Reading and writing Research

British Journal of Educational Psychology (2021), 91, 1456–1480

© 2021 The British Psychological Society

www.wileyonlinelibrary.com

Relations among motivation, behaviour, and performance in writing: A multiple-group structural equation modeling study

AnaCamacho1,2* , Rui A.Alves1, FienDeSmedt3,HildeVanKeer3

and Pietro Boscolo4

1Faculty of Psychology and Education Sciences, University of Porto, Portugal 2School of Health, Polytechnic of Porto, Portugal 3Department of Educational Studies, Ghent University, Belgium 4Department of Developmental Psychology and Socialization, University of Padova, Italy

Background. Writing is a particularly demanding activity, which poses unique

motivational challenges for students. Despite the wealth of research on the relation

between writing motivation and writing performance, little is known about the role of

students’ writing frequency in writing motivation and writing performance.

Aims. We aimed to: (1) examine structural relations among two motivational variables

(i.e., self-efficacy and attitudes), a behavioural variable (i.e., writing frequency), andwriting

performance; and (2) inspect whether these relations varied across two text genres (i.e.,

narrative and opinion texts) and across two educational levels (i.e., students in grades 5–6 and grades 7–8).

Sample. Six hundred and five students from grades 5–8 participated in this study.

Methods. Students completed self-report scales andwrote narrative and opinion texts.

We conducted multiple-group structural equation modeling to analyse the data.

Results. Regarding narrative texts, digital writing frequency was significantly associated

with text quality for students in grades 7–8, but this relationwas not significant in students from grades 5–6. Both attitudes and self-efficacy for self-regulation made a direct

contribution to narrative text quality across educational levels. In addition, attitudeswere

associated with both literary and digital writing frequency across educational levels.

Concerning opinion texts, no significant differences emerged in terms of educational

level. Attitudes contributed to both literary and digital writing frequency as well as to

opinion text quality across educational levels.

Conclusions. This study underlines the fundamental contribution of motivational

variables to students’ writing performance. Accordingly, teachers need to adopt

motivation-enhancing practices in writing instruction across grade levels.

Teaching students to write effectively is a key aim of mandatory education. This aim is

consistentwith the cross-cutting role ofwriting inmany facets of everyday life.Writing is a

*Correspondence should be addressed to Ana Camacho, Faculdade de Psicologia e de Cîencias da Educac�~ao, Universidade do Porto, Rua Alfredo Allen, 4200-135 Porto, Portugal (email: [email protected]).

DOI:10.1111/bjep.12430

1456

fundamental skill to enhance learning and achievement in school, to performdaily tasks at

the workplace, to communicate with other people, and to exert civic rights and duties

(Graham & Harris, 2019; Graham, Wijekumar, et al., 2019). However, many students do

not develop strong writing skills, which may limit their attainments in knowledge-based societies (Graham, Harris, & Santangelo, 2015). Furthermore, students often report low

motivation to engage in writing tasks (Boscolo & Hidi, 2007), which may hinder writing

performance. In this regard, a recent study showed that autonomous motivation to write

declines over the school years (De Smedt, Rogiers, Heirweg, Merchie, & Van Keer, 2020).

Determining which factors bolster writing performance is thus a crucial task for

researchers. According to the recent writer(s)-within-community model (WWC), writing

is a social activity that is shaped by writing communities and by the writers’ cognitive

characteristics, capacities, and physical actions (Graham, 2018). In the current study, we focused on the writers’ characteristics and actions mentioned in theWWCmodel. To this

end, we surveyed Portuguese fifth- to eighth-grade students on their writing motivation

and writing frequency and evaluated their writing performance.

Definition of writing

Decades of writing research from a cognitive lens have shown that writing is a highly

complex activity (Alves, 2019; Hayes, 1996; Kellogg, 1994; Kim, 2020b). The cognitive load imposed by the writing process creates unique motivational challenges for students

(Bruning & Horn, 2000). Moreover, writing does not get easier over the school career as

teachers assign more demanding writing tasks (Boscolo & Hidi, 2007), and therefore,

students pursue more challenging writing goals (Bereiter & Scardamalia, 1987).

In the WWC model, Graham (2018) combined both cognitive and sociocultural

theoretical frameworks of writing, thus conceivingwriting as ‘simultaneously shaped and

constrained by the context, the capabilities, and perceptions of writers and collaborators,

and the interaction between the two’ (p. 258). According to this model, the writing process entails several cognitive mechanisms (viz., motivational beliefs; knowledge;

control mechanisms; production processes; and modulators). Among the motivational

beliefs identified by Graham (2018), our study will focus on beliefs about one’s

competence inwriting (i.e., self-efficacy) and beliefs involvingwhether or not one likes to

write (i.e., attitudes).

We will also focus on the actions performed by writers (i.e., writing frequency). The

WWCmodel postulates that one’s beliefs and beliefs about thewriting task fuel a variety of

cognitive and physical actions, including whether a student engages in writing and how much effort will be committed.

The role of self-efficacy in writing performance

Self-efficacy is the most studied motivational construct in writing research (Camacho,

Alves, & Boscolo, 2021; Pajares, 2003). Self-efficacy for writing refers to the confidence

that one can perform successfully in writing (Bandura, 1982; Bruning, Dempsey,

Kauffman, McKim, & Zumbrunn, 2013). Bruning et al. (2013) found empirical evidence for a three-dimensionalmodel ofwriting self-efficacy, comprising self-efficacy for ideation,

conventions, and self-regulation. Self-efficacy for ideation refers to the confidence in

generating ideas for writing; self-efficacy for conventions pertains to the confidence in

relying on linguistic skills to express one’s ideas; and self-efficacy for self-regulation refers

to the confidence in self-managing behaviour and affect while writing.

Relations among motivation, behaviour, and performance in writing 1457

Prior research has focused on the link between self-efficacy and writing performance.

Research has consistently shown that self-efficacy is one of the strongest motivational

predictors of writing performance (Camacho et al., 2021; Klassen, 2002; Zumbrunn,

Broda, Varier, & Conklin, 2020). However, few researchers examined self-efficacy towards different text genres. An exception is the study by De Smedt, Van Keer, and

Merchie (2016), which showed that self-efficacy for ideation was associated with both

narrative and informational text quality among fifth and sixth graders.

Research by Pajares and associates showed that writing self-efficacy was linked to

writing-related measures such as essay quality (Pajares, 2007; Pajares, Hartley, & Valiante,

2001; Pajares & Valiante, 1997, 1999). Most notably, self-efficacy stood out as the only

motivational predictor of teacher ratings of students’ writing competence when other

motivational variables were considered (Pajares & Valiante, 1999). Writing self-efficacy was also found to make an independent contribution to narrative text quality (Graham,

Harris, Kiuhara, & Fishman, 2017) and to essay writing performance (Pajares & Valiante,

1997).

Previous research has also examined grade-level differences in writing self-efficacy.

Some studies indicated that students from lower grade levels reported higher self-efficacy

for writing, whereas other studies indicated that students from higher grade levels

reported stronger self-efficacy beliefs in writing (Camacho et al., 2021; Klassen, 2002;

Pajares, Johnson, & Usher, 2007).

The role of attitudes in writing performance

Attitudes are the second most studied motivational construct in writing research

(Camacho et al., 2021). The conceptualization of attitudes has varied greatly across

previous studies (Ekholm, Zumbrunn, & DeBusk-Lane, 2018). In the current study, we

adopted the definition put forth by Graham, Berninger, and Fan (2007), who defined

writing attitudes as an ‘affective disposition involving how the act of writing makes the author feel, ranging from happy to unhappy’ (p. 518). Recently, Graham, Daley, Aitken,

Harris, and Robinson (2018) conceptualized writing attitudes as a multidimensional

construct, comprising attitudes towards academic and recreational writing in print and

digital contexts. Attitudes towards academic and recreational writing in digital contexts

acted as unique predictors of writing performance.

Prior research has pointed out the relevant role of writing attitudes in enhancing

writing performance (Camacho et al., 2021; Ekholm et al., 2018; Graham et al., 2007;

Graham, Wijekumar, et al., 2019). Ekholm et al. (2018) reviewed empirical research on writing attitudes and concluded that this motivational constructwas positively associated

with quantitative measures of writing performance, such as text quality. This systematic

review also showed that positive writing attitudes tend to decline over the school years

(Ekholm et al., 2018).

Two recent studies examined the contribution of writing skills, writing knowledge,

strategic writing behaviours, attitudes, and self-efficacy to fifth-grade students’ persuasive

text quality (Graham, Wijekumar, et al., 2019; Wijekumar et al., 2019). Graham,

Wijekumar, et al. (2019) found that attitudes and self-efficacy together made a statistically significant unique contribution to text quality beyond the other predictors in both fall and

spring. Conversely, Wijekumar et al. (2019) found that attitudes, self-efficacy for writing

mechanics, and self-efficacy for regulation did not uniquely predict text quality at both

timepoints. Wijekumar and colleagues argued that the disparate results may be partially

1458 ANA CAMACHO et al.

explained by differences in participants, writing prompts (i.e., science vs. social sciences

writing topics), and self-efficacy measures (i.e., single vs. two self-efficacy factors).

The role of writing frequency in writing performance

We focused on a specific behavioural component – writing frequency – which we

conceptualized as how often students carry out literary and digital writing activities.

Studies examining writing frequency are scarce (Troia, Harbaugh, Shankland, Wolbers, &

Lawrence, 2013). An exception is the study by Troia et al. (2013). In this study, the authors

tested a first model where writing frequency was placed before motivational beliefs (i.e.,

moderator) and a second model where writing frequency was placed between

motivational beliefs and text quality (i.e., mediator). In the first model, writing frequency made a direct contribution tomotivational beliefs, which in turn positively contributed to

text quality. In the second model, motivational beliefs made a minimal contribution to

writing frequency,which in turn did not contribute to text quality. However, bothmodels

were not directly compared. In a recent review, Graham (2019) observed that one of the

indicators of insufficient writing instruction nowadays is that students do not write

frequently in a typical class. This is troublesome considering that when students write

more frequently, there is a 12 percentile-point increase in writing quality (Graham &

Harris, 2016). In the reading domain, studies examining the relations among motivational,

behavioural, and performance variables are more common (e.g., De Naeghel, Keer,

Vansteenkiste, & Rosseel, 2012; El-Khechen, Ferdinand, Steinmayr, & McElvany, 2016;

Troyer, Kim, Hale, Wantchekon, & Armstrong, 2019) than in writing research. Across

these studies, authors typically placed reading frequency as a mediator of themotivation– performance relationship. For example, De Naeghel et al. (2012) examined structural

relations among motivational variables, reading frequency, reading engagement, and

reading comprehension in fifth graders. The authors found that motivational variables were associated with reading frequency, reading engagement, and reading comprehen-

sion. Reading engagement was significantly associated with reading comprehension,

while this was not the case for writing frequency.

The present study

Research linkingwritingmotivation andwritingperformance has blossomedover thepast

decades (Camacho et al., 2021; De Smedt, 2019; Graham et al., 2007). By contrast, evidence on the role of writing frequency in writing performance is still scarce (Troia

et al., 2013). Therefore, there are noteworthy gaps in the literature that warrant further

empirical inquiry. First, there is little empirical evidence on the role of writing frequency

in the relations amongmotivational variables andwriting performance (Troia et al., 2013).

Second, limited research has examined whether motivational, behavioural, and perfor-

mance variables differ across text genres (Camacho et al., 2021; Korat & Schiff, 2005;

Olinghouse, Graham, & Gillespie, 2015; Troia et al., 2013). Third, to the best of our

knowledge, no study has analysed how the relations among self-efficacy, attitudes,writing frequency, and writing performance vary across students in different educational levels

using structural equation modeling.

In the current study, we sought to address these shortcomings. Specifically, we aimed

to: (1) examine structural relations among motivational variables (i.e., self-efficacy and

attitudes), a behavioural variable (i.e., writing frequency), and writing performance (i.e.,

Relations among motivation, behaviour, and performance in writing 1459

text quality), while controlling for handwriting fluency; and (2) analyse whether and how these relations vary or remain stable across two text genres (i.e., narrative text andopinion

text) and across two educational levels (i.e., students in grades 5–6 and grades 7–8). Based on the reviewed literature, we examined five hypotheses (see Figure 1):

Hypothesis 1. We hypothesized that writing self-efficacy and attitudes would be signifi-

cantly and positively associated with writing frequency since Troia et al.

(2013) found thatmotivational beliefsmade a contribution – even if minimal – to writing frequency. In addition, a large body of reading research

indicated that students’ reading motivation was positively associated with

reading frequency (De Naeghel et al., 2012; Troyer et al., 2019). Therefore,

we expected the same pattern of relations in writing since reading and

writing are different but related skills that draw on common cognitive and

language skills as well as socio-emotional factors (Kim, 2020a).

Hypothesis 2. We predicted that writing self-efficacy would be significantly and positively

related to text quality across text genres as previous research showed

significant, positive relations between self-efficacy and narrative text quality

(De Smedt et al., 2016; Graham et al., 2017), opinion essay quality (Limpo &

Alves, 2017), and informational text quality (De Smedt et al., 2016).

Hypothesis 3. We anticipated that writing attitudes would be significantly and positively related to text quality across text genres considering that prior literature

indicated that attitudes were positively related with narrative text quality

(Graham, Berninger, Berninger, & Abbott, 2012; Graham et al., 2007) and

persuasive essay quality (Graham, Harris, et al., 2019).

Wri�ng a�tudes

Self-efficacy for wri�ng

Mo�va�onal component

Behavioural component

Wri�ng performance

Wri�ng frequency

Text quality

Hypothesis 1

Hypothesis 1 Hypothesis 3

Hypothesis 2

Hypothesis 4

Handwri�ng fluency

Hypothesis 5

Figure 1. Hypothesized relational model relating motivational and behavioural components to writing

performance.

1460 ANA CAMACHO et al.

Hypothesis 4. We hypothesized that writing frequency would be significantly and

positively related to text quality across text genres considering that prior

literature indicated thatwhen studentswrite frequently, there is an increase

in writing quality (Graham, 2019; Graham & Harris, 2016).

Hypothesis 5. We anticipated that handwriting fluency (i.e., covariate) would significantly

contribute to text quality (Alves & Limpo, 2015; Feng, Lindner, Ji, &

Malatesha Joshi, 2019; Kent & Wanzek, 2016).

We refrained from formulating a hypothesis concerning educational level differences

in the studied relations as this aim was not strongly supported by previous empirical evidence.

Method

Educational context

The Portuguese educational system is divided into different educational levels: first cycle or primary school (grades 1–4), second cycle (grades 5–6), third cycle (grades 7–9), and secondary education (grades 10–12). Transitions from one cycle to another aremarked by

curricular and contextual changes, which may hinder students’ motivation and school

attainment (Abrantes, 2005).

Participants

Six hundred and five students from grades 5, 6, 7, and 8 from three Portuguese public schools participated in this study. The schools were located in the Porto metropolitan

area. The sample encompassed students in the second cycle (grades 5–6; n = 342) and in

the third cycle (grades 7–8; n = 263) of the Portuguese educational system. Gender

distribution was balanced (nfemale = 310, nmale = 295). We obtained both consent from

parents and assent from students to participate in the study (see Table 1 for additional

demographic information).

Measures

We used motivational, behavioural, and performance measures, which are described

below.

Motivational measures

Self-efficacy. Self-efficacy for writing was assessed using the Self-Efficacy for Writing

Scale (SEWS; Bruning et al., 2013). The SEWS is a 16-item scale wherein students rate their

self-efficacy for writing on a 0 to 100 scale, ranging from no confidence to complete

confidence. Students were asked to rate their self-confidence in three dimensions: self-

efficacy for ideation items asked students about their confidence in the ability to generate ideas (e.g., ‘I can think of many ideas for my writing’); self-efficacy for conventions items

prompted students to judge on their beliefs about the compliance with writing

conventions (e.g., ‘I can spell my words correctly’); and self-efficacy for self-regulation

items asked students about their beliefs on the use of self-regulatory behaviours while

writing (e.g., ‘I can avoid distractions while I write’.).

Relations among motivation, behaviour, and performance in writing 1461

The Confirmatory Factor Analysis (CFA) of the self-efficacy scale for narrative texts

showed a good fit of the expected three-factor model to the data, Satorra–Bentler (SB) v2(101) = 256.50, p < .001, CFI = 0.94, RMSEA = 0.07, SRMR = 0.05. Internal consis-

tencies of the three self-efficacy subscales were high (Bentler’s qideation = 0.84; Bentler’s qregulation = 0.81; Bentler’s qconventions = 0.81). A good model fit was also found for the

self-efficacy scale for opinion texts, SB v2(101) = 233.42, p < .001, CFI = 0.95,

RMSEA = 0.06, SRMR = 0.04. Internal consistencies of the three self-efficacy

subscales were high (Bentler’s qideation = 0.88; Bentler’s qregulation = 0.82; Bentler’s

qconventions = 0.84).

Attitudes. Attitudes towardswritingweremeasured using a self-report scale comprising five items (Graham et al., 2017). Students indicated their level of agreement with five

statements on a 5-point Likert scale, ranging from strong disagreement to strong

agreement (e.g., ‘I like to write at school’).

The CFA showed a reasonable fit of a single-factor model to the data, SB v2(5) = 24.20,

p <.001, CFI = 0.97, RMSEA = 0.12, SRMR = 0.03. The scale was highly reliable

(Bentler’s q = 0.87).

Behavioural measure

Writing frequency. We adapted a self-report scale to assess writing frequency (Bruning

et al., 2013; ). Students rated howoften they carriedout eightwriting activities on a5-point Likert scale, ranging from never to once a day or more: writing poetry or song lyrics; a

Table 1. Student demographic characteristics

Demographic variable

Second cycle

Grades 5–6 (n = 342)

Third cycle

Grades 7–8 (n = 263)

Total

(N = 605)

M SD M SD M SD

Age in years 10.73 0.8 13 1 11.80 1.5

School mark in Portuguese (1–5) 3.40 0.8 3.21 0.7 3.32 0.8

Demographic variable n % n % n %

Female 176 51.5 134 51 310 51.2

Male 166 48.5 129 49 295 48.8

Special educational needs 4 1.2 7 2.7 11 1.8

Portuguese as the native language 330 96.5 242 92 572 94.5

Mother’s educational level

No educational level 2 0.6 5 1.9 7 1.2

Grade 4 or less 20 5.8 20 7.6 40 6.6

Grade 6 or less 33 9.6 40 15.2 73 12.1

Grade 9 or less 70 20.5 58 22.1 128 21.2

Grade 12 or less 97 28.4 66 25.1 163 26.9

University degree 92 26.9 47 17.8 139 22.9

Unknown 28 8.2 27 10.3 55 9.1

1462 ANA CAMACHO et al.

personal diary or journal; notes or letters to others; narrative texts; opinion texts; emails;

text messages; messages or publications in social media networks.

We first conducted an Exploratory Factor Analysis (EFA) to assess the factorial

structure of the writing frequency scale. We used maximum-likelihood extraction with promax rotation. Results revealed that the promax rotation accounted for 50.90% of the

total scale variance,with two factors having an eigenvalue higher than 1.0. Factor loadings

revealed a two-factor structure, representing literary writing frequency comprising five

items (e.g., writing narrative texts; writing opinion texts) and digital writing frequency

comprising two items (e.g.,writing textmessages). The itemon emailwriting showed low

factor loadings and was therefore excluded from further analyses. Loadings of the

remaining items on the targeted factors ranged from .61 to .89. We then conducted a CFA

with a two-factor solution, showing a reasonable fit to the data, SB v2(21) = 642.30, p < .001, CFI = 0.92, RMSEA = 0.09, SRMR = 0.05. Internal consistencies for the two

factors were acceptable (Bentler’s qliterary writing = 0.69; Bentler’s qdigital writing = 0.77).

Performance measure

Text quality. To avoid testing overload, students wrote a narrative text (‘Tell a story

about a child who found a wounded animal’) and an opinion text (‘What is your opinion

about children practicing sport every day?’) one week apart. The quality of both text

genres was assessed bymeans of a holistic rating scale (Cooper, 1977). Handwritten texts

were typed, and spelling, punctuation, and capitalization errors were corrected to

minimize presentation biases (Graham, Harris, & Hebert, 2011). Eight independent research assistants scored text quality. We formed four pairs of

raters, and eachpair assessed text quality of students froma specific grade level. The raters

were trained by the first author to use a holistic rating scale ranging from 1 (lowquality) to

7 (high quality), grounded on four equally important criteria: relevance of ideas or

arguments; coherence; syntax; and vocabulary. The raters were provided with anchor

texts representing low, average, and high text quality for each text genre and grade level.

Inter-rater reliability using Pearson r was high for narrative (rgrade 5 = .89; rrade 6 = .83;

rgrade 7 = .87; rgrade 8 = .80) and opinion texts (rgrade 5 = .87; rrade 6 = .86; rgrade 7 = .90; rgrade 8 = .84). The final quality score was the average score between raters.

Covariate of writing performance

Handwriting fluency. Handwriting fluency was measured using the alphabet task

(Berninger, Mizokawa, & Bragg, 1991). Students wrote the alphabet in lowercase letters

the fastest they could,while keeping legibility, during 60s. The final scorewas the number

of correct letters. Previous research has shown that handwriting fluency predicts writing

performance (Alves & Limpo, 2015; Berninger & Swanson, 1994; Jones & Christensen,

1999). The first author coded the alphabet task of all students, and a research assistant

independently recoded 40% of the cases (r = .99).

Procedure

Two research assistants collected the data in February–March of 2019 during two classes

of 50 min each in the presence of a teacher. One research assistant read aloud the

instructions and students performed the tasks individually. The self-efficacy scale was

Relations among motivation, behaviour, and performance in writing 1463

administered twice – before writing the narrative text and before writing the opinion text

– as Bandura pleaded for specificity when assessing self-efficacy (Bandura, 1997).

Data analysis overview

The statistical analyseswere computedwith the R software (RCore Team, 2019).Weused

the lavaan package for R (Rosseel, 2012) and the lavaan.survey package to take the

clustered nature of the data into account as students were nested in 44 classes (Muth�en &

Satorra, 1995; Oberski, 2014; Stapleton, 2006). We conducted multiple-group structural

equation modeling (MG-SEM). The method of estimation was maximum likelihood with a

SB-scaled chi-square test statistic (Chou, Bentler, & Satorra, 1991; Oberski, 2014; Yuan &

Bentler, 2000). We performed preparatory analyses before computing the MG-SEM models, namely

confirmatory factor analyses (CFA), reliability analyses, andmeasurement invariance tests.

We also computed an exploratory factor analysis (EFA) for the frequency scale since the

factorial structure was not previously established.

As to the main analyses, we compared two MG-SEM models to examine the relations

among self-efficacy, attitudes, writing frequency, and text quality across two educational

levels (grades 5–6 and grades 7–8). We did this comparison separately for narrative and

opinion texts. We compared a model fixing factor loadings and intercepts across groups (Model 1) with a model fixing factor loadings, intercepts, and regressions across groups

(Model 2). Significant differences between these models indicated significant differences

in the regression coefficients across educational levels.

Following the recommendations of Hu and Bentler (1999) and Kline (2005), we relied

on four fit statistic measures to evaluate model fit: (1) chi-square test statistic (v2) and p-

value (p); (2) comparative fit index (CFI); (3) root-mean-square error of approximation

(RMSEA); and (4) root-mean-square error of approximation (SRMR). A value above .90 for

CFI (Browne & Cudeck, 1992), a value close to .06 for RMSEA (Hu & Bentler, 1999), and a value equal to or lower than .08 for SRMR (Hu & Bentler, 1999; Schreiber, Nora, Stage,

Barlow, & King, 2006) indicate adequate model fit.

Results

Measurement invariance analyses We examined multiple-group measurement invariance of motivational and behavioural

scales across educational levels. We compared three models: configural invariance (i.e.,

no constraints), weak invariance (i.e. equal factor loadings across educational levels), and

strong invariance (i.e., equal factor loadings and intercepts across educational levels). We

found changes in the comparative fit index (DCFI) equal or smaller than 0.01 (Cheung &

Rensvold, 2002) for motivational scales. Only for writing frequency, the changes were

slightly above the recommended threshold (DCFImodel 2–3 = 0.02; see Table 2).

Descriptive statistics

Table 3 displays descriptive statistics and Table 4 depicts educational level differences in

the mean structure of the factors. The results showed that students in grades 7–8, as compared with students in grades 5–6, reported significantly lower self-efficacy for

conventions in opinion texts, more negative attitudes towards writing, and performing

1464 ANA CAMACHO et al.

T a b le

2 . M u lt ip le -g ro u p m e as u re m e n t in va ri an ce

te st in g m o ti va ti o n al an d b e h av io u ra ls ca le s

Sc al e

O ve ra ll re su lt s

C o m p ar e d m o d e ls

M o d e ld iff e re n ce

re su lt s

SB v2

df p

C FI

R M SE A

SR M R

D SB

v 2

D df

p D C FI

D R M SE A

D SR

M R

Se lf- e ffi ca cy

fo r w ri ti n g n ar ra ti ve

te x ts

C o n fi gu ra li n va ri an ce

3 7 1 .0 6 1

2 0 2

.0 0 0

0 .9 3 4

0 .0 6 7

0 .0 5 2

- –

– –

– –

– W

e ak

in va ri an ce

3 8 6 .5 5 9

2 1 5

.0 0 0

0 .9 3 4

0 .0 6 5

0 .0 5 8

M o d e l1

vs .m

o d e l2

1 5 .5 0

1 3

.2 7 7

0 .0 0 0

0 .0 0 2

0 .0 0 6

St ro n g in va ri an ce

4 0 7 .9 5 7

2 2 8

.0 0 0

0 .9 3 3

0 .0 6 4

0 .0 5 9

M o d e l2

vs .m

o d e l3

2 1 .4 0

1 3

.0 6 5

0 .0 0 1

0 .0 0 1

0 .0 0 1

Se lf- e ffi ca cy

fo r w ri ti n g o p in io n te x ts

C o n fi gu ra li n va ri an ce

3 6 0 .8 1 4

2 0 2

.0 0 0

0 .9 4 4

0 .0 6 8

0 .0 5 1

- –

– –

– –

– W

e ak

in va ri an ce

3 8 2 .5 9 1

2 1 5

.0 0 0

0 .9 4 1

0 .0 6 7

0 .0 6 4

M o d e l1

vs .m

o d e l2

2 1 .7 8

1 3

.0 5 9

0 .0 0 3

0 .0 0 1

0 .0 1 3

St ro n g in va ri an ce

4 1 9 .5 9 5

2 2 8

.0 0 0

0 .9 3 5

0 .0 6 9

0 .0 6 6

M o d e l2

vs .m

o d e l3

3 7 .0 0

1 3

.0 0 0

0 .0 0 6

0 .0 0 2

0 .0 0 2

A tt it u d e s to w ar d s w ri ti n g

C o n fi gu ra li n va ri an ce

4 0 .5 7 7

1 0

.0 0 0

0 .9 7 0

0 .1 2 2

0 .0 3 1

- –

– –

– –

– W

e ak

in va ri an ce

4 7 .3 0 2

1 4

.0 0 0

0 .9 7 0

0 .1 0 3

0 .0 3 9

M o d e l1

vs .m

o d e l2

6 .7 3

4 .1 5 1

0 .0 0 0

0 .0 1 9

0 .0 0 8

St ro n g in va ri an ce

6 1 .1 8 4

1 8

.0 0 0

0 .9 6 4

0 .1 0 1

0 .0 4 7

M o d e l2

vs .m

o d e l3

1 3 .8 8

4 .0 0 8

0 .0 0 6

0 .0 0 2

0 .0 0 8

W ri ti n g fr e q u e n cy

C o n fi gu ra li n va ri an ce

7 6 .2 4 8

2 6

.0 0 0

0 .9 1 1

0 .0 9 2

0 .0 5 7

- –

– –

– –

– W

e ak

in va ri an ce

8 6 .9 4 4

3 1

.0 0 0

0 .9 0 2

0 .0 8 8

0 .0 6 2

M o d e l1

vs .m

o d e l2

1 0 .7 0

5 .0 5 7

0 .0 0 9

0 .0 0 4

0 .0 0 5

St ro n g in va ri an ce

1 0 2 .9 7 9

3 6

.0 0 0

0 .8 8 0

0 .0 9 1

0 .0 6 8

M o d e l2

vs .m

o d e l3

1 6 .0 4

5 .0 0 7

0 .0 2 2

0 .0 0 3

0 .0 0 6

Relations among motivation, behaviour, and performance in writing 1465

less often literary writing activities (all ps < .05). Students in grades 7–8 reported carrying out more frequently digital writing activities than students in grades 5–6 (p < .001).

Table 4. Educational level differences in the structure of the factors for self-efficacy forwriting, attitudes

towards writing, and writing frequency

Variable Mean factor score SE p Standardized factor score

Narrative texts

Self-efficacy for ideation �0.34 0.25 .170 �0.20

Self-efficacy for conventions �0.23 0.23 .331 �0.16

Self-efficacy for self-regulation �0.19 0.24 .423 �0.12

Attitudes �0.32 0.09 .000*** �0.40

Literary writing frequency �0.27 0.06 .000*** �0.60

Digital writing frequency 0.44 0.10 .000*** 0.66

Opinion texts

Self-efficacy for ideation �0.27 0.26 0.29 �0.16

elf-efficacy for conventions �0.45 0.21 0.03* �0.30

Self-efficacy for self-regulation �0.36 0.27 0.18 �0.20

Attitudes �0.32 0.09 .000*** �0.41

Literary writing frequency �0.28 0.06 .000*** �0.62

Digital writing frequency 0.47 0.11 .000*** 0.73

Note. Students in grades 5–6 were the reference category.

*p < .05. ***p < .001.

Table 3. Descriptive statistics concerning motivational, behavioural, and performance variables

Variable

Second cycle

Grades 5–6 (n = 342)

Third cycle

Grades 7–8 (n = 263)

All students

(N = 605)

M SD M SD M SD

Motivational variables

Self-efficacy for narrative texts

Self-efficacy for ideation 71.90 19.42 68.36 18.33 70.36 19.02

Self-efficacy for conventions 72.39 18.68 69.67 17.20 71.20 18.09

Self-efficacy for self-regulation 69.93 20.70 67.65 19.12 68.94 20.05

Self-efficacy for opinion texts

Self-efficacy for ideation 72.88 21.16 70.45 18.74 71.82 20.16

Self-efficacy for conventions 75.18 17.51 70.38 117.76 73.09 17.77

Self-efficacy for self-regulation 72.01 20.67 68.74 20.55 70.59 20.67

Attitudes 3.43 0.87 3.12 0.87 3.29 0.89

Behavioural variables

Literary writing frequency 2.16 0.83 1.74 0.64 1.98 0.78

Digital writing frequency 4.06 1.20 4.54 0.86 4.27 1.09

Writing performance

Narrative text quality 3.93 1.17 4.11 1.29 4.01 1.22

Opinion text quality 3.36 1.30 3.47 1.41 3.41 1.35

Covariate

Handwriting fluency 43.62 18.83 61.56 21.89 51.45 22.08

1466 ANA CAMACHO et al.

Main analyses

MG-SEM examining educational level differences in narrative texts

MG-SEM models. The results showed that there were no significant educational level

differences betweenbothMG-SEMmodels, SBv2(15) = 8.87,p = .884. This indicates that

no significant differences emerged in the regression coefficients between students in

grades 5–6 and students in grades 7–8. However, we put forth an adaption of Model 2 as

small differences could have been invisible in specific regressions since we estimated all

regression coefficients simultaneously.

We adapted Model 2 by allowing one specific regression to vary across groups to

further examine educational level differences. The difference between the log-likelihood values associated with both models has approximately a chi-square distribution with one

degree of freedom, subject to the scaling correction factors of the two models (Satorra &

Bentler, 2001). Therefore, we tested each specific regression in the MG-SEM model to

examine significant educational level differences in the regression coefficients. The

results showed a significant educational level difference in the regression between

digital writing frequency and narrative text quality, SB v2(1) = 4.84, p = .028 (see

Table 5).

As a result, we decided to proceed with a final MG-SEM model in which: (1) all intercepts, factor loadings, and regressions were set to be equal across educational levels;

and (2) the specific regression betweendigitalwriting frequency andnarrative text quality

was set to vary across educational levels. The final model showed an acceptable fit to the

data, SB v2(831) = 1,336.33, p < .001, CFI = 0.91, RMSEA = 0.05, SRMR = 0.07. This

model respectively accounted for 21% and 22% of the variance of narrative text quality in

grades 5–6 and grades 7–8 (see Figure 2).

Relations between motivational and behavioural variables. Attitudes were signifi-

cantly related to literary writing frequency (b = .43, p < .001) and digital writing

frequency (b = .10, p < .05). None of the self-efficacy factors were significantly related to

writing frequency (all ps > .05).

Relations between motivational variables and writing performance. Attitudes made

an independent contribution to narrative text quality (b = .21, p < .01). Of the self- efficacy factors, only self-efficacy for self-regulation was significantly associated with text

quality (b = .33, p < .05).

Relations between behavioural variable and writing performance. Digital writing

frequencywas significantly related to text quality (b = .18, p < .05) for students in grades

7–8. Sobel tests (Sobel, 1982) further indicated that digital writing frequency did not

mediate the relation between any of the motivational factors and writing performance in grades 7–8 (all ps > .05). By contrast, the relation between digital writing and text quality

was not significant for students in grades 5–6 (p = .959). Literary writing frequency did

not make a significant contribution to text quality (p = .087) in any of the educational

levels.

Relations among motivation, behaviour, and performance in writing 1467

T a b le

5 . M u lt ip le -g ro u p st ru ct u ra l e q u at io n m o d e lin g: co m p ar is o n o f d iff e re n t m o d e ls ac ro ss

e d u ca ti o n al le ve ls fo r n ar ra ti ve

te x t

M o d e l

SB v 2

df C o m p ar e d m o d e ls

D SB

v 2

D df

p

1 a

1 ,3 3 2 .3 0

8 1 7

� �

� �

2 b

1 ,3 4 1 .1 7

8 3 2

M o d e l1

vs . M o d e l2

8 .8 7

1 5

.8 8 4

A d ap ti o n s o f m o d e l2 :A

llo w in g o n e sp e ci fi c re gr e ss io n to

va ry

ac ro ss

e d u ca ti o n al cy cl e s

W ri ti n g M o ti va ti o n ?

W ri ti n g Fr e q u e n cy

Se lf- e ffi ca cy

fo r id e at io n ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 4 0 .9 0

8 3 1

vs . m o d e l2

0 .2 7

1 .6 0 6

Se lf- e ffi ca cy

fo r co n ve n ti o n s ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 4 0 .8 8

8 3 1

vs . m o d e l2

0 .2 9

1 .5 9 1

Se lf- e ffi ca cy

fo r se lf- re gu la ti o n ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 4 0 .8 7

8 3 1

vs . m o d e l2

0 .3 0

1 .5 8 1

A tt it u d e s ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 4 0 .5 5

8 3 1

vs . m o d e l2

0 .6 2

1 .4 3 0

Se lf- e ffi ca cy

fo r id e at io n ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 4 0 .8 7

8 3 1

vs . m o d e l2

0 .3 1

1 .5 8 1

Se lf- e ffi ca cy

fo r co n ve n ti o n s ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 4 1 .1 6

8 3 1

vs . m o d e l2

0 .0 1

1 .9 1 6

Se lf- e ffi ca cy

fo r se lf- re gu la ti o n ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 4 1 .0 6

8 3 1

vs . m o d e l2

0 .1 2

1 .7 3 5

A tt it u d e s ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 4 0 .5 5

8 3 1

vs . m o d e l2

0 .6 2

1 .4 3 0

W ri ti n g m o ti va ti o n ?

W ri ti n g P e rf o rm

an ce

Se lf- e ffi ca cy

fo r id e at io n ?

N ar ra ti ve

te x t q u al it y

1 ,3 3 9 .6 6

8 3 1

vs . m o d e l2

1 .5 1

1 .2 1 9

Se lf- e ffi ca cy

fo r co n ve n ti o n s ?

N ar ra ti ve

te x t q u al it y

1 ,3 4 0 .3 7

8 3 1

vs . m o d e l2

0 .8 0

1 .3 7 0

Se lf- e ffi ca cy

fo r se lf- re gu la ti o n ?

N ar ra ti ve

te x t q u al it y

1 ,3 4 0 .5 5

8 3 1

vs . m o d e l2

0 .6 2 1

1 .4 3 1

A tt it u d e s ?

N ar ra ti ve

te x t q u al it y

1 ,3 4 1 .8 0

8 3 1

vs . m o d e l2

0 .6 2 5

1 .4 2 9

W ri ti n g Fr e q u e n cy

? W

ri ti n g P e rf o rm

an ce

L it e ra ry

w ri ti n g fr e q u e n cy

? N ar ra ti ve

te x t q u al it y

1 ,3 4 1 .4 1

8 3 1

vs . m o d e l2

0 .2 4

1 .6 2 4

D ig it al w ri ti n g fr e q u e n cy

? N ar ra ti ve

te x t q u al it y

1 ,3 3 6 .3 3

8 3 1

vs . m o d e l2

4 .8 4

1 .0 2 8

C o va ri at e ?

W ri ti n g P e rf o rm

an ce

H an d w ri ti n g fl u e n cy

? N ar ra ti ve

te x t q u al it y

1 ,3 4 1 .7 9 9

8 3 1

vs . m o d e l2

0 .6 2 1

1 .4 3 1

a E q u al fa ct o r lo ad in gs

an d e q u al in te rc e p ts ac ro ss

e d u ca ti o n al le ve ls .; b E q u al fa ct o r lo ad in gs , e q u al in te rc e p ts , an d e q u al re gr e ss io n co e ffi ci e n ts ac ro ss

e d u ca ti o n al

le ve ls .

1468 ANA CAMACHO et al.

Covariate of writing performance. Handwriting fluency significantly contributed to

text quality (b = .18, p < .001).

MG-SEM examining educational level differences in opinion texts

MG-SEM models. The results indicated no significant educational level differences

between both MG-SEM models, SB v2(15) = 5.99, p = .980. We followed the same procedure put forth for narrative texts and adapted Model 2 by allowing one specific

regression to vary across educational levels. These analyses showed no significant

educational level differences in the regression coefficients (see Table 6). Therefore, we

decided to proceed with Model 2 (i.e., equal intercepts, factor loadings, and regressions

across educational levels) as the final model for grades 5–6 and 7–8, which showed an

acceptable fit to the data, SB v2(832) = 1,304.20, p < .001, CFI = 0.92, RMSEA = .05,

SRMR = .07. The final model respectively accounted for 16% and 15% of the variance of

opinion text quality in grades 5–6 and grades 7–8 (see Figure 3).

Self-efficacy for idea�on

Mo�va�onal component

Behavioural component

Wri�ng performance

Narra�ve text quality

Self-efficacy for conven�ons

Self-efficacy for self-regula�on

A�tudes

Literary wri�ng frequency

Digital wri�ng frequency

Handwri�ng fluency

R2grades 5-6 = .21 R2grades 7-8 = .22

R 2

grades 5-6 = .22

R 2

grades 7-8 = .36

.10*

.33*

.43***

R 2

grades 5-6 = .02

R 2

grades 7-8 = .04

.21** .18***

.18*

Figure 2. Significant standardized parameter estimates of the structural model for narrative texts.Note.

The dotted lines indicate a significant difference in the regression between students in grades 5–6 and

students in grades 7–8. ***p < .001; **p < .01; *p < .05.

Relations among motivation, behaviour, and performance in writing 1469

T a b le

6 . M u lt ip le -g ro u p st ru ct u ra l e q u at io n m o d e lin g: co m p ar is o n o f d iff e re n t m o d e ls ac ro ss

e d u ca ti o n al le ve ls fo r o p in io n te x t

M o d e l

SB v 2

df C o m p ar e d m o d e ls

D SB

v 2

D df

p

1 a

1 ,2 9 8 .2 1 1

8 1 7

– –

– –

2 b

1 ,3 0 4 .2 0 0

8 3 2

M o d e l1

vs .M

o d e l2

5 .9 9

1 5

.9 8 0

A d ap ti o n s o f m o d e l2 : A llo w in g o n e sp e ci fi c re gr e ss io n to

va ry

ac ro ss

e d u ca ti o n al cy cl e s

W ri ti n g M o ti va ti o n ?

W ri ti n g Fr e q u e n cy

Se lf- e ffi ca cy

fo r id e at io n ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 0 3 .1 0 1

8 3 1

vs .m

o d e l2

1 .1 0

1 .2 9 4

Se lf- e ffi ca cy

fo r co n ve n ti o n s ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 0 3 .3 7 7

8 3 1

vs .m

o d e l2

0 .8 2

1 .3 6 4

Se lf- e ffi ca cy

fo r se lf- re gu la ti o n ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 0 3 .2 9 2

8 3 1

vs .m

o d e l2

0 .9 1

1 .3 4 1

A tt it u d e s ?

L it e ra ry

w ri ti n g fr e q u e n cy

1 ,3 0 3 .5 0 0

8 3 1

vs .m

o d e l2

0 .7 0

1 .4 0 3

Se lf- e ffi ca cy

fo r id e at io n ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 0 4 .1 6

8 3 1

vs .m

o d e l2

0 .0 4

1 .8 4 1

Se lf- e ffi ca cy

fo r co n ve n ti o n s ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 0 4 .2 3

8 3 1

vs .m

o d e l2

0 .0 3

1 .8 6 9

Se lf- e ffi ca cy

fo r se lf- re gu la ti o n ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 0 4 .1 6

8 3 1

vs .m

o d e l2

0 .0 4

1 .8 4 0

A tt it u d e s ?

D ig it al w ri ti n g fr e q u e n cy

1 ,3 0 3 .6 6

8 3 1

vs .m

o d e l2

0 .5 4

1 .4 6 4

W ri ti n g m o ti va ti o n ?

W ri ti n g P e rf o rm

an ce

Se lf- e ffi ca cy

fo r id e at io n ?

O p in io n te x t q u al it y

1 ,3 0 3 .9 5

8 3 1

vs .m

o d e l2

0 .2 5

1 .6 1 6

Se lf- e ffi ca cy

fo r co n ve n ti o n s ?

O p in io n te x t q u al it y

1 ,3 0 3 .4 7

8 3 1

vs .m

o d e l2

0 .7 3

1 .3 9 2

Se lf- e ffi ca cy

fo r se lf- re gu la ti o n ?

O p in io n te x t q u al it y

1 ,3 0 3 .8 0

8 3 1

vs .m

o d e l2

0 .4 0

1 .5 2 7

A tt it u d e s ?

O p in io n te x t q u al it y

1 ,3 0 3 .9 3

8 3 1

vs .m

o d e l2

0 .2 7

1 .6 0 5

W ri ti n g Fr e q u e n cy

? W

ri ti n g P e rf o rm

an ce

L it e ra ry

w ri ti n g fr e q u e n cy

? O p in io n te x t q u al it y

1 ,3 0 4 .3 0

8 3 1

vs .m

o d e l2

0 .1 0

1 7 5 3

D ig it al w ri ti n g fr e q u e n cy

? O p in io n te x t q u al it y

1 ,3 0 3 .1 0

8 3 1

vs .m

o d e l2

1 .1 0

1 .2 9 4

C o va ri at e ?

W ri ti n g P e rf o rm

an ce

H an d w ri ti n g fl u e n cy

? O p in io n te x t q u al it y

1 ,3 0 7 .1 5

8 3 1

vs .m

o d e l2

2 .9 5

1 .0 8 6

a E q u al fa ct o r lo ad in gs

an d e q u al in te rc e p ts ac ro ss

e d u ca ti o n al le ve ls .; b E q u al fa ct o r lo ad in gs , e q u al in te rc e p ts , an d e q u al re gr e ss io n co e ffi ci e n ts ac ro ss

e d u ca ti o n al

le ve ls .

1470 ANA CAMACHO et al.

Relations between motivational and behavioural variables. Attitudes made signifi-

cant contributions to both literary writing frequency (b = .53, p < .001) and digital

writing frequency (b = .13, p < .05). Of the self-efficacy factors, only self-efficacy for

ideation was significantly associated with literary writing frequency (b = .41, p < .05).

Relations between motivational variables and writing performance. Attitudes

significantly contributed to opinion text quality (b = .28, p < .001). None of the self-

efficacy factors were significantly related to text quality (all ps > .05).

Relations betweenbehavioural variable andwriting performance. Neither of the two writing frequency factors made a significant contribution to text quality (both ps > .05).

***p < .001; *p < .05.

Self-efficacy for idea�on

Mo�va�onal component

Behavioural component

Wri�ng performance

Opinion text quality

Self-efficacy for conven�ons

Self-efficacy for self-regula�on

A�tudes

Literary wri�ng frequency

Digital wri�ng frequency

R 2

grades 5-6 = .16

R 2

grades 7-8 = .15

R 2

grades 5-6 = .25

R 2

grades 7-8 = .38

R 2

grades 5-6 = .01

R 2

grades 7-8 = .02

.53***

.13*

.28***

Handwri�ng fluency

.20***

.41*

Figure 3. Significant standardized parameter estimates of the structural model for opinion texts.

***p < .001; *p < .05.

Relations among motivation, behaviour, and performance in writing 1471

Covariate of writing performance. Handwriting fluency significantly contributed to

text quality (b = .20, p < .001).

Discussion

In this study, we examined the structural relations among self-efficacy, attitudes, writing

frequency, and text quality, while controlling for handwriting fluency. Simultaneously,

we inspected whether these relations varied across two text genres and across two

educational levels. Our hypothesized relational model was partially confirmed.

Relations between motivational and behavioural variables

In linewith ourHypothesis 1, writing attitudes and self-confidence in generating ideas for

opinion texts directly contributed to how frequently students write. These results are

consistent with the WWC model, which posits that one’s beliefs and beliefs about the

writing task affect whether one engages in writing (Graham, 2018). These results also

extend the findings of Troia et al. (2013) by showing that attitudes contribute directly to

writing frequency. In addition, we found that the significant, positive relations among motivation and reading frequency (e.g., De Naeghel et al., 2012) were also verified in the

writing domain.

Relations between motivational variables and writing performance

Our Hypothesis 2 was only partially confirmed. Of the self-efficacy factors, only self-

efficacy for self-regulation made a significant contribution to narrative text quality across

educational levels. This result corroborates evidence indicating that self-efficacy for self- regulation significantly correlates with writing grades and performance in state-wide

writing assessments (Bruning et al., 2013). Interestingly, perceived competence in more

cognitive and linguistic skills (i.e., self-efficacy for ideation and for conventions) was not

associated with writing performance. Considering the automatization of transcription

skills and the progress in writing fluency throughout the development (e.g., Alves &

Limpo, 2015), students enrolled in our study –middle school students – probably did not

worry as much with the generation of ideas and the compliance with conventions as they

did with the regulation of their behaviour and affect while writing (e.g., avoiding distractions, dealing with frustration, persisting in the face of difficult writing tasks;

Bruning et al., 2013).

Unexpectedly, the significant relation between self-efficacy for self-regulation and text

quality was not observed for opinion texts. This result does not concur with previous

research, which showed that self-efficacy for self-regulation was significantly related to

opinion text quality (Limpo & Alves, 2017). A possible explanation for this result is that

students may have experienced difficulties in determining their self-confidence for

writing opinion texts, since it is a genre with which they are less familiar. Students form their self-efficacy beliefs based on the interpretation of information from four sources:

prior performance, social persuasion, vicarious experience, and physiological states

(Bandura, 1997; Pajares, 2003). If students enrolled in our study had limited experience

writing opinion texts (as opposed to narrative texts), they possibly had less information to

form accurate self-efficacy beliefs about opinion writing. Indeed, knowledge and

strategies for opinion writing are usually acquired later when compared with narrative

1472 ANA CAMACHO et al.

writing (e.g., Olive, Favart, Beauvais, & Beauvais, 2009). In the Portuguese educational

context, the narrative text is introduced long before the opinion text (Buescu, Morais,

Rocha, &Magalh~aes, 2015). In addition, in line with the narrative text model, self-efficacy

for ideation and for conventions did not contribute to opinion text quality. Consistent with Hypothesis 3, attitudes emerged as a cornerstone factor contributing

to text quality across text genres and across educational levels. This result concurs with

previous research showing that attitudes are related to text quality (Camacho et al., 2021;

Camacho, Silva, Silva, Santos, Jacques, & Alves, 2020; Ekholm et al., 2018; Graham et al.,

2007). The leading role of writing attitudes in our models highlights the need to nurture

positivewriting attitudes in the classroom. In this regard, the influential article by Bruning

and Horn (2000) emphasizes, for example, the importance of teachers developing

students’ positive beliefs about the writing task, which implies positive writing attitudes.

Relation between behavioural variable and writing performance

Hypothesis 4 was only partially confirmed. Of the two writing frequency factors, only

digital writing frequency made a significant contribution to narrative text quality for

students in grades 7–8. The same relation was, however, not found in grades 5–6. These disparate results may be partially explained by the possibility that older students were

more experienced in using digital technologies to write when compared to younger students (European Commission/EACEA/Eurydice, 2019). Additionally, this finding

suggests a promising role of digital writing frequency for writing performance, which is

consistent with studies reporting the benefits of digital tools for the writing process (e.g.,

Camacho et al., 2021; Karchmer-Klein, 2019; Yamac�, €Ozt€urk, & Mutlu, 2020).

Both literary and digital writing frequency factors did not show a significant relation

with opinion text quality for both younger and older students nor with narrative text

quality for younger students. This result does not imply that frequency of writing is not

important for improving text quality across grade levels. In fact, prior evidence shows that themore frequently studentswrite, the better theywill perform inwriting (Graham, 2019;

Graham&Harris, 2016; Graham, McKeown, McKeown, Kiuhara, & Harris, 2012; Graham

& Perin, 2007; Hillocks, 1986).

In our study, the use of a limited self-reportwriting frequencymeasuremay explain the

lack of significant relations between writing frequency and text quality (except for the

significant, positive relation between digital writing frequency and narrative text quality

for older students). Thewriting frequencymeasureweused tapped a restricted number of

writing activities, while the frequency of other important writing activities was not assessed (e.g., writing expository texts, writing syntheses, writing in response to reading,

blogging). Therefore, a comprehensive writing frequency measure that covers a wide

range of digital and non-digital writing activities in both academic and recreative contexts

is highly needed to further examine the relations among motivation, writing frequency,

and text quality. In the future, writing researchers can partially overcome the

shortcomings of self-report measures with the use of alternative research methods – such as diary studies.

Covariate of writing performance

As predicted in Hypothesis 5, we found a significant contribution of handwriting fluency

to narrative and opinion texts across educational levels, which aligns with previous

Relations among motivation, behaviour, and performance in writing 1473

research (Alves & Limpo, 2015; Berninger & Swanson, 1994; Jones & Christensen, 1999;

Kent & Wanzek, 2016).

Limitations

Thisstudyhassomelimitations.First,weadoptedacross-sectionaldesign,whichprecludes

us fromdrawingconclusions aboutcausality. Second,we left out otherpotentially relevant

motivational constructs that contribute to writing frequency and performance (e.g.,

interest and autonomous motivation). Third, we relied on a self-report scale to assess

writing frequency. Nevertheless, this scale was shown to be valid and reliable.

Educational implications

At least three implications for writing instruction may stem from this study. First and

foremost, teachers need to be mindful about the critical role of motivational variables in

writing proficiency (Camacho et al., 2021; De Smedt, 2019; Graham, 2018). Specifically,

teachers need to nurture positive attitudes towards writing in the classroom as these

contribute directly to how well lower and upper grade students perform across text

genres. At the same time, teachers should be aware of students’ self-efficacy in regulating

behaviours and feelings while writing as this factor was found to contribute directly to narrative text quality. Thus, teachers should promote the use of self-regulation strategies

in writing (Harris & Graham, 1992, 2017), which might boost students’ self-efficacy for

this aspect.

Second, teachers need to provide many opportunities for students to write and

simultaneously make use of these opportunities to intentionally promote students’

writing-related motivation, skills, knowledge, and strategies (Graham, 2018, 2019;

Graham & Harris, 2016, 2019). This seems especially important in the case of opinion

texts. Third, our study showed a specific contribution of digital writing frequency to narrative text quality in seventh and eighth graders. Therefore, teachers may consider the

integration of paper-based with digital writing tools into the classroom.

These educational implications need to be interpreted considering that our study

adopted a correlational research design. Longitudinal research is highly needed to draw

causal relations between the studied variables. Additionally, true experimental research is

necessary to identify whether instructional practices prompting stronger self-efficacy

beliefs for writing, more positive writing attitudes, and more writing frequency result in

better writing performance (Camacho, Silva, et al., 2020).

Directions for future research

This study opens avenues for future research. Concerning motivational measures, future

studies could use the multidimensional conceptualization of writing attitudes (i.e.,

attitudes towards paper-based and digital writing in academic and recreational contexts)

applied by Graham et al. (2018). Regarding behavioural measures, researchers could

examine how both writing frequency (i.e., a matter of quantity) and writing engagement (i.e., a matter of quality) contribute to writing performance. Researchers need also to

address how the frequency of other types of literary writing (e.g., writing informative

texts,writing syntheses,writing in response to reading) and digital writing (e.g., blogging,

wikis, word processing software; Karchmer-Klein, 2019) contribute to writing perfor-

mance.

1474 ANA CAMACHO et al.

In addition, researchers could examine the relations among motivation, behaviour,

and performance across other educational levels (e.g., primary and secondary school) and

text genres (e.g., descriptive and informational texts). Longitudinal research is also

needed to observe how motivational, behavioural, and writing performance factors are inter-related over the school years. Finally, experimental and quasi-experimental studies

are warranted to determine which instructional practices promote stronger self-efficacy

beliefs and more positive attitudes towards writing.

Conclusion

In this study, we delved into the relations of motivational and behavioural components to

writing performance across two text genres and two educational levels. A positive contribution of attitudes to both writing frequency and performance was verified across

text genres and educational levels. A positive contribution of self-efficacy for self-

regulation across educational levels was found for narrative writing performance. These

findings suggest that teachers need to adopt motivation-enhancing practices in writing

instruction (Bruning & Horn, 2000) across grade levels. Our results, however, did not

reveal a significant role of writing frequency in writing performance, except for the

specific contribution of digital writing frequency for narrative texts in older students.

Thus, further research is warranted to clarify how several types of writing frequency in digital and non-digital contexts (i.e., a matter of quantity) as well as writing engagement

(i.e., a matter of quality) contribute to students’ writing performance.

Acknowledgements

This work was supported by a grant attributed to the first author from the Portuguese

Foundation for Science and Technology (grant SFRH/BD/116281/2016). The authors thank

Mariana Silva and Susana Santos for their assistance in data collection. The authors also thank all

teachers and students involved in this study.

Conflicts of interest

All authors declare no conflict of interest.

Author contributions

Ana Camacho (Conceptualization; Data curation; Formal analysis; Investigation; Method-

ology; Writing – original draft) Rui A. Alves (Conceptualization; Supervision; Writing – review & editing) Fien De Smedt (Formal analysis; Writing – review & editing) Hilde Van

Keer (Formal analysis; Writing – review & editing) Pietro Boscolo (Supervision; Writing – review & editing).

Data availability statement

Thedataset supporting the results of this study is available from thecorresponding author upon

request.

Relations among motivation, behaviour, and performance in writing 1475

References

Abrantes, P. (2005). As transic�~oes entre ciclos de ensino: Entre problema social e objecto sociol�ogico [Transitions between educational cycles: Between the social problem and the sociological

object]. Interacc�~oes, 1.25–53. https://doi.org/10.25755/int.281 Alves, R. A. (2019). The early steps in becoming awriter: Enabling participation in a literateworld. In

J. S. Horst & J. V. K. Torkildsen (Eds.), International handbook of language acquisition (pp.

567–590). London, UK: Routledge. Alves, R. A., & Limpo, T. (2015). Progress in written language bursts, pauses, transcription, and

written composition across schooling. Scientific Studies of Reading, 19, 374–391. https://doi. org/10.1080/10888438.2015.1059838

Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122– 147. https://doi.org/10.1037/0003-066X.37.2.122

Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY: Freeman.

Bereiter, C., & Scardamalia, M. (1987). . The psychology of written composition, Hillsdale, NJ:

Lawrence Erlbaum Associates. https://doi.org/10.4324/9780203812310

Berninger, V. W., Mizokawa, D. T., & Bragg, R. (1991). Theory-based diagnosis and remediation of

writing disabilities. Journal of School Psychology, 29, 57–79. https://doi.org/10.1016/0022- 4405(91)90016-K

Berninger, V.W., & Swanson, H. L. (1994). Modifying Hayes and Flower’smodel of skilledwriting to

explain beginning and developing writing. In E. Butterfield (Ed.), Children’s writing: Toward a

process theory of development of skilled writing (pp. 57–81). Greenwich, CT: JAI Press.

Boscolo, P., & Hidi, S. (2007). The multiple meanings of motivation to write. In S. Hidi & P. Boscolo

(Eds.), Writing and motivation (pp. 1–14). Amsterdam, The Netherlands: Elsevier.

Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit. Sociological Methods

& Research, 21, 230–258. https://doi.org/10.1177/0049124192021002005 Bruning, R., Dempsey, M., Kauffman, D. F., McKim, C., & Zumbrunn, S. (2013). Examining

dimensions of self-efficacy for writing. Journal of Educational Psychology, 105(1), 25–38. https://doi.org/10.1037/a0029692

Bruning, R., & Horn, C. (2000). Developing motivation to write. Educational Psychologist, 35(1),

25–37. https://doi.org/10.1207/S15326985EP3501_4 Buescu, H. C., Morais, J., Rocha,M. R., &Magalh~aes, V. F. (2015).Programa emetas curriculares de

português do ensino b�asico. Lisboa: Minist�erio da Educac�~ao e Ciência. Retrieved June 15, 2020, from https://www.dge.mec.pt/sites/default/files/Basico/Metas/Portugues/pmcpeb_julho_

2015.pdf

Camacho, A., Alves, R. A., & Boscolo, P. (2021).Writingmotivation in school: A systematic review of

empirical research in the early twenty-first century. Educational Psychology Review, 33(1),

213–247. https://doi.org/10.1007/s10648-020-09530-4 Camacho, A., Silva, M., Santos, M., Jacques, T., & Alves, R. A. (2020). Atitudes face �a escrita em

adolescentes portugueses [Attitudes towards writing of Portuguese adolescents]. Indagatio

Didactica, 12(2), 143–157. https://doi.org/10.34624/id.v12i2.17466 Cheung,G.W.,&Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testingmeasurement

invariance. Structural Equation Modeling: A Multidisciplinary Journal, 9, 233–255. https:// doi.org/10.1207/S15328007SEM0902_5

Chou, C.-P., Bentler, P. M., & Satorra, A. (1991). Scaled test statistics and robust standard errors for

non-normal data in covariance structure analysis: A Monte Carlo study. British Journal of

Mathematical and Statistical Psychology, 44, 347–357. https://doi.org/10.1111/j.2044-8317. 1991.tb00966.x

Cooper, C. R. (1977). Holistic evaluation of writing. In C. R. Cooper, & L. Odell (Eds.), Evaluating

Writing: Describing, measuring, judging (Vol. 3–31). Urbana IL: National Council of Teachers of English.

De Naeghel, J., Keer, H. V., Vansteenkiste, M., & Rosseel, Y. (2012). The relation between

elementary students’ recreational and academic reading motivation, reading frequency,

1476 ANA CAMACHO et al.

engagement, and comprehension: A self-determination theory perspective. Journal of

Educational Psychology, 104, 1006–1021. https://doi.org/10.1037/a0027800 De Smedt, F. (2019). Cognitive and motivational challenges in writing: The impact of explicit

instruction and peer-assisted writing in upper-elementary grades (Unpublished doctoral

dissertation). Ghent, Belgium: University of Ghent.

De Smedt, F., Rogiers, A., Heirweg, S., Merchie, E., & Van Keer, H. (2020). Assessing and mapping

reading andwritingmotivation in third to eight graders: A self-determination theory perspective

[Original Research]. Frontiers in Psychology, 11, 1678. https://doi.org/10.3389/fpsyg.2020.

01678

De Smedt, F., Van Keer, H., & Merchie, E. (2016). Student, teacher and class-level correlates of

Flemish late elementary school children’s writing performance. Reading and Writing: An

Interdisciplinary Journal, 29, 833–868. https://doi.org/10.1007/s11145-015-9590-z Ekholm, E., Zumbrunn, S., & DeBusk-Lane, M. (2018). Clarifying an elusive construct: A systematic

review of writing attitudes. Educational Psychology Review, 30, 827–856. https://doi.org/10. 1007/s10648-017-9423-5

El-Khechen, W., Ferdinand, H. D., Steinmayr, R., & McElvany, N. (2016). Language-related values,

reading amount, and reading comprehension in students with migration backgrounds. British

Journal of Educational Psychology, 86, 256–277. https://doi.org/10.1111/bjep.12102 European Commission/EACEA/Eurydice (2019). Digital education at school in Europe. Eurydice

Report. Luxembourg: Publications Office of the European Union. Retrieved June 15, 2020, from

https://eacea.ec.europa.eu/national-policies/eurydice/sites/default/files/en_digital_educa

tion_n.pdf

Feng, L., Lindner, A., Ji, X. R., &Malatesha Joshi, R. (2019). The roles of handwriting and keyboarding

in writing: A meta-analytic review. Reading and Writing, 32(1), 33–63. https://doi.org/10. 1007/s11145-017-9749-x

Graham, S. (2018). A revised writer(s)-within-community model of writing. Educational

Psychologist, 53, 258–279. https://doi.org/10.1080/00461520.2018.1481406 Graham, S. (2019). Changing howwriting is taught. Review of Research in Education, 43(1), 277–

303. https://doi.org/10.3102/0091732X18821125

Graham, S., Berninger, V., & Abbott, R. (2012). Are attitudes toward writing and reading separable

constructs? A study with primary grade children. Reading & Writing Quarterly, 28(1), 51–69. https://doi.org/10.1080/10573569.2012.632732

Graham, S., Berninger, V., & Fan, W. (2007). The structural relationship between writing attitude

and writing achievement in first and third grade students. Contemporary Educational

Psychology, 32, 516–536. https://doi.org/10.1016/j.cedpsych.2007.01.002 Graham, S., Daley, S. G., Aitken, A. A., Harris, K. R., & Robinson, K. H. (2018). Do writing

motivational beliefs predict middle school students’ writing performance? Journal of Research

in Reading, 41, 642–656. https://doi.org/10.1111/1467-9817.12245 Graham, S., & Harris, K. R. (2016). A path to better writing: Evidence-based practices in the

classroom. The Reading Teacher, 69(4), 359–365. https://doi.org/10.1002/trtr.1432 Graham, S.,&Harris, K. R. (2019). Evidence-basedpractices inwriting. In S.Graham,C.MacArthur&

M. Hebert (Eds.), Best practices in writing instruction (pp. 3–28). New York, NY: The Guilford

Press.

Graham, S., Harris, K. R., Fishman, E., Houston, J., Wijekumar, K., Lei, P.-W., & Ray, A. B. (2019).

Writing skills, knowledge, motivation, and strategic behavior predict students’ persuasive

writing performance in the context of robust writing instruction. The Elementary School

Journal, 119, 487–510. https://doi.org/10.1086/701720 Graham, S., Harris, K. R., & Hebert, M. (2011). It is more than just the message: Analysis of

presentation effects in scoring writing. Focus on Exceptional Children, 44, 1–12. https://doi. org/10.17161/fec.v44i4.6687

Graham, S., Harris, K. R., Kiuhara, S. A., & Fishman, E. J. (2017). The relationship among strategic

writing behavior, writingmotivation, andwriting performance with young, developing writers.

Elementary School Journal, 118(1), 82–104. https://doi.org/10.1086/693009

Relations among motivation, behaviour, and performance in writing 1477

Graham, S., Harris, K. R., & Santangelo, T. (2015). Research-basedwriting practices and the common

core: Meta-analysis and meta-synthesis. Elementary School Journal, 115, 498–522. https://doi. org/10.1086/681964

Graham, S., McKeown, D., Kiuhara, S., & Harris, K. R. (2012). A meta-analysis of writing instruction

for students in the elementary grades. Journal of Educational Psychology, 104, 879–896. https://doi.org/10.1037/a0029185

Graham, S., & Perin, D. (2007).Whatwe know,whatwe still need to know: Teaching adolescents to

write. Scientific Studies of Reading, 11(4), 313–335. https://doi.org/10.1080/

10888430701530664

Graham, S., Wijekumar, K., Harris, K. R., Lei, P.-W., Fishman, E. J., Ray, A. B., & Houston, J. (2019).

Writing skills, knowledge, motivation, and strategic behavior predict students’ persuasive

writing performance in the context of robust writing instruction. The Elementary School

Journal, 119, 487–510. https://doi.org/10.1086/701720 Harris, K. R.,&Graham, S. (1992). Self-regulated strategy development: Apart of thewritingprocess.

InM. Pressley, K. R. Harris & J. T. Guthrie (Eds.), Promoting academic competence and literacy

in school (p. 277–309). Academic Press.

Harris, K. R., & Graham, S. (2017). Self-regulated strategy development: Theoretical bases, critical

instructional elements, and future research. In R. Fidalgo, K. R. Harris & M. Braaksma (Eds.),

Design principles for teaching effective writing (pp., 119–151). Hershey, PA: Brill Editions. Hayes, J. R. (1996). A new framework for understanding cognition and affect inwriting. InC.M. Levy

& S. Ransdell (Eds.), The science of writing: Theories, methods, individual differences, and

applications (pp. 1–27). Mahwah, NJ: Lawrence Erlbaum Associates.

Hillocks, G. (1986). Research on written composition: New directions for teaching. Urbana, IL:

National Council of Teachers of English.

Hu, L. T., & 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), 1–55. https://doi.org/10.1080/10705519909540118 Jones, D., & Christensen, C. A. (1999). Relationship between automaticity in handwriting and

students’ ability to generate written text. Journal of Educational Psychology, 91(1), 44–49. https://doi.org/10.1037/0022-0663.91.1.44

Karchmer-Klein, R. (2019). Writing with digital tools. In S. Graham, C. A. MarArthur, & M. Hebert

(Eds.),Best practices inwriting instruction (3rd ed., pp. 185–208). NewYork,NY: TheGuilford

Press.

Kellogg, R. T. (1994). The psychology of writing. Oxford, UK: Oxford University Press.

Kent, S. C., &Wanzek, J. (2016). The relationship between component skills andwriting quality and

production across developmental levels: A meta-analysis of the last 25 years. Review of

Educational Research, 86, 570–601. https://doi.org/10.3102/0034654315619491 Kim, Y.-S.-G. (2020a). Interactive dynamic literacy model: An integrative theoretical framework for

reading-writing relations. In R. A. Alves, T. Limpo, & R. M. Josh (Ed.), Reading-writing

connections: Towards integrative literacy science. Berlin, Germany: Springer.

Kim, Y.-S.-G. (2020b). Structural relations of language and cognitive skills, and topic knowledge to

written composition: A test of the direct and indirect effectsmodel ofwriting.British Journal of

Educational Psychology, 90, 910–932. https://doi.org/10.1111/bjep.12330 Klassen, R. (2002). Writing in early adolescence: A review of the role of self-efficacy beliefs.

Educational Psychology Review, 14, 173–203. https://doi.org/10.1023/A:1014626805572 Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York,

NY: The Guilford Press.

Korat, O., & Schiff, R. (2005). Do children who read more books know ’what is goodwriting’ better

than children who read less? A comparison between grade levels and SES groups. Journal of

Literacy Research, 37, 289–324. https://doi.org/10.1207/s15548430jlr3703_2 Limpo, T., & Alves, R. A. (2017). Relating beliefs in writing skill malleability to writing performance:

The mediating role of achievement goals and self-efficacy. Journal of Writing Research, 9, 97– 125. https://doi.org/10.17239/jowr-2017.09.02.01

1478 ANA CAMACHO et al.

Muth�en, B. O., & Satorra, A. (1995). Complex sample data in structural equation modeling.

Sociological Methodology, 25, 267–316. https://doi.org/10.2307/271070 Oberski, D. (2014). lavaan.survey: An R package for complex survey analysis of structural

equation models. Journal of Statistical Software, 57(1), 27. https://doi.org/10.18637/jss.v057.

i01

Olinghouse, N. G., Graham, S., & Gillespie, A. (2015). The relationship of discourse and topic

knowledge to fifth graders’ writing performance. Journal of Educational Psychology, 107, 391.

https://doi.org/10.1037/a0037549

Olive, T., Favart, M., Beauvais, C., & Beauvais, L. (2009). Children’s cognitive effort and fluency in

writing: Effects of genre and of handwriting automatisation. Learning and Instruction,19, 299– 308. https://doi.org/10.1016/j.learninstruc.2008.05.005

Pajares, F. (2003). Self-efficacy beliefs, motivation, and achievement in writing: A review of the

literature. Reading & Writing Quarterly, 19(2), 139–158. https://doi.org/10.1080/

10573560308222

Pajares, F. (2007). Empirical properties of a scale to assess writing self-efficacy in school contexts.

Measurement and Evaluation in Counseling andDevelopment, 39, 239–249. https://doi.org/ 10.1080/07481756.2007.11909801

Pajares, F., Hartley, J., & Valiante, G. (2001). Response format in writing self-efficacy assessment:

Greater discrimination increases prediction.Measurement and Evaluation in Counseling and

Development, 33, 214–221. https://doi.org/10.1080/07481756.2001.12069012 Pajares, F., Johnson,M. J., &Usher, E. L. (2007). Sources ofwriting self-efficacy beliefs of elementary,

middle, and high school students. Research in the Teaching of English, 42(1), 104–120. http:// www.jstor.org/stable/40171749

Pajares F., ValianteG. (1997). Influence of self-efficacy on elementary students’ writing. The Journal

of Educational Research, 90, (6), 353–360. http://dx.doi.org/10.1080/00220671.1997.

10544593

Pajares, F., & Valiante, G. (1999). Grade level and gender differences in the writing self-beliefs of

middle school student. Contemporary Educational Psychology, 24, 390–405. https://doi.org/ 10.1006/ceps.1998.0995

R Core Team (2019). R: A language and environment for statistical computing. R Foundation for

Statistical Computing. https://www.R-project.org/

Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical

Software, 48(2), 1-36. https://doi.org/10.18637/jss.v048.i02

Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic for moment structure

analysis. Psychometrika, 66, 507–514. https://doi.org/10.1007/BF02296192 Schreiber, J. B., Nora, A., Stage, F. K., Barlow, E. A., & King, J. (2006). Reporting structural

equation modeling and confirmatory factor analysis results: A Review. The Journal of

Educational Research, 99, 323–338. https://doi.org/10.3200/JOER.99.6.323-338

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural

equation models. Sociological Methodology, 13, 290–312. https://doi.org/10.2307/270723 Stapleton, L. M. (2006). An Assessment of practical solutions for structural equation modeling with

complex sample data. Structural Equation Modeling: A Multidisciplinary Journal, 13(1), 28– 58. https://doi.org/10.1207/s15328007sem1301_2

Troia, G. A., Harbaugh, A. G., Shankland, R. K., Wolbers, K. A., & Lawrence, A. M. (2013).

Relationships between writing motivation, writing activity, and writing performance: Effects of

grade, sex, and ability. Reading and Writing: An Interdisciplinary Journal, 26(1), 17–44. https://doi.org/10.1007/s11145-012-9379-2

Troyer,M., Kim, J. S., Hale, E.,Wantchekon, K. A., &Armstrong, C. (2019). Relations among intrinsic

and extrinsic reading motivation, reading amount, and comprehension: A conceptual

replication. Reading and Writing - an Interdisciplinary Journal, 32, 1197–1218. https://doi. org/10.1007/s11145-018-9907-9

Wijekumar, K., Graham, S., Harris, K. R., Lei, P.-W., Barkel, A., Aitken, A., . . .Houston, J. (2019). The roles of writing knowledge, motivation, strategic behaviors, and skills in predicting elementary

Relations among motivation, behaviour, and performance in writing 1479

students’ persuasive writing from source material. Reading and Writing: An Interdisciplinary

Journal, 32, 1431–1457. https://doi.org/10.1007/s11145-018-9836-7 Yamac�, A., €Ozt€urk, E., &Mutlu,N. (2020). Effect of digitalwriting instructionwith tablets onprimary

school students’ writing performance and writing knowledge. Computers & Education, 157,

https://doi.org/10.1016/j.compedu.2020.103981103981

Yuan, K.-H., & Bentler, P. M. (2000). Three likelihood-based methods for mean and covariance

structure analysis with nonnormal missing data. Sociological Methodology, 30(1), 165–200. https://doi.org/10.1111/0081-1750.00078

Zumbrunn, S., Broda, M., Varier, D., & Conklin, S. (2020). Examining the multidimensional role of

self-efficacy for writing on student writing self-regulation and grades in elementary and high

school.British Journal of Educational Psychology,90, 580–603. https://doi.org/10.1111/bjep. 12315

Received 13 August 2020; revised version received 12 March 2021

1480 ANA CAMACHO et al.

Copyright of British Journal of Educational Psychology is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.