Development and Validation of The Situated Academic Writing Self-Efficacy Scale (SAWSES)

Assessing Writing 48 (2021) 100524
Contents lists available at ScienceDirect
Assessing Writing
journal homepage: www.elsevier.com/locate/asw
Development and validation of the Situated Academic Writing

Self-Efficacy Scale (SAWSES)
Kim M. Mitchell a, b, *, Diana E. McMillan a, c, Michelle M. Lobchuk a,
Nathan C. Nickel d, Rasheda Rabbani a, d, e, Johnson Li f
a
University of Manitoba, College of Nursing, Rady Faculty of Health Sciences, 89 Curry Place, Winnipeg, Manitoba, R3T 2N2, Canada
b
Red River College, Department of Nursing, Faculty of Health Sciences and Community Services, C-608 – 2055 Notre Dame Ave., Winnipeg,
Manitoba, R3H 0J9, Canada
c
Clinical Chair, Health Sciences Centre, 750 McDermot Ave., Winnipeg, Manitoba, R3E 0T5, Canada
d
University of Manitoba, Department of Community Health Sciences, Rady Faculty of Health Sciences, 750 Bannatyne Ave, Winnipeg, Manitoba,
R3E 0W2, Canada
e
George and Fay Yee Center for Health Innovation, 753 McDermot Ave., Winnipeg, Manitoba, R3E 0T6, Canada
f
University of Manitoba, Department of Psychology, Faculty of Arts, 190 Dysart Rd., Winnipeg, Manitoba, R3T 2N2, Canada
A R T I C L E I N F O A B S T R A C T
Keywords: Existing writing self-efficacy instruments have assessed the concept through mechanical and
Writing self-efficacy process features of writing to the neglect of the influence of situated context. The purpose of this
Scale development study was to develop and test the Situated Academic Writing Self-Efficacy Scale (SAWSES) based
Factor analysis
on Bandura’s self-efficacy theory and a model of socially constructed writing. A sequential
Developmental writing
multimethod approach constituted the methods. A Delphi panel of 15 expert scholars conducted a
Situated perspectives
theoretical evaluation of the scale and the items were piloted with 20 nursing undergraduate
students using cognitive interviews. The scale was validated in two studies with independent
samples of 255 nursing students (Study 1), and in an interdisciplinary sample of undergraduate
(N = 543) and graduate students (N = 264) (Study 2). The three identified factors present a
structure to the questionnaire which is developmental and has the potential to detect gaps in
student self-assessed ability to master various facets of disciplinary writing: 1) Writing-Essentials
– synthesis, emotional control, language; 2) Relational-Reflective – relationship building with
writing facilitators (teachers, academic sources) and the self through reflection; and 3) Creative
Identity – exploring gaps in student achievement of transformative writing (creativity, voice, and
disciplinary identity), where confidence can help identify the most engaged writers.
1. Introduction
In 1986, David Bartholomae wrote that students entering post-secondary education and adapting to new writing contexts had to
“invent the university for the occasion” (p. 4). By inventing the university, students are repeatedly trying on new identities in order to
think, reflect, argue, evaluate, and use language when writing. Writing research has demonstrated over more than 40 years of study
* Corresponding author at: Red River College, Department of Nursing, Faculty of Health Sciences and Community Services, C-608 – 2055 Notre
Dame Ave., Winnipeg, Manitoba, R3H 0J9, Canada.
E-mail addresses: [email protected] (K.M. Mitchell), [email protected] (D.E. McMillan), [email protected]
(M.M. Lobchuk), [email protected] (N.C. Nickel), [email protected] (R. Rabbani), [email protected] (J. Li).
https://doi.org/10.1016/j.asw.2021.100524
Received 10 August 2020; Received in revised form 19 January 2021; Accepted 20 January 2021
Available online 31 January 2021
1075-2935/© 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
K.M. Mitchell et al. Assessing Writing 48 (2021) 100524
that perception of what constitutes an acceptable written text changes from discipline to discipline (Lea & Street, 1998), as students
progress across curriculums (Mitchell & McMillan, 2018), and from teacher to teacher in individual classes (Mitchell, 2020f). With
contexts in a continuous state of flux, student writing approaches are also in a constant state of flux. Writing becomes more complex as
a writer moves from one level of education to another (e.g. high school to undergraduate to graduate writing). Simultaneously,
students must also learn to self-assess their ability to communicate to a remote audience (Plakhotnik & Rocco, 2016), pay attention to
numerous rhetorical demands of the genre of writing assigned, employ strategies of cognition and metacognition, and manage their
emotions (Mitchell, McMillan, & Lobchuk, 2019). This cognitive and situative multitasking inevitably affects students’ confidence in
their writing.
Self-efficacy, “people’s beliefs in their ability to produce desired effects by their actions” (Bandura, 1997, p. vii), is a motivational
construct that functions to promote human agency. In a learning environment, agency provides individuals with power to act to
improve learning, grades, or any high priority motivating factor in their context. Low writing self-efficacy and struggles with writing
ability have been connected to delayed student progression through academic disciplinary programs (Mitchell & McMillan, 2018).
Through human agency, writing self-efficacy has the potential to bridge knowledge transfer between the classroom and practice
environments (Bandura, 2006; Jefferies et al., 2018; Mitchell, 2018).
Writing is disciplinary, which means that different academic disciplines have forged their own expectations as to what defines
effective writing and how knowledge should be communicated among disciplinary members (Hyland, 2003). Writing is social; a writer
is unable to produce an effective written product without knowledge of the norms and conventions of the environment within which
they write. While Bandura’s self-efficacy theory acknowledges the social context that surrounds people’s development of self-beliefs in
their abilities, the social context is relegated to the periphery and emphasizes individual goals and motivations. Self-efficacy as a
construct is typically examined as the plight of an individual battling internal angels and demons or, as Paré (2014) has observed,
reduces the writer to one, “struggling … alone in a loft, seeking inspiration” (p. A-90). Those angels and demons either prevent or
facilitate the writer’s ability to develop personal agency.
In studies of writing self-efficacy, measurement of the construct has also focused on the individual writer and their ability to
accomplish concrete tasks contained to the cognitive process categories originally developed by Flower and Hayes (1981) of planning,
translating (composing), and revising stages of writing (Mitchell, Harrigan, Stefansson, & Setlack, 2017; Mitchell, Rieger, & McMillan,
2017). In process and cognitive perspectives, writing contexts are acknowledged but their influence on the development of writing
ability, and, consequently, writing self-efficacy, are not addressed (Russell, Lea, Parker, Street, & Donahue, 2009). Writing context is
central in the more modern situated perspectives on writing that have been present in the literature since the 1980s. Situated per
spectives acknowledge, that successful writing emerges through activities that are not visible in the texts produced and those activities
are more complex than individualistic writing process strategies. Writing functions to develop the identity of a writer in a discipline,
facilitates the maintenance of disciplinary norms and values within genres, and helps communities develop an understanding of how
meaning is negotiated between readers and writers (Mitchell, 2018; Paré, 2014). Thus, new approaches to measuring writing
self-efficacy that consider that social and situated context are needed.
Despite the refocusing of writing scholarship on these contextual and relational aspects of writing, the assessment of writing self-
efficacy has not kept pace with the epistemological evolution of writing theory (Mitchell et al., 2019). Until recently, the social
cognitive literature and the sociocultural literature have operated in silos with few scholars recognizing the potential benefits to their
merger (Graham, 2018; Mitchell et al., 2019; Turner & Nolen, 2015). Graham (2018) notes that incorporating both cognitive and
sociocultural considerations of writing can result in “richer fuller understanding of writing,” and provides “a more comprehensive
explanation, one that goes beyond just context or the capabilities of writers and their collaborators” (p. 258). Considering social
perspectives on writing in the measurement of writing-self efficacy could help researchers and educators understand how contexts
work to limit self-efficacy development. Hence, this project began with this goal in mind: to develop a measure of writing self-efficacy
for post-secondary use that stimulates student reflection on aspects of their writing context that affects their writing self-efficacy
development.
2. Current measures of writing self-efficacy
In a previous exploration of writing self-efficacy measures used in post-secondary contexts, 11 existing writing self-efficacy
measures were identified (Mitchell, Rieger et al., 2017). More related instruments have emerged since that publication, for
example, Teng, Peigian, and Xu (2018) who explored a new instrument to examine English language learner’s writing self-efficacy, and
Golombek, Klingsieck, and Scharlau (2019) who developed an instrument to assess self-efficacy for self-regulation of academic writing.
In an in progress systematic review of 49 papers that collectively used 51 measures of writing self-efficacy in English post-secondary
contexts (unpublished data), no single measure has emerged as the gold standard. One classic measure (Shell, Murphy, & Bruning,
1989) is one of the more frequently used (11.8 % of studies), however focuses on self-efficacy for grammatical and mechanical aspects
of writing. The next most common measure was Schmidt and Alexander’s (2012) scale developed for writing center contexts (9.8 % of
studies). Instead researchers tend to develop study-specific sets of items not intended to be combined into total scores (19.6 % of
studies). Ten other tools in varying stages of validation have been developed and used only once and then never appear in the literature
again or are used exclusively by the developing researcher over a small number of studies without adoption by other research teams
(31.4 % of studies). Others create their own instruments by patchworking together various items from existing instruments (17.6 %), or
measures targeted to children (9.8 %). Inconsistent attention to construct validation of measured concepts has been observed in a
review of social and personality research where 46 % of studies were found to have used unvalidated tools created “on the fly” (Flake,
Pek, & Hehman, 2017, p. 374). “On the fly” use of tools for measuring writing self-efficacy appears to be the norm, rather than the
2
exception, in this body of research. That the writing self-efficacy research community has not settled on a gold standard instrument
may be a sign that contextual differences leave researchers feeling dissatisfied with existing instruments. Researchers appear to prefer
to create new instruments rather than use something they feel inadequately represents the process tasks applicable to their context of
writing.
Given the complexity of writing in the post-secondary context, it is unlikely that existing writing self-efficacy measures are
capturing the complete writing experience. Acontextual assessments of writing are especially problematic in a disciplinary profession
such as nursing which uses writing to explain relationships between patients, health, and the nursing context (Mitchell, 2018). Nursing
was an ideal place to begin an exploration of writing self-efficacy. Nursing’s unique issues with writing instruction, such as its practice
focus and tendency to devalue writing over patient care activities, amplified particular academia-wide contextual problems such as
rigidity with writing expectations, strictly enforced objectivity, and faculty uncertain about their skills to guide writing (Yagelski,
2011). To improve generalizability of the final scale, the study was extended to further validate the instrument with an interdisci
plinary sample.
3. Theoretical premise of the scale development
The Situated Academic Writing Self-Efficacy Scale (SAWSES) was developed from within two theoretical perspectives, Bandura’s
self-efficacy theory (Bandura, 1997), and a socially constructed model of writing (Mitchell, 2018). Bandura’s theory contributed to the
item development process by adhering to recommendations for self-efficacy scale construction and developing items with various
degrees of difficulty (Pajares & Valiante, 2006). The strength of Bandura’s cognitive perspective is his recognition of how emotions are
stimulated (emotional arousal) in self-efficacy experiences, and how significant others (peers/teachers) can influence self-perceptions
through modelling (vicarious experiences) and feedback provision (verbal persuasion).
Because Bandura’s theory does not provide guidance as to how to evaluate the influence of the writer’s social context on self-
efficacy appraisals, a model of writing from a situated and socially constructed perspective, which considers the influence of disci
plinary and contextual aspects of writing, also guided item development using five intertwined domains (Mitchell, 2018).
1) Identity: Understanding discipline specific language, voice, and use of metacognition to reflect upon the relationship between
writing and their disciplinary values, defines identity.
2) Creativity: Discipline-specific definitions of novelty and originality inform creativity. Idea generation, synthesis, genre adaptations,
and interpretive abilities all require creativity. Self-efficacy will be impacted if students feel they are not allowed to be creative or if
their version of creativity is not shared by the faculty who assigns the writing.
3) Emotions: Ability to control emotions at all phases of the writing process, write in the face of difficulties, and degree of validation of
emotions within a context impact writing self-efficacy.
4) Relationality: Writers form relationships with the sources they incorporate through citation, inspiration, or interpretation. Students
form relationships with their teachers and peers during pedagogical processes and feedback interactions.
5) Context: Aspects of the environment for writing define context, including the impact of disciplinary and/or faculty tacit and explicit
beliefs about writing. Writing self-efficacy will be impacted if students feel unsupported, devalue the assigned writing, or if they do
not feel their actions in their writing context are likely to lead to success within the writing genre assigned.
The intertwined nature of the theory suggests there are overlaps between categories. Throughout this validation project, there was
never any intention to force items to remain within the five apriori domains or force a five-factor instrument. As items were edited
(Delphi and Cognitive Interview phases), they were not moved from their original theoretical category; hence, even if an edit changed
the nature of the item to a degree that it might be a better fit for a different theoretical domain, it was not re-assigned. This decision was
made with the assumption that the factor analysis phase of the study would cluster items where they were most statistically similar.
4. Methods and results
The overall scale development project uses a multimethod sequential design. Items were drafted based on a review of the literature,
personal experience with writing and writing instruction, and the model reported in Mitchell (2018). A Delphi panel of nurse educators
and writing scholars supplied expert analysis of the relevance of the developed items. Items were then piloted with undergraduate
nursing students using cognitive interview procedures (Willis, 2005). Two survey studies followed:
1 Study 1 piloted the questionnaire with undergraduate nursing students along with several existing instruments chosen for con
current validity assessments: two previously established writing self-efficacy scales (Mitchell, Harrigan et al., 2017; Shell et al.,
1989), writing apprehension (Daly & Miller, 2013), and a measure of nursing identity (Adams, Hean, Sturgis, & Macleod Clark,
2006). Writing context perceptions were assessed with single item measures. The main analysis was an exploratory factor analysis
and a structural equation model of situated writing.
2 Study 2 tested the scale with an interdisciplinary sample of undergraduate and graduate students from various educational contexts
and multigroup confirmatory factor analysis was performed.
Ethical approval was obtained in two separate applications. The first for the Delphi study and the second for the student re
cruitments which required additional ethical approvals at all participating sites. All participants were volunteers. The Delphi
3
Table 1
Demographic Characteristics of the Samples for Study 1 (N = 234) and Study 2 Undergraduates (N = 543); graduate students (N = 264).
Study 1 Study 2 Study 2
Nursing Baccalaureate Interdisciplinary Undergraduate Interdisciplinary Graduate
Only Students Students
n(%) n(%) n(%)
Institution
Institution 1 116 (49.6 %) 10(1.8 %) 0
Institution 2 77 (32.9 %) 422(77.7 %) 171(64.8 %)
Institution 3 41 (17.5 %) 87(16.0 %) 6(2.3 %)
Other Canadian 8(1.5 %) 15(5.7 %)
USA 11(2.0 %) 40(15.2 %)
International 7(1.3 %) 32(12.1 %)
Country
Canada 234(100 %) 526(96.9 %) 192(72.7 %)
USA 11(2.0 %) 40(15.2 %)
Great Britain/Ireland 2(0.4 %) 15(5.7 %)
Other 4(0.7 %) 17(6.4 %)
Discipline
Nursing 234(100 %) 28(5.2 %) 28(10.6 %)
STEM disciplines 106(19.5 %) 40(15.1 %)
Education 34(6.3 %) 25(9.5 %)
Psychology 50(9.2 %) 18(6.8 %)
Social Work 24(4.4 %) 0
Undecided/generalist 68(12.5 %) 0
Other Health Sciences 0 33(12.5 %)
Other 233(42.9 %) 120(45.5 %)
Year in program (Nursing)

1 92(39.3 %)
2 58(24.8 %)
3 84(35.9 %)
Years in post-secondary (interdisciplinary undergraduate)

1 or less 107(19.7 %)
2 100(18.4 %)
3 89(16.4 %)
4 83(15.3 %)
5 or more 72(13.3 %)
Second or subsequent degree 87(16.0 %)
Other 5(0.9 %)
Level of graduate studies

Masters taking coursework 72(27.3 %)
Masters thesis/capstone 55(20.8 %)
PhD taking coursework 34(12.9 %)
PhD pre-candidacy 19(7.2 %)
PhD candidate 70(26.5 %)
PhD defense ready 14(5.3 %)
Writing Attitude (N ¼ 255)

Generally, likes writing 44(17.3 %) 237(43.6 %) 124(47.0 %)
Somewhere in between 108(42.4 %) 219(40.3 %) 111(42.0 %)
Generally, dislikes writing 103(40.4 %) 87(16 %) 29(11.0 %)
Gender
Female 215(91.9 %) 388(71.5 %) 194(73.5 %)
Male 18(7.7 %) 140(25.8 %) 61(23.1 %)
Non-binary (other) 1(0.4 %) 15(2.8 %) 9(3.4 %)
Learned to Speak in English

Under 5 years of age 142(60.7 %) 363(66.9 %) 177(67.0)
Spoke English and another language under 5 43(18.4 %) 94(17.3 %) 35(13.3 %)
years
Learned English between the ages of 5 and 12 37(15.8 %) 59(10.9 %) 23(8.7 %)
Learned English between the ages of 13 and 17 10(4.3 %) 18(3.3 %) 22(8.3 %)
Learned English as an adult (18+) 2(0.9 %) 8(1.5 %) 5(1.9 %)
Other unspecified 0 1(0.2 %) 2(0.8 %)
(continued on next page)
4
Table 1 (continued )
Nursing Baccalaureate Interdisciplinary Undergraduate Interdisciplinary Graduate
Only Students Students
n(%) n(%) n(%)
Learned to Write in English

Elementary School prior to grade 4 217(92.7 %) 511(94.1 %) 226(85.6 %)
Middle School (grades 5− 9) 9(3.8 %) 15(2.8 %) 17(6.4 %)
High school (grades 10− 12) 7(3.0 %) 13(2.4 %) 11(4.2 %)
After graduating from high school 1(0.4 %) 4(0.7 %) 9(3.4 %)
Other(unspecified) 0 0 1(0.4 %)
Mean age: 25.07 Mean age: 25.29 Mean age: 33.43
Age
Range 18.0− 51.5 Range 17.0− 72 Range 21.0− 73.83
STEM = Science, Technology, Engineering, Mathematics.
participants consented to releasing their anonymity and were offered a $25 CAD Amazon gift card for their time. The cognitive
interview participants were given a $50 cash honorarium. Study 1 participants were provided the opportunity of five draws for $100
each for a gift card of choice. No honorariums were offered in interdisciplinary Study 2.
5. Item development and revision
Forty-one (41) items for the proposed Situated Academic Writing Self-Efficacy Scale (SAWSES) were initially developed. The
evolution of the items across study phases is presented in the Supplemental material. Items were presented according to processes
described in Bandura (1997) and Pajares and Valiante (2006). These authors advised that items be presented as a 0− 100 rating scale
and phrased as “I can” statements reflecting current ability (rather than “I will” reflecting a future state). An electronic sliding scale was
anchored with the phrases, “completely sure I cannot” to “completely sure I can.” The initial items were first vetted by the author team,
one writing scholar who teaches a discipline-specific writing course for nurses, and at an international nursing conference poster
presentation (Mitchell et al., 2018).
The Delphi and cognitive interview studies were then conducted with three objectives: reduce the number of items, ensure
theoretical relevance, and ensure correct interpretation of the items.
5.1. Delphi panel
A more detailed report of the methods used in the Delphi panel are reported in (Mitchell, 2020f). The Delphi participants completed
two online surveys in the Qualtrics platform. Fifteen scholars from five countries (Australia, Canada, Germany, United Kingdom, and
United States of America) completed both surveys (7 nurses, 8 writers). Round 1 asked the scholars to rate the original 41 items for
relevance (0− 100) within its theoretical domain (e.g. identity). Qualitative comment boxes were provided for each item. Items were
edited or deleted based on feedback and presented for additional feedback in round 2. In total, 25 items were kept, edited or newly
created based on the Delphi feedback and then presented in the cognitive interview phase to students.
5.2. Cognitive interviews
Twenty nursing students (16 female, 4 male), ranging in age from 20 to 42 years, from two Baccalaureate nursing programs were
asked to think aloud about their personal experiences with writing while reading and scoring each of the 25 items (Mitchell, 2020f).
Students from all years of both Baccalaureate programs participated in the interviews. Racially, most students were persons of colour
(55 %) or identified as Indigenous (10 %). Three students (15 %) reported learning English age 13+, and of the remaining students
most spoke two or more languages (55 %). There were equal proportions of students reporting that they liked (20 %) or disliked (20 %)
writing but most felt “somewhere in between” about writing for school (60 %). Previous degrees or diplomas were obtained by 30 % of
the sample but all the students had previous university course experience before being accepted to nursing. Interview procedures
followed recommendations from Willis (2005). Concurrent prompting was used throughout the interview to encourage out loud
thinking, clarify understanding of terminology, and gain deeper explanations of the stories students described. Items were edited,
watched, or deleted based on observations related to interpretation, degree of student interest in the item, and scoring patterns (Knafl
et al., 2007). The goal was to sample until three consecutive interviews produced minimal problems with items. The 25 items were
edited at periodic intervals after 3, 6, 12, and 17 interviews. Item deletion was considered if the students felt indifferent to the issue
presented in the item, the item produced a ceiling effect in chosen scores, or if issues of item wording failed to make the item un
derstandable. At the end of the cognitive interview phase, 20 items moved forward to statistical testing.
6. Study 1: exploratory factor analysis and hypothesis testing
The primary goals of Study 1 were to evaluate the questionnaire for its factor structure and test a hypothesized structural equation
model that predicated writing apprehension, nursing identity, and three measures of writing context would predict scores on the
5
SAWSES.
6.1. Participants
Table 1 presents the demographic characteristics of the samples that participated in Study 1 and Study 2. All participants in Study 1
were nursing students in three Baccalaureate nursing programs located in a prairie province in Canada. Responses were included up to
point of completion. The sample size for the exploratory factor analysis was 255. Due to a large number of measures being included in
this study, some response burden was present on participants. Participants who did not complete tools for apprehension, identity,
writing context, and the demographic survey were excluded listwise from the structural equation model for final N of 234.
6.2. Procedures
Faculty were emailed a description of the study from a representative within the nursing departments and were asked to consider
inviting the researcher to come and speak to their class about the survey. If a faculty member agreed, a designated person within the
program emailed the Qualtrics survey link to the specific class on the date of the in-class presentation. In order to ensure that all
students were sent the survey it was emailed to the entire nursing student population at a later date or posted on student message
boards in the program learning management system. In two classrooms (in first- and third-year at Institution 1) students were provided
a paper version of the developing SAWSES in order to assess the reliability indicators between the electronic version and a parallel
paper and pencil version. The paper version included a visual analog scale where students placed an X on a line to indicate their score
between the two anchor phrases of “completely sure I cannot” and “completely sure I can.” The paper form was completed first and the
electronic form was completed immediately after. Each paper survey was assigned a number that students entered into the Qualtrics
survey to facilitate matching. The paper form required ruler measurements and manual entry of data. A two person check of mea
surements and data entry was completed. Errors of greater than 1 mm were observed in 12 item responses and they underwent a third
check. The two rater scores were averaged for correlational analysis. Missing data was minimal in this study (<1%). Missing responses
to SAWSES were replaced with the average score of the participants’ remaining SAWSES responses.
6.3. Instruments
Six instruments were included within the main survey in the order described:
1 The developing SAWSES instrument – The 20 items that remained after the cognitive interviews were presented. The questions
were ordered on the survey by approximate degree of difficulty of the item as assessed based on conversations that took place
during the cognitive interviews.
2 The Self-Efficacy Scale for Academic Writing – The SESAW was included in order to provide concurrent validity evidence. The
SESAW was previously developed and tested in a nursing context and represents a writing process-focused scale (Mitchell, Harrigan
et al., 2017). Concurrent validity with the Post-Secondary Writerly Self-Efficacy Scale (Schmidt & Alexander, 2012) ranged from
.76 to .81 (Mitchell & McMillan, 2018). In this study a 9-item scale had a Cronbach’s alpha of .85.
3 Writing Self-Efficacy Instrument (WSEI) – Developed by Shell et al. (1989) the 8-item skills subscale which focuses on grammar and
mechanics was used to test a hypothesized discriminant relationship between situated writing self-efficacy and writing mechanics
self-efficacy. The tool is structured with a 0–100 scale with 0 meaning “no chance” and 100 meaning “completely certain.” In this
study, Cronbach’s alpha equaled .93.
4 Writing Apprehension Test (WAT) –A 20-item version of the Writing Apprehension Test (Daly & Miller, 1975) was retrieved from
the Measurement Instrument Database for Social Science (Daly & Miller, 2013) and presented as a 5-point Likert scale (strongly
disagree to strongly agree). This scale contains both positive and negative worded items. To avoid negative values for the structural
equation model tests, the scoring was reversed so that a high score meant low apprehension. The Cronbach’s alpha for the current
study was .92.
5 Nursing Identity –The 8-item Macleod Clark professional identity questionnaire (PIQ) (Adams et al., 2006) was presented in a
5-point Likert format (strongly disagree to strongly agree) to assess sense of identification with the nursing profession. Wor
thington, Salamonson, Weaver, and Cleary (2013) were able to use the scale to predict student retention in a nursing program after
12-months. Cronbach’s alpha for the current study was .83.
6 Context items – Three author developed items assessing students’ appraisal of their writing context were included. These items
were rated on a 0–100 sliding scales:
Support: This class and/or my institution can support me in the way I need to be successful with writing this essay assignment
(“not at all true” to “completely true”).
Stakes: It is important for my future that I get the grade I need on this writing assignment so I do well in this course (“not at all
important” to “completely important”).
Value: I feel that the writing assignment in this class will be important for my learning in the nursing program (“not at all
important” to “completely important”).
6
6.4. Data analysis
The Statistical Package for the Social Sciences (SPSS v.24) was used for the descriptive analysis. Exploratory factor analysis was
conducted in RStudio (v. 1.2.5033) using the Psych package (Revelle, 2017). The Lavaan package (Rosseel, 2017) was used for the
structural equation model. Data files and R code are publicly available (Mitchell, 2020b, 2020c; Mitchell, 2020e).
6.5. Results
6.5.1. Exploratory factor analysis

To assess factorability of the scale the Kaiser-Meyer-Olkin (KMO) index of sampling adequacy was calculated at .95 indicating high
suitability for factor analysis (Revelle, 2017). Each individual item produced high KMO index scores >.90 ranging from .91 to .98.
Parallel analysis produced the scree diagram in Fig. 1 that suggested that both 1 and 3 factor solutions should be tested. Principal axis
methods were used with oblique rotation for greatest interpretability (Revelle, 2017).
Numerous models were tested and four are reported in Table 2 which includes key goodness of fit indexes: Root Mean Square Error
of Approximation (RMSEA, excellent fit <.06, acceptable fit < .08), Tucker Lewis Index (TLI ideal >.95), Comparative Fit Index (CFI
ideal >.95), and Root Mean Square of the Residuals (RMSR ideal <.05) (Beaujean, 2014). Items were retained on a factor if the factor
loading was >.30. Cross loading items were considered for deletion in combination with knowledge of participant reaction to that item
based on the cognitive interviews. A single factor model demonstrated that all 20 items produced factor loadings greater than .50 with
the exception of item 2 (.42). This model had less than adequate fit: RMSEA = .096. TFI = .86, and CFI = .87. A three factor 20-item
model had adequate fit but items 4 and 9 demonstrated cross loadings on two factors and item 14 did not produce factor loadings >.30.
A third model removed items 4, 9, and 14 with slight improvement to fit. Item 2 (“If I get stuck, I can contact a teacher or tutor to help
me move forward with my writing ideas), the lowest loading item in the single factor model, was considered for removal. It had also
elicited confusion from students who felt they could contact a teacher but likely wouldn’t. A fourth model with items 2, 4, 9, and 14
removed explained 61 % of variance and provided the best fit with all remaining items loading >.45 on an independent factor.
Goodness of fit indices were acceptable to excellent on this final model (Model 4) with RMSEA = .068 (90 % CI .052–.081), TLI = .95,
CFI = .97, and RSMR = .03.
Despite five theoretical domains guiding the development of the questionnaire items, three factors were considered sufficient, as it
was accepted from the onset that theoretical dimensionality and statistical dimensionality may not match (Lock & Johnston, 2016).
Items clustered statistically with theoretical relevance. Factor 1 (8-items) was labeled Relational-Reflective Writing. The items
described interactions with others in the writing context, responses to feedback, audience awareness, relationship to source materials
when reading, adapting writing to the assignment (which often meant adapting to teachers according to student statements in the
cognitive interviews but could also mean adapting to different genres), or represented reflective elements of writing process. Factor 2
(5-items) was labelled Creative Identity. Items represented writing acts describing integrating creativity in a prescripted writing
environment, developing personal voice, or learning the language of a discipline. Factor 3 (3-items) included a perceptibly diverse set
of items we labelled Writing Essentials because ability to move forward despite difficulties, ability to integrate and synthesize
numerous sources, and ability to use academic language suitable for post-secondary writing are essential skills required to produce an
effective academic writing assignment.
6.5.2. Reliability testing

Parallel form analysis compared an electronic version of the questionnaire and a paper form of the questionnaire completed by the
same participants (n = 56). Correlations between the paper and electronic versions of the scale produced reliability coefficients of .91
(total scale), .88 (Relational-Reflective), .88 (Creative-Identity), and .85 (Writing Essentials). The mean, standard deviation, and 95 %
Fig. 1. Scree Diagram for Exploratory Factor Analysis.
7
Table 2
Four exploratory factor analysis models SAWSES Nursing student sample (N = 255), with goodness of fit analysis.
Exploratory Factor Analysis Factor Loadings
Situated Academic Writing Self- Model 1 Model 2 Model 3 Model 4

Efficacy Scale (Model Testing) 1-Factor 20- 3-Factor 20-item scale 3-Factor 17-item scale 3-Factor 16-item scale
item scale
Factor Factor Factor Factor Factor Factor Factor Factor Factor
1 2 3 1 2 3 1 2 3
1 Even when the writing is hard, I can .62 .55 .63 .58
find ways to overcome my writing
difficulties.
2 If I get stuck, I can contact a teacher .42 .48 .45 Item removed as lowest
or tutor to help me move forward loading item in Model 1
with my writing ideas
3 I can successfully use scholarly .62 .63 .82 .82
academic words and phrases when
writing in my courses.
4 I can write a major writing .57 .31 .38 Item removed due to cross loading
assignment without letting my
emotions prevent me from making
progress.
5 I can combine or synthesize .66 .57 .52 .48
multiple sources I’ve read to create
an original product or text.
6 I can use creativity when writing an .77 .62 .64 .62
academic paper.
7 When I write, I can think about my .75 .42 .43 .49
audience and write so they clearly
understand my meaning.
8 When I receive feedback on my .60 .74 .84 .82
writing, no matter how it makes me
feel, I can use that feedback to
improve my writing in the future.
9 Writing assignments relevant to .55 .62 − .33 Item removed due to cross loading
nursing can help me to feel part of
the nursing profession.
10 When I reflect on what I am writing .76 .80 .71 .75
I can make my writing better.
11 I feel I can give my writing a .77 .75 .73 .72
creative spark and still sound
professional.
12 When I read articles about my topic, .75 .70 .67 .69
the connections I feel with the ideas
of other authors can inspire me to
express my own ideas in writing.
13 When I look at the overall picture .84 .60 .63 .75
I’ve presented in my writing, I can
assess how all the pieces tell the
complete story of my topic or
argument.
14 I can write without worrying about .52 No No No Item removed for not loading on any factor
my ability to correctly apply the load load load
rules of the required style guideline
(e.g. APA format).
15 I feel I can develop my own writing .75 .91 .92 .91
voice (ways of speaking in my
writing that are uniquely me).
16 Even with very specific assignment .76 .85 .85 .87
guidelines, I can find ways of
writing my assignment to make it
original or unique.
17 I can comfortably express nursing .81 .55 .48 .47
concepts, language, and values in
my writing assignments.
18 I can recognize when I’ve wandered .71 .32 .33 .45
away from writing what my
audience needs to know and have
begun writing about interesting, but
unrelated, ideas.
19 .81 .39 .43 .51
8
Exploratory Factor Analysis Factor Loadings
Situated Academic Writing Self- Model 1 Model 2 Model 3 Model 4

Efficacy Scale (Model Testing) 1-Factor 20- 3-Factor 20-item scale 3-Factor 17-item scale 3-Factor 16-item scale
item scale
Factor Factor Factor Factor Factor Factor Factor Factor Factor
1 2 3 1 2 3 1 2 3
With each new writing assignment, I

can adapt my writing to meet the
needs of that assignment.
20 When I seek feedback on my .60 .34 .44 .53
writing, I can decide when that
feedback should be ignored or
incorporated into a revision in my
writing.
Eigen Values 9.60 4.48 4.17 2.43 3.95 3.69 2.65 4.26 3.47 2.06
Variance Explained (Cumulative) .48 .22 .43 .55 .23 .45 .59 .27 .48 .61
RMSEA (90 % CI) .096 .067 (.054− .076) .068 (.053− .079) .068 (.052− .081)
(.085− 1.02)
Tucker Lewis Index .86 .93 .94 .95
CFI .87 .95 .96 .97
RMSR .06 .03 .03 .03
CI for the total scale, subscales, and individual items are present in Table 3 for both Study 1 and 2. Cronbach’s alpha scores are
indicated and range from .79 for writing essentials to .95 for the complete scale.
6.5.3. Concurrent validity

The SESAW and WSEI instruments were completed in the survey by all participants at the same time as the SAWSES scale. The
SESAW validity co-efficient was r = .59 and the WSEI was r = .50 indicating some conceptual differences between these established
scales and the SAWSES. Descriptive data for the SESAW, WSEI and measures of writing apprehension, nursing identity, and three
writing context items are presented in Table 4.
6.5.4. Structural equation modelling

A structural equation model (SEM) was hypothesized for Study 1 which proposed that measures of nursing identity, writing
apprehension, and assessments of writing context would predict the SAWSES. SAWSES items showed visible negative skews to the
data, therefore the maximum likelihood robust (MLR) estimator was chosen in Lavaan to account for lack of normality (Rosseel, 2017).
The hypothesized model was tested and decisions for additional models were made based on the parameters observed in this model.
Three models will be reported here (Table 5).
Model 1 was developed with three latent variables based on the factors established in the EFA procedures (Fig. 2). A higher-order
latent variable, SAWSES, was created from the three factor latent variables. Apprehension, identity, and three context items (perceived
support, stakes, and value) were added as composite variables predicting the higher-order latent variable. This model had adequate fit
but examining any modification indices >10 provided theoretically logical options for adding error covariance between items 5 and 6
(two items developed as creativity items in the theoretical premise), 3 and 17 (two items referring to use of words and language), 8 and
20 (two feedback items), 6 and 11 (two creativity items). These same covariances emerged as suggested modifications in all tested
models thus were coded into the analysis for model 2 and 3.
The context-value variable (β = − 0.013, z-value = − 0.21; p = .83) was non-significant in the composite analysis and was an
insignificant contributor to model 2 (Fig. 3). Nursing identity (β = 0.12; z-value = 2.28; p = .02) and context-stakes (β = 0.14; z-value
= 2.74; p = .01) were significant, however were weak contributors to the predicting SAWSES. Model 3 (Fig. 4) removed context-value,
context-stakes, and nursing identity. In model 3 apprehension was the strongest predictor of SAWSES (β = .625; z-value = 9.63; p <
.001) with context-support making a small contribution (β = .271; z-value = 4.64; p < .001). The parameter estimates and factor
loadings for the latent variables for model 3 are presented in Table 6. The table show that the z scores for each item on its latent
variable were statistically significant. Unstandardized (B) factor loadings act like a regression coefficient and reflect the linear change
in response of the item to a unit change in the predictor; thus for each 1 standard deviation increase in the latent variable writing
essentials we can expect a 9.59 unit increase in the score for item 1. For comparison, standardized factor loadings are presented in the
Beta column and are a representation of loading of each item on its associated latent variable with a higher Beta meaning a stronger
relationship between the item and its latent variable.
6.6. Discussion
The SAWSES tool tested in a single discipline (nursing) demonstrated properties which support its initial reliability and validity.
First, it captures unique aspects of the writing context which are discriminant from other tools that focus on process or mechanics as
evidenced by the moderate correlations with previously validated writing self-efficacy tools (SESAW, r = .59; WSEI r = .50). Second,
9
Table 3
Mean, Standard Deviations, and 95 % Confidence Intervals (CI) for Scale, Subscales, and Items on the Situated Academic Writing Self-Efficacy Scale
(SAWSES): Study 1 Nursing sample (N = 255); Study 2 Interdisciplinary Undergraduate Students (N = 543); Interdisciplinary Graduate Students
(N = 264) (Possible Range of Scores 0-100).
Nursing Sample Interdisciplinary Undergraduate Interdisciplinary Graduate Sample
Items Item# Sample
Mean SD 95 %CI Mean SD 95 %CI Mean SD 95 % CI
Total Scale (α ¼ .95, .94, .95) 63.49 18.52 61.20¡65.77 72.68 17.24 71.23¡74.13 74.27 17.77 72.11¡76.42
Creative Identity (α ¼ .91, .88, 56.84 22.66 54.05¡59.64 68.02 21.09 66.24¡69.80 69.03 22.41 66.31¡71.74
.90)
I can use creativity when writing an 6 56.36 26.32 53.11− 59.61 69.91 25.21 67.78− 72.03 69.55 25.49 66.46− 72.64
academic paper.
I feel I can give my writing a creative 11 55.87 25.58 52.72− 59.03 64.52 27.50 62.20− 66.83 63.63 28.89 60.13− 67.13
spark and still sound
professional.
I feel I can develop my own writing 15 57.54 28.43 54.04− 61.05 69.42 26.29 67.21− 71.64 68.20 27.76 64.84− 71.57
voice (ways of speaking in my
writing that are uniquely me).
Even with very specific assignment 16 53.57 25.92 50.37− 56.77 63.17 26.20 60.96− 65.38 67.26 27.05 63.98− 70.54
guidelines, I can find ways of
writing my assignment to make
it original or unique.
I can comfortably express nursing
concepts, language, and values
in my writing assignments.
I can comfortably express the
17 60.89 25.40 57.76− 64.02 73.10 23.60 71.11− 75.09 76.50 22.28 73.80− 79.20
concepts, language, and values
of my discipline or major in my
writing assignments
(Modification for Study 2)
Relational Reflective (α ¼ .91,
65.92 18.99 63.58¡68.26 73.76 17.57 72.28¡75.24 74.91 17.84 72.74¡77.07
.88, .89)
When I write, I can think about my
audience and write so they 7 65.76 23.71 62.84− 68.69 72.11 23.42 70.14− 74.08 72.75 23.49 69.91− 75.60
clearly understand my meaning.
When I receive feedback on my
writing, no matter how it makes
me feel, I can use that feedback 8 75.61 24.29 72.61− 78.60 80.67 22.85 78.75− 82.60 82.50 19.04 80.19− 84.81
to improve my writing in the
future.
When I reflect on what I am writing I
10 68.51 23.59 65.60− 71.42 77.36 23.08 75.42− 79.31 78.50 23.71 75.63− 81.38
can make my writing better.
When I read articles about my topic,
the connections I feel with the
ideas of other authors can 12 60.41 25.56 57.27− 63.56 70.64 25.75 68.46− 72.81 73.09 25.91 69.95− 76.24
inspire me to express my own
ideas in writing.
When I look at the overall picture
I’ve presented in my writing, I
can assess how all the pieces tell 13 63.86 23.44 60.97− 66.75 72.34 22.76 70.42− 74.26 75.31 22.52 72.58− 78.03
the complete story of my topic
or argument.
I can recognize when I’ve wandered
away from writing what my
audience needs to know and 18 61.70 25.10 58.61− 64.80 69.30 25.24 67.17− 71.43 67.65 26.33 64.46− 70.84
have begun writing about
interesting, but unrelated, ideas.
With each new writing assignment, I
can adapt my writing to meet 19 66.79 23.09 63.94− 69.64 74.34 22.56 72.44− 76.24 74.70 23.39 71.87− 77.53
the needs of that assignment.
When I seek feedback on my writing,
I can decide when that feedback
should be ignored or 20 64.73 25.83 61.55− 67.92 73.29 24.80 71.20− 75.38 74.74 25.01 71.71− 77.77
incorporated into a revision in
my writing.
Writing Essentials (α ¼ .79, .81,
68.07 19.83 65.62¡70.52 77.60 18.41 76.05¡79.15 81.29 17.18 79.21¡83.38
.80)
1 69.98 22.53 67.20− 72.75 77.44 20.78 75.69− 79.19 76.89 21.65 74.27− 79.51
10
Nursing Sample Interdisciplinary Undergraduate Interdisciplinary Graduate Sample
Items Item# Sample
Mean SD 95 %CI Mean SD 95 %CI Mean SD 95 % CI
Even when the writing is hard, I can

find ways to overcome my
writing difficulties.
I can successfully use scholarly 3 66.62 24.93 63.54− 69.70 76.95 22.79 75.03− 78.87 83.18 20.26 80.72− 85.63
academic words and phrases
when writing in my courses.
(courses/program for grad
student survey)
I can combine or synthesize multiple 5 67.62 23.54 64.71− 70.52 78.41 21.46 76.60− 80.22 83.81 18.83 81.53− 86.09
sources I’ve read to create an
original product or text.
Items removed from Scale after EFA

If I get stuck, I can contact a teacher 2 66.51 28.89 62.95− 70.07
or tutor to help me move
forward with my writing ideas
I can write a major writing 4 61.92 26.62 58.64− 65.20
assignment without letting my
emotions prevent me from
making progress.
Writing assignments relevant to 9 58.35 29.36 54.73− 61.97
nursing can help me to feel part
of the nursing profession.
I can write without worrying about 14 53.61 30.40 49.86− 57.36
my ability to correctly apply the
rules of the required style
guideline (e.g. APA format).
Table 4
Mean, Standard Deviation of Supporting Scales (N = 228).
Scales Possible range in Mean SD
scores
Writing Self-Efficacy Instrument (WSEI) (α ¼ .93) 0− 800 616.2 154.4

Self-Efficacy Scale for Academic Writing (SESAW) (α ¼ .85) 9− 36 26.54 3.78
Writing Apprehension Test (WAT) (α ¼ .92) 20− 100 60.63 12.91
Professional Identity Questionnaire (PIQ) (α ¼ .83) 8− 40 33.55 4.09
This class and/or my institution can support me in the way I need to be successful with writing this essay assignment 0− 100 65.81 23.93
(Context perceived support item)
It is important for my future that I get the grade I need on this writing assignment so I do well in this course. (Context 0− 100 82.30 19.88
perceived stakes item)
I feel that the writing assignment in this class will be important for my learning in the nursing program. (Context 0− 100 63.76 28.63
perceived value item)
Table 5
Model Goodness of fit for three SEM models with MLR estimator N = 228.
Model Model Details X2 df p- CFI TLI RMSEA (90 %
value CI)
1 Hypothesized model with apprehension, nursing identity, and the context items (support, 318.74 200 <.001 .93 .92 .066
stakes, value) inputted as composite predictors of the higher order SAWSES latent (.054− .078)
variable. (Fig. 2)
Hypothesized model with modification indices considered from Model 1 (Fig. 3) .056
2 269.79 200 <.001 .96 .95
Modification covariance: Items 5 and 6, 3 and 17, 8 and 20, 6 and 11 (.042− .068)
Only the two strongest predictors (Writing apprehension and context-support) and .052
3 189.45 152 <.001 .97 .96
modification indices considered. (Fig. 4) (.036− .067)
initial reliability assessments demonstrate Cronbach’s alpha scores for all subscales that confirm its internal consistency. Parallel form
analysis showed the scale holds similar properties when completed by paper and pencil which may be beneficial for researchers who
are unable to deliver the questionnaire electronically. Exploratory factor analysis demonstrated three practical and relevant subscales
– Writing Essentials, Relational-Reflective writing, Creative Identity – potentially useful for diagnosing aspects of self-efficacy
11
Fig. 2. Hypothesized Structural Equation Model 1 in Nursing Student Sample.
theoretically important to assessing writing in classroom settings. In testing convergent validity, the structural equation models
demonstrated that writing apprehension is predictive of the latent SAWSES variable. Previous research has also shown writing
apprehension to be related to other measures of writing self-efficacy (Goodman & Cirka, 2009; Huerta, Goodson, Beigi, & Chlup, 2017;
Pajares & Johnson, 1994; Sanders-Reio, Alexander, Reio, & Newmann, 2014) and supports construct validity of SAWSES. Nursing
identity and items assessing writing context were contributors in the hypothesized model (Model 1 and 2), however, only the
context-support item made a meaningful contribution to the final model (Model 3). The identity tool may be too nursing
practice-oriented to assess nursing knowledge identity, however, that it produced a small effect on writing self-efficacy is promising
initial evidence of the proposed theoretical relationship. Perceptions of the importance of feeling supported in the writing environment
were confirmed in student statements during the cognitive interviews. Understanding the teacher and their expectations was an
important factor in student self-beliefs in their writing ability (Mitchell, 2020f). These exploratory findings warrant further
investigation.
Given these successes, it was necessary to confirm if the factor structure was still relevant in a broader population of interdisci
plinary undergraduate and graduate students who may experience different writing concerns not captured in the nursing population.
7. Study 2: multigroup confirmatory factor analysis
Undergraduate and graduate students write for different purposes and face different writing expectations within their disciplinary
contexts which may impact the structure and functioning of the items on the SAWSES scale and enlighten researchers to possible
discipline-specific features of the tool.
7.1. Participants
The characteristics of participants in Study 2 are presented in Table 1. Undergraduate and graduate students were recruited via two
routes: social media recruitment through Twitter to gain an international sample of mainly graduate students (13.2 % of the sample),
and from the same three institutions where nursing students had been sampled in Study 1 (86.8 % of the sample). Sample sizes from the
survey included 543 undergraduate and 264 graduate student surveys for a total sample of 807. The interdisciplinary population had a
higher proportion of male respondents compared to nurses. The mean age of graduate students was 8 years older than the
12
Fig. 3. Hypothesized Model 2 with Modification Indices Considered in Nursing Student Sample.
undergraduates. Both the undergraduate and graduate interdisciplinary populations had higher proportions of students reporting that
they “generally liked writing” and a much smaller proportion who “generally disliked writing” compared to the undergraduate
nursing-only sample. Over 40 specialties or research focuses were identified for each of the undergraduate and graduate groups. The
most common specialties are reported in Table 1.
7.2. Procedures
The survey was created in Qualtrics and contained the 16 SAWSES items retained in the EFA along with a demographic survey. An
open-ended text box was included to allow for student comments about the items. The social media posts were sent from the first
author’s personal academic Twitter account with over 10,000 followers during the month of December 2019. Registered students were
sent an email with the survey link via a third party within each participating institution. Both the email and social media posts
introduced the study with a graphic asking students the question, “What defines you as a student writer?” A) I like writing. B) I hate
writing. C) It depends. The phrasing was chosen to signal the students who disliked writing, a population of students presumed to feel
less enthusiastic about participating, that their opinions about writing were valid and were important as respondents to the survey.
Missing data on SAWSES was <1 %. Because the survey was publicly available on social media, for quality control, survey respondents
were deleted if they did not disclose their institution of study.
7.3. Data analysis
Multigroup confirmatory factor analysis (MGCFA) was conducted in RStudio (v. 1.2.5033) primarily using the Lavaan package
(Rosseel, 2017) and the equaltestMI package (Jiang, Mai, & Yuan, 2017). The processes for analysis, interpretation, and reporting were
drawn from recommendations from Beaujean (2014); Putnick and Bornstein (2016), and Cheung and Rensvold (2002). Data and R
13
Fig. 4. Best Fit Model 3 Apprehension and Context-Support as Predictors and Modification Indices Considered in Nursing Student Sample.
Table 6
Parameter Estimates and Factor Loadings for Model 3.
Latent Factor Indicator (Item) B SE Z p-value Beta
Writing Essentials 1 9.59 1.14 8.39 <.001 .75

3 10.73 1.36 7.89 <.001 .76
5 10.20 1.21 8.45 <.001 .74
Relational Reflective 7 3.88 1.75 2.21 .03 .77
8 3.14 1.48 2.12 .03 .61
10 3.95 1.83 2.16 .03 .79
12 4.10 1.89 2.16 .03 .75
13 4.37 2.03 2.15 .03 .87
18 4.04 1.90 2.13 .03 .75
19 3.96 1.78 2.22 .03 .80
20 3.35 1.59 2.11 .04 .62
Creative Identity 6 8.48 1.70 4.98 <.001 .79
11 8.65 1.76 4.91 <.001 .81
15 10.01 2.09 4.79 <.001 .84
16 9.11 2.00 4.57 <.001 .84
17 8.80 1.80 4.90 <.001 .83
SAWSES Writing Essentials 0.91 0.13 6.87 <.001 .81
Relational Self-Reflective 2.95 1.47 2.01 .04 .98
Creative Identity 1.40 0.31 4.48 <.001 .91
14
code files are publicly available (Mitchell, 2020a, 2020d).
7.4. Results
The MGCFA analysis used processes of invariance testing to explore differences between groups using several steps (Table 7). The
factor structure that emerged from the EFA in Study 1 was the model explored for the full sample (N = 807) CFA. This model
demonstrated reasonable fit: CFI of .942 and RMSEA = .074 (90 % CI .066− .082). Since model fit was adequate, modification indices
were not considered in this phase, so the focus turned to identifying how the two groups differed from each other on responses to the
individual items. MGCFA procedures found slightly poorer fit of the model in graduate students. A hierarchy of stacked constraints was
then placed on the two model groups simultaneously to establish if the groups differed in any meaningful way in their response to the
individual items. While a non-significant chi square difference test is most valued in model testing, this parameter is highly sensitive to
sample size and was therefore not able to be achieved in this study (Beaujean, 2014). Change in CFI scores from step to step was
monitored for a drop in index >.01 (Cheung & Rensvold, 2002). Additionally, recommendations that RMSEA changes no more than
.015 and SRMR changes no more than .015–.03 have also been suggested as ideal (Putnick & Bornstein, 2016). Constraints were added
in the following order: configural invariance explores how the groups differ in factor structure (configural model), metric invariance
forces equivalent factor loadings on both groups (metric model), scalar invariance forces equivalent item intercepts (scalar model), and
strict residual models forces equivalent item residuals and error variances (strict model). CFI scores did not drop in excess of .01
indicating that the differences in item responses between undergraduate and graduate students were not statistically meaningful.
7.5. Discussion
The results of the MGCFA on an independent large sample of interdisciplinary undergraduate and graduate students provides
preliminary evidence that the factor structure of the Situated Academic Writing Self-Efficacy Scale may be generalizable to a more
heterogeneous student population of varying levels of experience in writing; thus the scale demonstrates properties of validity in
populations beyond nursing students where it was initially developed. The slightly poorer model fit in graduate students when
considered independently without invariance constraints, does suggest that additional testing of the instrument in graduate students
may provide additional understanding about the factor structure of the scale.
8. Exploring differences between study 1 and study 2 samples
Additional tests to assess patterns of responses in known groups (nursing and interdisciplinary undergraduate and graduate stu
dents) were undertaken. These included exploring differences in student response based on student type and writing attitude, and an
exploration of item discrimination patterns between samples.
Undergraduate nursing students appeared to self-assess lower means on all scales and subscales (Table 3). One-way analysis of
variance procedures identified statistically significant differences in reported means scores between student types on all factors: Total
SAWSES F(2, 1061) = 29.93, p < .001; Writing Essentials F(2, 1061) = 36.30, p < .001; Relational Reflective F(2, 1061) = 20.55, p <
.001; Creative Identity F(2, 1061) = 27.15, p < .001. Post hoc analysis (Tukey HSD) confirms the hypothesis that nursing students had
lower mean scores than both the interdisciplinary undergraduates (p < .001 for all subscales and total SAWSES) and graduate students
(p < .001 for all subscales and total SAWSES). Between interdisciplinary undergraduate and graduate students, statistical significance
was only observed on the Writing Essentials subscale (p = .021).
8.1. Writing attitude
Demographic differences in student writing attitudes (whether they liked, disliked, or felt somewhere in between about writing)
were evident with nurses reporting higher rates of disliking writing. These relationships were tested using a chi-square test for in
dependence which found a significant difference between student type and attitude toward writing, X2 (4, 1062) = 104.9, p < .001
with a medium effect size Cramer’s V = .222 (p < .001). Isolating the two most comparable groups (nursing undergraduates and
interdisciplinary undergraduates) the chi-square test of independence identified significance X2 (2, 798) = 77.8, p < .001 with a
medium effect size Cramer’s V = .312 (p < .001). These differences in groups prompted the exploration of patterns of means in re
sponses to SAWSES by subscale. The expectation was that students who disliked writing would report lower mean scores than the other
Table 7
Multigroup CFA Tests of Goodness of Fit and Measurement Invariance on SAWSES.
Model X2 df RMSEA (90 %CI) CFI SRMR ΔRMSEA ΔCFA ΔSRMR
Overall Model 376.97 101 0.074 (.066− .082) 0.942 0.043 – – –

Undergraduate Model 261.72 101 0.068 (.058− .078) 0.947 0.041 – – –
Graduate Model 257.39 101 0.093 (.079− .107) 0.919 0.057 – – –
Configural Invariance 519.30 202 0.077 (.069− .085) 0.937 0.046 – – –
Metric Invariance 547.56 215 0.076 (.068− .084) 0.935 0.055 .001 .002 − .009
Scalar Invariance 594.49 228 0.076 (.069− .084) 0.930 0.057 0 .005 − .002
Strict Invariance 628.47 244 0.076 (.069− .083) 0.926 0.058 0 .004 − .001
15
two groups. SPSS was used to conduct a one-way ANOVA and confirmed differences between attitude groups in all samples. On post
hoc analysis, all attitude categories produced statistically different SAWSES means by subscale (p < .001) (Table 8).
8.2. Exploring item discrimination
The items on the SAWSES were developed to be of varying complexity, requiring evolving skill levels to master. Exploring trends in
mean scores on the three identified subscales (Table 3) demonstrated patterns in means where all groups reported their highest scores
in the Writing Essentials subscale and their lowest scores in the Creative Identity subscale indicating the possibility that participants
viewed the writing essentials items as easier to endorse with higher scores than the Creative Identity items. This pattern suggested
there may be varying degrees of discriminant ability among the items between the three subscales and in the different populations. The
SjPlot package in Rstudio (Lüedecke, 2020) was used to assess item discrimination for each item. Discrimination refers to an item’s
ability to discriminate between individuals where those who score high on an item are also more likely to obtain a higher score for the
scale. Discrimination scores can range between .20 and 1.00 and higher scores are an indication of an item that is more discriminating.
The pattern among mean scores for the three subscales was suggestive that the Writing Essentials subscale was less discriminating and
more likely to elicit higher scores from participants despite how a participant may feel about their Relational-Reflective or Creative
Identity self-efficacy levels. An average was recorded for each subscale to explore differences in discrimination between subscales
(Table 9).
The Writing Essentials category contains the least discriminating items. The Creative Identity subscale contains the most
discriminating items. Scores on the Relational-Reflective subscale trended toward falling between Writing Essentials and Creative
Identity; however, discrimination scores for the undergraduate and graduate interdisciplinary students had similar scores to writing
essentials.
9. Discussion
This project represents a large sample size instrument development study only limited by convenience sample recruitment. The
Situated Academic Writing Self-Efficacy Scale was designed to detect situated (or contextual) influences on students’ self-beliefs about
their writing. Overall SAWSES demonstrated strong reliability scores in nursing and interdisciplinary post-secondary students.
Construct validity through predicted relationships with writing apprehension, previously established measures of writing self-efficacy,
and perception of contextual writing support in nursing students is also evident.
The SAWSES scale was able to detect differences in writing self-efficacy in students who report liking, disliking, or feeling
somewhere in between about writing. Interdisciplinary confirmation of the scale structure is promising evidence of expanded
generalizability of the scale beyond the nursing student population where the scale was developed. As writing genres, expectations,
and disciplinary influences are known to change from context to context and influence writing self-efficacy in the process (Mitchell &
McMillan, 2018; Sanders-Reio et al., 2014; Van de Poel & Gasiorek, 2012), the instrument’s structure and function would benefit from
testing in more homogeneous populations of disciplinary students. The fit of the scale structure to a generalized population of graduate
students was less adequate than for undergraduates which may indicate the scale requires additional revisions to better capture the
writing self-efficacies of graduate students. Graduate students often face higher expectations in terms of reach of their work through
publication, expectations that they create new knowledge and communicate with knowledge users and non-academics, thus, they often
write for a higher variety of audiences and in higher stakes contexts.
The developmental patterns evident in the discrimination score findings between the subscales is a pragmatic finding that may be
useful to researchers and writing instructors. Different writing demands are made in every writing context (Slomp, 2012), meaning
that scores on each subscale could fluctuate from context to context. The Writing Essentials subscale includes a parsimonious set of
items that represent some of the essential skills for managing the academic writing environment at the post-secondary level. The
Relational-Reflective subscale includes actions that writers need to consider outside of the act of writing, including connecting with
others, engaging with source materials, and being self-reflective about process. The Creative Identity subscale represents items where
writing’s transformative benefits are assessed – where mastery may only be seen in writers who have achieved the highest expertise in
writing and disciplinary identity development. High scores on the Creative Identity factor may identify the most engaged writers. Low
scores among students in any factor could help writing faculty target areas for intervention within disciplinary populations. For
example, in nursing, it is not uncommon for the students entering a graduate program to be 10 years out of their undergraduate degree.
This is due to practice requirements in the nursing profession. If this group of students entering a new academic program score
themselves low on writing essentials, time away from writing at the post-secondary level might have stimulated self-doubt about their
capabilities to meet expectations. A review of what academic language and synthesis means and being responsive to writing emotions
might be a first step toward supporting novice student needs. Scoring themselves low on relational reading and incorporating of
sources, might simulate a need to address how to read texts within a particular substantive area. Confusion over allowances for
creativity could start a discussion about what creativity might mean within the discipline or to a specific instructor. Many contextual
conditions are tacit. What may seem obvious to an experienced faculty member may be baffling, unknown, or unfamiliar to a novice
student (Lea & Street, 1998).
The item discrimination findings should also be viewed with the knowledge that students in previous writing self-efficacy studies
have exhibited tendencies to overinflate scores when facing contexts that are unfamiliar (Jones, 2008; Meier, McCarthy, & Schmeck,
1984; Stewart, Seifert, & Rolheiser, 2015; Zimmerman & Kitsantas, 2002). A student can only appraise their own abilities with
reflection upon past experience. The tendency to underestimate the writing knowledge required in a new context may be most
16
Table 8
Mean, Standard Deviation, and One-Way ANOVA Results for Self-Reported Writing Attitude and SAWSES in all Student Samples.
Likes Writing Somewhere In Between Dislikes Writing F* (df = 2)
Nursing (N ¼ 255) n = 44 n = 108 n = 103

Total Scale 79.92 (15.2) 65.47 (16.4) 54.39 (16.5) 39.53
Writing Essentials 84.92 (15.2) 69.29 (17.7) 59.60 (18.9) 31.63
Relational Reflective 80.95 (14.4) 67.49 (17.4) 57.85 (18.1) 28.53
Creative Identity 75.25 (20.5) 59.95 (20.8) 45.73 (19.2) 35.49
Interdisciplinary Undergraduate (N ¼ 543) n = 237 n = 219 n = 87

Total Scale 81.52 (13.5) 69.72 (15.0) 56.10 (16.9) 102.55
Graduate (N ¼ 264) n =124 n = 111 n = 29

Total Scale 81.24 (14.4) 70.82 (16.0) 57.67 (22.5) 29.54
For comparison purposes, each item scored 0− 100; Scale and subscale means calculated as total score/#items on scale.
*
All F values p < .001.
Table 9
Item Discrimination for all student groups.
Item
Discrimination
N U G
Writing Essentials
Even when the writing is hard, I can find ways to overcome my writing difficulties. .64 .63 .61
I can successfully use scholarly academic words and phrases when writing in my courses. (courses/program for grad student survey) .66 .64 .69
I can combine or synthesize multiple sources I’ve read to create an original product or text. .62 .69 .66
Subscale Average .64 .65 .65
Relational Reflective Writing

When I write, I can think about my audience and write so they clearly understand my meaning. .71 .70 .74
When I receive feedback on my writing, no matter how it makes me feel, I can use that feedback to improve my writing in the future. .62 .55 .56
When I reflect on what I am writing I can make my writing better. .73 .67 .64
When I read articles about my topic, the connections I feel with the ideas of other authors can inspire me to express my own ideas in writing. .70 .60 .51
When I look at the overall picture I’ve presented in my writing, I can assess how all the pieces tell the complete story of my topic or argument. .83 .73 .75
I can recognize when I’ve wandered away from writing what my audience needs to know and have begun writing about interesting, but .72 .59 .71
unrelated, ideas.
With each new writing assignment, I can adapt my writing to meet the needs of that assignment. .75 .75 .80
When I seek feedback on my writing, I can decide when that feedback should be ignored or incorporated into a revision in my writing. .63 .57 .63

Creative Identity
I can use creativity when writing an academic paper. .76 .69 .79
I feel I can give my writing a creative spark and still sound professional. .79 .74 .80
I feel I can develop my own writing voice (ways of speaking in my writing that are uniquely me). .79 .69 .84
Even with very specific assignment guidelines, I can find ways of writing my assignment to make it original or unique. .81 .79 .77
I can comfortably express nursing concepts, language, and values in my writing assignments.
.76 .62 .61
I can comfortably express the concepts, language, and values of my discipline or major in my writing assignments (Modification for Study 2)
N = Nursing, UG = Undergraduate, G = Graduate.
pronounced as students move from discipline to discipline, switch between different writing genres (e.g. academic writing to reflective
writing or creative writing), or from one level of writing to another, such as when starting a graduate program. Students develop
abilities for sophisticated thinking about their writing and their use of disciplinary language as they progress through programs or
begin targeting additional audiences – a context that is often the case for graduate students seeking publication of their work for the
first time. Increased complexity in writing means new writing struggles can be revealed (Camp, 2012). This effect can be seen in how
the results of this study demonstrate only minor differences between undergraduate and graduate interdisciplinary students. They
were similar not just on patterns in the way they responded to the items as demonstrated in the MGCFA but also on mean scores for the
individual items, and item discrimination. As new contexts for writing are introduced, new uncertainties about how to manage those
contexts emerge. The instrument can be a conversation starter to help students understand how to navigate increasing complexities in
17
writing and develop realistic self-appraisals of their writing as they move from undergraduate to graduate school. Broadly, more
research is needed to understand how patterns of over inflation in self-assessments of writing ability function in undergraduate and
graduate student contexts.
One notable marker of contextual understanding of writing was how students perceived creativity in their writing environment.
Some nursing student participants in the cognitive interviews felt they were not allowed to be creative. Comments about inability to
use creativity in academic writing were also common in the open-ended box provided in the interdisciplinary survey in Study 2.
Knowing what defines creativity and where it is appropriate within a discipline or within a specific writing genre requires engagement
within a disciplinary community and self-identification within that community (McVey, 2008; Mitchell, 2018). The creative identity
subscale could be a useful tool for starting discussions with students about the definitions of creativity in academic writing in their
discipline and stimulating reflections on writing voice.
The cognitive interviews conducted when piloting the instrument items with a nursing student population, provided evidence that
the items triggered students to closely examine the conditions of writing in their discipline, and within the various experiences of
individual classrooms when providing an explanation for how they scored items (Mitchell, 2020f). Thus, the tool holds potential
beyond its measurement capabilities as a reflective device for classroom instructors, writing centre tutors, graduate student advisors,
and others who wish to understand the lived experience of student writing within their post-secondary context.
10. Conclusion
This validation study has demonstrated that the SAWSES instrument meets acceptable thresholds for reliability and validity for use
in interdisciplinary post-secondary writing contexts and research studies. The tool has the potential to detect domains of contextual
writing requiring intervention in areas such as basic writing tasks, navigating relational and reflective writing, or starting conver
sations about what creativity means to identity development within a discipline. SAWSES could serve as a reflective tool while working
with individual students to identify beliefs about writing within post-secondary contexts or identify where students are under
estimating the complexities of their writing context.
Funding
This research was funding by the Western and Northern Region Canadian Association of Schools of Nursing Graduate Student
Research Grant and a University of Manitoba College of Nursing Endowment Fund Graduate Student Research Grant.
CRediT authorship contribution statement
Kim M. Mitchell: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing - original draft, Writing -
review & editing, Project administration, Funding acquisition. Diana E. McMillan: Conceptualization, Validation, Writing - review &
editing, Supervision, Funding acquisition. Michelle M. Lobchuk: Conceptualization, Validation, Writing - review & editing, Super
vision. Nathan C. Nickel: Conceptualization, Validation, Writing - review & editing, Supervision. Rasheda Rabbani: Validation,
Formal analysis, Writing - review & editing. Johnson Li: Conceptualization, Validation, Writing - review & editing, Supervision.
Declaration of Competing Interest
The authors report no declarations of interest.
Acknowledgements
The authors would like to acknowledge the contribution of the 15 expert Delphi panelists (Jo-Anne Andre, University of Calgary,
Canada Vasiliki Betihavas, University of Sydney, Australia Bridget Blankely, University of Southampton, UK Trevor Day, Royal Lit
erary Fund, UK Paul Glew, Western Sydney University, Australia Caroline Havery University of Technology Sydney, Australia Fiona
Henderson, Victoria University, Australia Diana Jefferies, Western Sydney University, Australia Andrea Klien, Independent Scholar,
Germany Sabine Korin, University of Applied Sciences Hamburg, Germany Lynne Kuhl, Viterbo University, USA Linda McCloud-
Bondoc, Athabasca University, Canada Brenna Quinn, University of Massachusetts Lowell, USA Coleen Toronto, Curry College, USA
Christina West, University of Manitoba, Canada) who participated in this research and graciously gave permission for their name to be
presented in the acknowledgements of this work.
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the online version, at doi: https://doi.org/10.1016/j.asw.2021.
100524.
18
References
Adams, K., Hean, S., Sturgis, P., & Macleod Clark, J. (2006). Investigating the factors influencing professional identity of first-year health and social care students.
Learning in Health and Social Care, 5(2), 55–68. https://doi.org/10.1111/j.1473-6861.2006.00119.x
Bandura, A. (1997). Self-efficacy: The exercise of control. W. H Freeman.
Bandura, A. (2006). Toward a psychology of human agency. Perspectives on Psychological Science, 1(2), 164–180. https://doi.org/10.1111/j.1745-6916.2006.00011.x
Bartholomae, D. (1986). Inventing the university. Journal of Basic Writing, 5(1), 4–23.
Beaujean, A. A. (2014). Latent variable modeling using R: A step-by-step guide. Routledge.
Camp, H. (2012). The psychology of writing development – and its implications for assessment. Assessing Writing, 17, 92–105. https://doi.org/10.1016/j.
asw.2012.01.002
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9(2), 233–255. https://
doi.org/10.1207/S15328007SEM0902_5
Daly, J. A., & Miller, M. D. (1975). The empirical development of an instrument to measure writing apprehension. Research in the Teaching of English, 9(3), 242–249.
https://www.jstor.org/stable/40170632.
Daly, J. A., & Miller, M. D. (2013). Writing apprehension test (WAT). Measurement instrument database for the social science. http://www.midss.org/sites/default/files/
writing_apprehension_test.pdf.
Flake, J. K., Pek, J., & Hehman, E. (2017). Construct validation in social and personality research: Current practice and recommendations. Social Psychology and
Personality Science, 8(4), 370–378. https://doi.org/10.1177/1948550617693063
Flower, L., & Hayes, J. R. (1981). A cognitive process theory of writing. College Composition and Communication, 32(4), 365–387. https://doi.org/10.2307/356600
Golombek, C., Klingsieck, K. B., & Scharlau, I. (2019). Assessing self-efficacy for self-regulation of academic writing: Development and validation of a scale. European
Journal of Psychological Assessment, 35(5), 751–761. https://doi.org/10.1027/1015-5759/a000452
Goodman, S. B., & Cirka, C. C. (2009). Efficacy and anxiety: An examination of writing attitudes in a first-year seminar. Journal on Excellence in College Teaching, 20(3),
5–28.
Graham, S. (2018). A revised writer(s)-within-community model of writing. Educational Psychologist, 53(4), 258–279. https://doi.org/10.1080/
00461520.2018.1481406
Huerta, M., Goodson, P., Beigi, M., & Chlup, D. (2017). Graduate students as academic writers: Writing anxiety, self-efficacy and emotional intelligence. Higher
Education Research & Development, 36(4), 716–729. https://doi.org/10.1080/07294360.2016.1238881
Hyland, K. (2003). Genre-based pedagogies: A social response to process. Journal of Second Language Writing, 12(2003), 17–29. https://doi.org/10.1016/S1060-3743
(02)00124-8
Jefferies, D., McNally, S., Roberts, K., Wallace, A., Stunden, A., D’Souza, S., et al. (2018). The importance of academic literacy for undergraduate nursing students and
its relationship to future professional clinical practice: A systematic review. Nurse Education Today, 60, 84–91. https://doi.org/10.1016/j.nedt.2017.09.020
Jiang, G., Mai, Y., & Yuan, K.-H. (2017). EqualtestMI: Examine measurement invariance via equivalence testing and projection method [R package version 0.1.0]. https://
CRAN.R-project.org/package=equaltestMI.
Jones, E. (2008). Predicting performance in first-semester college basic writers: Revisiting the role of self-beliefs. Contemporary Educational Psychology, 33, 209–238.
https://doi.org/10.1016/j.cedpsych.2006.11.001
Knafl, K., Deatrick, J., Gallo, A., Holcombe, G., Bakitas, M., Dixon, J., et al. (2007). The analysis and interpretation of cognitive interviews for instrument
development. Research in Nursing & Health, 30, 224–234. https://doi.org/10.1002/nur.20195
Lea, M. R., & Street, B. V. (1998). Student writing in higher education: An academic literacies approach. Studies in Higher Education, 23(2), 157–172. https://doi.org/
10.1080/03075079812331380364
Lock, T., & Johnston, M. (2016). Developing an individual and collective self-efficacy scale for the teaching of writing in high schools. Assessing Writing, 28, 1–14.
https://doi.org/10.1016/j.asw.2016.01.001
Lüedecke, D. (2020). SjPlot: Data visualization for statistics in social science [R package version 2.8.3]. https://doi.org/10.5281/zenodo.1308157
McVey, D. (2008). Why all writing is creative writing. Innovations in Education and Teaching International, 45(3), 289–294. https://doi.org/10.1080/
1470329082176204
Meier, S., McCarthy, P. R., & Schmeck, R. R. (1984). Validity of self-efficacy as a predictor of writing performance. Cognitive Therapy and Research, 8(2), 107–120.
https://doi.org/10.2307/357865
Mitchell, K. M. (2018). Constructing writing practices in nursing. Journal of Nursing Education, 57(7), 399–407. https://doi.org/10.3928/01484834-20180618-04
Mitchell, K. M. (2020a). Interdisciplinary undergraduate and graduate student data (Version V1) [Data set]. Harvard Dataverse. https://doi.org/10.7910/DVN/M07HQ7
Mitchell, K. M. (2020b). Nursing student data (Version V2) [Data set]. Harvard Dataverse. https://doi.org/10.7910/DVN/MQ8EP0
Mitchell, K. M. (2020c). Replication data for: R code for exploratory factor analysis SAWSES validation (Version V1) [Data set]. https://doi.org/10.7910/DVN/CHJKJM
Mitchell, K. M. (2020d). Replication data for: R code for multi-group CFA SAWSES validation (Version V1) [Data set]. https://doi.org/10.7910/DVN/UDXOGA
Mitchell, K. M. (2020e). Replication data for: R code for structural equation model SAWSES validation (Version V1) [Data set]. https://doi.org/10.7910/DVN/ZKFYT6
Mitchell, K. M. (2020f). The theoretical construction and measurement of writing self-efficacy. Doctoral thesis,. University of Manitoba https://mspace.lib.umanitoba.ca/
handle/1993/35135.
Mitchell, K. M., & McMillan, D. E. (2018). A curriculum-wide assessment of writing self-efficacy in a baccalaureate nursing program. Nurse Education Today, 70,
20–27. https://doi.org/10.1016/j.nedt.2018.08.003
Mitchell, K. M., McMillan, D. E., & Lobchuk, M. M. (2019). Applying the “social turn” in writing scholarship to perspectives on writing self-efficacy. Journal of Learning
Development in Higher Education, 15(2019). https://journal.aldinhe.ac.uk/index.php/jldhe/article/view/512/pdf.
Mitchell, K. M., McMillan, D. E., Lobchuk, M., & Nickel, N. (2018). Psychometric Testing of a New Instrument to Measure Writing Self-Efficacy: A Proposed Mixed-Methods
Delphi Study. International Nursing Research Congress – Sigma Theta Tau –. Melbourne Australia https://sigma.nursingrepository.org/handle/10755/624591.
Mitchell, K. M., Harrigan, T., Stefansson, T., & Setlack, H. (2017). Exploring self-efficacy and anxiety in first-year nursing students enrolled in a discipline-specific
scholarly writing course. Quality Advancement in Nursing Education, 3(1), 4. https://doi.org/10.17483/2368-6669.1084
Mitchell, K. M., Rieger, K. L., & McMillan, D. E. (2017). A template analysis of writing self-efficacy measures. Journal of Nursing Measurement, 25(2), 205–223. https://
doi.org/10.1891/1061-3749.25.2.205
Pajares, F., & Johnson, M. J. (1994). Confidence and competence in writing: The role of self-efficacy, outcome expectancy, and apprehension. Research in the Teaching
of English, 28(3), 313–331. http://www.jstor.org/stable/40171341.
Pajares, F., & Valiante, G. (2006). Self-efficacy beliefs and motivation in writing development. In C. A. MacArthur, S. Graham, & J. Fitzgerald (Eds.), Handbook of
writing research (pp. 158–170). Guilford Press.
Paré, A. (2014). Rhetorical genre theory and academic literacy. Journal of Academic Language and Learning, 8(1), A83–A94.
Plakhotnik, M. S., & Rocco, T. S. (2016). Increasing writing self-efficacy of adult learners: Different approaches, different results. Adult Learning, 27(4), 160–167.
https://doi.org/10.1177/1045159515611879
Putnick, D. L., & Bornstein, M. H. (2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research.
Developmental Review, 41, 71–90. https://doi.org/10.1016/j.dr.2016.06.004
Revelle, W. (2017). Package ‘psych’. September 9, https://cran.r-project.org/web/packages/psych/psych.pdf.
Rosseel, Y. (2017). The lavaan tutorial. December 18, http://lavaan.ugent.be/tutorial/tutorial.pdf.
Russell, D., Lea, M., Parker, J., Street, B., Donahue, T., et al. (2009). Exploring notions of genre in academic literacies and writing across the curriculum: Approaches
across countries and contexts. In C. Bazerman (Ed.), Genre in a changing world: Perspectives on writing (pp. 459–491). WAC Clearinghouse.
Sanders-Reio, J., Alexander, P. A., Reio, T. G., & Newmann, I. (2014). Do students’ beliefs about writing relate to their writing self-efficacy, apprehension, and
performance? Learning and Instruction, 33, 1–11. https://doi.org/10.1016/j.learninstruc.2014.02.001
19
Schmidt, K. M., & Alexander, J. E. (2012). The empirical development of an instrument to measure writerly self-efficacy in writing centers. Journal of Writing
Assessment, 5(1), 62. http://www.journalofwritingassessment.org/article.php?article=62.
Shell, D. F., Murphy, C. C., & Bruning, R. H. (1989). Self-efficacy and outcome expectancy mechanisms in reading and writing achievement. Journal of Educational
Psychology, 81(1), 91–100. https://doi.org/10.1037/0022-0663.81.1.91
Slomp, D. H. (2012). Challenges in assessing the development of writing ability: Theories, constructs and methods. Assessing Writing, 17(2012), 81–91. https://doi.
org/10.1016/j.asw.2012.02.001
Stewart, G., Seifert, T. A., & Rolheiser, C. (2015). Anxiety and self-efficacy’s relationship with undergraduate students’ perceptions of the use of metacognitive writing
strategies. The Canadian Journal for the Scholarship of Teaching and Learning, 6(1), 4. https://doi.org/10.5206/cjsotl-rcacea.2015.1.4
Teng, L. S., Peigian, P. S., & Xu, L. (2018). Conceptualizing writing self-efficacy in English as a foreign language contexts: Scale validation through structural equation
modeling. TESOL Quarterly, 52(4), 911–942. https://doi.org/10.1002/tesq.432
Turner, J. C., & Nolen, S. B. (2015). Introduction: The relevance of the situative perspective in educational psychology. Educational Psychologist, 50(3), 167–172.
https://doi.org/10.1080/00461520.2015.1075404
Van de Poel, K., & Gasiorek, J. (2012). Effects of an efficacy-focused approach to academic writing on students’ perceptions of themselves as writers. Journal of English
for Academic Purposes, 11, 294–303. https://doi.org/10.1016/j.jeap.2012.07.003
Willis, G. B. (2005). Cognitive interviewing: A tool for improving questionnaire design. Sage.
Worthington, M., Salamonson, Y., Weaver, R., & Cleary, M. (2013). Predictive validity of the Macleod Clark professional identity scale for undergraduate nursing
students. Nurse Education Today, 33, 187–191. https://doi.org/10.1016/j.nedt.2012.01.012
Yagelski, R. (2011). Writing as a way of being: Writing instruction, nonduality, and the crisis of sustainability. Hampton: Press.
Zimmerman, B. J., & Kitsantas, A. (2002). Acquiring writing revision and self-regulatory skill through observation and emulation. Journal of Educational Psychology, 94
(4), 660–668. https://doi.org/10.1037//0022-0663.94.4.660
Dr. Kim M. Mitchell is the Research and Quality Assurance coordinator in the Department of Nursing, Red River College This paper represents the culmination of her
doctoral work with the College of Nursing, Rady Faculty of Health Sciences.
Dr. Diana E. McMillan is Associate Professor in the College of Nursing, Rady Faculty of Health Sciences, University of Manitoba and Clinical Chair, Health Sciences
Centre.
Dr. Michelle M. Lobchuk is Associate Professor in the College of Nursing, Rady Faculty of Health Sciences, University of Manitoba.
Dr. Nathan Nickel is Associate Professor in the Department of Community Health Sciences, Rady Faculty of Health Sciences, and Associate Director and Research
Scientist at the Manitoba Centre for Health Policy, University of Manitoba.
Dr. Rasheda Rabbani is a Biostatistician at the Manitoba Center for Nursing and Health Research, College of Nursing, the George and Fay Yee Centre for Health
Innovation, and Assistant Professor in the Department of Community Health Sciences, Rady Faculty of Health Sciences, University of Manitoba.
Dr. Johnson Li is Associate Professor in Quantitative Psychology, Faculty of Arts, University of Manitoba.
20

Development and Validation of The Situated Academic Writing Self-Efficacy Scale (SAWSES)

Uploaded by

Copyright:

Available Formats

Development and Validation of The Situated Academic Writing Self-Efficacy Scale (SAWSES)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Development and Validation of The Situated Academic Writing Self-Efficacy Scale (SAWSES)

Uploaded by

Copyright:

Available Formats

What is the purpose of developing and validating the Situated Academic Writing Self-Efficacy Scale?

What is the purpose of developing and validating the Situated Academic Writing Self-Efficacy Scale?

What are the three factors identified in the scale?

What are the three factors identified in the scale?

Assessing Writing 48 (2021) 100524

Contents lists available at ScienceDirect

Development and validation of the Situated Academic Writing

2. Current measures of writing self-efficacy

3. Theoretical premise of the scale development

4. Methods and results

Year in program (Nursing)

Years in post-secondary (interdisciplinary undergraduate)

Level of graduate studies

Writing Attitude (N ¼ 255)

Learned to Speak in English

Learned to Write in English

STEM = Science, Technology, Engineering, Mathematics.

5. Item development and revision

5.1. Delphi panel

5.2. Cognitive interviews

6. Study 1: exploratory factor analysis and hypothesis testing

6.4. Data analysis

6.5.1. Exploratory factor analysis

6.5.2. Reliability testing

Fig. 1. Scree Diagram for Exploratory Factor Analysis.

Situated Academic Writing Self- Model 1 Model 2 Model 3 Model 4

Situated Academic Writing Self- Model 1 Model 2 Model 3 Model 4

With each new writing assignment, I

6.5.3. Concurrent validity

6.5.4. Structural equation modelling

Mean SD 95 %CI Mean SD 95 %CI Mean SD 95 % CI

Mean SD 95 %CI Mean SD 95 %CI Mean SD 95 % CI

Even when the writing is hard, I can

Items removed from Scale after EFA

Writing Self-Efficacy Instrument (WSEI) (α ¼ .93) 0− 800 616.2 154.4

Fig. 2. Hypothesized Structural Equation Model 1 in Nursing Student Sample.

7. Study 2: multigroup confirmatory factor analysis

7.3. Data analysis

Writing Essentials 1 9.59 1.14 8.39 <.001 .75

code files are publicly available (Mitchell, 2020a, 2020d).

8. Exploring differences between study 1 and study 2 samples

8.1. Writing attitude

Overall Model 376.97 101 0.074 (.066− .082) 0.942 0.043 – – –

8.2. Exploring item discrimination

Nursing (N ¼ 255) n = 44 n = 108 n = 103

Interdisciplinary Undergraduate (N ¼ 543) n = 237 n = 219 n = 87

Graduate (N ¼ 264) n =124 n = 111 n = 29

Relational Reflective Writing

Subscale Average .71 .65 .67

N = Nursing, UG = Undergraduate, G = Graduate.

CRediT authorship contribution statement

Declaration of Competing Interest

The authors report no declarations of interest.

Appendix A. Supplementary data

You might also like