EMOTIONS, ACHIEVEMENT, AND BUOYANCY 1

Achievement Emotions and Academic Achievement: Reciprocal Relations and the

Moderating Influence of Academic Buoyancy

David W. Putwain1, Peter Wood1 and Reinhard Pekrun2,3

1
School of Education, Liverpool John Moores University
2
Department of Psychology, University of Essex
3
Institute of Positive Psychology and Education, Australian Catholic University

Journal of Educational Psychology, in press

Author Note

David W. Putwain https://orcid.org/0000-0001-5196-4270

Peter Wood https://orcid.org/0000-0002-2727-9342

Reinhard Pekrun https://orcid.org/0000-0003-4489-3827

We have no known conflict of interest to disclose. Our work was funded by a grant

from the Bowland Trust. We thank Emma Rainbird for her assistance with data collection.

Correspondence concerning this article should be addressed to: David W. Putwain,

School of Education, Liverpool John Moores University, IM Marsh Campus, Mossley Hill

Rd, Liverpool, L17 6BD. Phone: +44 (0)151 231 5270. Email: [email protected]
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 2

Abstract

Control-value theory proposes that achievement emotions impact achievement, and

that achievement outcomes (i.e., success and failure) reciprocally influence the development

of achievement emotions. Academic buoyancy is an adaptive response to minor academic

adversity, and might, therefore, offer protection from achievement being undermined by

negative achievement emotions. At present, however, there is little empirical evidence for

these hypothesized relations. In this study we examined reciprocal relations between three

achievement emotions (enjoyment, boredom, and anxiety) and test performance in the

context of mathematics, and whether academic buoyancy moderated relations between these

emotions and test performance. Data were collected from 1,242 primary school students

(mean age = 9.3 years) over four waves within one school year. Achievement emotions (T1

and T3) and test performance (T2 and T4) were measured alternately. Academic buoyancy was

measured at T3. A structural equation model showed negative relations of anxiety to

subsequent test performance, and negative relations of test performance to subsequent

anxiety. Test performance also predicted enjoyment and boredom, but not vice versa. A

latent-interaction structural equation model showed buoyancy moderated relations between

anxiety and test performance. Test performance was highest when anxiety was low and

buoyancy high. Practitioners should consider using interventions to reduce anxiety and

downstream effects on achievement.

Keywords: control-value theory, achievement emotions, academic achievement,

anxiety, buoyancy
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 3

Educational Impact and Implications Statement

In classroom settings, multiple emotions such as enjoyment, boredom, and anxiety may

occur. Among these emotions, anxiety is especially important for students’ achievement in

mathematics according to this study with elementary school children. Reducing anxiety

would be beneficial for students’ achievement, for example through fostering their adaptive

responses to failure and increasing perceptions of control. The findings also suggest that

developing academic buoyancy can benefit the achievement of students with mild forms of

anxiety.
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 4

Achievement Emotions and Academic Achievement: Reciprocal Relations and the

Moderating Influence of Academic Buoyancy

Mathematics learning in elementary or primary school is generally considered to be of

critical importance to the person and for society at large. Functional numeracy skills are

vitally important in adult personal and work life (e.g., managing personal finances), and

failure to master basic mathematics skills is associated with subsequent unemployment and

lower earning potential (e.g., Hanushek & Woessmann, 2008). Furthermore, economic

national competitiveness requires a highly skilled science, technology, engineering, and

mathematics (STEM) workforce (Kärkkäinen & Vincent-Lancrin, 2013). Unfortunately,

many students fail to learn these fundamental numeracy skills. For example, in the United

States, 19% of children were judged to show below basic mathematics skills at the end of

Grade 4 in 2019 (National Center for Educational Statistics, 2019). Similarly, 21% of

children left primary school in England at the age of 11 years in 2019 without reaching the

expected minimum standard in mathematics (Department for Education, 2019).

Achievement emotions have been found to be critically important for students’

academic achievement (e.g., Loderer et al., 2018; Tze et al., 2016; von der Embse et al.,

2018). However, few studies have been conducted that take into consideration the combined

influence of multiple achievement emotions. Furthermore, the impact of possible protective

factors has been neglected. The present longitudinal study with Year 5 elementary school

children targets these deficits by examining joint relations between three emotions that

students commonly experience in the classroom, namely enjoyment, boredom, and anxiety,

and their performance in mathematics. In addition, we investigated whether an asset-driven

psychological attribute, namely academic buoyancy, is influential in protecting students’

mathematics test performance from the detrimental influences of boredom and anxiety.

Achievement Emotions
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 5

Concept of Achievement Emotions

Achievement emotions were defined by Pekrun (2017) as “…emotions that relate to

achievement activities (e.g., participating in a competition) or achievement outcomes (e.g.,

success and failure)” (p. 252). Achievement emotions are multifaceted, containing affective,

cognitive, physiological, and motivational components, and are distinct from moods which

are of lower intensity, longer lasting, and often with no specific referent (Linnenbrink, 2006;

Pekrun, 2006). Achievement emotions can be experienced in a variety of achievement-

oriented settings including the classroom, tests and exams, and homework. In the present

study we focused on three classroom-related emotions: Enjoyment, boredom, and anxiety.

These emotions can be differentiated by valence and physiological activation. Enjoyment is a

positive activating emotion, boredom is a negative deactivating emotion, and anxiety is a

negative activating emotion.

Relations between Achievement Emotions and Academic Achievement: Control-Value

Theory

CVT (Pekrun, 2006, 2018, in press; Pekrun & Perry, 2014; Pekrun et al., 2002, 2019) is

a theoretical framework that incorporates the antecedents and outcomes of achievement

emotions. Appraisals of value (how and why an achievement activity or outcome is

important) and control (expectancy for future, and attributions of past success and failure) are

considered as proximal antecedents of achievement emotions. Achievement emotions are

thought to influence cognitive and motivational processes that, in turn, underpin performance

and academic achievement. According to CVT, not only will achievement emotions influence

performance and achievement, but performance and achievement outcomes can also

influence achievement emotions in a cycle of reciprocal causation.

Enjoyment. In CVT, enjoyment is theorized to enhance academic achievement though

promoting interest and intrinsic motivation, maintaining cognitive resources, focusing

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 6

attention on the task at hand, and supporting use of flexible and deep learning strategies as

well as self-regulation of learning. These different motivational and cognitive mechanisms

can interact. For example, the influence of positive affect on cognition and attention might

differ according to the degree of motivational intensity (Gable & Harmon-Jones, 2010).

When positive affect is combined with less intense approach motivation (e.g., in

contentment), cognition and attention can be broadened, but when combined with high-

intensity approach motivation (e.g., in enjoyment), cognition and attention can be narrowed

to focus on the most salient task details in order to facilitate goal pursuit. As such, enjoyment

should be especially conducive to academic learning. Furthermore, effects on achievement

may also depend on interactions of these mechanisms with task demands. Specifically, effects

on achievement mediated by different styles of processing may depend on the match between

processing style and type of task (see, e.g., Fiedler & Beier, 2014). For example, as proposed

in CVT, anxiety can facilitate rigid rehearsal of learning materials, whereas enjoyment can

enhance more creative ways of studying, implying that these two emotions can promote

different kinds of task performance. Given the interplay between mediating mechanisms and

their interactions with type of task, the effects of emotions on students’ learning may be

complex. However, for resulting academic achievement, it is reasonable to expect that greater

enjoyment typically results in greater achievement. In turn, academic success would, all

things being equal, strengthen control and positive value appraisals resulting in greater

enjoyment (a positive reciprocal cycle; for supporting evidence, see Pekrun et al., 2017).

Boredom. In CVT boredom is theorized to impair performance and achievement by

undermining interest and intrinsic motivation, reducing cognitive resources, and promoting

superficial learning. Types of boredom characterized by very low arousal (indifferent and

apathetic) may be more damaging for learning than those (e.g., searching and reactant)

characterized by an active search for less boring alternatives (Goetz & Hall, 2014). Students
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 7

in English Year 5 primary school mathematics lessons do not have the option to choose

alternative (and potentially less boring) activities. It is likely that the more damaging types of

boredom (indifferent and apathetic) would be experienced. Lower performance and success

would, all things being equal, weaken value appraisals, resulting in greater boredom (a

positive reciprocal cycle).

Anxiety. In CVT anxiety is theorized to have dual effects on performance and

achievement. Interest and intrinsic motivation can be undermined and working memory

processes and executive functions disrupted (also see Derakshan & Eysenck, 2011; Eysenck

et al., 2007). However, anxiety can also facilitate more rigid information processing, such as

simple rehearsal, and can increase motivation to invest effort to avoid failure. Rigid

information processing is unlikely to benefit students on the mathematics test used in the

present study as the items required not only the recall of previously learnt mathematical

reasoning but the application of that reasoning to a novel question. While increased effort

will likely reduce the negative effects of anxiety, the overall effects of anxiety on academic

achievement are negative in the vast majority of students; a recent meta-analysis of 238

studies showed a mean correlation of r = -.24 between test anxiety and achievement (von der

Embse et al., 2018). Lower performance and success would, all things being equal, weaken

control and negative value appraisals resulting in greater anxiety (a positive reciprocal cycle).

Empirical Studies of Relations between Achievement Emotions and Academic

Achievement

When considered in isolation, enjoyment shows positive, and boredom and anxiety

show negative, relations with academic achievement (for meta-analyses see Loderer et al.,

2018; von der Embse et al., 2018; Tze et al. 2016). Few studies have modeled enjoyment,

boredom, and anxiety together, however, to account for the concurrent relations between

these three emotions and their unique contributions to predicting achievement. Three notable
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 8

exceptions have included enjoyment, boredom, and anxiety, together in single analytic

models. In secondary school students, Ahmed et al. (2013) found unique statistically

significant relations with mathematics achievement for enjoyment (positively), and boredom

and anxiety (negatively). However, Raccanello et al. (2019; elementary school students) and

Putwain et al. (2020; primary school students) found that only enjoyment and anxiety, but not

boredom, remained statistically significant unique predictors of subsequent mathematics

achievement.

A few studies that have included enjoyment, boredom, or anxiety, alongside other

emotions not included in the present study, further underscore how the predictive value of

discrete emotions can differ when considered together with other emotions. When considered

alongside pride, enjoyment positively and anxiety negatively predicted mathematics

achievement in undergraduates (Villavicencio, & Bernardo, 2016). Neither enjoyment nor

boredom or anxiety were significantly related to the mathematics achievement of elementary

school students when considered alongside surprise, curiosity, confusion, and frustration

(Muis et al., 2015). Another study of elementary school students that included the same

achievement and epistemic emotions found that boredom negatively predicted mathematics

achievement, mediated by lower use of cognitive learning strategies (DiLeo et al., 2019).

Achievement emotions are often moderately to highly intercorrelated (e.g., rs = -.62 to

.82 in Pekrun et al., 2011) due to the shared appraisal antecedents (Pekrun, 2006). The co-

linearity between emotions reduces the predictive power for each discrete emotion; only

those with stronger relations to performance and achievement, or that have relatively weaker

co-linearity with other emotions, will remain statistically significant predictors. The practical

corollary in classroom settings is that when achievement emotions co-occur, not all may exert

effects of the same strength on performance and achievement; some emotions may be more

practically significant for performance than others. Therefore, it is vital for empirical work to
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 9

combine achievement emotions to competitively identify the most robust predictors of

performance and achievement.

Reciprocal Relations between Achievement Emotions and Academic Achievement

Studies testing reciprocal relations between achievement emotions and achievement

are largely lacking, with three notable exceptions. Over ten waves of measurement reciprocal

relations were shown between boredom and test performance (βs -.10 to -.23) in

undergraduate students taking an introductory psychology course (Pekrun et al., 2014). Over

five waves of measurement, Pekrun et al. (2017) showed reciprocal relations with

mathematics achievement for enjoyment (βs .11 to .13), boredom (βs -.06 to -.09), and

anxiety (βs -.07 to -.14) in secondary school students. Only one study (Putwain, Becker, et

al., 2018; primary school students) has modelled reciprocal effects for more than a single

emotion (enjoyment and boredom) simultaneously. Over four waves, reciprocal relations

were shown between enjoyment and achievement (βs .12 to .30), and boredom and

achievement (βs -.07 to -.36); enjoyment and boredom were shown to have unique effects.

No studies thus far have included more than two emotions simultaneously in a single analytic

model. In the present study we address this limitation of the literature.

Academic Buoyancy

What is it, and how does it differ from Cognate Constructs?

Academic buoyancy is the ability to respond adaptively to the everyday challenges,

setbacks, and pressures experienced by students during their studies (Martin & Marsh, 2009).

Examples are periods of poor performance; dips in confidence, motivation, and engagement;

receiving negative feedback from teachers; and the demands of tests and assessments.

Academic buoyancy can be contrasted with academic resilience which refers to adaptive

responses to major adversities such as chronic underachievement and failure, school truancy

and refusal, and clinical levels of anxiety or depression (Martin & Marsh, 2009). In short,
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 10

academic buoyancy is the ability to ‘bounce back’ from minor adversities whereas academic

resilience is the ability to ‘bounce back’ from major adversities.

Fong and Kim (2019) showed that academic buoyancy was distinct from other

cognate constructs, including grit (i.e., persistency of effort and consistency of interest;

Duckworth et al., 2007) and future time perspective (i.e., the perceived connection between

present activities and future goals; Lens & Seginer, 2015), in a sample of undergraduate

students. Items for academic buoyancy loaded separately from those of grit and future time

perspective in factor analysis, and correlations between academic buoyancy and the other two

constructs were small (rs < .27).

An unresolved question in the literature is the extent to which some level of exposure

to adversity is necessary for persons to build adaptive responses (e.g., Brooks, 2006; Compas,

2004). The types of adversities that academic buoyancy is theorized to protect against are

those experienced during routine schooling by the majority of students (Martin & Marsh,

2009). Indeed the very utility of the academic buoyancy construct is founded on this point;

unlike resilience it has relevance to the majority of students. Evidence has shown that

children in primary or elementary school can experience and overcome difficulties in reading,

writing, and numeracy (e.g., Holmes & Dowker, 2013; O’Connor et al., 2015), may receive

negative feedback from teachers (Hattie & Timperley, 2007), negatively compare themselves

to higher achieving classmates (Marsh, 2007), and be exposed to the pressures of testing

(e.g., von der Embse & Witmer, 2014). These are the types of everyday academic adversities

that buoyancy is theorized to protect against, and they are adequately captured by the

academic buoyancy scale (Martin & Marsh, 2008).

Relations with Achievement

Although academic buoyancy shows positive relations with adaptive beliefs, affect,

and behaviors in primary and secondary school students (e.g., Martin et al., 2010, 2013;
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 11

Hirvonen, Yli-Kivistö, et al., 2019; Hirvonen, Putwain, et al. 2019; Putwain et al., 2012,

2015) relations between academic buoyancy and achievement are equivocal. Putwain et al.

(2016) showed that domain general academic buoyancy positively predicted performance in

aggregated scores for English, science and mathematics secondary school exit examinations

(β = .16), after controlling for concurrent relations with test anxiety. Martin (2014) showed

that greater academic buoyancy predicted higher achievement (β = .07) on standardized

secondary school numeracy and literacy tests, after controlling for socio-demographical

variables and ‘big-five’ personality traits.

However, other studies have shown that buoyancy did not always predict achievement

when included in more complex models with multiple predictors. In studies of secondary

school students that have included control, academic buoyancy did not predict academic

achievement on standardized school numeracy and literacy tests (Collie et al., 2015) or

secondary school exit examinations in English, mathematics, and science (Putwain &

Aveyard, 2018). In the aforementioned study by Fong and Kim (2019), academic buoyancy

was not significantly related with self-reported GPA in undergraduates after controlling for

grit and future time perspective.

While the bivariate correlations in these studies (Collie et al., 2015; Fong & Kim,

2019; Putwain & Aveyard, 2018) were positive (rs = .10 to .15), in the presence of related

variables the predictive value of academic buoyancy was reduced. This reduction of direct

effects may be due to the effects of buoyancy being mediated by other variables (e.g.,

buoyancy bolstering perceived control, and control influencing performance in the studies by

Collie et al., 2015, and Putwain & Aveyard, 2018). Alternatively, the reason may be construct

overlap reducing the individual predictive power of buoyancy when combined with, for

instance, the perseverance component of grit. Furthermore, reduced direct effects do not rule

out the possibility that buoyancy interacts with other variables, especially those that pose
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 12

academic adversity (such as low perceived control). It is also possible that small or

statistically non-significant relations between academic buoyancy and achievement also arise

from using domain-general measures of academic buoyancy and achievement or mismatching

domain-general measures of academic buoyancy with domain-specific measures of

achievement (see Swann et al., 2007).

Only two studies have used domain-specific measures of academic buoyancy to

examine relations with achievement. Yun et al. (2018) showed that academic buoyancy in

second language (L2) acquisition predicted end-of-course L2 exam scores (β = .31) after

controlling for prior achievement in a sample of undergraduate students. Colmar et al. (2019),

however, found only small relations between academic buoyancy for mathematics and

mathematics test performance (r = .10), and between academic buoyancy for reading and

reading test performance (r = .09), in primary school students. Yun et al.’s (2018) findings

suggest there is some merit in the idea that relations between academic buoyancy and

achievement are stronger when analyzed in a domain-specific fashion. Accordingly, in the

present study we adopted a domain-specific approach.

The Buffering Effect of Academic Buoyancy for Adaptive Educational Outcomes

Academic buoyancy is linked to adaptive responses to adversity, including

strengthened positive and reduced negative emotions. In addition, it is plausible that

academic buoyancy would not only lessen the intensity of emotions such as anxiety and

boredom but also reduce their educational detrimental impact. Low levels of academic

buoyancy would be expected to have little impact on the negative relations between boredom

and anxiety, on the one hand, and achievement, on the other. As buoyancy increases, it would

be expected to buffer against the detrimental impact of boredom and anxiety such that the

negative relation would be weakened. An interaction between academic buoyancy and

boredom or anxiety would therefore be expected. At low boredom or anxiety, there would be
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 13

little difference in the achievement of low and high academically buoyant students. As

boredom and anxiety increase, however, high academically buoyant students will show

higher achievement that their low academically buoyant counterparts.

Few studies have examined the possible moderating role of academic buoyancy.

Putwain et al. (2016) showed that the negative relation between test anxiety and task-focused

coping was reduced in academically buoyant secondary school students. Symes et al. (2015)

found that the negative relation between teachers’ use of failure-avoidance messages prior to

a forthcoming high-stakes exam, on the one hand, and the appraisal of that exam as

threatening, on the other, was reduced in academically buoyant secondary school students.

Finally, Martin and Marsh (2019) found a marginally significant effect (β = -.13, p = .10) of

the interaction between academic buoyancy and academic adversity on subsequent academic

adversity in a sample of secondary school students. The positive relation between prior and

later academic adversity, a year apart, was weaker at higher academic buoyancy in keeping

with its theorized adaptive nature.

No studies have examined how academic buoyancy buffers the effects of performance

impairing negative classroom emotions, such as boredom and anxiety. In the present study

we address this gap in the literature. As low enjoyment is not typically considered as

academically adverse, we would not expect enjoyment to interact with academic buoyancy

but keep this as an exploratory question.

Aims and Hypotheses

Given the age of the participants in our study, namely students in Year 5 in their

penultimate year of primary education (Year 5) aged 9-10 years, we chose to measure

classroom emotions specifically. In England, where there study was based, students take

standardized National Curriculum Tests (NCTs) in reading (two papers) and mathematics

(two papers) at the end of Year 2, aged 6-7 years (Key Stage 1 NCTs), and in English (three
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 14

papers) and mathematics (three papers) at the end of primary schooling, Year 6, aged 10-11

years (Key Stage 2 NCTs). Key Stage 1 NCTs are administered informally and marked by

teachers whereas Key Stage 2 NCTs are administered formally and marked by an external

agency.

Students in Year 5, therefore, have fewer experiences of formal testing than in other

countries (notably the United States) and have less homework (or self-study) than students in

secondary or higher education. To capture the typical affective learning experiences of

English students at this age it is preferable to focus on classroom experiences. We focused on

enjoyment, boredom, and anxiety partly as they are three of the most commonly experienced

achievement emotions (see Pekrun et al., 2002a). Furthermore, the only instrument presently

available for measuring achievement emotions in elementary/ primary school children

(Lichtenfeld et al., 2012) contains scales for these three emotions. When combined with the

need to limit the number of items required by participating schools, and use high-quality age-

appropriate instruments, we took the pragmatic decision to focus on enjoyment, boredom,

and anxiety.

The aim of the study was twofold. First, we examined reciprocal relations between

achievement emotions and achievement, with multiple emotions entered simultaneously in

the same analytical model. More specifically, we aimed to test a model of reciprocal relations

between three classroom achievement emotions (enjoyment, boredom, and anxiety) and test

performance in a sample of Year 5 primary school students (aged 9-10 years) in the domain

of mathematics, over four waves. We alternated the assessment of emotions and test

performance over waves as sequential models are suited to test reciprocal relations (Little et

al., 2007; Pekrun et al., 2014; Rosel & Plewis, 2008). Thus the first novel contribution of this

study was to examine enjoyment, boredom, and anxiety, simultaneously using a four-wave

design in a sample of primary school students. Despite the (often) greater ethical and
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 15

logistical challenges involved, in order to ensure a robust and generalizable evidence base for

CVT is it critical that empirical work uses samples of pre-secondary students as well as those

in secondary and higher education. The present study contributes to the paucity of empirical

studies to use samples of younger students in pre-secondary education.Second, we examined

whether academic buoyancy moderated relations between achievement emotions and test

performance. The second novel contribution of our study was, therefore, to test if academic

buoyancy could protect achievement from emotions like boredom and anxiety. Like with

achievement emotions, to ensure a robust evidence base for the buoyancy construct, it is

essential for empirical work to use younger students in primary (or elementary) schooling as

well as older students in secondary or university education. Only one study to date (Colmar et

al., 2019) has examined academic buoyancy in primary school students. The present study,

therefore, makes a noteworthy contribution to the understanding of academic buoyancy by

also using a sample of students in primary education.

Succinctly stated, we tested the following hypotheses (see Figure 1):

Hypothesis 1. Enjoyment and academic buoyancy are positively related, and boredom

and anxiety are negatively related, to subsequent test performance.

Hypothesis 2. Test performance is positively related to subsequent enjoyment and

negatively related to subsequent boredom and anxiety.

Hypothesis 3. Academic buoyancy attenuates the negative relations between boredom

and subsequent test performance, and between anxiety and subsequent test performance

Method

Participants and Procedure

Data were collected over four waves; self-report data at T1 and T3 and mathematics

test performance at T2 and T4 (see Figure 1). All data were collected in the participants’

classrooms at school by the regular classroom teacher following a standardized script. The
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 16

survey items and mathematics tests were hosted online and prompted students where they had

missed an answer. This was to minimize missing data arising from participants inadvertently

missing an item. T1 and T2 data were collected in December 2018, and T3 and T4 data in June

2019. Mathematics tests were scheduled for approximately one week after the surveys. The

project was approved by an institutional research ethics committee (19/EHC/01) at the first

author’s university. Written consent was provided by the head teacher of each participating

school and the parent or carer of each participating student. Individual verbal assent for each

participant was sought at each wave of data collection. The script required teachers to check

the voluntary participation of each student, verbally, and provide an alternative activity if the

students declined to participate. Students were informed that all survey and test responses

would remain anonymous and not be seen by teachers or parents. These points were also

explained on the online survey and mathematics tests. Although timed, the classroom setting

and anonymity of results would characterize these tests as being low-stakes.

At Time 1, data were collected from 1,242 students (633 male, 609 female; mean age

= 9.3 years, SD = .49) from 24 English primary schools (45 different classrooms). The ethnic

backgrounds of participants were Asian = 246 (19.8%), black = 58 (4.7%), white = 876

(70.5%), Chinese = 11 (0.9%), other = 22 (1.8%), and mixed heritage = 29 (2.3). Data could

not be collected for the socio-economic backgrounds of individual students due to private

data protection reasons. However, the schools were located in two of twelve nationally

designated ‘opportunity’ areas characterized by relatively high levels of deprivation

(Department for Education, 2017).

There was attrition at subsequent waves of data collection (T2 n = 979, T3 n = 863,

and T4 n = 734) resulting from participants either being absent from school at the time of data

collection or from choosing not to participate. To assess potential bias in missing data we

conducted an omnibus test for missing completely at random (MCAR; Little’s test, Little,
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 17

1988). This was followed by a series of t-tests comparing mean values of T1 and T3 age and

emotions, and T2 and T4 mathematics test performance, as well as logistic regressions for T1

and T3 frequencies of gender, for participants with complete versus incomplete data. Little’s

test was statistically significant (p <.001) indicating that MCAR could not be assumed.

Participants who scored lower on the T2 mathematics test were less likely to participate at T3,

t(977) = 4.87, p < .001, and T4, t(734) = 6.58, p < .001. All other differences were not

statistically significant (ps >.05). Since missing data could be accounted for by T2

mathematics test performance, they were treated as missing at random (MAR) and handled

using full-information-maximum-likelihood (FIML) estimation. FIML has been found to

result in trustworthy, unbiased estimates for MAR when the variable causing missingness is

included in the model (Nicholson, Deboeck, & Howard, 2017), even in the case of a high

amount of missing values (Enders, 2010), and to be an adequate method to manage missing

data in longitudinal studies (Jeličič, Phelps, & Lerner, 2009).

Measures

Emotions

Achievement emotions were measured at T1 and T3, using the 12 items from the

Achievement Emotions Questionnaire-Elementary School classroom emotions scales (AEQ-

ES: Lichtenfeld et al., 2012). These scales measure three achievement emotions (enjoyment,

boredom, and anxiety) experienced in classroom settings with four items each (e.g., ‘I look

forward to maths lessons’ for enjoyment; ‘I find maths lessons so boring I would rather do

something else’) for boredom; ‘When I think about maths lessons, I get nervous’ for anxiety).

All items were mathematics-specific and adapted to use parlance typical for the English

context (e.g., ‘class’ changed to ‘lesson’). Participants responded to items on a 5-point scale

(1 = not at all, 5 = very much). Internal consistency coefficients for the present study were

excellent (Table 1).

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 18

Buoyancy

Academic buoyancy was measured at T3 using the Academic Buoyancy Scale (ABS;

Martin & Marsh, 2008). The ABS comprises of four items that were adapted to be

mathematics-specific and the example included in the one item simplified to make

appropriate to the age of the target sample (‘I'm good at dealing with setbacks in maths at

school, e.g., getting a question wrong’). Participants responded to these items on a five-point

scale (1 = not at all, 5 = very much). The internal consistency of the scale in the present study

was excellent (Table 1).

Performance

Mathematics test performance was measured at T2 and T4 using items pooled from the

2016, 2017, and 2018 Key Stage 2 National Curriculum Test (NCT) reasoning papers

(Standards and Testing Agency, 2016a, 2016b, 2017a, 2017b, 2018a, 2018b). NCTs are tests

taken by English schoolchildren at the end of primary schooling (Year 6) covering the

curriculum taught from Years 3 to 6 (Key Stage 2; age 7 to 11 years). There are three

mathematics NCTs: one 30-minute arithmetic paper and two 40-minte reasoning papers

scheduled over two days using a paper and pencil format. Each paper consisted of a series of

closed-response questions worth between one and three marks each that used constructive

and substantive styles of reasoning (Bohn-Gettler, 2009; Forgas, 2008). Questions required

convergent analytical thinking (reasoning with one correct solution) rather than divergent and

more creative thinking. The maximum score was 40 for the arithmetic paper and 35 for each

of the reasoning papers (the exact number of questions differs in each paper depending on the

number of marks allocated to each question)1.

1
Papers and mark schemes can be found at https://www.gov.uk/government/collections/national-curriculum-
assessments-practice-materials.
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 19

All schools in England follow a prescribed national curriculum which at Key Stage 2

covers arithmetic, measurement, geometry, fractions and statistics, ratio and proportion, and

simple algebra (Department for Education, 2013). As participants in the present study were in

Year 5, we did not include arithmetic questions (typically taught in Years 3 and 4) as these

are easier and may have resulted in a ceiling effect. We asked two primary school

mathematics teachers unrelated to this study to select all those items from the 2016, 2017,

and 2018 reasoning papers that would be appropriate for a Year 5 student (questions on

measurement, geometry, fractions and statistics, were included). The resulting pool of items

was subsequently confirmed as being appropriate for Year 5 students by two mathematics

learning specialists unconnected to the present study.

Items were randomly selected from this pool to create two tests. Each test was timed

for 40 minutes (to correspond to that of a NCT reasoning paper) and contained one- and two-

point questions worth twenty marks in total (the first test comprised of 17 questions and the

second test 18 questions). Responses required students to either provide a numerical value, or

choose one or more answers from a list or menu of options. Unlike NCTs, marks were

awarded for a correct answer only and no marks were awarded for correct reasoning when an

incorrect answer was given. An exemplar item is: “A box contains 2.6kg of washing powder.

Jack used 65grams of powder per wash. He uses all of the powder. How many washes did

Jack do?” Participants were not provided with feedback on their test score. The internal

consistency for the two tests was excellent (Table 1).

Demographic variables

We controlled for self-reported gender (0 = male, 1 = female) and age in the analysis.

Data Analysis

A latent variable modeling approach was adapted using Mplus v.8 (Muthén &

Muthén, 2017). Confirmatory factor analysis (CFA) was employed to check the properties of
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 20

measurement models, check for measurement invariance, and estimate latent bivariate

correlations. Structural equation modeling (SEM) was used to estimate reciprocal relations

between emotions and achievement, and a latent interaction structural equation model (LI-

SEM) was used to estimate the moderating effects of academic buoyancy on the relations

between emotions and achievement. The ‘type = complex’ command was used to adjust

standard errors for the clustering of data. Using single-level modeling and adjusting standard

errors is recommended to account for nestedness when relations between variables at higher

levels are not the target of investigation (Wu & Kwok, 2012).

All latent models were estimated using maximum-likelihood with robust standard

errors (MLR) and evaluated using the following model fit indices: Root mean error of

approximation (RMSEA), standardized root mean residual (SRMR), confirmatory fit index

(CFI), and Tucker-Lewis index (TLI). A good model fit is indicated by RMSEA <.05, SRMR

<.06, and CFI and TLI >.95 (Hu & Bentler, 1999). Caution must be used, however, when

applying guidance derived from simulations studies to more complex studies, such as ours,

using naturalistic data (Heene et al., 2011; Lance et al., 2006).

Results

Descriptive Statistics

Participants reported relatively high levels of enjoyment and academic buoyancy and

low levels of boredom and anxiety (see Table 1). T1 enjoyment showed a slightly negative

and T1 boredom a slightly positive skew (hence the decision to use the MLR estimator). The

internal consistency of self-report measures (Cronbach’s αs and McDonalds’ ωs ≥ .79) and

mathematics tests (αs ≥ .79 and ωs ≥.81) was good, and items loaded substantively on their

target factors (λs ≥ .64) in CFA (see Table 1). Intraclass correlation coefficients (ICC1 or ρI)

showed that the proportion of variance attributable to the school level was relatively small for
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 21

the classroom achievement emotions and academic buoyancy (approximately 3 – 7%) and

somewhat larger for the mathematics test performance (approximately 12 – 14%).

Latent Bivariate Correlations

To estimate latent bivariate correlations, a measurement model was built that included

achievement emotions (4 items each for enjoyment, boredom, and anxiety at T1 and T3),

academic buoyancy (4 items at T3), and mathematics test performance at T2 and T4. The

corresponding indicators for classroom achievement emotions at T1 and T3 were allowed to

correlate. Mathematics test performance was treated as a single-item latent variable with σε =

.1 in line with estimates derived from previous empirical studies (Hoy et al., 2006; Watkins et

al., 2007). Gender (0 = male, 1 = female) and age were added as covariates and treated as

manifest variables.

The CFA showed a good fit to the data, χ2 (298) = 589.30, p <.001, RMSEA = .026,

SRMR = .028, CFI = .982, and TLI = .977, and so we proceeded to inspect correlation

coefficients (see Table 2). Enjoyment and buoyancy were positively, and boredom and

anxiety negatively, correlated with mathematics test performance. Female students reported

lower enjoyment, higher anxiety and boredom, lower buoyancy, and lower mathematics test

performance. Older students reported lower enjoyment and higher boredom, and they showed

higher mathematics test performance.

The intercorrelations between T1 emotions (rs = -.68 to .64) and T3 emotions and

academic buoyancy (rs = -.78 to .62) indicate the possibility of multicollinearity in

subsequent analyses that model constructs simultaneously. T4 mathematics test performance

was regressed onto T1 and T3 classroom emotions, and T2 mathematics test performance, and

T2 mathematics test performance regression into on T1 and T3 classroom emotions, in SPSS.

Tolerance statistics were > .33 and variance inflation factors < 3.00, suggesting that

multicollinearity would not unduly bias parameter estimates.

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 22

Measurement Invariance

A prerequisite for the modelling of longitudinal relations is temporal measurement

invariance (Widaman et al., 2010). Accordingly, we tested a series of models for classroom

achievement emotions that included successively strict constraints (Meredith, 1993). The

configural model specified the measurement model at T1 and T3 and included correlations

between the corresponding indicators at the two time points. The metric invariance model

constrained factor loadings of items to be equal, the scalar invariance model additionally

constrained intercepts of items to be equal, and the strict invariance model additionally

constrained item residual variances of items to be equal. Tests of measurement invariance are

reported in Table 3. A decline in model fit of ΔRMSEA > .015 or ΔCFI/ TLI > .01, from one

model to the next, indicates non-invariance (Chen, 2007; Cheung & Rensvold, 2002). All

classroom achievement emotions showed strict measurement invariance, and so we

proceeded to examine longitudinal relations.

Structural Equation Modelling of Reciprocal Effects

A SEM was used to test the fully forward reciprocal relations model (RRM) shown in

Figure 1. This model was tested competitively (see Table 4) against three alternative models:

A baseline model where all directional paths were constrained to zero, a unidirectional model

where paths from classroom achievement emotions to test performance were freely estimated

but those from test performance to achievement emotions were constrained to zero, and a

unidirectional model where paths from test performance to achievement emotions were freely

estimated but those from achievement emotions to test performance were constrained to zero.

All models included age and gender as covariates. The RRM showed a good fit to the data

that was superior to all other models (Table 4). Furthermore, the RRM showed a significantly

better fit than the other models using the Satorra–Bentler scaled χ2 difference test (TRd;

Bryant & Satorra, 2012) and an improved relative fit on the Akaike Information Criterion
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 23

(AIC). Lower AIC values are indicative of a better model fit and ΔAIC > 10 indicates a

substantive change in model fit (Hix-Small et al., 2004). The RRM (Figure 2) was accepted

as the better fitting model. Standardized path coefficients and standard errors are reported in

Table 5 and were interpreted as βs ≤ .09 indicating small effects, .10 to .24 moderate effects,

and >.25 large effects (Keith, 2006).

Relations from T1 Emotions to T2 Test Performance

T1 anxiety negatively predicted T2 mathematics test performance. T1 enjoyment and

T1 boredom were not significantly related to T2 performance.

Relations from T2 Test Performance to T3 Emotions

T2 mathematics test performance positively predicted T3 enjoyment over and above

the auto-lagged relation with prior T1 enjoyment and the cross-lagged relations with prior T1

boredom and T1 anxiety. T2 mathematics test performance negatively predicted T3 boredom

over and above the auto-lagged relation with prior T1 boredom and the cross-lagged relations

with prior T1 enjoyment and T1 anxiety. T2 mathematics test performance also negatively

predicted T3 anxiety over and above the auto-lagged relation with prior T1 anxiety and the

cross-lagged relations with prior T1 enjoyment, and T1 boredom.

Relations from T3 Emotions to T4 Test Performance

T3 anxiety negatively predicted T4 mathematics test performance over and above the

auto-lagged relation with prior T2 performance. T3 enjoyment, T3 boredom, T1 enjoyment, T1

boredom, and T1 anxiety were not significantly related to T4 test performance.

Relations with Covariates

Gender showed a significantly negative relation with T1 enjoyment and significantly

positive relations with T1 boredom and anxiety. Age was significantly negatively related to T1

enjoyment and positive related to T1 boredom.

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 24

Moderating Effects of Academic Buoyancy in the Relation between Emotions and Test

Performance

Interactions for Academic Buoyancy × Enjoyment, Academic Buoyancy × Boredom,

and Academic Buoyancy × Anxiety, were estimated in a single model using the latent

moderated structural equation modeling (LMS) approach (Klein & Moosbrugger, 2000)

implemented in Mplus v.8 (Muthén & Muthén, 2017). Absolute model fit indices are not

provided in the LMS approach. However, it is possible to establish whether a model

including an interaction term offers a better fit to the data using relative fit indices

(Maslowsky et al., 2015): Akaike Information Criterion (AIC), sample-size adjusted

Bayesian information criterion (aBIC), the log likelihood ratio test, and the change in the

proportion of variance (ΔR2) explained in the outcome variable (i.e., T4 mathematics test

performance).

A better model fit is indicated by smaller AIC and aBIC values (Hix-Small et al.

2004) and a larger R2 in explaining the variance of the outcome. A statistically significant log

likelihood ratio test (D) would indicate a worse fit for the more parsimonious model (i.e., the

model without the interaction term). Due to the computational power required for the LMS

approach 5,000 Monte Carlo Integration points were used.

Age and gender were included as covariates. As we were predicting T4 test

performance, autoregressive relations with T2 test performance were also controlled for.

Because T3 emotions were predictor variables in these analyses, there would have been no

benefit to including T1 emotions in the models. Hence, to keep the models as parsimonious as

possible, T1 emotions were not included. In the log likelihood ratio test the three additional

parameters, one per interaction term, equate to three degrees of freedom.

A model without an interaction term showed a good fit to the data: χ2 (146) = 422.44,

p <.001, RMSEA = .036, SRMR = .031, CFI = .971, and TLI = .962. However, the model
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 25

including three interaction terms (Academic Buoyancy × Enjoyment, Academic Buoyancy ×

Boredom, and Academic Buoyancy × Anxiety) showed an improved fit (ΔAIC = -4.31,

ΔaBIC = -3.97) and explained a greater proportion of variance in T4 test performance (ΔR2 =

.01). Furthermore, a statistically significant likelihood ratio test, D(3) = 22.26, p < .001,

indicated a worse fit for the model without the three interaction terms. Structural coefficients

are shown in Table 6. A statistically significant interaction was shown for Academic

Buoyancy × Anxiety, but not Academic Buoyancy × Enjoyment or Academic Buoyancy ×

Boredom.

To probe the Academic Buoyancy x Anxiety interaction simple slopes were estimated

at ±1SD buoyancy (Figure 3a; please note that the Mplus software estimates simple slopes as

unstandardized regression coefficients only). At +1SD buoyancy a negative relation was

shown between anxiety and performance (B = -3.91, SE = 1.27). This relation was weaker at

mean buoyancy (B = -1.38, SE = .34) and at -1SD buoyancy became non-significant (B =

1.16, SE = 1.25). To facilitate interpretation of the interaction, we also plotted relations

between academic buoyancy and test performance at ±1SD anxiety (see Figure 3b). At +1SD

anxiety, a negative relation was shown between academic buoyancy and test performance (B

= -2.80, SE = 1.38). This relation was not significant at mean anxiety (B =- .01, SE = .21) and

became positive at -1SD anxiety (B = 2.20, SE = 1.33). The findings suggest that

performance benefited from lower anxiety combined with higher buoyancy. At higher levels

of anxiety the performance-enhancing influence of buoyancy declined.

Discussion

Using a longitudinal design with four alternating waves of data collection, we

investigated the relations between three classroom achievement emotions (enjoyment,

boredom, and anxiety) and test performance in the context of mathematics at primary school.

Supporting Hypothesis 1, anxiety negatively predicted subsequent mathematics test scores

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 26

after controlling for gender, age, concurrent classroom achievement emotions, and

autoregressive relations with prior achievement. Enjoyment and boredom, however, were not

significantly related to subsequent test scores. In line with Hypothesis 2, test scores positively

predicted subsequent enjoyment and negatively predicted boredom and anxiety after

controlling for prior emotions and the covariates. Thus, a positive feedback loop over time

was shown between anxiety and test scores; higher anxiety was related to subsequent lower

test scores, and lower test scores, in turn, to higher subsequent anxiety. Overall, this pattern

of findings suggests that when multiple achievement emotions are considered simultaneously,

anxiety exerts a stronger effect on performance than enjoyment and boredom. It is notable

that despite low mathematics test scores at T2, there was sufficient variance with which to

model relations with previous and subsequent classroom emotions.

We also investigated whether academic buoyancy moderated relations between

negative emotions that are detrimental to learning (boredom and anxiety) and protected test

scores from these emotions. The model including interaction terms showed a superior fit

relative to a model not including interactions and a significant effect was shown for the

Academic Buoyancy × Emotion interaction. In partial support of Hypothesis 3, test

performance was highest when anxiety was low and buoyancy high. With higher anxiety, the

benefit for performance shown for high academic buoyancy diminished.

Relations between Achievement Emotions and Academic Achievement

CVT proposes reciprocal relations between achievement emotions and indicators of

achievement, such as test scores, as measured in the present study. Emotions exert cognitive-

motivational effects that influence learning and performance, and learning and performance,

in turn, reinforce the control-value appraisals that underpin emotions (Pekrun, 2006, 2017;

Pekrun & Perry, 2014). Previous studies have shown how achievement emotions relate to

subsequent mathematics achievement in primary/ elementary, secondary, and undergraduate

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 27

students (e.g., Ahmed et al., 2013; Raccanello et al., 2019; Villavicencio, & Bernardo, 2016)

but far fewer have examined how emotions and mathematics achievement are reciprocally

related over time (Pekrun et al., 2014, for an undergraduate psychology course, Pekrun et al.,

2014, for secondary school students, and Putwain, Becker, et al., 2018, for primary school

students).

A question of conceptual and applied importance is whether achievement emotions

exert unique effects when considered simultaneously. Notably, only one study thus far

examined reciprocal relations with mathematics achievement for more than a single emotion

in tandem; Putwain, Becker, et al. (2018) examined enjoyment and boredom simultaneously

and found that unique reciprocal effects for both of these emotions. The findings of the

present study offer novel insights into the question of unique reciprocal effects by

considering three classroom emotions (enjoyment, boredom, and anxiety) simultaneously in a

single analytic model. When the variance shared between emotions measured concurrently

was controlled for in this way, reciprocal relations with mathematics achievement were only

shown for anxiety. Although mathematics test performance predicted subsequent enjoyment

and boredom (in the expected direction), enjoyment and boredom were not significantly

related to subsequent test performance.

Thus, when these three emotions are pitted competitively against each other, anxiety

exerts the strongest predictive power for achievement. It is important to clarify that we do not

argue against the presence of reciprocal relations between enjoyment and achievement, and

between boredom and achievement. Rather, the findings suggest that anxiety emerges as the

most significant (statistically and practically) of the three emotions when they are modeled

together. Analytically speaking, this is a combined result of anxiety being related more

strongly to test performance than enjoyment or boredom and of the intercorrelations between
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 28

enjoyment, boredom, and anxiety. The result is that insufficient unique variance remained in

enjoyment and boredom to predict test performance.

Theoretically speaking, the implication is that the cognitive-motivational mechanisms

affected by anxiety, specifically interference with working memory capacity and function

(Derakshan & Eysenck, 2011; Eysenck et al. 2007), may exert a stronger influence on

performance than the cognitive-motivational mechanisms affected by enjoyment and

boredom, such as interest, intrinsic motivation, and depth of learning (Fredrickson, 2001;

Fredrickson & Branigan, 2005). Practically speaking, when multiple emotions are considered

simultaneously in achievement settings, or when the presence of unpleasant emotions is

associated with the absence of pleasant emotions, it may be more important to focus on

reducing anxiety in the first instance than to attempt to reduce boredom and foster enjoyment.

In summary, we found support for our second hypothesis, that test performance would

be positively related to subsequent enjoyment and negatively related to subsequent boredom

and anxiety. There was, however, only mixed support for our first hypothesis, that enjoyment

would be positively, and boredom and anxiety negatively, related to subsequent test

performance. Our findings build on the body of evidence to show that anxiety is detrimental

for mathematics achievement and that it is not just an outcome or epiphenomenon of prior

performance, but can predict lower achievement beyond the influence of prior performance.

A minor, but nonetheless intriguing, point is that T1 enjoyment was a stronger

predictor of T3 boredom than T1 boredom. We suspect that this finding is due to students’

enjoyment being more stable over time as compared with boredom (see the autoregressive

effects for the two emotions from T1 to T3, .59 and .24, respectively; Table 5). The high

stability of enjoyment may have made it possible for enjoyment to more continuously

influence students’ subsequent boredom. The exact reason why enjoyment was more than

stable than boredom is unclear, but could be related to the differential role of perceived
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 29

control as an antecedent of the two emotions. Perceived control will result in greater

enjoyment and will also lead to lower boredom in most cases, but can alternatively induce

higher boredom due to a lack of challenge (e.g., Pekrun, 2006, 2018; Pekrun & Perry, 2014),

thus explaining lower stability of boredom.”

It could be questioned whether findings from the present study would generalize to

non-testing situations typically encountered by students during classroom learning. There is

evidence that domain-specific emotions related to learning (e.g., in math, science, and L2

learning) can influence classroom achievement (for reviews see Horwitz, 2001; Maloney,

2016; Sinatra et al., 2016). For emotions related to classroom activities specifically, class-

related anxiety has been shown to relate negatively to both class grades and test scores in

mathematics (rs = -.21 to -.32; Lichtenfeld et al., 2012; Peixoto et al., 2016; Putwain et al.,

2020; Racanello et al., 2019; for L2 class-related anxiety, see Shao et al., 2020). In the study

by Lichtenfeld et al. (2012), the magnitude of correlations between classroom-related anxiety

and two different measures of math achievement (class grades vs. test scores) did not differ (z

= -.40 and .43 for 2nd and 3rd grade samples, respectively; ps >.05). Overall, these findings

would suggest the results of the present study for relations between class-related emotions,

notably anxiety, and test scores would be applicable to classroom achievement, such as

grades based on classroom activities, as well.

The Moderating Effect of Academic Buoyancy

Academic buoyancy is the capacity to respond effectively to minor academic

adversities (Martin & Marsh, 2009). Studies have shown that higher domain general

academic buoyancy is related to adaptive educational outcomes in secondary school students

(e.g., Malmberg et al., 2013; Martin et al., 2013) including higher positive, and lower

negative, achievement emotions in primary and secondary school students (e.g., Hirvonen et

al., 2019; Martin et al., 2010; Putwain et al., 2012). Studies linking domain general academic
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 30

buoyancy to aggregated achievement in mathematics and reading have shown equivocal

results, however, with some reporting statistically significant relations but others not (e.g.,

Collie et al., 2015; Martin, 2014). We reasoned this may be partly an artifact of reduced

predictive power arising from lack of domain specificity and used a mathematics-specific

measure of academic buoyancy in the present study (also ensuring matching specificity with

domain-specific achievement emotions).

Conceptually, higher levels of buoyancy could moderate (i.e., buffer) the impact of

academic adversities on subsequent outcomes (e.g., Martin & Marsh, 2019; Putwain et al.,

2016; Symes et al, 2015). Accordingly, we theorized that higher levels of buoyancy could

protect test performance from boredom and anxiety. The model including the interaction

terms between academic buoyancy and emotions showed a superior fit to the model not

including these interactions and a statistically significant interaction between Academic

Buoyancy and Anxiety. Interactions between Academic Buoyancy and Enjoyment, and

between Academic Buoyancy and Boredom, were not statistically significant.

The pattern of the observed interaction partially supported our third hypothesis.

Rather than protecting test performance at higher levels of anxiety, higher academic

buoyancy amplified test performance at lower levels of anxiety. With higher anxiety, the

differential benefit to performance offered by higher academic buoyancy diminished. This

finding suggests that academic buoyancy did not protect test performance against the adverse

effects of high classroom anxiety. Furthermore, there is the intriguing finding that high

buoyancy may even be detrimental for performance when anxiety is also high. We offer three

considerations to explain this pattern of effects.

First, buoyancy may not have protected performance because anxiety was too

adverse. Although academic anxiety is often considered a minor form of adversity in

comparison to clinical anxiety (Martin, 2013a), it is known that high levels of academic
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 31

anxiety overlap with clinical anxiety (e.g., Herzer et al., 2014; Warren et al., 1996; see also

Pekrun & Loderer, in press). It may be that the higher levels of anxiety measured in the

present study constituted more of a major than a minor adversity to students, in which case

buoyancy may not have been a sufficiently strong factor to offer protection. Classroom

anxieties may have been furthered by the subject material; mathematics learning and

reasoning is, for many students, a source of anxiety (e.g., Maloney, 2016). Martin (2013a)

and Martin and Marsh (2008, 2009) argue that resilience, rather than buoyancy, is required to

respond to chronic or stronger academic adversities.

Second, an alternative interpretation would be that buoyancy did not protect against

performance because the classroom measure of anxiety underestimated the level of anxiety

experienced during the test. Although the mathematics test used in the present study was

characterized as low-stakes, it is possible that a greater degree of anxiety was experienced

during the test than was typically experienced during classroom learning. In this respect the

classroom measure may have reported anxiety as not being sufficiently adverse for academic

buoyancy to effectively offer protection. To investigate this possibility, future research could

include a measure of test-related emotions occurring during test taking. Furthermore,

measures of engagement and learning during mathematics lessons (e.g., on-task behaviors,

cognitive strategies used, and tasks completed) may have been more sensitive to the

protection offered by buoyancy against classroom learning anxiety, than test performance.

Third, it is known that children can over-estimate their abilities (see Muenks et al.,

2018; Salles et al., 2016). It is possible that participants may have over-estimated their

academic buoyancy and believed they had a greater capacity to bounce back than was

actually the case. This may account for the diminishing protection offered by academic

buoyancy as anxiety increased; there was less buoyant than was reported. This over-

estimation could also account for why very high (+2SD) academic buoyancy became
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 32

detrimental to performance. If participants anticipated they would be highly able to

effectively respond to difficulties encountered during the test but experienced the opposite,

they could be highly de-motivated, reduce effort, or even give up completely. This is

somewhat akin to ‘choking’ when the pressure from performing mathematics tasks becomes

exaggerated by concerns about one’s performance overloading working memory (e.g.,

Beilock et al., 2004).

Fourth, a curious finding reported by Martin and Marsh (2019) was that academic

buoyancy exerted a greater protective role for academic adversities reported 12 months later

than with current academic adversities. This finding is in line with the view that some

exposure to adversity is necessary in order for persons to build adaptive responses. It is

therefore possible that the benefits of buoyancy play out, or accumulate, over time rather than

contemporaneously. If this were the case, then we would have been unlikely to observe a

protective effect on a test taken one week after measurements of self-reported anxiety and

academic buoyancy. In this respect academic buoyancy can be considered as a malleable

trait, a lasting attribute that can develop over time and that is responsive to intervention

(Martin, 2013b). Beneficial effects would occur downstream at a later point in time. We

conclude the theoretical proposition that academic buoyancy can protect subsequent

outcomes from minor adversity may be valid despite having been confirmed only partially in

the present study.

Although we focused on mathematics in the present study, as we wanted to adopt a

domain-specific approach, other studies on buoyancy have considered English and reading in

research with primary and secondary students (e.g., Colmar et al., 2015; Putwain & Aveyard,

2018). There are no theoretical reasons for academic buoyancy to differentially relate to

achievement and adaptive beliefs, emotions, and behaviors, in varying academic subjects.

While mathematics may be anxiety-provoking for some students, it is known that literacy
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 33

also presents a challenge for students, and reading motivation begins to decline at the end of

elementary school (see Wigfield, 1997). These are adversities that buoyancy is theorized to

protect against. However, we are mindful that some subjects do present unique challenges for

students and it would be beneficial for academic buoyancy studies to broaden the repertoire

of academic subjects considered.

Despite a positive bivariate correlation between academic buoyancy and subsequent

test performance, there was no first-order effect of academic buoyancy on performance after

controlling for current relations with emotion, the interaction with emotion, and prior

achievement. The prediction made in Hypothesis 1 that buoyancy would be positively related

to achievement was not supported in these models. It is possible that emotions mediate

relations between academic buoyancy and subsequent achievement (see Putwain et al., 2015).

Although we could not formally test this hypothesis because academic buoyancy and

achievement emotions were assessed simultaneously at T3, this interpretation is supported by

the strong relations between buoyancy and emotions (Table 1), combined with the relations

between emotions and subsequent test performance. Hence, the relations with emotions may

have reduced the direct predictive power of academic buoyancy. In this case, academic

buoyancy would represent a relatively rare example of the same variable operating as both a

mediator and a moderator (for a related example for self-efficacy as a mediator and

moderator, see Dicke et al., 2014).

When considering findings, it is important to acknowledge that the mathematics tests

used in the present study contained only closed response questions that required reproductive

styles of reasoning. Some students may have employed the correct approach to solving a

question but ultimately arrived at an incorrect answer. Including open questions, where

students could receive marks for showing their reasoning, as well as their final answer could

potentially show a different pattern of relations with classroom emotions and academic
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 34

buoyancy. Similarly, questions requiring creative and global processing that requires making

new connections between concepts, which are potentially more challenging, could also relate

differently to classroom emotions and academic buoyancy than questions requiring

convergent analytical reasoning.

We would also like to briefly comment on the missing data in our study and how it

was treated. Analyses suggested that the cause of the missing data were T2 mathematics test

scores (and hence missing data were treated as MAR). The implication of MAR is that

missingness was not completely random, but could be treated as random after T2

mathematics scores were controlled for (see Little & Rubin, 2002). This situation is typical of

applied research where substantive study variables or socio-demographic correlates may

influence decisions whether to continue participation or not (e.g., Lamers et al., 2012).

Attrition may be higher for groups of participants characterized by vulnerability, low

motivation, or where the study may pose emotional distress. In the present study, it would

seem plausible that students who experienced the T2 mathematics test as more difficult (and

hence performed worse) may have been less motivated to continue participating as this may

have reminded them of their difficulties with mathematics.

When the cause of the missing data is ignored, there is a danger that model estimates

may be biased. In the present study, the risk would be an under-representation of participants

with lower mathematics scores. However, simulation studies have shown that when the cause

of missing data is included in the algorithm used for handling missing data (FIML was used

in the present study), disproportional participant attrition can be corrected for to yield

unbiased estimates (e.g., Collins et al., 2001; Nicholson et al., 2017). We followed a strategy

to identify the cause of the missing data (T2 mathematics scores) and included that variable

in our analytic models. Hence, we are confident that parameter estimates are applicable to the

entire sample despite reduced representation of participants with lower T2 test scores. Future
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 35

studies may follow a similar approach to test and report assumptions of MCAR and MAR

more completely and openly.

Limitations and Directions for Future Research

Despite the novel theoretical contributions of this study, and the use of a relatively

understudied age group in achievement emotion and academic buoyancy research, there are

limitations that should be considered and can be used to suggest directions for subsequent

studies. First, we were able to measure data across two alternate assessments of emotions and

performance. Thus, we were able to test the predictive power of emotions on subsequent

performance twice, but reciprocal predictive effects of test performance on subsequent

emotions only once. Although this design is sufficient to test reciprocal relations (Little et al.,

2007; Rosel & Plewis, 2008), additional alternating waves of emotions and achievement

would allow for multiple assessments of reciprocal relations between emotions and

achievement. The present design only permitted a single test for the moderating effect of

academic buoyancy while controlling for previous achievement. Additional alternating waves

of academic buoyancy (alongside emotions) and achievement would also allow for multiple

tests of protective effects of buoyancy on relations between adverse emotions and

achievement.

A related point is that there were unequal time intervals between assessments. T1 and

T2, and T3 and T4, were spaced apart by one week. T2 and T3, however, were spaced apart by

approximately seven months. We were constrained by school administration to schedule data

collection this way in order to minimize impact on routine teaching and learning. Thus, we

cannot make direct comparisons of the size of paths from emotions to test performance (with

the one week time interval) to the size of the paths from test performance to emotions (with

the seven month time interval).

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 36

Second, we measured three classroom emotions (enjoyment, boredom, and anxiety).

Although being the first study to model reciprocal relations in all three simultaneously, there

are other achievement emotions likely to co-occur in classroom settings (e.g., hope and

hopelessness). It would be useful to include additional emotions to address the question of

which emotions are the most meaningful predictors of achievement, when considered

together. We were somewhat limited in that the only validated measure of achievement

emotions available for the age group of participants used in the present study (AEQ-ES;

Lichtenfeld et al., 2012) measures enjoyment, boredom, and anxiety specifically. It would be

a useful extension of this measure to include additional emotions.

Third, from a CVT perspective, emotions influence subsequent achievement through

cognitive-motivational mechanisms, and achievement influences subsequent emotions

through control-value appraisals. Although it is useful to first test for reciprocal relations

solely between emotions and achievement, partly as these relations are meaningful in their

own right, and partly as a precursor to investigating mediating processes, future studies

should additionally include cognitive-motivational mechanisms and control-value appraisals.

A study that combined tests of reciprocal effects in conjunction with the presumed mediating

mechanisms would offer an even more robust test of CVT.

Fourth, we speculated two reasons for the inconsistent relations shown been academic

buoyancy and achievement in the extant literature; low domain specificity or mismatch

between measures of academic buoyancy and achievement, and the presence of additional

variables that may either overlap with academic buoyancy (e.g., grit or future time

perspective; Fong & Kim, 2019) or mediate relations between academic buoyancy and

achievement (e.g., perceived control; Collie et al., 2015, and Putwain & Aveyard, 2018). The

correlations between domain-specific academic buoyancy in mathematics and students’

mathematics achievement found in the present study (rs = .23 and .25) were stronger than
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 37

correlations between domain-general academic buoyancy and achievement in previous

studies (e.g., rs = .13 – .17; Martin, 2014; Putwain et al., 2016). This would lend credibility

to the view that high domain specificity between academic buoyancy and achievement

strengthens relations. Furthermore, in the LI-SEM, the first-order effects of academic

buoyancy were negligible; this indicates a possible mediating role of emotions. However,

from the present study we cannot establish whether domain-general variables that overlap

with academic buoyancy, such as grit and future time perspective, would reduce the

magnitude of relations between academic buoyancy and achievement when using domain-

specific measures of academic buoyancy. Future studies should continue to explore how

academic buoyancy (especially when considered as a domain-specific construct) can be

differentiated from cognate constructs and how relations with achievement can be used to

inform understanding of overlap between similar constructs.

Fifth, we did not account for learning disabilities (e.g., dysgraphia or dyscalculia) in

the present study. It is likely that they would have been meaningful covariates and might

explain variance in both predictors and outcomes. In order to collect accurate data for

learning disabilities for children aged 9 to 10 years, it would be necessary to use official

school records rather than to rely on participant self-report; not all students at this age may

understand if they have been diagnosed with a learning disability or if they have, what that

learning disability is. In order to keep data collection anonymous for ethical reasons, we were

unable to match participant self-report data with school records. However, it would be

desirable for future studies, where ethical protocols permit, to include information about

learning disabilities.

Finally, our test of the moderating effect of buoyancy on the relations between

academic adversity and subsequent outcomes was limited to test performance. There are

other salient outcomes that academic buoyancy could protect from adversity, including
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 38

attendance, compliance with school behavioral policy, positive relationships with peers and

staff, adaptive motivation (e.g., intrinsic motivation), and behavioral engagement (e.g.,

participation in lessons and extracurricular activities). Future research should consider such

outcomes, and where possible use official school recorded data (e.g., attendance) to

complement self-report data on these outcomes. It would also be useful to include measures

of emotional self-regulation to establish whether the moderating influence of academic

buoyancy is related to differential use of regulatory strategies.

Insights for Practice

In classroom settings, when emotions such as enjoyment, boredom, and anxiety may

co-occur, anxiety emerges as the strongest (negative) emotional predictor of achievement

according to the present findings. Attempts to reduce anxiety would therefore be beneficial in

helping students’ learning and performance. Typically interventions have focused on test

anxiety (von der Embse et al., 2013), school phobia, and school refusal (Lauchlan, 2003).

There are fewer interventions for classroom or learning anxiety, and these are focused on

specific forms of anxiety such as math anxiety (Schaeffer et al., 2018) and statistics anxiety

(Smith & Capuzzi, 2019). Math anxiety interventions are germane to the present study with

the substantive focus on classroom emotions in mathematics (cognate, although not identical

with math anxiety).

Math anxiety interventions have focused broadly on either building subject mastery,

reducing negative appraisals of competence or physiological arousal, or the normalization of

failure as part of learning (Ramirez et al., 2018). In CVT (Pekrun, 2016, 2017; Pekrun &

Perry, 2014) and the integrated model of emotion regulation in achievement situations

proposed by Harley et al. (2019), mastery, reappraisal, and failure-response beliefs

correspond to ways of building perceived control or regulating emotion through cognitive

change. These theoretically derived mechanisms of reducing anxiety are not solely the
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 39

province of psychologists and specialist interventions. There are practical ways in which

instructors can incorporate mastery practice and adaptive responses to failure in both

mathematics and other subjects through directing student attributions for success and failure

(Perry at al., 2014) and creating a classroom culture whereby failure is defined as a normal

part of the learning process (Murphy & Dweck, 2010).

Conclusion

When the relations between three classroom emotions (enjoyment, boredom, and

anxiety) and test performance were modelled simultaneously over four waves of data

collection, reciprocal relations were shown for anxiety (in terms of negative reciprocal

relations). High test performance predicted higher enjoyment and lower boredom, but

enjoyment and boredom were not significantly related to subsequent test performance. Thus,

when the shared variance between these emotions is considered (as is likely to happen in

classroom situations where discrete emotions may co-occur), anxiety emerged as the emotion

that is more important for achievement. We also investigated whether academic buoyancy

might protect performance against anxiety. We found that this was partially the case.

Buoyancy protected performance at lower levels of anxiety, suggesting that buoyancy can

help students coping at least with mild forms of negative emotion.

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 40

References

Ahmed, W., van der Werf, G., Kuyper, H., & Minnaert, A. (2013). Emotions, self-regulated

learning, and achievement in mathematics: A growth curve analysis. Journal of

Educational Psychology, 105, 150–161. doi: 10.1037/a0030160

Beilock, S. L., Kulp, C. A., Holt, L. E., & Carr, T. H. (2004). More on the fragility of

performance: Choking under pressure in mathematical problem solving. Journal of

Experimental Psychology: General, 133, 584–600.doi: 10.1037/0096-3445.133.4.584

Bohn-Gettler, C.M. (2019). Getting a grip: The PET framework for studying how reader

emotions influence comprehension. Discourse Processes, 56, 386-401.

https://doi.org/10.1080/0163853X.2019.1611174

Brooks, J. E. (2006). Strengthening resilience in children and youths: Maximizing

opportunities through the schools. Children and Schools, 28, 69–76.

doi:10.1093/cs/28.2.69

Bryant, F. B., & Satorra, A. (2012). Principles and practice of scaled difference chi-square

testing. Structural Equation Modeling, 19, 372–398.

doi:10.1080/10705511.2012.687671

Collins, L. M., Schafer, J. L., & Kam, C. M. (2001). A comparison of inclusive and

restrictive strategies in modern missing data procedures. Psychological Methods, 6,

330–351.doi: 10.1037/1082-989X.6.4.330

Colmar, S., Liem, G.A.D., Connor, J. &. Martin, A.M. (2019). Exploring the relationships

between academic buoyancy, academic self-concept, and academic performance: a

study of mathematics and reading among primary school students. Educational

Psychology, 39, 1068-1089. doi: 10.1080/01443410.2019.1617409

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 41

Compas, B. E. (2004). Processes of risk and resilience during adolescence: Linking contexts

and individuals. In R. M. Lerner & L. Steinberg (Eds.), Handbook of adolescent

psychology (pp. 263–296). John Wiley & Sons Inc. doi:10.1002/9780471726746.ch9

Cheng, C. (2001). Assessing coping flexibility in real-life and laboratory settings: a

multimethod approach. Journal of Personality and Social Psychology, 80, 814-833.doi:

10.1037/0022-3514.80.5.814

Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing

measurement invariance. Structural Equation Modeling, 9, 233–255.

doi:10.1207/S15328007SEM0902_5

Collie, R. J., Martin, A. J., Malmberg, L.-E., Hall, J., & Ginns, P. (2015). Academic buoyancy,

student’s achievement, and the linking role of control: A cross-lagged analysis of high

school students. British Journal of Educational Psychology, 85, 113–130.doi:

10.1111/bjep.12066

Department for Education (2013). Mathematics programmes of study: Key Stages 1 and 2

National Curriculum in England. Her Majesty's Stationery Office.

Department for Education (2017). Opportunity areas selection methodology. Her Majesty's

Stationery Office.

Department for Education (2019). National curriculum assessments at Key Stage 2 in England,

2019. Her Majesty's Stationery Office.

Derakshan, N. & Eysenck, M. W. (2011). New perspectives in attentional control theory.

Personality and Individual Differences, 50, 955-960. doi: 10.1016/j.paid.2010.08.019

Dicke, T., Parker, P. D., Marsh, H. W., Kunter, M., Schmeck, A., & Leutner, D. (2014). Self-

efficacy in classroom management, classroom disturbances, and emotional exhaustion:

A moderated mediation analysis of teacher candidates. Journal of Educational

Psychology, 106, 569-583. doi: 10.1037/a0035504

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 42

Di Leo, I., Muis, K. R., Singh, C., and Psaradellis, C. (2019). Curiosity Confusion? Frustration!

The role and sequencing of emotions during mathematics problem solving.

Contemporary Educational Psychology, 58, 121–137. doi:

10.1016/j.cedpsych.2019.03.001

Duckworth, A. L., Peterson, C.,Matthews, M. D., & Kelly, D. R. (2007). Grit: Perseverance

and passion for long-term goals. Journal of Personality and Social Psychology, 92,

1087–1101. doi: 10.1037/0022-3514.92.6.1087

Enders, C. K. (2010). Applied missing data analysis. Guilford.

Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety and cognitive

performance: attentional control theory. Emotion, 7, 336-353.doi: 10.1037/1528-

3542.7.2.336

Fiedler, K., & Beier, S. (2014). Affect and cognitive processes in educational contexts. In R.

Pekrun & L. Linnenbrink-Garcia (Eds.), International handbook of emotions in

education (pp. 36-55). Taylor & Francis.

Forgas, J. P. (2008). Affect and cognition. Perspectives on Psychological Science, 3,94–

101.https://doi.org/10.1111/j.1745-6916.2008.00067.x.

Fong, C.J. & Kim, Y.W. (2019). A clash of constructs? Re-examining grit in light of academic

buoyancy and future time perspective. Current Psychology. Advance online publication.

doi:10.1007/s12144-018-0120-4

Fornell, C., & Larcker, D.F (1981). Evaluating structural equation models with unobservable

variables and measurement error, Journal of Marketing Research, 28, 39-50.doi:

10.2307/3151312

Gable, P., & Harmon-Jones, E. (2010). The motivational dimensional model of affect:

Implications for breadth of attention, memory, and cognitive categorisation. Cognition

and Emotion, 24, 322-337. https://doi.org/10.1080/02699930903378305

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 43

Goetz, T., & Hall, N. (2014). Academic boredom. In R. Pekrun and L. Linnenbrink-Garcia

(Eds.). International handbook of emotions in education (pp. 311-330). Routledge.

Harley, J.M., Pekrun, R., Taxer, J.L., & Gross, J.J. (2019). Emotion regulation in achievement

situations: An integrated model. Educational Psychologist, 54, 106-126. doi:

10.1080/00461520.2019.1587297

Hanushek, E. A., & Woessmann, L., (2008). The role of cognitive skills in economic

development. Journal of Economic Literature, 43, 607-668.doi: 10.1257/jel.46.3.607

Hattie, J., & H. Timperley. (2007). The Power of Feedback. Review of Educational Research

77, 81–112. doi:10.3102/003465430298487

Heene, M., Hilbert, S., Draxler, C., Ziegler, M., & Bühner, M. (2011). Masking misfit in

confirmatory factor analysis by increasing unique variances: a cautionary note on the

usefulness of cutoff values of fit indices. Psychological Methods, 16, 319-336. doi:

10.1037/a0024917

Herzer, F., Wendt, J. & Hamm, A. O. (2014). Discriminating clinical from nonclinical

manifestations of test anxiety: A validation study. Behavior Therapy, 45, 222-231. doi:

10.1016/j.beth.2013.11.001

Hirvonen, R., Yli-Kivistö, L., Putwain, D. W., Ahonen, T., & Kiuru, N. (2019). School-

related stress among sixth-grade students–Associations with academic buoyancy and

temperament. Learning and Individual Differences, 70, 100–108. doi:

10.1016/j.lindif.2019.01.012.

Hirvonen, R., Putwain, D.W., Määttä, S., Ahonen, T., & Kiuru, N. (2019). The role of

academic buoyancy and emotions in students’ learning-related expectations and

behaviours in primary school. British Journal of Educational Psychology. Advance

online publication. doi: 10.1111/bjep.12336

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 44

Hix-Small, H., Duncan, T. E., Duncan, S. C., & Okut, H. (2004). A multivariate associative

finite growth mixture modeling approach examining adolescent alcohol and marijuana

use. Journal of Psychopathology and Behavioral Assessment, 26, 255-270. doi:

10.1023/B:JOBA.0000045341.56296.fa

Holmes, W., & Dowker, A. (2013). Catch up numeracy: a targeted intervention for children

who are low-attaining in mathematics. Research in Mathematics Education, 15, 249–

265. doi:10.1080/14794802.2013.803779

Horwitz, E. K. (2001). Language Anxiety and Achievement. Annual Review of Applied

Linguistics, 21, 112-127. http://dx.doi.org/10.1017/S0267190501000071

Hoy, W. K., Tarter, C. J., & Hoy, A. W. (2006). Academic optimism of schools: A force for

student achievement. American Educational Research Journal, 43, 425-446. doi:

10.3102/00028312043003425

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure

analysis: Conventional criteria versus new alternatives. Structural Equation Modeling:

A Multidisciplinary Journal, 6, 1–55. doi: 10.1080/10705519909540118

Jeličič, H., Phelps, E., & Lerner, R. M. (2009). Use of missing data methods in longitudinal

studies: The persistence of bad practices in developmental psychology. Developmental

Psychology, 45, 1195-1199.

Kärkkäinen, K. & Vincent-Lancrin, S. (2013). Sparking Innovation in STEM Education with

Technology and Collaboration: A Case Study of the HP Catalyst Initiative. OECD

Education Working Papers, No. 91. OECD Publishing. doi:10.1787/5k480sj9k442-en.

Keith, T. Z. (2006). Multiple regression and beyond. Pearson Education.

Klein, A., & Moosbrugger, H. (2000). Maximum likelihood estimation of latent interaction

effects with the LMS method. Psychometrika, 65, 457-474. doi:10.1007/BF02296338

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 45

Lamers, F., Hoogendoorn, A.W., Smit, J.H., van Dyck, R., Zitman, F.G., Nole, W.A.,

Penninx, B.W., 2012. Sociodemographic and psychiatric determinants of attrition in the

Netherlands Study of Depression and Anxiety. Comprehensive Psychiatry 53, 63-70.

https://doi.org/10.1016/j.comppsych.2011.01.011

Lance, C. E., Butts, M. M., & Michels, L. C. (2006). The sources of four commonly reported

cutoff criteria what did they really say? Organizational Research Methods, 9, 202–220.

doi: 10.1177/1094428105284919

Lauchlan, F. (2003). Responding to chronic non-attendance: A review of intervention

approaches. Educational Psychology in Practice, 19, 133–146.

doi:10.1080/02667360303236

Lens, W., & Seginer, S. (2015). Future time perspective and motivation. International

Encyclopedia of the Social & Behavioral Sciences, 2, 561–566. doi: 10.1016/B978-0-

08-097086-8.24098-1

Lichtenfeld, S., Pekrun, R., Stupnisky, R. H, Reiss, K. & Murayama, K. (2012). Measuring

students' emotions in the early years: The Achievement Emotions Questionnaire-

Elementary School (AEQ-ES). Learning and Individual Differences, 22, 190-201. doi:

10.1016/j.lindif.2011.04.009

Little, R. J. A. (1988). A test of missing completely at random for multivariate data with

missing values. Journal of the American Statistical Association, 83, 1198–1202. doi:

10.2307/2290157

Little, R. J., & Rubin, D. B. (2002). Statistical analysis with missing data (2nd Ed). Wiley-

Interscience.

Linnenbrink, E. A. (2006). Emotion research in education: theoretical and methodological

perspectives on the integration of affect, motivation and cognition. Educational

Psychology Review, 18, 307-314. doi: 10.1007/s10648-006-9028-x

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 46

Linnenbrink-Garcia, L. & Barger, M. M. (2014). Achievement goals and emotions. In R.

Pekrun and L. Linnenbrink-Garcia (Eds.), International handbook of emotions in

education (pp. 142-161). Routledge.

Linnenbrink, E. A., & Pintrich, P. R. (2002). Achievement goal theory and affect: An

asymmetrical bidirectional model. Educational Psychologist, 37, 69–78.

doi:10.1207/s15326985ep3702_2

Little, T. D., Preacher, K. J., Selig, J. P., & Card, N. A. (2007). New developments in latent

variable panel analyses of longitudinal data. International Journal of Behavioral

Development, 31, 357–365. doi: 10.1177/0165025407077757

Loderer, K., Pekrun, R., & Lester, J. (2018). Beyond cold technology: A systematic review

and meta-analysis on emotions in technology-based learning environments. Learning

and Instruction. Advance online publication.doi: 10.1016/j.learninstruc.2018.08.002

Loose, F., Régner, I., Morin, A.J.S., & Dumas, F. (2012). Are academic discounting and

devaluing double-edged swords? Their relations to global self-esteem, achievement

goals and performance among stigmatized students. Journal of Educational

Psychology, 104, 713-725. doi: 10.1037/a0027799

Malmberg, L.-E., Hall, J., & Martin, A. J. (2013). Academic buoyancy in secondary school:

Exploring patterns of convergence in English, mathematics, science, and physical

education. Learning and Individual Differences, 23, 262–266.

doi:10.1016/j.lindif.2012.07.014

Maloney, E.A. (2016). Math anxiety: Causes, consequences, and remediation. In In K.R.

Wentzel and D. B. Miele (Eds.) Handbook of motivation at school (pp. 408-424).

Oxon: Routledge.

Marsh, H.W. (2007). Self-Concept theory, measurement and research into practice: The role

of self concept in educational psychology. British Psychological Society.

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 47

Martin, A.M. (2013a). Academic buoyancy and academic resilience: Exploring ‘everyday’

and ‘classic’ resilience in the face of academic adversity. School Psychology

International, 34, 488-500. doi: 10.1177/0143034312472759

Martin, A.J. (2013b). The Personal Proficiency Network: Key self-system factors and

processes to optimize academic development. In D.M. McInerney, H.W. Marsh., R.G.

Craven, & F. Guay (Eds). Theory driving research: New wave perspectives on self-

processes and human development (pp.251-286). Information Age Publishing.

Martin, A. M. (2014). Towards buoyancy and academic outcomes: Towards a further

understanding of students with attention-deficit/hyperactivity disorder (ADHD),

students without ADHD, and academic buoyancy. British Journal of Educational

Psychology, 84, 86–104. doi:10.1111/bjep.12007

Martin, A.,J., Colmar, S.H., Davey, L.A., & Marsh , H.W. (2010) Longitudinal modelling of

academic buoyancy and motivation: Do the ‘5Cs’ hold up over time? British Journal of

Educational Psychology, 80, 473-496. doi: 10.1348./000709910X486376.

Martin, A. J., Ginns, P., Brackett, M. A., & Malmberg, L.-E. (2013). Academic buoyancy and

psychological risk: Exploring reciprocal relationships. Learning and Individual

Differences, 27, 128–133. doi:10.1016/j.lindif.2013.06.006

Martin, A.J. & Marsh, H.W. (2008) Academic buoyancy: towards an understanding of

students’ everyday academic resilience, Journal of School Psychology, 46, 53–83.

doi:10.1016/j.jsp.2007.01.002

Martin, A. J., & Marsh, H. W. (2009). Academic resilience and academic buoyancy:

Multidimensional and hierarchical conceptual framing of causes, correlates and cognate

constructs. Oxford Review of Education, 35, 353–370. doi:

10.1080/03054980902934639
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 48

Martin, A.M., & Marsh, H.W. (2019). Investigating the reciprocal relations between

academic buoyancy and academic adversity: Evidence for the protective role of

academic buoyancy in reducing academic adversity over time. International Journal of

Behavioral Development. Advance online publication. doi:

10.1177/0165025419885027

Maslowsky, J., Jager, J., & Hemken, D. (2015). Estimating and interpreting latent variable

interactions: A tutorial for applying the latent moderated structural equations method.

International journal of Behavioral Development, 39, 87-96.doi:

10.1177/0165025414552301

Meredith, W. (1993). Measurement invariance, factor analysis and factorial invariance.

Psychometrika, 58, 525–543. doi:10.1007/BF02294825

Muenks, K., Wigfield, A., & Eccles, J. S. (2018). I can do this! The development and

calibration of children’s expectations for success and competence beliefs.

Developmental Review, 48, 24-39. doi: 10.1016/j.dr.2018.04.001

Muis, K. R., Psaradellis, C., Lajoie, S. P., Di Leo, I., & Chevrier, M. (2015). The role of

epistemic emotions in mathematics problem solving. Contemporary Educational

Psychology, 42, 172–185. doi: 10.1016/j.cedpsych.2015.06.003

Murphy, M. C., & Dweck, C. S. (2010). A culture of genius: How an organization’s lay

theory shapes people’s cognition, affect, and behavior. Personality and Social

Psychology Bulletin, 36, 283–296.doi:10.1177/0146167209347380

Muthén, L. K., & Muthén, B. O. (2017). Mplus user’s guide (8th ed.). Muthén & Muthén.

National Center for Educational Statistics (2019). 2019 National assessment of educational

progress: Highlighted results at grades 4 and 8 for the nation, states, and districts. US

Department for Education. Retrieved 29th February from

https://www.nationsreportcard.gov/highlights/mathematics/2019
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 49

Nicholson, J. S., Deboeck, P. R., & Howard, W. (2017). Attrition in developmental

psychology: A review of modern missing data reporting and practices. International

Journal of Behavioural Development, 41, 143-153. doi:10.1177/0165025415618275

O’Connor, R.E., Beach, K., Sanchez, V. Bocian, K., and Flynn, L. (2015). Building BRIDGES:

A Design Experiment to Improve Reading and United States History Knowledge ofPoor

Readers in 8th Grade. Exceptional Children, 81, 399-425.

https://doi.org/10.1177/0014402914563706

Pekrun, R. (2006). The control-value theory of achievement emotions: Assumptions,

corollaries, and implications for educational research and practice. Educational

Psychology Review, 18, 315–341. doi: 10.1007/s10648-006-9029-9

Pekrun, R. (2017). Achievement emotions. In A.J. Ellior, C.S. Dweck, & D.S. yeager (Eds.)

Handbook of Competence and Motivation Second Edition (pp. 251-271). Guilford

Press.

Pekrun, R. (2018). Control-value theory: A social-cognitive approach to achievement

emotions. In G. A. D. Liem & D. M. McInerney (Eds.), Big theories revisited 2: A

volume of research on sociocultural influences on motivation and learning (pp. 162-

190). Information Age Publishing.

Pekrun, R. (in press). Self-appraisals and emotions: A control-value approach. In T. Dicke, F.

Guay, H. W. Marsh, R. G. Craven, & D. M. McInerney (Eds). Self – a multidisciplinary

concept. Information Age Publishing.

Pekrun, R. & Perry, R.P. (2014). Control-value theory of achievement emotions. In R. Pekrun

and L. Linnenbrink-Garcia (Eds.). International handbook of emotions in education

(pp. 120-141). Routledge.

Pekrun, R., Goetz, T., Frenzel, A.C., Barchfeld, P., & Perry, R. P (2011). Measuring

emotions in students’ learning and performance: The Achievement Emotions

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 50

Questionnaire (AEQ). Contemporary Educational Psychology, 36, 36-48. doi:

10.1016/j.cedpsych.2010.10.002

Pekrun, R., Goetz, T., Titz, W., & Perry, R. P. (2002) Academic emotions in students' self-

regulated learning and achievement: A program of qualitative and quantitative research.

Educational Psychologist, 37, 91-105. doi: 10.1207/S15326985EP3702_4

Pekrun, R., Hall, N. C., Goetz, T., & Perry, R. P. (2014). Boredom and academic

achievement: Testing a model of reciprocal causation. Journal of Educational

Psychology, 106, 696-710. doi: 10.1037/a0036006

Pekrun, R., Lichtenfeld, S., Marsh, H. W., Murayama, K., & Goetz, T. (2017). Achievement

emotions and academic performance: Longitudinal models of reciprocal effects. Child

Development, 88, 1653-1670. doi: 10.1111/cdev12704

Pekrun, R., & Loderer, K. (in press). Control-value theory and students with special needs:

Achievement emotion disorders and their links to behavioral disorders and academic

difficulties. In A. J. Martin, R. A. Sperling, and K. J. Newton (Eds.), Handbook of

educational psychology and students with special needs. Taylor & Francis.

Pekrun, R., Murayama, K., Marsh, H. W., Goetz, T., & Frenzel, A. C. (2019). Happy fish in

little ponds: Testing a reference group model of achievement and emotion. Journal of

Personality and Social Psychology, 117, 166-185. doi: 10.1037/pspp0000230

Perry, R. P., Chipperfield, J. G., Hladkyj, S., Pekrun, R., & Hamm, J. M. (2014). Attribution-

based treatment interventions in some achievement settings. In S. Karabenick & T. C.

Urdan (Eds.), Advances in motivation and achievement (Vol 18, pp. 1-35). Emerald

Group Publishing Limited.

Peixoto, F., Sanches, C., Mata, L., & Monteiro, V. (2016). “How do you feel about math?”:

Relationships between competence and value appraisals, achievement emotions and

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 51

academic achievement. European Journal Of Psychology Of Education, 32, 385-405.

https://doi.org/10.1007/s10212-016-0299-4

Putwain, D.W., & Aveyard, B. (2018). Is perceived control a critical factor in understanding

the negative relationship between cognitive test anxiety and examination performance?

School Psychology Quarterly, 33, 65-74. doi: 10.1037/spq0000183

Putwain, D. W., Becker, S., Symes, W., & Pekrun, R. (2018). Reciprocal relations between

academic enjoyment, boredom, and achievement over time. Learning and Instruction,

54, 73-81. doi: 10.1016/j.learninstruc.2017.08.004

Putwain, D. W., Connors, L., Symes, W., & Douglas-Osborn, E. (2012). Is academic

buoyancy anything more than adaptive coping? Anxiety, Stress & Coping, 25, 349–358.

doi: 10.1080/10615806.2011.582459

Putwain, D. W., Chamberlain, S., Daly, T., & Saddredini, S. (2015). Academically buoyant

students are less anxious about and perform better in high-stakes examinations. British

Journal of Educational Psychology, 85, 247‒263. doi: 10.1111/bjep.12068

Putwain, D. W., Daly, T., Chamberlain, S., & Saddredini, S. (2016). “Sink or swim”:

buoyancy and coping in the test anxiety and academic performance relationship.

Educational Psychology, 36, 1807–1825. doi: 10.1080/01443410.2015.1066493

Putwain, D. W., Pekrun, R., Nicholson, L. J., Symes, W., Becker, S., & Marsh, H. W. (2018).

Control-value appraisals, enjoyment, and boredom in mathematics: A longitudinal

latent interaction analysis. American Educational Research Journal, 55, 1339–1368.

doi: 10.3102/0002831218786689

Putwain, D. W., Schmitz, E. A., Wood, P., & Pekrun, R. (2020). The role of achievement

emotions in primary school mathematics: Control-value antecedents and achievement

outcomes. Advance online publication. https://doi.org/10.1111/bjep.12367

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 52

Raccanello, D., Brondino, M., Moe, A., Stupnisky, R., & Lichtenfeld, S. (2019). Enjoyment,

boredom, and anxiety in elementary schools in two domains: Relations with

achievement. Journal of Experimental Education, 87, 449-469. doi:

10.1080/00220973.2018.1448747

Ramirez, G., Shaw, S. T., & Maloney, E. A. (2018). Math anxiety: Past research, promising

interventions, and a new interpretation framework. Educational Psychologist, 53, 145-

164. doi: 10.1080/00461520.2018.1447384

Régner, I., & Loose, F. (2006). Relationship of socio-cultural factors and academic self-

esteem to school grades and school disengagement in North African French

adolescents. British Journal of Social Psychology, 45, 777–797.

doi:10.1348/014466605X83610

Rosel, J., & Plewis, I. (2008). Longitudinal data analysis with structural equations.

Methodology, 4, 37–50. doi:10.1027/1614-2241.4.1.37

Salles, A., Ais, J., Semelman, M., Sigman, M., & Calero, C. I. (2016). The metacognitive

abilities of children and adults. Cognitive Development, 40, 101–110.

Doi:10.1016/j.cogdev.2016.08.009.

Schaeffer, M. W., Rozek, C. S., Berkowitz, T., Levine, S. C., & Beilock, S. L. (2018).

Disassociating the relation between parents’ math anxiety and children’s math

achievement: Long-term effects of a math app intervention. Journal of Experimental

Psychology: General, 147, 1782–1790. Doi: 10.1037/xge0000490

Shao, K., Pekrun, R., Marsh, H. W., & Loderer, K. (2020). Control-value appraisals,

achievement emotions, and foreign language performance: A latent interaction analysis.

Learning and Instruction. Advance online publication.

https://doi.org/10.1016/j.learninstruc.2020.101356
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 53

Sinatra, G.M., Broughton, S.H., & Lombardi, D. (2016). Emotions in science education. In R.

Pekrun and L. Linnenbrink-Garcia (Eds.). International handbook of emotions in

education (pp. 415-436). Routledge.

Smith, T.F., & Capuzzi, G. (2019). Using a mindset intervention to reduce anxiety in the

statistics classroom. Psychology Teaching and Learning. Online advance publication.

doi: 10.1177/1475725719836641

Standards and Testing Agency (2016a). 2016 National Curriculum Tests, Key Stage 2

Mathematics Paper 2: Reasoning. Her Majesty's Stationery Office.

Standards and Testing Agency (2016b). 2016 National Curriculum Tests, Key Stage 2

Mathematics Paper 3: Reasoning. Her Majesty's Stationery Office.

Standards and Testing Agency (2017a). 2017 National Curriculum Tests, Key Stage 2

Mathematics Paper 2: Reasoning. Her Majesty's Stationery Office.

Standards and Testing Agency (2017b). 2017 National Curriculum Tests, Key Stage 2

Mathematics Paper 3: Reasoning. Her Majesty's Stationery Office.

Standards and Testing Agency (2018a). 2018 National Curriculum Tests, Key Stage 2

Mathematics Paper 2: Reasoning. Her Majesty's Stationery Office.

Standards and Testing Agency (2018b). 2018 National Curriculum Tests, Key Stage 2

Mathematics Paper 3: Reasoning. Her Majesty's Stationery Office.

Swann, W., Chang-Schneider, C., & McClarty, K. (2007). Do people’s self-views matter? Self-

concept and self-esteem in everyday life. American Psychologist, 62, 84–94.

doi:10.1037/0003-066X.62.2.84

Symes, W., Putwain D. W., & Remedios, R. (2015). Academic buoyancy and the appraisal of

fear appeals, used prior to a high-stakes examination, as threatening or challenging. School

Psychology International, 36, 605–619.doi: 10.1177/0143034315610622

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 54

Tze, V. M., Daniels, L. M., & Klassen, R. M. (2016). Evaluating the relationship between

boredom and academic outcomes: A meta-analysis. Educational Psychology Review,

28, 119 –144. doi:10.1007/s10648-015-9301-y

Villavicencio, F. T., & Bernardo, A. B. I. (2016). Beyond math anxiety: Positive emotions

predict mathematics achievement, self-regulation and self-efficacy. The Asia-Pacific

Education Researcher, 25, 415–422. doi: 10.1007/s40299-015-0251-4

von der Embse, N., Barterian, J., & Segool, N. (2013). Test anxiety interventions for children

and adolescents: A systematic review of treatment studies from 2000-2010. Psychology

in the Schools, 50, 57-71. doi: 10.1002/pits.21660

von der Embse, N.P., Jester, D., Roy, D., & Post, J. (2018). Test anxiety effects, predictors,

and correlates: A 30-year meta-analytic review. Journal of Affective Disorders, 227,

483-493. doi: 10.1016/j.jad.2017.11.048

Watkins, M. W., Lei, P. W., & Canivez, G. L. (2007). Psychometric intelligence and

achievement: A cross-lagged panel analysis. Intelligence, 35, 59-68. doi:

10.1016/j.intell.2006.04.005

von der Embse, N.P. & Witmer, S. (2014). High-stakes accountability: Student anxiety and

large-scale testing. Journal of Applied School Psychology, 2, 1-24. doi:

10.1080/15377903.2014.888529

Warren, M.K., Ollendick, T.H., & King, N.J. (1996). Test anxiety in girls and boys: A

clinical-developmental analysis. Behaviour Change, 13, 157-170. doi:

10.1017/S0813483900004939

Widaman, K. F., Ferrer, E., & Conger, R. D. (2010). Factorial invariance within longitudinal

structural equation models: Measuring the same construct across time. Child

Development Perspectives, 4, 10-18. doi: 10.1111/j.1750-8606.2009.00110.x

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 55

Wigfield, A. (1997). Reading motivation: a domain specific approach to motivation.

Educational Psychologist, 32, 59-68. doi: 10.1207/s15326985ep3202_1

Wu, J-W., & Kwok, O-W. (2012). Using SEM to analyze complex survey data: A

Comparison between design-based single-level and model-based multilevel approaches.

Structural Equation Modeling, 19, 16-35. 10.1080/10705511.2012.634703

Yun, S., Hiver, P., & Al-Hoorie, A. H. (2018). Academic buoyancy: Exploring learners’

everyday resilience in the language classroom. Studies in Second Language

Acquisition, 40, 805–830. doi: 10.1017/S0272263118000037

 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 56

Table 1
Descriptive Statistics and Item Factor Loadings
Factor
Mean SD α/ω ρI Skewness Kurtosis
loadings

T1 Enjoyment 16.24 4.48 .92 / .92 .07 -1.09 0.21 .84 - .90
T1 Boredom 7.87 4.38 .91 / .91 .04 1.12 0.16 .79 - .88
T1 Anxiety 8.30 4.38 .82 / .82 .03 0.99 0.17 .70 - .75
T3 Enjoyment 15.40 4.75 .93 / .93 .06 -0.86 -0.27 .84 - .96
T3 Boredom 7.78 4.38 .91 / .92 .04 0.94 -0.11 .86 - .89
T3 Anxiety 8.23 4.27 .82 / .83 .04 0.99 0.24 .79 - .90
T3 Buoyancy 15.58 3.85 .79 / .79 .03 -0.51 -0.28 .64 - .78
T2 Mathematics Test Performance 4.64 3.67 .79 / .81 .12 0.82 0.15 —
T4 Mathematics Test Performance 9.55 4.72 .85 / .85 .14 0.11 -0.73 —

Note. ω = McDonalds omega. ρI = intraclass correlation coefficient (ICC1).

 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 57

Table 2
Correlations between the Study Variables
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

1. T1 Enjoyment — -.68** -.36*** .60*** -.48*** -.24*** .44*** .14** .13** -.17*** -.07*
2. T1 Boredom -.64*** — .64*** -.40*** .45*** .28*** -.28*** -.18*** -.17*** .07* .11***
3. T1 Anxiety -.33*** .56*** — -.24*** .25*** .51*** -.38*** -.32*** -.36*** .12** .01
4. T3 Enjoyment .57*** -.41*** -.27*** — -.78*** -.40*** .62*** .18*** .16** -.15*** -.01
5. T3 Boredom -.48*** .56*** .27*** -.75*** — .58*** -.43*** -.19*** -.19*** .11** .05
6. T3 Anxiety -.27*** .31*** .45*** -.38*** .53*** — -.47*** -.34*** -.40*** .09* .01
7. T3 Academic Buoyancy .37*** -.26*** -.33*** .52*** -.28*** -.36*** — .25*** .23*** -.10** .02
8. T2 Test Performance .15*** -.17*** -.29*** .21*** -.21*** -.32*** .22*** — .67*** -.08* .07
9. T4 Test Performance .18*** -.19*** -.33*** .20*** -.21*** -.38*** .20*** .69*** — -.06 .10**
10. Gender -.16*** .05 .10** -.16*** .11*** .09** -.08** -.08** -.08* — —

11. Age -.06* .10*** .01 -.01 .03 .02 .01 .07* .11** — —

Note. Latent bivariate correlations above the diagonal, manifest Pearson’s correlations below the diagonal.
*p <.05. **p <.01. ***p <. 001.
 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 58

Table 3
Tests of Measurement Invariance for Classroom Achievement Emotions
χ2(df) RMSEA SRMR CFI TLI Δ RMSEA ΔCFI ΔTLI

Enjoyment
T1 3.16(2) .022 .006 1.000 .999
T3 3.47(2) .026 .005 .999 .997
Configural 17.89(15) .011 .012 1.000 .999
Metric Invariance 22.14(18) .012 .020 .999 .999 +.001 -.001 <.001
Scalar Invariance 44.72(22) .026 .037 .996 .995 +.014 -.003 -.004
Residual Invariance 38.56(26) .018 .034 .998 .998 -.008 +.002 +.003

Boredom
T1 1.42(2) .000 .004 1.000 1.001
T3 24.81(2) .100 .020 .981 .944
Configural 67.33(15) .048 .023 .988 .977
Metric Invariance 76.91(18) .047 .027 .986 .978 -.001 -.002 -.001
Scalar Invariance 85.82(22) .044 .036 .985 .981 -.003 -.001 +.003
Residual Invariance 88.38(26) .040 .035 .985 .984 -.004 <.001 +.003

Anxiety
T1 2.18(2) .008 .007 1.000 1.000
T3 8.95(2) .055 .018 .990 .997
Configural 40.78(15) .034 .023 .988 .987
Metric Invariance 42.26(18) .030 .024 .989 .983 -.004 +.001 -.004
Scalar Invariance 42.87(22) .025 .025 .991 .988 -.005 +.002 +.005
Residual Invariance 54.03(26) .027 .030 .987 .986 +.002 -.004 -.002
 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 59

Note. χ2 statistic for all models statistically significant at p <.001.

 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 60

Table 4
Comparison of the Reciprocal Relations Model to the Baseline and Unidirectional Relations Models
χ2 (df) RMSEA SRMR CFI TLI AIC ΔAIC TRd(df)

Baseline Model 729.48 (315)*** .033 .082 .972 .967 76777.58 209.96 188.28 (18)***
Unidirectional Model A 609.64 (306)*** .028 .052 .980 .975 76647.67 80.05 79.89 (9)***
Unidirectional Model B 686.71 (312)*** .031 .070 .975 .969 76734.32 166.70 140.07 (15)***
Reciprocal Relations Model 526.24 (297)*** .025 .027 .985 .980 76567.62 — —

Notes. (a) Unidirectional Model A: Relations of emotions to subsequent performance freely estimated, relations of performance to subsequent
emotions constrained to zero. Unidirectional Model B: Relations of performance to subsequent emotions freely estimated, relations of emotions
to subsequent performance constrained to zero. (b) χ2 statistic for all models statistically significant at p < .001.
 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 61

Table 5
Standardized Path Coefficients for the Fully-Forward Reciprocal Relations Model (Standard Errors in Parentheses)
T1 T2 Test T3 T4 Test
T1 Boredom T1 Anxiety T3 Boredom T3 Anxiety
Enjoyment Performance Enjoyment Performance

T1 Enjoyment .09 (.06) .59 (.05) -.31 (.07) -.19 (.08) .05 (.05)
T1 Boredom -.11 (.07) .04 (.06) .24 (.07) -.18 (.07) .08 (.08)
T1 Anxiety -.36 (.06) -.03 (.05) -.05 (.06) .50 (.07) -.13 (.08)
T2 Test Performance .12 (.04) -.14 (.03) -.19 (.04) .60 (.03)
T3 Enjoyment -.05 (.06)
T3 Boredom -.01 (.05)
T3 Anxiety -.16 (.07)
Gender -.17 (.03) -.07 (.03) .12 (.04) -.05 (.03) -.03 (.04) .02 (.03) -.01 (.03) .01 (.04)
Age -.07 (.03) .11 (.03) .01 (.04) .06 (.04) .02 (.03) .01 (.03) .03 (.04) .05 (.04)
 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 62

Table 6
Standardized Path Coefficients for the LI-SEM to Predict T4 Test Performance from Interactions between Academic Buoyancy and Enjoyment,
Boredom, and Anxiety (Standard Errors in Parentheses)

T2 Test Academic T4 Test

Predictor Enjoyment Boredom Anxiety
Performance Buoyancy Performance

Enjoyment (ENJ) .04 (.06)

Boredom (BOR) -.08 (.05)
Anxiety (ANX) -.25 (.06)
Academic Buoyancy (B) .01 (.04)
B × ENJ -.05 (.06)
B × BOR -.03 (.07)
B × ANX -.10 (.04)
T2 Test Performance .19 (.04) -.19 (.04) -.34 (.04) .25 (.05) .61 (.03)
Gender -.06 (.05) .21 (.07) .08 (.09) .06(.07) .03 (.05) .02 (.06)
Age .06 (.04) .06 (.09) .10 (.08) .02 (.06) .01 (.07) .10 (.07)
 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 63

Figure 1

The Hypothesized Fully Forward Reciprocal Model Including Autoregressive, Cross-Lagged, and Concurrent Relations between Achievement
Emotions and Mathematics Test Performance

T1 Enjoyment T3 Enjoyment

T1 Boredom T2 Maths Test T3 Boredom T4 Maths Test

T1 Anxiety T3 Anxiety

Note. Paths for gender and age are not depicted.

 EMOTIONS, ACHIEVEMENT, AND BUOYANCY 64

Figure 2
Significant Autoregressive, Cross-lagged, and Concurrent Relations Between Achievement Emotions and Test Performance

T1 Enjoyment .59 T3 Enjoyment

-.31 .12

-.68
-.68

.60
-.36

T1 Boredom T2 Maths Test -.14 T3 Boredom T4 Maths Test

-.16

-.35
-.18

.60
.64

-.36

.24 -.19

T1 Anxiety .50 T3 Anxiety

-.19

Note. Paths for gender and age are not depicted.

EMOTIONS, ACHIEVEMENT, AND BUOYANCY 65

Figure 3a
The Model-implied Effect of the T3 Academic Buoyancy × Anxiety interaction on T4
Mathematics Test Performance

12
10
T4 Mathematics Test Performance

8
6
4 T3 Academic Buoyancy

2 Low (-1SD)
Mean
0
High (+1SD)
-2
-4
-6
-8
-2 SD -1.5 SD -1 SD -0.5 SD Mean +0.5 SD +1 SD +1.5 SD +2 SD
T3 Anxiety

Note. Anxiety represented on the x axis and slopes plotted for ±1SD academic buoyancy.
EMOTIONS, ACHIEVEMENT, AND BUOYANCY 66

Figure 3b
The Model-implied Effect of the T3 Academic Buoyancy × Anxiety interaction on T4
Mathematics Test Performance
T4 Mathematics Test Performance

3 T3 Anxiety
Low (-1SD)
-2 Mean
High (+1SD)

-7

-12
-2 SD -1.5 SD -1 SD -0.5 SD Mean +0.5 SD +1 SD +1.5 SD +2 SD
T3 Academic Buouancy

Note. Academic represented on the x axis and slopes plotted for ±1SD anxiety.