, specifically exploratory structural equation modeling (ESEM; Marsh et al.
2014), were undertaken with Mplus 8 (Muthén and Muthén 2017) using robust maximum
Educational Psychology Review
Individual
Student
Self-Concept
Class-Average
Student
Achievement
Individual
Student
Achievement
aBig-Fish-Little-Pond Effect (BFLPE)
Individual
Student
Self-Concept
Class-Average
Student
Achievement
Individual
Student
Achievement
b
Individual Ach X
Class-Average Ach
Interaction
Bright-Student Hypothesis: A Moderator of the BFLPE
Individual
Student
Self-Concept
Class-Average
Student
Achievement
Individual
Student
Achievement
c
Moderator X
Class-Avg Ach
Interaction
Moderator
Achievement Motivation: Potential
BFLPE Moderators
Fig. 1 aConceptual model of the BFLPE. bBFLPE moderated by individual student achievement. cBFLPE
moderated by motivation moderator and individual achievement
Educational Psychology Review
likelihood estimation (MLR). We used SET-ESEM (Dicke et al. 2018; Marsh et al. 2019a,b)
to alleviate confounding at the item level between measures of self-beliefs (treated as the
outcome variables) and the set of motivational moderator variables listed in Table 1(the
predictor variables posited to moderate the BFLPE). In SET-ESEM, based on the a priori
model, items are allowed to cross-load on factors within the same set but not on factors from
different sets, thus avoiding confounding for constructs within the same set. A set was defined
by a group of constructs based on the same motivation theory. For example, the set for
achievement goal consisted of a mastery goal, performance-approach goal, and performance-
avoidance goal. In preliminary analyses, we applied SET-ESEM with target rotation to test the
factor structure of these 15 a priori latent factors—the two outcomes (MSC and self-efficacy)
factors and the 13 achievement motivation variables listed in Table 1. The factor analysis
based on 59 indicators designed to measure these factors provided a good fit to the data
according to guidelines of goodness-of-fit (e.g., Marsh et al. 1988). Target factor loadings
show that all factors are well-defined (see Supplemental Materials for Mplus syntax and
further discussion of the psychometric analyses). All subsequent analyses used factor scores
based on this preliminary factor analysis.
Multilevel Models
We performed moderation analysis using multilevel modeling with random intercept, fixed
slope estimation, with the commercially available MlwiN (Rasbash et al. 2004; also see Marsh
2016) program. This allowed us to accommodate the two-level hierarchical structure of the
data: students (L1) nested within classes (L2). Fixed effects considered in different models (see
Fig. 1) include the first-order (“main”) effects of individual student (linear and quadratic)
achievement, class-average achievement, and factor scores representing each of the 13
achievement motivation factors (see Table 1). Interaction effects included the multiplicative
terms between class-average achievement and each of the potential moderators of the BFLPE
(individual achievement and the 13 achievement motivation factors listed in Table 1). Random
effects included the intercepts at the two levels to evaluate class-to-class variation in individual
and class-average achievement. Separate analyses were done for each of the 13 achievement
motivation moderators to test the moderation of the BFLPE.
To facilitate interpretation of results in relation to a standard effect size metric, we
standardized individual student scores (M=0, SD = 1), including academic achievement and
all variables based on factor scores. However, none of the multiplicative effects (quadratic
achievement, interactions of class-average achievement with individual achievement, or any of
the achievement motivation moderators) or aggregated variables (class-average achievement)
were re-standardized; thus, they were kept in the same metric as the individual student
variables. This total-group standardization is important because it provides a common metric
with which to compare each class, as opposed to within-class standardization (e.g., within-
class centering, transforming the mean of each class to be zero).
Results
Here, our focus is on potential moderators of the BFLPE in relation a priori predictions based on
parsimonious predictions (based on BFLPE empirical research and Darwinian-economic theory) that
none of these achievement motivation constructs would moderate the BFLPE, and more nuanced
Educational Psychology Review
predictions based on our interpretations of motivation theories underlying the SAL and achievement
motivation constructs (Table 1). We begin by testing the BFLPE for our two self-belief measures
(MSC and math self-efficacy) and then test potential moderators of the BFLPE listed in Table 1.
Big-Fish-Little-Pond-Effect
Model 1A (Table 2) shows that individual student achievement is substantially related to both
math self-concept (.39) and math self-efficacy (.34). In the basic model of the BFLPE (Model
1B, also see Fig. 1a), the negative effect of class-average achievement is substantial for both
math self-concept (−.34) and math self-efficacy (−.27). In Model 1C, the quadratic compo-
nent of individual achievement is added to the model. Although the quadratic component is
statistically significant, adding it has little effect on the size of the BFLPE. In Model 1D, the
effect of individual student achievement was made random at the class level. However, this
random effect was not significant, indicating that the effect of achievement on both self-belief
constructs was consistent across the 47 different classes.
Moderation of the BFLPE by Individual Achievement (the “Bright Student”
Hypothesis)
In Model 1E (Table 2; also see Fig. 1b), we added the cross-level interaction between
individual and class-average achievement to test the “bright student”hypothesis (that the
Table 2 BFLPEs: Effects of Individual and Class-average Achievement on Math Self-concept and Math Self-
efficacy
Mod1A Mod1B Mod1C Mod1D Mod1E
Est SE Est SE Est SE Est SE Est SE
Self-Concept Fixed Part
BFLPE -.34 .07 -.33 .07 -.33 .07 -.35 .07
L1-Achievement-Linear .39 .03 .43 .03 .45 .03 .45 .03 .45 .03
L1-Achievement-Quad .05 .02 .06 .02 .08 .02
L1xL2Achievement -.07 .05
Random Part
L2: Class Intercept .10 .02 .06 .02 .06 .02 .06 .02 .06 .02
L2: L1-Ach-Lin .00 .01
L1: Student Intercept .74 .02 .73 .02 .73 .02 .73 .02 .73 .02
-2*log likelihood: 4949 4928 4919 4918 4917
Self-Efficacy Fixed Part
BFLPE -.27 .07 -.27 .07 -.27 .07 -.28 .07
L1-Achievement-Linear .34 .03 .38 .03 .39 .03 .40 .03 .39 .03
L1-Achievement-Quad .04 .02 .05 .02 .07 .03
L1xL2Achievement -.08 .05
Random Part
L2: Class Intercept .08 .02 .05 .02 .05 .02 .06 .02 .06 .02
L2: L1-Ach-Lin/CONS .00 .01 .00 .01
L1: Student Intercept .76 .03 .76 .03 .76 .03 .76 .03 .76 .03
-2*log likelihood: 5012 4997 4992 4990 4988
Note. Separate analyses were done for self-concept and self-efficacy. L1 = student level. L2= class level. Ach =
achievement (linear and quadratic components. BFLPE= big-fish-little-pond effect, the effect of class average
(L2) achievement. Parameter estimates in italic are statistically significant (p<.05).N= 47 classes, 1921
students.
Educational Psychology Review
BFLPE should be substantially smaller for the brightest students). However, this interaction
(−.07) was not statistically significant, and even the direction of the non-significant interaction
was not consistent with the hypothesis. Consistent with a priori predictions based on theory
and previous research (see earlier discussion), there is no support for the bright student
hypothesis.
Moderation of the BFLPE by Achievement Motivation Constructs
We tested the moderating effect of each of the 13 achievement motivation constructs (see
Table 1for the constructs and a priori hypotheses) in a series of 13 separate models (Model
3A–Model 3M in Table 3; also see Fig. 1c). Results indicate that none of the 13 interactions
between class-average achievement and the 13 moderators were statistically significant.
The BFLPEs in Table 3are the effects of class-average achievement after controlling
for moderators and interaction effects (Fig. 1c). Because many of the moderators are
substantially correlated with math self-concept, the effect of class-average achievement
on MSC (the BFLPE; Table 3) is smaller after controlling for them. For example, math
self-concept and math interest are highly correlated (.86, see Supplemental Materials), so
that the BFLPE after controlling for interest (−.11) is substantially less than in the
corresponding model without controlling for interest (−.33, Table 2). However, even
after controlling for interest, the effect of class-average achievement on MSC is still
highly significant (−.11, SE = .02). Importantly, for all 13 moderation models, the direct
effect of class-average achievement (the BFLPE) remains significantly negative, ranging
in size from −.11 to −.38 (M=−.21).
Discussion
Moderation of the BFLPE
Our primary focus is on the ability of a diverse set of motivational variables to moderate the
BFLPE or, conversely, the generalizability of the BFLPE in relation to these variables. The
findings are easy to summarize in that none of the interactions were statistically significant.
These results provide clear support for the extremely parsimonious prediction of no interac-
tions based on Darwinian-economic theory and previous BFLPE research. The results provide
no support for more nuanced predictions based on achievement motivation theories (Table 1;
also see related discussion by Dai and Rinn 2008).
The central rationale of our study is that social comparison is the basis of the BFLPE.
Indeed, previous BFLPE research (Huguet et al. 2009; Marsh et al. 2014) has shown that
controlling for social comparison largely eliminated the BFLPE. Reviews (e.g., Marsh and
Seaton 2015) of previous empirical BFLPE studies suggest that there are no substantial
moderators of the BFLPE at the level of the individual student, even going so far as to
suggest that it is a universal, pan-human phenomenon. Frank’s(2012) evolutionary per-
spective argued that social comparison tendencies are universal. Thus, from these perspec-
tives and consistent with our findings, achievement motivation variables considered here
are unlikely to alter social comparison tendencies and, thus, are unlikely to moderate the
BFLPE. Hence, these distinct research disciplines are consistent with each other as well as
the results of the present investigation.
Educational Psychology Review
Table 3 ModerationofBFLPEbyeachof13Covariates
Model Model3A Model3B Model3C Model3D Model3E Model3F Model3G Model3H Model3I Model3J Model3K Model3L Model3M M
Moderator Interest Fixed Importance Mastery PerfAppr PerfAvd Memory Persist Cooperat Compet Useful Deep Strategy
Fixed Effects Est SE Est SE Est SE Est SE Est SE Est SE Est SE Est SE Est SE Est SE Est SE Est SE Est SE
BFLPE -.11 .02 -.29 .05 -.35 .06 -.37 .06 -.21 .05 -.22 .06 -.32 .06 -.38 .06 -.35 .06 -.27 .05 -.35 .059 -.29 .05 -.32 .06 -.21
L1-Ach-Lin .08 .01 .31 .03 .40 .03 .38 .02 .27 .02 .30 .03 .45 .03 .37 .02 .46 .03 .29 .02 .40 .025 .34 .02 .39 .03 .34
L1-Ach-Quad .01 .01 .06 .02 .06 .02 .0 6 .02 .05 .01 .03 .02 .05 .02 .05 .02 .06 .02 .04 .01 .05 .017 .04 .02 .05 .02 .05
Moderator .93 .010 -.74 .03 .40 .02 .46 .02 .67 .02 -.46 .02 .31 .02 .46 .02 .27 .02 .65 .02 .43 .02 .52 .02 .46 .02 .34
Interaction .02 .01 -.01 .05 .04 .03 .03 .03 -.02 .02 -.05 .03 -.06 .03 .02 .03 -.01 .03 .02 .02 .01 .03 .02 .03 .01 .03 .00
Random Part
L2-Class .01 0.00 .04 .01 .05 .01 .04 .01 .03 .01 .04 .01 .05 .01 .04 .01 .05 .01 .03 .01 .05 .012 .03 .01 .04 .01 .04
L1-Student .12 .01 .57 .02 .61 .02 .56 .02 .37 .01 .59 .02 .65 .02 .57 .02 .68 .02 .38 .01 .59 .019 .52 .02 .58 .02 .52
Note. L1 = student level. L2 = class level. Ach = achievement (linear and quadratic components. BFLPE = big-fish-little-pond effect, the effect of class average (L2) achievement (in
italic). M = mean of effects across all 13 models. Parameter estimates shaded in gray are statistically significant (p< .05). Separate analyses were done for each of the achievement-
motivation moderators (see Table 1for more information on the achievement motivations listed under each model). Thus, for Model3A the moderator is interest and the interaction is
the effect of the interest by class-average achievement interaction. N= 47 classes, 1921 students.
Educational Psychology Review
Strengths, Weaknesses, and Directions for Further Research
It is essential to evaluate the strengths and limitations of the present investigation in relation to the
corpus of BFLPE studies (see reviews by Fang et al. 2018; Marsh and Seaton 2015; Marsh et al.
2017). BFLPE studies are now routinely based on appropriate multilevel models that more clearly
separate the effects of achievement at the levels of the individual student, the class, and the school
(see Marsh 2007; Marsh and Seaton 2015 for reviews of the essential design and statistical
requirements of BFLPE studies). Most BFLPE studies are based on a single wave of data for a
single group, which makes interpretations of generalizability and causality problematic. More robust
support for generalizability comes from the growing number of PISA studies showing that BFLPE
results generalize over many countries. More defensible support for causality comes from longitu-
dinal designs in which variables are collected in multiple waves (e.g., Marsh 1991; Marsh et al.
2000,2001,2019b;Pekrunetal.2019; also see review by Marsh and Seaton 2015).
In a few longitudinal studies (Marsh et al. 2000,2019b;alsoseeMarshetal.1995), analyses
were based on the individual student achievement measures from the end of primary school that
were used to assign students to different ability tracks. These pre-transition measures provide
more robust controls for pre-existing differences and the ordering of variables than do achieve-
ment indicators measured after the start of the transition, at the same time as ASC. Several studies
described as quasi-experimental (see review by Marsh and Seaton 2015; Marsh et al. 2018b)
provided additional controls in relation to alternative interpretations. Finally, in BFLPE laboratory
studies based on true random assignment, Zell and Alicke (2009; see also Alicke et al. 2010)
found support for the BFLPE when they experimentally manipulated the frame of reference in
relation to feedback given to participants about how their performances compared with those of
others. In summary, there is a convergence of support for the BFLPE interpretations from a wide
variety of different studies using multiple methods.
Generalizability of BFLPE over Motivation Constructs
A particular strength of our study was the theoretically diverse range of achievement motivation
measures based upon the OECD-SAL instrument that provides a standard set of motivation
measures that have been validated across the world. In their presentation of SAL, Marsh et al.
(2006) specifically noted the usefulness of this instrument for testing theoretical predictions in
relation to a standard set of measures, as we have done here. In this sense, we provide tests of
potential moderators of the BFLPE representing important motivation constructs and associated
theoretical models from which they are derived. We recognize, of course, that there are inevitably
additional motivation-related constructs that we could have considered (e.g., achievement
emotions; Pekrun et al. 2019). Nevertheless, the comprehensive, systematic, and diverse set of
constructs that we have considered is clearly a strength. We also recognize the need to replicate
these results in relation to additional data sets representing other countries and different educa-
tional systems. However, we do note that our results here are largely consistent with other
empirical research based on the BFLPE, including large cross-national studies based in PISA
data (see summaries by Fang et al. 2018; Marsh and Seaton 2015; Marsh et al. 2017).
Design Features
An important design feature of our study is the use of a high-stakes measure of achievement
collected before students actually started high school. Because this test had important
Educational Psychology Review
implications for subsequent tracking of students, it was likely to be taken more seriously by the
students than low-stakes tests, which have little or no implications for individual students (e.g.,
standardized tests in cross-national studies such as PISA). We also note that the inclusion of all
students from intact classes avoids many of the limitations associated with sampling variability
when school-/class-average measures are based on a sample of students. Nevertheless, our
study is based on a relatively small number of classes (47) from a single region (Hong Kong)
and a single age group over the transition to secondary school. From this perspective, there is a
need to evaluate the generalizability of the results with larger, more diverse samples of
students. We also note that although the design of the study is longitudinal in relation to
achievement and self-concept, it is cross-sectional in relation to self-belief and moderating
variables other than prior achievement. From this perspective, it would be useful to collect
multiple waves of data and apply more sophisticated statistical models based on the temporal
ordering of the variables that more fully test moderation, mediation (e.g., the extent to which
intervening variables can explain the BFLPE), and moderated mediation in relation to prox-
imal variables like those considered here as well as more distal outcomes.
Limitations and Directions for Future Research
It is important to emphasize that neither our study nor any finite set of studies can prove that
there are no student-level moderators of the BFLPE. Indeed, we have not examined all
potentially relevant moderators of the BFLPE, even at the individual student level. Even
within our more limited focus on motivational moderators, there is room for consideration of
additional constructs. Thus, for example, within the expectancy-value framework, our research
did not evaluate whether the perceived cost of an activity moderated the BFLPE. Also, we
have not thoroughly evaluated self-regulation processes (but see Seaton et al. 2010, for some
relevant research). We also note that different studies are not always consistent in their results.
Thus, for example, positive student-teacher relationships were found to be a substantial
moderator of the BFLPE in a study by Schwabe et al. (2019), even though a large PISA
study by Seaton et al. (2010) showed that it did not moderate the relationship. Hence, even for
apparently the same moderator, there is a need for systematic reviews of studies providing
inconsistent results. Nevertheless, note that positive social relationships with peers and
teachers warrant further research as a potential class-level moderator of the BFLPE. Hence,
even for apparently the same moderator, there is a need for systematic reviews of studies
providing inconsistent results. Furthermore, student-teacher relationships might be seen as a
teacher- or class-level variable, rather than, or in addition to, an individual student variable.
Hence, in further pursuit of this issue, it would be useful to separate the class- and student-level
components of student-teacher relationships in appropriate multilevel models (e.g., Marsh
et al. 2012) and to relate these to the BFLPE.
In the early stages of planning for this study, the collected research team sought the most
appropriate dataset to pursue these issues. Critical requirements were that the database had to
provide appropriate data to test the BFLPE, and had to provide a reasonable representation of
key constructs posited in motivation theory. Although we failed to identify any ideal database,
we deemed the database used here to be appropriate. In particular, it provided reasonable data
to test the BFLPE and the extent to which motivation constructs—including achievement goal
theory constructs—moderated the BFLPE. In particular, the database paralleled the constructs
collected in the PISA2000 SAL instrument, described as “OECD’s brief self-report measure of
educational psychology’s most useful affective constructs”(Marsh et al. 2005;alsoseeOECD
Educational Psychology Review
2004). We note, however, that the data were based on the trichotomous model of achievement
goals, rather than subsequent extensions of this model (e.g., Elliot and Hulleman 2017;Elliot
et al. 2011). Nevertheless, in evaluating the suitability of the data set, the team (including two
authors of the extended achievement goal theory models) thought that using the trichotomous
model was adequate because it focuses on the three most commonly studied and most broadly
applicable goals in the achievement goal literature—mastery-approach goals, performance-
approach goals, and performance-avoidance goals. These goals are most relevant to our target
sample of secondary school students. Nevertheless, we acknowledge that this is a potential
limitation of the present investigation and a possible direction for further research.
The rationale used in the present investigation, consistent with much previous research, is to
look for substantial moderators. We operationally defined “substantial”as moderators that
would neutralize or change the direction of the BFLPE. Thus, high levels of anxiety and
neuroticism exacerbate the BFLPE, but even low-anxious and low-neurotic students still suffer
the BFLPE. In this sense, the direction—but not the size—of the BFLPE generalizes over
levels of anxiety and neuroticism. However, we do not argue that moderation of the BFLPE in
relation to anxiety and neuroticism is unimportant. Hence, more emphasis in future research
(and systematic reviews) should be placed on the generalizability of effect sizes as well as the
direction of the BFLPE.
The focus of BFLPE research and particularly this study has been on psychological
moderators of the BFLPE. The major exception to this is individual-student-level achievement
that also failed to moderate the BFLPE. There are, of course, many other non-psychological
variables that might moderate the BFLPE. For example, in the present investigation, the
students were of similar ages; thus, we were not able to systematically evaluate the general-
izability of the BFLPE over an extensive age range (i.e., age as a moderator of the BFLPE).
Nevertheless, based on previous research, we know that student age affects the BFLPE.
Although the BFLPE has been demonstrated even for very young students at the start of
primary school (e.g., Tymms 2001), there is also evidence that the size of the BFLPE is
systematically smaller for very young students (Marsh et al. 2015; Salchegger 2016). Some
research suggests that this reflects developmental-cognitive differences in the ability of young
students to form accurate self-perceptions of their competence and the competence of others
that are the basis of social comparisons (as suggested by Marsh et al. 2015). However, it also
reflects the structure of schooling such that secondary schools are more likely to be stratified as
a function of ability than primary schools (as suggested by Parker et al. 2018; also see Lohbeck
and Möller 2017; Salchegger 2016). Further research into how the BFLPE and the social
comparison processes underpinning it vary with age is an important direction for further
research.
We also stress that the failure to find student-level moderators of the BFLPE interactions
does not mean that there are no group-level interactions. Indeed, the theoretical model
underpinning the BFLPE assumes that the size of the BFLPE is a function of the extent to
which there are systematic between-school differences in school-average achievement; if there
are no between-school differences in achievement (i.e., the variance of school-average
achievement is zero), the BFLPEs would be predicted to be zero. Thus, Parker et al. (2018)
demonstrated that the size of BFLPEs across different countries (and over different time waves
within the same country) varied systematically with various measures of between-school
variation within each country. Hence, the level of differentiation among schools is a moderator
of the BFLPE, consistent with the BFLPE theory. Nevertheless, even if school-average
achievement were the same for all schools, students would still use social comparison
Educational Psychology Review
processes. However, the BFLPE would be eliminated because the frame-of-reference (school-
average achievement) would be the same for all students in that country. Hence, an important
direction for further research is to explore further the effects of reducing the ability stratifica-
tion of schools within a system on the BFLPE and educational outcomes more generally.
BFLPE studies have been based mostly on academic self-belief measures, mainly MSC.
Although not our focus, a relevant question is to what extent does the negative effect of class-
average achievement (the BFLPE) generalize to other constructs, notably the achievement
motivation constructs considered here. In supplemental analyses (Supplemental Materials,
Section 4,Table3), we show that the effect of class-average achievement varies substantially
across the achievement motivation constructs considered here; the BFLPE is only statistically
significant for three of 13 motivation constructs: interest (−.24), performance-approach goals
(−.18), and performance-avoidance goals (.25; i.e., higher performance-avoidance goals for
children who attended classes with higher average achievement). Marsh (2007;Marshand
Seaton 2015) previously proposed that the size of the BFLPE for different constructs is logically
related (in the opposite direction) to the size of the effect of individual student achievement on
the construct as an outcome variable. The rationale for this proposal is that if a construct is not
systematically related to achievement, then social comparison in relation to achievement is
unlikely to have much effect on the achievement motivation variable. Hence, the largest
BFLPEs occur for the constructs that are also most highly related to individual-student-level
achievement (Supplemental Materials, Section 4). Thus, for example, the BFLPE is the largest
for self-concept and self-efficacy, and these arethe two constructs that are most highly related to
individual achievement. Indeed, across all 15 constructs (including self-concept and self-
efficacy), the correlation between effects of individual student achievement and the BFLPE
(the negative effect of class-average achievement) is a remarkable average r= .86. Exploring
further this relation between the size of the BFLPE with other constructs and how strongly
related the construct is to achievement is an important direction for further research.
Integrating our Theory-Integrating Approach and the Traditional Delphi
Method
The name Delphi is based on the Oracle of Delphi, who was able to foresee the future. The
traditional Delphi method (e.g., Linstone and Turoff 1975;Roweetal.1991; Sourani and Sohail
2015) is used to forecast the future, often contrasting these predictions with those based on theory,
time-series trends, and prior research. The process provides optimal integration of diverse
perspectives in relation to complex issues where consensus does not exist. The Delphi method
assumes that group judgments are more accurate than those of individuals in these situations.
The approach we used to integrate competing perspectives (see earlier discussion) has many
similarities to the hallmarks of the Delphi approach; bringing together a group of experts,
offering competing theoretical predictions, revising the predictions over several rounds based
on written feedback from the other experts, and a summary of the final results. Although the
original applications of the Delphi approach were to seek consensus on forecasts for the future,
there are many variations with quite different aims. Thus, in the classic overview of the Delphi
method, Adler and Ziglio (1996) state that the Delphi method is an exercise in group
communication. It intends systematically to enhance informed decision-making by enabling
decision-makers to plan based on a broad reservoir of knowledge, experience, and expertise.
Alternative aims include generating new ideas, problem-solving, forecasting, policy
Educational Psychology Review
development, and consensus-building. Thus, for example, Turoff (1970) described the Policy
Delphi method to assess social policy and public health in which competing policy alternatives
are the focus rather than forecasts of the future. LeBlanc and Baranoski (2011) described a
variation of the Delphi method in which the focus was to develop a consensus policy statement
in relation to best-practice medical advice. Boyer et al. (2019) described the application of the
Delphi approach to develop the core curriculum in a nursing program. Pezaro and Clyne
(2015) used the Delphi approach to develop an intervention to support midwives in distress.
De Vet et al. (2005) used a variation of the Delphi approach to evaluate determinants of
theoretical predictions that led to hypotheses worthy of further examination. A typical aim of
the Delphi approach is to reach consensus, but this is not always the case. Thus, in the
Argument Delphi technique (Seker 2015), the focus is to ask experts to create new arguments
and critique the arguments of other experts. In this sense, we see our Theory-Integrating
Approach as a variation of the traditional Delphi approach, along with the host of variants of
the Delphi approach that were developed for specific purposes.
Jingle-Jangle Fallacies in Educational Psychology Research
In the present investigation, latent correlations among several of the potential moderators are
substantial. This suggests a potential lack of discriminant validity and the possibility of jingle-
jangle fallacies (i.e., two constructs that have similar labels might be measuring different
constructs, and two constructs that have different labels might be measuring the same
construct). In educational psychology, there has been considerable conceptual convergence
on the operationalization of constructs such as those considered here. However, there is also an
ongoing debate about the degree of overlap between apparently distinct constructs. This is
particularly the case for measures coming from different theoretical frameworks and primarily
used by different “camps”of researchers who typically do not systematically evaluate how
their measures of constructs are related to those used by other researchers. Thus, for example,
Marsh (1994) evaluated the factor structure based on two different motivation instruments. The
mastery goal scales from the two instruments were highly related and reflected a common
underlying factor. However, the competition scale from one instrument reflected a perfor-
mance orientation primarily, but the competition scale from the other instrument reflected
more of a task orientation than a performance orientation. Thus, Marsh (1994; also see
Heyman and Dweck 1992; see Marsh et al. 2003) warned researchers to beware of jingle-
jangle fallacies, and to pursue construct validity studies to test interpretations of the measures
more vigorously. Similarly, Bong (1996) cautioned that “many researchers are too quick to
invent their own set of labels without carefully examining those found in the literature,”thus
creating “what can be aptly called ‘a conceptual mess’for those who try to draw a coherent
whole out of the relevant literature”(p. 151).
Given this history, it is not surprising that several of the constructs considered here are
substantially correlated (see latent correlation matrix in Supplemental Materials,butalsothe
wording of the items). Not unexpected, perhaps, were the high correlations between academic
self-concept and self-efficacy (see Marsh et al. 2019b, on the murky distinction between self-
concept and self-efficacy), and between performance-approach goals (e.g., “It is important for
me to do better than the other students in this subject”) and competitive learning (e.g., “Iliketo
try to be better than other students”—see Marsh et al. 2003). However, there were also high
correlations among mastery goals (e.g., “It is important for me to understand the content of this
Educational Psychology Review
subject as thoroughly as possible”), importance (e.g., “For me, being good at this subject is
very important”), and utility value (e.g., “Compared to most of my other activities, what I learn
in this subject is very useful”). Although not the focus of the present investigation, we note the
need to better clarify the theoretical and predictive distinctiveness of key constructs in
educational psychology by evaluating support for convergent and discriminant validity (and
jingle-jangle fallacies) when competing theoretical constructs are juxtaposed within the same
study. Ideally, this is best accomplished by collaboration among researchers from different
theoretical camps seeking to clarify conceptual issues in the measurement and application of
different constructs (e.g., Marsh et al. 1997; also see Marsh et al. 2019b). We suggest that this
might be accomplished by using a systematic approach such as our Theory-Integrating Delphi
method to sort our issues in the conceptual overlap and the appropriate measurement of critical
constructs.
Conclusions and Implications
In the present investigation, we have stressed the theoretical implications of integrating the
extensive BFLPE research literature with Frank’s(2012) Darwinian-economic perspective and
juxtaposing these with more nuanced predictions based on key motivational theories. The
results of the present investigation add to the growing research literature on the robustness of
the BFLPE.More specifically, we demonstrated that a range of student motivation variables that
might have been predicted to moderate the BFLPE based on achievement motivation theories
(Table 1) failed to do so. These results are in line with previous BFLPE research, which has
demonstrated the robustness of BFLPEs in relationto potential student-level moderators (Marsh
et al. 2017; Marsh and Seaton 2015). However, our paper does this more systematically in
relation to constructs based on achievement motivation theories. Importantly, we also provide a
synthesis with Darwinian-economic perspectives on this robustness. Indeed, even though there
has been little cross-referencing between this economic research and BFLPE studies, both have
a similar basis in terms of social comparison processes. The BFLPE provides further empirical
support for the Darwinian-economic perspective, and the Darwinian-economic perspective
provides an evolutionary theoretical basis for the robustness of the BFLPE.
There are many important implications associated with the BFLPE, social comparison
processes, and frame-of-reference effects more generally. We extend theory by integrating
theoretical perspective from economics (Darwinian) and educational (BFLPE) research disci-
plines. We extent BFLPE research by showing the BFLPE generalizes across a diverse set of
achievement motivation variables, contributing to the claim that it is a universal phenomenon
(Marsh et al. 2017; Marsh and Seaton 2015)—at least in relation to the achievement motiva-
tion moderators considered here. Concerning educational policy, in many school systems
worldwide, high-achieving students are increasingly being taught in academically selective
schools. However, the collected body of BFLPE research reviewed here—as well as our
results—suggests that this may not be the optimal environment for such students, at least in
terms of ASC. Indeed, there is a growing body of research suggesting that the ability
stratification that drives the BFLPE also has negative consequences for student achievement
and long-term educational attainment (Marsh 1991;MarshandO’Mara 2008). Thus, for
example, Parker et al. (2018) combined five cycles of PISA data to demonstrate that countries
with high levels of ability stratification had lower levels of achievement. Furthermore,
countries that increased ability stratification over this period also had decreasing levels of
academic achievement.
Educational Psychology Review
The evolutionary basis for the social comparison processes that underpin the BFLPE has
important practical implications. The search for student-level moderators of the BFLPE has
been prompted at least in part by the hope that these findings would lead to personalized
interventions that would counteract some of the negative consequences of social comparison
and the BFLPE (e.g., Dai and Rinn 2008). Alternatively, Frank (2012) suggests that social
comparison processes are inherent, but that there is a need to build environments in which its
negative consequences are reduced. Thus, for example, there is clear evidence from PISA
studies that the size of the BFLPE in different countries is strongly related to the extent of
ability stratification that exists in the different countries as noted earlier (Parker et al. 2018).
Taken to the extreme, if the average ability level is the same in all schools and classes, then
there should be no BFLPEs. The lesson to be learned from the Darwinian-economic perspec-
tive is that interventions aimed at reducing the negative consequences of the social comparison
processes should be aimed at the level of the class, school, school system, or even the whole
country rather than trying to modify the social comparison tendencies of individual students.
Supplementary Information The online version contains supplementary material available at https://doi.
org/10.1007/s10648-020-09583-5.
References
Adler, M., & Ziglio, E. (1996). Gazing into the oracle: The Delphi method and its application to social policy
and public health. Jessica Kingsley Publishers.
Alicke, M. D., Zell, E., & Bloom, D. (2010). Mere categorization and the frog pond effect. Psychological
Science, 2(2), 174–177. https://doi.org/10.1177/0956797609357718.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. New Jersey: Prentice-
Hall.
Bong, M. (1996). Problems in academic motivation research and advantages and disadvantages of their solutions.
Contemporary Educational Psychology, 21(2), 149–165. https://doi.org/10.1006/ceps.1996.0013.
Boyer, S., Mann-Salinas, E., Valdez-Delgado, K., & VanFosson, C. (2019). Using the Delphi technique to
determine core components of a nurse competency program. Journal for Nurses in Professional
Development, 35(5), 261–267. https://doi.org/10.1097/NND.0000000000000569.
Cheng, C. R., McInerney, D. M., & Mok, M. M. C. (2014). Does big-fish–little-pond effect always exist?
Investigation of goal orientations as moderators in the Hong Kong context. Educational Psychology, 34(5),
561–580. https://doi.org/10.1080/01443410.2014.898740.
Clark, A. E. (2018). Four decades of the economics of happiness: where next? Review of Income and Wealth,
64(2), 245–269. https://doi.org/10.1111/roiw.12369.
Coleman, J. M., & Fults, B. A. (1985). Special class placement, level of intelligence, and the self-concepts of
gifted children: a social comparison perspective. Remedial and Special Education, 6(2), 7–11. https://doi.
org/10.1111/roiw.12369.
Dai, D. Y., & Rinn, A. N. (2008). The big-fish-little-pond effect: what do we know and where do we go from
here? Educational Psychology Review, 20(3), 283e317–283e317. https://doi.org/10.1007/s10648-008-9071-
x.
Davis, J. A. (1966). The campus as a frog pond: an application of the theory of relative deprivation to career
decisions of college men. American Journal of Sociology, 72(1), 17–31. https://doi.org/10.1086/224257.
De Vet, E., Brug, J., De Nooijer, J., Dijkstra, A., & De Vries, N. K. (2005). Determinants of forward stage
transitions: a Delphi study. Health Education Research, 20(2), 195–205. https://doi.org/10.1093/her/cyg111.
Dicke, T., Marsh, H. W., Parker, P. D., Pekrun, R., Guo, J., & Televantou, I. (2018). Effects of school-average
achievement on individual self-concept and achievement: unmasking phantom effects masquerading as true
compositional effects. Journal of Educational Psychology, 110(8), 1112–1126. https://doi.org/10.1037
/edu0000259.
Dweck, C. (2000). Self-theories: Their role in motivation, personality and development. Hove: Psychology Press.
Educational Psychology Review
Eccles, J. S. (2009). Who am I and what am I going to do with my life? Personal and collective identities as
motivators of action. Educational Psychologist, 44(2), 78–89. https://doi.org/10.1080/00461520902832368.
Elliot, A. J. (1999). Approach and avoidance motivation and achievement goals. Educational Psychologist,
34(3), 169–189. https://doi.org/10.1207/s15326985ep3403_3.
Elliot, A. J., & Church, M. A. (1997). A hierarchical model of approach and avoidance achievement motivation.
Journal of Personality and Social Psychology, 72(1), 218–232. https://doi.org/10.1037/0022-3514.72.1.218.
Elliot, A. J., & Fryer, J. W. (2008). The goal construct in psychology. In J. Shah & W. Gardner (Eds.), Handbook
of motivational science (pp. 235–249). New York: Guilford Press.
Elliot, A. J., & Hulleman, C. S. (2017). Achievement goals. In A. J. Elliot, C. S. Dweck, & D. S. Yeager (Eds.),
Handbook of competence and motivation: Theory and application (pp. 43–60). New York: Guilford Press.
Elliot,A.J.,Murayama,K.,&Pekrun,R.(2011).A3x2achievementgoalmodel.JournalofEducational
Psychology, 103(3), 632–648. https://doi.org/10.1037/a0023952.
Fang, J., Huang, X., Zhang, M., Huang, F., Li, Z., & Yuan, Q. (2018). The big-fish-little-pond effect on academic
self-concept: a meta-analysis. Frontiers in Psychology, 9, 1569. https://doi.org/10.3389/fpsyg.2018.01569.
Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117–140. https://doi.
org/10.1177/001872675400700202.
Frank, R. H. (1985). Choosing the right pond: Human behavior and the quest for status. Oxford: Oxford
University Press.
Frank, R. H. (2012). The Darwin economy: liberty, competition, and the common good. Princeton: Princeton
University Press. https://doi.org/10.1515/9781400844982.
Guo, J., Marsh, H. W., Morin, A. J. S., Parker, P. D., & Kaur, G. (2015a). Directionality of the associations of
high school expectancy-value, aspirations, and attainment: a longitudinal study. American Educational
Research Journal, 52(2), 371–402. https://doi.org/10.3102/0002831214565786.
Guo, J., Parker, P. D., Marsh, H. W., & Morin, A. J. S. (2015b). Achievement, motivation, and educational
choices: a longitudinal study of expectancy and value using a multiplicative perspective. Developmental
Psychology, 51(8), 1163–1176. https://doi.org/10.1037/a0039440.
Heyman, G. D., & Dweck, C. S. (1992). Achievement goals and intrinsic motivation: their relation and their role
in adaptive mo tivation. Motivation and Emotion, 16(3), 231–247. https://doi.org/10.1007/BF00991653.
Huguet, P., Dumas, F., Marsh, H. W., Regner, I., Wheeler, L., Suls, J., Seaton, M., & Nezlek. (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), 671–710. https://doi.org/10.1037/a0015558.
Johnson, D. W., & Johnson, R. T. (1999). Making cooperative learning work. Theory into Practice, 38(2), 67–
73. https://doi.org/10.1080/00405849909543834
Jonkmann, K., Becker, M., Marsh, H. W., Lüdtke, O., & Trautwein, U. (2012). Personality traits moderate the
big-fish-little-pond effect of academic self-concept. Learning and Individual Differences, 22(6), 736–746.
https://doi.org/10.1016/j.lindif.2012.07.020.
LeBlanc, K., & Baranoski, S. S. (2011). Skin tears: State of the ccience: consensus statements for the prevention,
prediction, assessment, and treatment of skin tears. Advances in Skin & Wound Care, 24(Supplement), 2–15.
https://doi.org/10.1097/01.ASW.0000405316.99011.95.
Linstone, H. A., & Turoff, M. (1975). The Delphi method: techniques and applications. Boston: Addison-
Wesley.
Lohbeck, A., & Möller, J. (2017). Social and dimensional comparison effects on math and reading self-concepts
of elementary school children. Learning and Individual Differences, 54,73–81. https://doi.org/10.1016/j.
lindif.2017.01.013.
Marsh, H. W. (1984). Self-concept: the application of a frame-of-reference model to explain paradoxical results.
Australian Journal of Education, 28(2), 165–181. https://doi.org/10.1177/000494418402800207.
Marsh, H. W. (1991). Failure of high-ability high schools to deliver academic benefits commensurate with their
students’ability levels. American Educational Research Journal, 28(2), 445–480. https://doi.org/10.3102
/00028312028002445.
Marsh, H. W. (1994). Sport motivation orientations: beware of jingle-jangle fallacies. Journal of Sport and
Exercise Psychology, 16(4), 365–380. https://doi.org/10.1123/jsep.16.4.365.
Marsh, H. W. (2007). Self-concept theory, measurement and research into practice: the role of self-concept in
educational psychology. Leicester: British Psychological Society.
Marsh, H. W. (2016). Cross-cultural generalizability of year in school effects: negative effects of acceleration and
positive effects of retention on academic self-concept. Journal of Educational Psychology, 108(2), 256–273.
https://doi.org/10.1037/edu0000059.
Marsh, H. W., Abduljabbar, A. S., Morin, A. J. S., Parker, P., Abdelfattah, F., Nagengast, B., & Abu-Hilal, M.
M. (2015). The big-fish-little-pond effect: generalizability of social comparison processes over two age
cohorts from Western, Asian, and Middle-Eastern Islamic countries. Journal of Educational Psychology,
107(1), 258–271. https://doi.org/10.1037/a0037485.
Educational Psychology Review
Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness-of-fit indexes in confirmatory factor analysis:
the effect of sample size. Psychological Bulletin, 103(3), 391–410. https://doi.org/10.1037/0033-
2909.103.3.391.
Marsh, H. W., Chessor, D., Craven, R., & Roche, L. (1995). The effects of gifted and talented programs on
academic self-concept: the big fish strikes again. American Educational Research Journal, 32(2), 285–319.
https://doi.org/10.3102/00028312032002285.
Marsh, H. W., & Craven, R. G. (2006). Reciprocal effects of self-concept and performance from a multidimen-
sional perspective: beyond seductive pleasure and unidimensional perspectives. Perspectives on
Psychological Science, 1(2), 133–163. https://doi.org/10.1111/j.1745-6916.2006.00010.x.
Marsh, H. W., Craven, R. G., Hinkley, J. W., & Debus, R. L. (2003). Evaluation of the big-two-factor theory of
academic motivation orientations: an evaluation of jingle-jangle fallacies. Multivariate Behavioral Research,
38(2), 189–224. https://doi.org/10.1207/S15327906MBR3802_3.
Marsh, H. W., Guo, J., Dicke, T., Parker, P. D., & Craven, R. G. (2019a). Confirmatory factor analysis (CFA),
exploratory structural equation modeling (ESEM) & Set-ESEM: optimal balance between goodness of fit
and parsimony. Multivariate Behavioral Research, 55(1), 102–119. https://doi.org/10.1080
/00273171.2019.1602503.
Marsh, H. W., Pekrun, R., Parker, P. D., Murayama, K., Guo, J., Dicke, T., & Arens, A. K. (2019b). The murky
distinction between self-concept and self-efficacy: beware of lurking jingle-jangle fallacies. Journal of
Educational Psychology, 111(2), 331–353. https://doi.org/10.1037/edu0000281.
Marsh, H. W., & Hau, K.-T. (2003). Big-fish-little-pond effect on ASC: a cross-cultural (26-country) test of the
negative effects of academically selective schools. American Psychologist, 58(5), 364–376. https://doi.
org/10.1037/0003-066X.58.5.364.
Marsh, H. W., Hau, K. T., Artelt, C., Baumert, J., & Peschar, J. L. (2006). OECD’s brief self-report measure of
educational psychology’s most useful affective constructs: cross-cultural, psychometric comparisons across
25 countries. International Journal of Testing, 6(4), 311–360. https://doi.org/10.1207/s15327574ijt0604_1.
Marsh, H. W., Hau, K.T., & Grayson, D. (2005). Goodness of fit in structural equation models. In A. Maydeu-
Olivares & J. J. McArdle (Eds.), Multivariate applications book series. Contemporary psychometrics: A
festschrift for Roderick P. McDonald (p. 275–340). Lawrence Erlbaum Associates Publishers.
Marsh, H. W., Köller, O., & Baumert, J. (2001). Reunification of East and West German school systems:
longitudinal multilevel modeling study of the big-fish-little-pond effect on ASC. American Educational
Research Journal, 38(2), 321–350. https://doi.org/10.3102/00028312038002321.
Marsh, H. W., Kong, C. K., & Hau, K. T. (2000). Longitudinal multilevel models of the big-fish-little-pond effect
on academic self-concept: counterbalancing contrast and reflected-glory effects in Hong Kong schools.
Journal of Personality and Social Psychology, 78(2), 337–349. https://doi.org/10.1037/0022-3514.78.2.337.
Marsh, H. W., Kuyper, H., Morin, A. J., Parker, P. D., & Seaton, M. (2014). Big-fish-little-pond social
comparison and local dominance effects: integrating new statistical models, methodology, design, theory
and substantive implications. Learning and Instruction, 33,50–66. https://doi.org/10.1016/j.
learninstruc.2014.04.002.
Marsh, H. W., Lüdtke, O., Nagengast, B., Trautwein, U., Morin, A. J. S., Abduljabbar, A. S., & Köller, O.
(2012). Classroom climate and contextual effects: conceptual and methodological issues in the evaluation of
group-level effects. Educational Psychologist, 47(2), 106–124. https://doi.org/10.1080
/00461520.2012.670488.
Marsh, H. W., Martin, A. J., Yeung, A., & Craven, R. (2017). Competence self-perceptions. In A. J. Elliot, C.
Dweck, & D. Yeager (Eds.), Handbook of competence and motivation (pp. 85–115). New York: Guilford
Press.
Marsh, H. W., & O'Mara, A. (2008). Reciprocal effects between academic self-concept, self-esteem, achieve-
ment, and attainment over seven adolescent years: Unidimensional and multidimensional perspectives of
self-concept. Personality & Social Psychology Bulletin : Journal of the Society for Personality and Social
Psychology, 34(4), 542–552. https://doi.org/10.1177/0146167207312313
Marsh, H. W., & Parker, J. W. (1984). Determinants of student self-concept: is it better to be a relatively large
fish in a small pond even if you don’t learn to swim as well? Journal of Personality and Social Psychology,
47(1), 213–231. https://doi.org/10.1037/0022-3514.47.1.213.
Marsh, H. W., Parker, P. D., & Pekrun, R. (2018a). Three paradoxical effects on academic self-concept across
countries, schools, and students: frame-of-reference as a unifying theoretical explanation. European
Psychologist, 24(3), 231–242. https://doi.org/10.1027/1016-9040/a000332.
Marsh, H. W., Pekrun, R., Murayama, K., Arens, K. A., Parker, P. D., Guo, J., & Dicke, T. (2018b). An
integrated model of academic self-concept development: Academic self-concept, grades, test scores, and
tracking over 6 years. Developmental Psychology, 54(2), 263–280. https://doi.org/10.1037/dev0000393
Educational Psychology Review
Marsh, H. W., Roche, L. A., Pajares, F., & Miller, D. (1997). Item-specific efficacy judgments in mathematical
problem solving: the downside of standing too close to trees in a forest. Contemporary Educational
Psychology, 22(3), 363–377. https://doi.org/10.1006/ceps.1997.0942.
Marsh, H. W., & Seaton, M. (2015). The big-fish–little-pond effect, competence self-perceptions, and relativity:
substantive advances and methodological innovation. Advances in Motivation Science, 2, 127–184.
https://doi.org/10.1016/bs.adms.2015.05.002.
Marsh, H. W., Seaton, M., Trautwein, U., Lüdtke, O., Hau, K. T., O’Mara, A. J., & Craven, R. G. (2008). The
big-fish-little-pond-effect stands up to critical scrutiny: implications for theory, methodology, and future
research. Educational Psychology Review, 20(3), 319–350. https://doi.org/10.1007/s10648-008-9075-6.
Morin, A. J. S., & Marsh, H. W. (2015). Disentangling shape from level effects in person-centered analyses: An
illustration based on university teachers’multidimensional profiles of effectiveness. Structural Equation
Modeling, 22(1), 39–59. https://doi.org/10.1080/10705511.2014.919825
Muthén, L. K., & Muthén, B. O. (2017). Mplus user’sguide(8th ed.). Los Angeles, CA: Author.
Nagengast, B., & Marsh, H. W. (2012). Big fish in little ponds aspire more: mediation and cross-cultural
generalizability of school-average ability effects on self-concept and career aspirations in science. Journal of
Educational Psychology, 104(4), 1033–1053. https://doi.org/10.1037/a0027697.
OECD. (2004). PISA 2003 technical report. Paris: Organisation for Economic Co-operation and Development.
Parker, P. D., Dicke, T., Guo, J., & Marsh, H. W. (2018). A macro context theory of academic self-concept:
ability stratification and the big-fish-little-pond effect. Psyarxiv.https://doi.org/10.31234/osf.io/bwy59.
Parker, P. D., Marsh, H. W., Jerrim, J., Guo, J., & Dicke, T. (2020). Trade-offs between equity and excellence in
academic performance: Evidence from 27 OECD countries. New York: American Education Research
Journal.
Pekrun, R., Murayama, K., Marsh, H. W., Goetz, T., & Frenzel, A. C. (2019). Happy fish in little ponds: Testing
a reference group model ofachievement and emotion. Journal of Personality and Social Psychology, 117(1),
166–185. https://doi.org/10.1037/pspp0000230.
Pezaro, S., & Clyne, W. (2015). Achieving consensus in the development of an online intervention designed to
effectively support midwives in work-related psychological distress: protocol for a delphi study. JMIR
Research Protocols, 4(3), e107. https://doi.org/10.2196/resprot.4766.
Rasbash, J., Steele, F., Browne, W., & Prosser, B. (2004). Auser’s guide to MLwiN–Version 2.0. Bristol:
University of Bristol.
Rowe, G., Wright, B., & Bolger, F. (1991). Delphi: a reevaluation of research and theory. Technological
Forecasting and Social Change, 39(3), 235–251. https://doi.org/10.1016/0040-1625(91)90039-I.
Salchegger, S. (2016). Selective school systems and academic self-concept: how explicit and implicit school-
level tracking relate to the big-fish–little-pond effect across cultures. Journal of Educational Psychology,
108(3), 405–423. https://doi.org/10.1037/edu0000063.
Schwabe,F., Korthals, R., & Schils, T. (2019). Positive social relationships with peers and teachers as moderators
of the big-fish-little-pond effect. Learning and Individual Differences, 70,21–29. https://doi.org/10.1016/j.
lindif.2018.12.006.
Seaton, M., Marsh, H. W., & Craven, R. G. (2009). Earning its place as a pan-human theory: universality of the
big-fish-little-pond effect across 41 culturally and economically diverse countries. JournalofEducational
Psychology, 101(2), 403–419. https://doi.org/10.1037/a0013838.
Seaton, M., Marsh, H. W., & Craven, R. G. (2010). Big-fish-little-pond-effect: generalizability and moderation—
two sides of the same coin. American Educational Research Journal, 47(2), 390–434. https://doi.
org/10.3102/0002831209350493.
Seker, S. E. (2015). Computerized argument Delphi technique. IEEE Access, 3, 368–380. https://doi.org/10.1109
/ACCESS.2015.2424703.
Sourani, A., & Sohail, M. (2015). The Delphi method: review and use in construction management research.
International Journal of Construction Education and Research, 11(1), 54–76. https://doi.org/10.1080
/15578771.2014.917132.
Tracey, D. K., Marsh, H. W., & Craven, R. G. (2003). Self-concepts of preadolescent students with mild
intellectual disabilities: Issues of measurement and educational placement. In H. W. Marsh, R. G. Craven, &
D. M. McInerney (Eds.), International Advances in Self Research (Vol. 1, pp. 203–230). Charlotte:
Information Age Publishing.
Turoff, M. (1970). The design of a policy DelphiII. Technological Forecasting and Social Change, 2(2), 149–
171. https://doi.org/10.1016/0040-1625(70)90161-7.
Tymms, P. (2001). A test of the big fish in a little pond hypothesis: an investigation into the feelings of seven-
year-old pupils in school. School Effectiveness and School Improvement, 12(2), 161–181. https://doi.
org/10.1076/sesi.12.2.161.3452.
Wigfield, A., & Eccles, J. S. (2000). Expectancy-value theory of achievement motivation. Contemporary
Educational Psychology, 25(1), 68–81. https://doi.org/10.1006/ceps.1999.1015.
Educational Psychology Review
Wouters, S., Colpin, H., Van Damme, J., & Verschueren, K. (2013). Endorsing achievement goals exacerbates
the big-fish-little-pond effect on academic self-concept. Educational Psychology, 35(2), 252–270.
https://doi.org/10.1080/01443410.2013.822963.
Zell, E., & Alicke, M. D. (2009). Contextual neglect, self-evaluation, and the frog-pond effect. Journal of
Personality and Social Psychology, 97(3), 467–482. https://doi.org/10.1037/a0015453.
Zimmerman, B. J. (2000). Attaining self-regulation: a social cognitive perspective. In M. Boekaerts, P. R.
Pintrich, & M. Zeidner (Eds.), Handbook of self-regulation (pp. 13–39). Cambridge: Academic Press.
Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Affiliations
Herbert W. Marsh
1,2
&Kate M Xu
3
&Philip D Parker
1
&Kit-Tai Hau
4
&
Reinhard Pekrun
1,5,6
&Andrew Elliot
7
&Jiesi Guo
1
&Theresa Dicke
1
&
Geetanjali Basarkod
1
*Herbert W. Marsh
Herb.Marsh@acu.edu.au
Kate M Xu
Kate.Xu@ou.nl
Philip D Parker
Philip.Parker@acu.edu.au
Kit-Tai Hau
kthau@cuhk.edu.hk
Reinhard Pekrun
pekrun@lmu.de
Andrew Elliot
andrew.elliot@rochester.edu
Jiesi Guo
Jiesi.Guo@acu.edu.au
Theresa Dicke
Theresa.Dicke@acu.edu.au
Geetanjali Basarkod
Geetanjali.Basarkod@acu.edu.au
1
Australian Catholic University, Level 10, 33 Berry Street, North Sydney, NSW 2060, Australia
2
Oxford University, Oxford OX1 2JD, UK
3
Open University of the Netherlands, Heerlen, Netherlands
4
The Chinese University of Hong kong, Hong Kong, Hong Kong
5
University of Essex, Colchester, UK
6
University of Munich, Munich, Germany
7
University of Rochester, Rochester, NY, USA
Educational Psychology Review
A preview of this full-text is provided by Springer Nature.
Content available from Educational Psychology Review
This content is subject to copyright. Terms and conditions apply.