This article was downloaded by: [Michigan State University]

On: 22 February 2015, At: 20:53

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

The Journal of Educational Research

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/vjer20

Influence of Self-Efficacy on Elementary Students'

Writing
a b
Frank Pajares & Gio Valiante
a
Emory University
b
Broward County Schools
Published online: 14 Nov 2012.

To cite this article: Frank Pajares & Gio Valiante (1997) Influence of Self-Efficacy on Elementary Students' Writing, The
Journal of Educational Research, 90:6, 353-360, DOI: 10.1080/00220671.1997.10544593

To link to this article: http://dx.doi.org/10.1080/00220671.1997.10544593

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the
Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Influence of Self-Efficacy on
Elementary Students' Writing
FRANK PAJARES GIO VALIANTE
Emory University Broward County Schools

which students obtain from the interpreted results of their

ABSTRACT Path analysis was used to test the influence of efforts. These previous results are obviously strong predic-
Downloaded by [Michigan State University] at 20:53 22 February 2015

writing self-efficacy, writing apprehension, perceived useful- tors of subsequent efforts, but because "people's percep-
ness of writing, and writing aptitude on the essay-writing per- tions of their efficacy touch, at least to some extent, most
formance of 218 fifth-grade students. A model that also includ-
ed sex accounted for 64 % of the variance in performance. As everything they do" (Bandura, 1984, p. 251 ), self-efficacy
hypothesized, self-efficacy beliefs made an independent contri- beliefs also predict academic performances and are impor-
bution to the prediction of performance despite the expected tant motivational factors. -
powerful effect of writing aptitude. Aptitude also had a strong Self-efficacy beliefs affect what students do by influenc-
direct effect on self-efficacy, which mediated the indirect effect ing the choices they make, the effort they expend, the per-
of aptitude on performance. Self-efficacy had direct effects on
apprehension and perceived usefulness. Girls and boys did not sistence and perseverance they exert in the face of adversi-
differ in performance, but girls reported higher writing self- ty, and the anxiety they experience. Believing that they are
efficacy, found writing more useful, and had lower apprehen- capable writers, for example, will serve students well when
sion. Results support the hypothesized role of self-efficacy in they attempt to write an essay, not because the belief itself
Bandura's (1986) social cognitive theory. increases writing competence, but because it helps create
greater interest in writing, more sustained effort, and greater
perseverance and resil iency when obstacles get in the way
of the task. When students have confidence in their capabil-

A ccording to self-efficacy theori sts, people 's judg-

ments of what they can accomplish are influential
arbiters in human agency and, as such, powerful determi-
ity to write essays, they will also feel less apprehensive
about writing. For these reasons, Bandura ( 1984, 1986)
described self-efficacy as a mediating mechanism of per-
nants of their behavior (Bandura, 1986, 1997). In large part, sonal agency-mediating between the influences that are
this is because these self-efficacy beliefs are said to act as the sources of its creation and subsequent behavior. Affec-
medi ato rs between other acknowledged influences on tive factors , such as the anxiety associated with specific
behavior-such as skill, ability, or prev ious accomplish- academic areas, and attitudes, such as how useful students
ments-and subsequent performance. The process of creat- consider the task itself, are considered common mechanisms
ing and usi ng beliefs is simple enough and rather intuitive: of personal agency. Like self-efficacy beliefs, they also
Individual s engage in behaviors, interpret the outcomes of influence academic outcomes. But, Bandura ( 1986) argued,
their actions, use the interpretations to develop beliefs about the influence of these common mechani sms on academic
their capabilities to engage in subsequent behaviors in the performances is due primarily to the sense of confidence
same domain, and behave in concert with the beliefs creat- with which students approach academic tasks.
ed. In school , for example, the beliefs that students develop Both the predictive and mediational roles of self-efficacy
about their academic capabilities help determine what they have received support from a growing body of findings in
do with the knowledge and skill s they possess. Consequent- diverse fields (see Multon, Brown, & Lent, 1991 , for a
ly, other influences on their academic performances are in meta-analysis of research on the relationship between self-
large part the result of what students actually believe they efficacy beliefs and academic outcomes). One academic
can accompli sh. Thi s helps explain why students' academic area that has received little attention from self-efficacy
outcomes may differ markedly even though they have sim- researchers, however, is written composition. This is an unfor-
ilar ability. tunate omission, given the important role that writing ski ll s
Students develop their beliefs of personal efficacy from play at all level s throughout the academic curriculum .
varied sources, including the observations they make of oth-
ers attempting similar tasks or the verbal messages they
receive from teachers, parents, and classmates. By far the Address correspondence to Frank Pajares, Educational Studies,
strongest source of efficacy information, however, is that Emory University, Atlanta, GA 30322.

353
 354 The Journal of Educational Research

Moreover, complex analyses with which to test hypothe- hood of success in writing (.59), and willingness to take
sized relationships-such as path analyses-have not been additional writing courses (.57). They also found that boys
conducted, and investigations have generally involved high were more apprehensive about writing than girls were. But
school students and college undergraduates. Researchers findings on the relationship between writing apprehension
have recommended that such studies be conducted at lower and writing have been inconsistent. Faigley, Daly, and Witte
academic levels, especially at those in which these sorts of ( 1981) found the relationship to be significant when writing
self beliefs are taking root (Bandura, 1989; Pajares & John- was assessed using standardized test results but not when an
son, 1996; Reyes , 1984; Schunk, 1991). essay was used (only one of two samples was significant).
McCarthy et al. ( 1985) did not find a relationship between
Beliefs About Writing apprehension and performance in the first of their studies .
Pajares and Johnson ( 1994, 1996) reported that writing
Researchers in the field of composition who have inves- apprehension correlates strongly with essay-writing scores.
tigated the cognitive processes that writers engage in as they When self-efficacy beliefs are controlled, however, the in-
compose text have primarily attempted to understand the fluence of apprehension diminishes or disappears.
thought processes underlying students' compositions (e.g., Students' perceived usefulness of writing has also been a
Downloaded by [Michigan State University] at 20:53 22 February 2015

Emig, 1979; Flower & Hayes, 1981 ; Scardamalia, Bereiter, focus of some studies. According to the expectancy-value
& Goelman, 1982; and see Faigley, 1990; Hairston , 1990). theory, motivation is primarily a result of individuals'
The more that researchers have learned about the relation- beliefs about the likely outcomes of their actions and of the
ship between cognition and writing, the more complex the incentive values they place on those outcomes (see McClel-
relationship seems to be (Hull & Rose, 1989). Some land, 1985; Rotter, 1982). Expectancy-value theori sts agree
researchers have addressed this complexity by investigating that self-efficacy judgments play an interactive role with
the affective factors involved in writing (e.g., Beach, 1989; valued outcomes in determining the tasks in which individ-
Elbow, 1993). Faigley, Cherry, Jolliffe, and Skinner ( 1985) uals will engage (see Wigfield & Eccles, 1992), but they
concluded that attitudes and beliefs play a key role in writ- emphasize the more prominent role of perceived value con-
ing. In particular, Beach ( 1989) suggested that self-efficacy structs. According to Bandura ( 1986), beliefs such as the
offered a particularly promising avenue of research for perceived usefulness of activities are related to efficacy
informing writing instruction. judgments because efficacy judgments in part determine the
Researchers who have explored the effect of self-efficacy perceived value of such activities. Individual s who expect
beliefs on writing agree that the two variables are related success in a particular enterprise tend to value those enter-
(e.g. , McCarthy, Meier, & Rinderer, 1985; Pajares & John- pri ses. Bandura argued that, because the outcomes people
son, 1994, 1996; Shell , Colvin, & Bruning, 1995; Shell, expect are largely dependent on their judgments of what
Murphy, & Bruning, 1989). For example, Meier, McCarthy, they can accomplish, beliefs such as outcome expectations
and Schmeck (1984) reported that writing self-efficacy pre- or perceived usefulness are unlikely to make a strong inde-
dicted the writing performance of college students, but they pendent contribution to predictions of behavior when self-
did not explore the nature of the relationships among vari- efficacy perceptions are controlled.
ables such as self-efficacy, apprehension, aptitude, and per- Shell et al. (1989) investigated the writing self-efficacy of
formance. McCarthy, Meier, and Rinderer identified 19 undergraduates and reported a significant correlation
writing skills and asked undergraduates whether they could between students' confidence in their writing skills and their
demonstrate them. They also assessed anxiety, locus-of- holistic scores on a 20-min essay (.32). They did not find sig-
control orientation, and cognitive processing. In two studies nificant correlations between the students' perceived useful-
conducted with the same students, self-efficacy-the confi- ness of writing and their essay scores (. 13). Pajares and John-
dence that students had in their writing skills-was related son ( 1994) reported that writing self-efficacy (.53), perceived
to holistically scored essay scores in the first study; self- usefulness (.55), and previous writing achievement (.57) cor-
efficacy and writing anxiety correlated with essay scores in related with the writing performance of undergraduates.
the second study. The relationship between self-efficacy However, in a multiple regression model that accounted for
and essay scores was moderate (.33) . Pajares and Johnson 68% of the variance in writing performance, only the stu-
( 1996) explored the writing self-efficacy of high school stu- dents' self-efficacy and aptitude were significant predictors.
dents and reported that both aptitude and self-efficacy As we have already noted, Bandura ( 1986) argued that
beliefs had strong direct effects on performance. Although both the anxiety students experience about an academic task
girls and boys did not differ in aptitude or performance, and their perceived usefulness of that task are in large part
girls had lower confidence in their writing skills. determined by their perceived efficacy. Students who are
The term writing apprehension was first used by Daly confident of their writing capabilities experience less appre-
and Miller ( 1975a) to describe a form of writing anxiety and hension when faced with a writing assignment and find
has since been the focu s of numerous studies. Daly and writing more useful than do students who believe they are
Miller ( 1975b) reported significant correlations between poor writers. Empirically, this means that self-efficacy
apprehension and SAT verbal scores (.19), perceived likeli- should continue to predict related academic performance
 July/August 1997 [Vol. 90(No. 6)] 355

when the effects of anxiety and perceived usefulness are such inferences may include biases and interpretations that
controlled, whereas the effect of anxiety and perceived use- can make the assessment an imperfect reflection of actual
fulness on performance should diminish or di sappear when writing ability. Researchers in the field of composition
self-efficacy judgments are controlled. believe that timed, in-class writing samples provide reliable
Our purpose in this study was to test the predictive and writing assessments and that holistic scoring by expert read-
mediational role of the writing self-efficacy of fifth-grade ers provides a reliable means to assess writing performance
students, using path analysis with a model containing vari- (Foster, 1983; Shell et al., 1989, 1995); is subject to interrater
ables that have been found to be rel ated to writing compe- reliability checks; and, when standardized procedures are fol-
tence. Specifically, our aim was to determine whether the lowed, provides consistent results (Hillocks, 1986). Conse-
confidence with which fifth-grade students approach writ- quently, one of the researchers and a second expert reader
ing an essay makes an independent contribution to the pre- used holi stic scoring with a 5-point scale to grade the essays
diction of a holi stically scored essay score when writing (see guidelines prescribed by Wolcott, 1989). Scorers were
apprehension , perceived usefulness of writing, writing apti- unaware of student identities. One of the scorers was a pro-
tude, and sex are part of the model. The analysis also tested fessor of education who spent 12 years as a teacher of Eng-
whether self-efficacy mediates the effect of prior determi- li sh composition; the second scorer was a professor of ele-
nants-writing aptitude and sex-on apprehension, per- mentary education with 20 years of previous experience as a
Downloaded by [Michigan State University] at 20:53 22 February 2015

ceived usefulness, and writing performance. teacher of composition. Consistent with guidelines of social
Controlling for writing aptitude provides a particularly cognitive theory, the criteria for scoring were the same as
stringent test of the predictive value of self-efficacy judg- those on which students were asked to assess their writing
ments, given that aptitude assessments are highly predictive self-efficacy, that is, in terms of students' demonstration of
of academic beliefs and attitudes and are generally acknowl- grammar, usage, composition, and mechanical skills. When
edged to be the strongest predictor of academic performance. scorers' judgments differed, a student's final score was the
Motivational and self-regulatory influences affect both prior average score. Interrater reliability was .87, which is within
and later pe1formance attainments (Bandura, 1986, 1997). As the range of interrater reliability estimates obtained for holi s-
a consequence, controlling for previous performance with tic scoring (.68 to .89) when adequate methods are used
aptitude assessments controls not only for aptitude but also (White, 1985).
for the prior impact of motivational determinants such as self- Writing self-efficacy. We made thi s a function of stu-
efficacy or anxiety on aptitude. These confounding influ- dents ' judgments of their confidence that they possess vari-
ences are not easily disentangled; hence they should be kept ous composition, grammar, usage, and mechanical skills.
in mind as results are interpreted . The Writing Skills Self-Efficacy scale, developed by Shell
et al. ( 1989), consists of eight items that ask students to rate
their confidence that they can perform writing skill s such as
Method and Data Source "correctly punctuate a one page passage" or "organize sen-
tences into a paragraph so as to clearly express a theme."
Participants and Procedures
Students may provide any score from 0 to I00 as a measure
Participants were 218 fifth-grade students in three public of their confidence in each skill. Shell et al. reported relia-
elementary schools (two in the South and one in the South- bility scores of .95 for the scale in a study of undergradu-
west). There were 115 girl s and I03 boys. Instruments were ates, and factor analysis showed positive correlations above
group administered in individual classes during two peri- .40 between items and scale scores. Pajares and Johnson
ods. During the first class period, students were asked to ( 1996) reported coefficient alpha reliability of .9 1 with high
compl ete the self-efficacy, perceived usefulness, and appre- school students. Shell et al. ( 1995), using a similar instru-
hension instruments. Directions and individual items were ment with a 7-point Likert scale, reported .76. We obtained
read aloud by the administrator. During the second class a Cronbach's alpha coefficient of .88 and positive correla-
period, students were asked to write the performance meas- tions above .68 between items and scale scores.
ure-a 30-min essay. Procedures were si milar to those used Writing apprehension. This describes "a person's tenden-
by writing self-efficacy researchers (e.g., Pajares & John- cies to approach or avoid situations perceived to potentially
son, 1994, 1996; Shell et al., 1989, 1995). require writing accompanied by some amount of perceived
evaluation" (Daly & Wilson, 1983, p. 327). We adapted Daly
and Miller's ( 1975) Writing Apprehension Test (WAT), a 26-
Instruments
item instrument regarded as a reliable measure of writing
Writing pe1jormance. Consistent with procedures used by anxiety (sample item : "I am afraid of writing essays when I
Shell et al. ( 1989, 1995), we asked students to write a 30-min know they will be evaluated.") Reed, Burton, and Vandett
essay titled "My Idea of a Perfect Day." We used holistic ( 1988) found the WAT reliable but suggested that the uncer-
scoring on a 5-point scale to grade the essays. Clearly, assess- tain item be removed from the 5-point scale. We reduced the
ing students' writing is not an objective task. It involves an 26-item instrument to 9 items for ease of administration with
inference by the reader of the quality of a written work, and elementary school students, increased the Likert scale to 8
 356 The Journal of Educational Research

points in accordance with social cognitive guidelines (see ing aptitude were exogenous variables hypothesized to
Bandura, 1997), and obtained an alpha coefficient of .83. influence the endogenous variables ; self-efficacy was
Using a similar scale with high school students, Pajares and hypothesized to influence apprehension, perceived useful-
Johnson (1996) obtained a Cronbach 's alpha coefficient of ness, and performance; apprehension and perceived useful-
.93. Exploratory factor analysis revealed that each item ness are considered common mechanisms of personal
loaded above .45 on a single factor. We regard the modest agency and were hypothesized to influence performance,
loss of reliability to be due primarily to the age of the students but they were placed between self-efficacy and performance
in the sample rather than to alteration of the scale (see Pajares in the model because of theoretical considerations ex-
& Urdan, 1996, for similar results with mathematics anxiety) . plained earlier. We refined this initial model by removing
The Perceived Usefulness of Writing scale. This scale nonsignificant paths in the order suggested by the stepwise
was adapted from the Writing Outcome Expectations Scale multivariate Wald test and retested each reduced model
(Shell et al., 1989). It consists of I0 items that assess stu- comparing goodness-of-fit indices (Bentler, 1990; Bentler
dents ' judgments of the importance of writing for success- & Chou, 1987). The final model reflects the results of this
fully accomplishing various academic and life endeavors process (see Figure I). All analyses were performed on the
such as making a good living, getting good grades, or get- variance-covariance matrix; we used the maximum likeli-
Downloaded by [Michigan State University] at 20:53 22 February 2015

ting a job. Shell et al. reported a Cronbach 's alpha coeffi- hood method of parameter estimation and the SAS system's
cient of .93 and positive item/total correlations above .40 CALIS procedure (SAS Institute, Inc., 1989). Prior to con-
for all items on the original scale used with undergraduates. ducting the path analysis, we also tested sex differences on
We obtained an alpha coefficient of .84 and similar factor the variables in the study with a multivariate analysis of
' loadings with our elementary schoo l sample. covariance (MANCOVA); teacher ratings was the covariate.
Writing aptitude. Teacher ratings of students ' writing apti-
tude is acknowledged as a reliable assessment of students ' Results
writing aptitude (Hoge & Butcher, 1984). Consequently, the
children's language arts teachers were asked to rate their stu- In Table 2 we present the means, standard deviations, and
dents ' writing aptitude on a 5-point scale, the same as that Pearson product-moment correlations for all variables in
used in the holistic scoring of the essays. This assessment the study. The magnitude of the correlati ons between the
was made toward the end of the first semester, after teachers independent variables and writing performance, as well as
had become familiar with students' writing and before stu- that between these variables and writing self-efficacy, was
dents wrote their essays. We again alert the reader to the con- consistent with those of previous investigations. The corre-
ceptual and empirical similarities between aptitude and per- lation between self-efficacy and performance was similar to
formance assessments and to the particularly stringent test of those obtained by Pajares and Johnson ( 1994, 1996) but
the predictive value of cognitive/affective judgments that the higher than some previously obtained. This was likely due
inclusion of such an assessment provides. to the correspondence between the self-efficacy items and
the criteria for scoring the essays, a procedure recommend-
Data Analysis ed by social cognitive theorists (see Bandura, 1986; Pajares,
Path analysis techniques are used to examine direct and 1996; Pajares & Miller, 1995). MANCOVA results revealed
indirect effects between variables, thus enabling causal a significant univariate effect for sex, Wilks's lambda= .88,
inferences to be made, if modestly and cautiously. In F(4, 212) = 7.44, p < .0001. There were sex differences on
essence, such analyses "allow us to move beyond simple or writing apprehension , F(2 , 215) = 20.06, p < .000 I, per-
multiple correlations to testing the causal ordering of these ceived usefulness, F(2, 215) = 4.87, p = .009, and writing
variables that is hypothesized on the basis of self-efficacy self-efficacy, F(2 , 215) = 18.84, p < .000 I. In each case, dif-
theory" (Hackett, 1985, p. 50). Cook and Campbell ( 1979) ferences favored girls. Sex differences in writing perfor-
suggested that they are especially appropriate when " theo- mance were a function of previous aptitude.
retical , empirical, and commonsense knowledge of a prob- Results of the structural equation modeling are presented
lem" (p. 307) provides a defensible mapping of the vari- in Figure I. The final model yielded a nonsignificant
ables in the model and their hypothesized links. Path X2 (6, N = 218) of 6.95, p = .33, a goodness-of-fit index
analyses are, therefore, appropriate in investigations in adjusted for degrees of freedom (AGFI) of .96; normed fit
which the tenets of social cognitive theory and previous index (NFI) of .98, and a non-normed fit index (NNFI) of
finding s are such that hypothesized relationships have .99. All are excell ent indices of goodness of fit. The inde-
strong theoretical and empirical support (see Table I). pendent variables accounted for 64% of the variability in
We began the analysis by testing the goodness of fit of writing performance. Chi-square tests demonstrated that the
the theoretical model posited by social cognitive theory. final model had significantly better fit than the initial model
Consistent with previous findings (e.g. , McCarthy et al. , or than the reduced models tested when nonsignificant
1985; Pajares & John son, 1994, 1996; Shell et al. , 1989) paths were removed.
and based on social cognitive theory (Bandura, 1986), the In spite of the expected powerful effect of aptitude (~ =
initial theoretical model tested was as follows: Sex and writ- .60 I), self-efficacy made an independent contribution to the
 July/August 1997 [Vol. 90(No. 6)] 357

Table I.-Decomposition of Effects From the Path Analyses

Parameter Standardized
Effect estimate SE estimate R2

On writing self-efficacy of . 14
Writing apti tude 5.18 1.07 .3 11 4.87
Sex - 6.1 7 2.28 -. 173 -2.7 1
On perceived useful ness of .09
Writ ing self-efficacy 0. 16 0.05 .230 3.45
Writing aptitude 115
Sex - 3.83 1.67 - . 152 -2.29
On writing apprehension of .33
Writing self- efficacy 0.36 0.05 -.452 -7.88
Writing aptitude 11s
Sex 7.28 1.63 .257 4.48
On writing performance of .64
Downloaded by [Michigan State University] at 20:53 22 February 2015

Writing apprehension 115

Perceived use fu lness 11s
Writing self-efficacy 0.02 0.00 .356 8. 15
Writing aptitude 0.62 0.05 .60 1 13.77
Sex 115

Table 2.-Means, Standard Deviations, and Zero-Order Correlations for Variables, by Sex

Girls Boys
Variable M SD 2 3 4 5 M SD

I. Performance 3.17 1.08 - .31 " .25" .53" .69" 2.68 1.09
2. Apprehension 32.04 12.96 -. 31 " - .2 1" -.52" - .2 1" 42 .35 13.51
3. Perceived use 65.05 10.66 .04 - .27" .26" . 10 59.85 13.96
4. Self-efficacy 79.25 16.36 .56" -.42" .1 8" .26 70.80 18.42
5. Aptitude 3.30 1.10 .73" - .13 .06 .36" 2.86 0.99

t'ore. For sex, girl s were coded 0 and boys were coded I.
p < .05.

prediction of wntmg performance ( ~ = .356), and thi s Discussion

resolves the substantive question of the study. Writing
apprehens ion and perceived usefulness of writing had no Our results demonstrate that elementary students' self-effi-
direct effects on performance. The effect of self-efficacy on cacy perceptions predict their writing performance and play
writing apprehension ( ~ = -.452) and on perceived useful - the mediational role that social cognitive theory hypothe-
ness ( ~ = .230) suggested that, as Bandura ( 1986, 1997) has sizes. Students' self-efficacy beliefs about their own writing
argued, com mon mechani sms of personal competence such capabi lity directly influenced their writing apprehension, per-
as anxiety and perceived usefulness are in part a by-product ceived usefulness of writing, and essay-writing perfonnance
of efficacy perceptions. The direct effects of aptitude and partially mediated the effects of sex and writing aptitude
(~ = .31 I) on self-efficacy reveal that, as hypothesized, pre- on their apprehension, perceived usefulness, and perfor-
vious achievements are important sources of efficacy infor- mance. As expected, aptitude also predicted writing perfor-
mation. As foreshadowed by MANCOVA results, sex did mance, but it made no independent contribution to the pre-
not have a direct effect on performance, but there were diction of apprehension or perceived usefulness.
direct effects from sex to perceived usefulness ( ~ = -.152), Inclusion of an academic-domain aptitude assessment
to apprehension (~ = .257), and to self-efficacy ( ~ =-. 173). provides a powerful test of the influence of variables such
These effects were consistent with results of the MAN COYA as self-efficacy in an investigation of academic perfor-
analysis showing that girls reported lower apprehension and mances (see Bandura, 1997; Pajares & Kranzler, 1995).
greater self-efficacy and perceived writing as more useful Recall that controlling for writing performance with apti -
than the boys did. tude scores that consist of teacher judgments of pupil
 358 The Journal of Educational Research

Figure I. Significant Path Coeffici ents for Variables in the Study

R2 =.33

WRITING
APTITUDE
Downloaded by [Michigan State University] at 20:53 22 February 2015

WRITING
SELF-EFFICACY

-.152

R2 =.15 R 2 =.64

achievement is potentially problematic because scores on about writing and express greater confidence in themselves
such assessments are partially composed of affective fac- as writers but that thi s confidence erodes as they progress
tors. Bandura ( 1997) cautioned that moti vational and self- through school despite no corresponding changes in writing
regulatory influences affect both prior and later perfor- competence. If this is so, it would be consistent with conclu-
mance attainments. As a consequence, controlling for past sions made by mathematics self-efficacy researchers in that
performance by using assessments of academic aptitude in area (e.g., Hackett & Betz, 1989).
the target subj ect controls not only for that aptitude but also Shell et al. ( 1995) reported sex differences in writing self-
for the prior impact of motivational determinants and self- efficacy but did not detect an interaction of sex and level in
beliefs such as self-efficacy or apprehension. The result is separate fourth-grade, middle school, and high school sam-
that the influence of affecti ve fac tors such as self-efficacy ples. A longitudinal study using the same cohort would help
on writing performance is potenti ally greater than the path cl arify these issues. Nonetheless, sex differences of any sort
obtained indicates . in self-beliefs that do not correspond to differences in aca-
Results from thi s investigation are especially noteworthy demic performance are differences in belief on ly. Whether
when contrasted with those obtai ned by Pajares and Johnson boys have lower self-efficacy early or gi rl s have lower self-
( 1996), who investigated the writing self-efficacy of enteri ng efficacy later or any particular group has lower self-efficacy
high school students. Pajares and Johnson reported that ninth- at any time, in all cases, altering inaccurate self-beliefs that
grade boys and gi rl s did not differ in writing performance, are unrealisticall y low and do not correspond to academic
but the boys reported higher self-efficacy. We also found no skill s should be beneficial. The effects of low self-efficacy
differences in perfom1ance, but the fifth-grade girl s in our beliefs may differ by sex and by age, but they are likely to
study reported higher self-efficacy and lower apprehension result in lessened effort, decreased persistence and persever-
and they perceived writing as more useful. It may be that, at ance, and lower optimism any time they are held (see Ban-
elementary school levels, girl s have more positive self-beliefs dura, 1997; Scheier & Carver, 1993).
 July/August 1997 [Vol. 90(No. 6)] 359

Two implications emerge from our findings. If it is true would test the causal rel ationships more directly and permit
that the writing self-efficacy of boys is lower during ele- interpretations of the reciprocal influence of self-efficacy
mentary school years and that the self-efficacy of girls beliefs and related academic outcomes.
erodes as they progress through school , teachers would be Findings from thi s study strengthen Bandura's ( 1986)
well advised to assess students' writing confidence as they claim that self-efficacy beliefs pl ay an influential role in
do their writing competence. Some self-efficacy researchers human agency. In addition, the contrast of these results with
have suggested that teachers should pay as much attention those of Pajares and Johnson ( 1994, 1996) suggests that
to students' perceptions of competence as to actual compe- there may be a developmental component to the creation
tence, for the perceptions may more accurately predict stu- and evolution of writing self-efficacy beliefs and that the
dents' motivation and future academic choices . Assessing predictive and mediational roles of self-efficacy may differ
stude nts ' se lf-efficacy beliefs can provide teachers with as a function of academic level and years of schooling. The
important insights. For example, researchers have demon- implication that arises is that researchers and school practi-
strated that self-efficacy be liefs strongly influence the tioners should investigate students' beliefs about their aca-
choice of majors and career decisions of college students demic capabilities as important predictors of other affective
(see Hackett & Betz, 1989). In many cases, inaccurate per- variables and of academic performance, and that efforts
Downloaded by [Michigan State University] at 20:53 22 February 2015

ceptions of ability, and not lack of skill , are responsible for should be made to identify these beliefs, because they are
avoidance of certain courses and careers. This phenomenon important components of motivation and behavior. This is
may be at work in the area of composition with students consistent with McLeod's ( 1987) observation that, because
who have unreali stically low writing self-efficacy. If so, writing is as much an emotional as a cognitive activity,
efforts to identify and alter inaccurate judgments should affective components strongly influence all phases of the
prove beneficial. And, if self-efficacy beliefs are a cause of writing process. She urged researchers to explore affective
variables such as writing apprehension, then interventions measures with an eye toward developing a "theory of affect"
designed to improve writing by decreasing anxiety may be to help students understand how their affective processes may
usefu l to the degree that they increase stude nts' confidence inform their writing. Given our findings, it seems warranted
in their writing ability. that students' self-efficacy perceptions should play a promi-
The second implication is that, if efficacy beliefs perform nent role in such a theory.
the functions posited by social cognitive theory, teachers
have the responsibility to increase students ' competence REFERENCES
and confidence as students prog ress through school. Ban-
dura ( 1986) argued that Bandura, A. ( 1984) Recycling misconceptions of perceived self-efficacy.
Cognitive Therapy and Research. 8, 23 1-255.
educational practices should be gauged not only by the skills Bandura, A. ( 1986). Social founda tions of thought and action: A social
and knowledge they impart for present use but also by what cognitive theory. Englewood Cliffs, NJ: Pre ntice Hall.
they do to children's beliefs about their capabilities, which Bandura, A. ( 1989). Human agency in soc ial cogn iti ve theory. American
Psyclzologist44, 11 75- 1184.
affects how they approach the future. Students who develop a
Bandura, A. ( 1997). Self-efficacy The exercise of contml. New York: Freeman.
strong sense of self-efficacy are well equipped to educate them- Beach, R. ( 1989). Showing students how to assess: Demonstrating tec h-
selves when they have to rely on their own initiative. (p. 417) niques for response in the writing conference. In C. M. Anson (Ed.),
Writin g and response (pp. 127- 148). Urbana, IL: NCTE.
Schunk ( 1991) and hi s associates have provided insights on Bentler, P. M. ( 1990). Comparative fit indexes in structural model s. Psy-
how this can be accomplished. clwlogical Bulletin, 107. 238- 246.
Be ntler, P. M. , & Chou , C. ( 1987). Practical issues in structural modelin g.
We caution the reader to some limitations of our study. Sociological Methods aud Resea rch, 16, 78- 187.
First, analyses on which inferences of causality are made Cook, T. D., & Campbell , D. T. ( 1979). Quasi-experimentation: Design
are not without controversy (see Freedman, 1987). Although and analysis issues for fie ld sellings. Boston: Houghton Mifnin.
Daly, J. A., & Miller, M. D. (1975a) . The emp irical development of an
they provide powerful statistical tools with which to explore instrume nt to measure wri ting apprehe nsion . Research in the Teaching
the nature of causal relationships in nonexperimental studies of English, 9. 272-289.
(see Bentler & Chou, 1987; Duncan, 1975), they are in large Daly, J. A. , & Miller, M. D. ( 1975b). Further studies in writing apprehen-
sion: SAT scores, success expec tati ons, willingness to take advanced
part at the mercy of the relationships hypothesized to exist courses, and sex differences. Research in the Teaching of English , 9,
before the model is constructed. As such, they reflect the the- 250-256.
oretical orie ntation that underg irds a study and the Dal y, J. A. , & Wil son, D. A. (1983). Writin g apprehensi on, se lf-esteem,
and personality. Resea rch in the Teaching of English, 17, 327-341 .
researcher's interpretati on of the theoretical directional inter- Duncan, 0 . D. ( 1975). Introduction to structural equation models. New
play among the variables . In the case of this study, the theo- York : Acade mic Press.
retical orientation was Bandura 's ( 1986) social cognitive the- Elbow, P. (1 993). Ranking. eval uating , and lik ing: Sorting out three form s
of judgment. College English, 55, 186-206.
ory. Interpretations must be made carefully and modestly. A Emig, J. ( 1979). Th e composing processes of twelfth graders. Urbana, JL:
replication using experimental designs and more powerful NCTE.
stati stical tool s to test competing theoretical models, such as Faigley, L. ( 1990). Competing theories of process: A critique and a pro-
posal. In R. L. Graves (Ed.), Rhetoric and composition (pp. 38-53).
structural equation mode ling, and additional motivational
Portsmouth, NH: Heinemann .
variables would be a reasonable nex t step. If causal model s Faigley, L. , Cherry, R. D., Jolliffe, D. A., & Skinner, A. M. ( 1985). Assess-
are to be used, repeated measures and longitudinal designs ing writers' knowledge and processes of composing. Norwood, NJ: Ablex.
 360 The Journal of Educational Research

Faigley, L. , Daly, J. A., & Witte, S. P. ( 1981 ). The effects of writing appre- The role of writing self-efficacy, outcome expectancy, and apprehension.
hens ion on writing performance and competence. Th e Journal of Educa- Research in the Teaching of English , 28, 3 13-33 I.
tional Research, 75, I 6-2 1. Pajares, F., & Johnson, M. J. ( 1996). Self-efficacy beliefs and the writing
Flower, L. S., & Hayes, J. R. (198 1). A cognit ive process theory of writing. perfonnance of entering high school students. Psychology in the Schools,
College Composition and Communication, 32, 365-387. 33, 163- 175.
Foster, D. ( 1983). A primer for writing teachers. Upper Montclair, NJ: Boyn- Pajares, F., & Kranzler, J. ( 1995). Self-efficacy beliefs and general mental
ton-Cook. abi lity in mathematical problem-solving. Contemporary Educational Psy-
Freedman, D. A. ( 1987). As others see us, a case study of path analysis. The chology, 20, 426-443.
Journal of Educational Statistics, 12, 101-128. Pajares, F., & Miller, M. D. ( 1995). Mathe matics self-efficacy and mathe-
Hackett, G. ( 1985). The role of mathematics self-efficacy in the choice of matical performances: The need for specificity of assessment. Journal of
math-related majors of college women and men: A path analysi s. Journal Counseling Psychology, 42, 190-1 98.
of Counseling Psychology, 32, 47-56. Pajares, F., & Urdan , T. C. ( I996). An exploratory factor analysis of the
Hackett, G. , & Betz, N. E. (1989). An exploration of the mathematics self- Mathe matics Anxiety Scale. Measurement and Evaluation in Counseling
efficacy/mathematics pe1fonnance correspondence. Journal for Research and Development, 29, 35- 147.
in Mathematics Education. 20. 26 1-273. Reed, W. M. , Burton, J. K. , & Vanden, N. M. ( 1988). Daly and Miller's writ-
Hairston , M. ( 1990). The winds of change: Thomas Kuhn and the revolution ing apprehension test and Hunt's t-unit analyses: Two measurement pre-
in the teach ing of writing. In R. L. Graves (Ed.), Rhetoric and composi- cautions in writing research. Journal of Research and Development in
tion (pp. 3- 15). Portsmouth, NH: Heine mann . Education, 2 1(2) , 1-8.
Hill ocks , G. ( 1986) . Resea rch on written composition. Urba na, IL: Rotter, J. B. (1982). Social learning theory. In N. T. Feather (Ed.), £\:pecta-
ERJC/NCT E. tions and actions: Expectancy -value models in psychology (pp. 24 1-260).
Downloaded by [Michigan State University] at 20:53 22 February 2015

Hoge, R. D. , & Butcher, R. ( 1984). Analysis of teacher judgments of pupil Hill sdale, NJ: Erlbaum .
achievement levels. Joumal of Educational Psychology, 76, 777- 78 I. SAS Institute, Inc. ( 1989). SAS/STA T users guide, version 6, fo urth edition,
Hull , G. , & Rose, M. ( 1989). Rethinking remediation: Toward a soc ial-cog- volume I . Cary, NC: SAS Institute Inc.
nitive understanding of problematic reading and wri ting. Written Commu- Scardamalia, M. , Bereiter, C., & Goelman, H. ( I982). The role of production
nication, 6, 139-154. factors in writing ability. In M. Nystrand (Ed.), What writers know: The
McCarthy, P., Meier, S. , & Rinderer, R. ( I985). Self-efficacy and writing. language, process, and structure ofwrillen discourse (pp. I 73-2 10). New
College Composition and Communication, 36, 465-47 I. York: Academic Press.
McClelland, D. C. (1985). How motives, skills, and values determine what Scheier, M. F. , & Carver, C. S. ( 1993). On the power of positive thinking:
people do. American Psychologist, 40. 8 I 2-825. The benefits of being optimi stic. Current Directions in Psychological Sci-
McLeod, S. ( I987). Some thoughts about feelings: The affective domain and ence, 2, 26-39.
the writing process. College Composition and Communication , 38, Schunk, D. H. (I 99 1). Self-efficacy and academic motivation. Educational
426-435 . Psychologist, 26. 207-23 I .
Me ier, S., McCarthy, P. R., & Schmeck, R. R. ( I984). Validity of self-effica- Shell, D. F., Colvin , C. , & Bruning, R. H. ( I995). Self-efficacy, attributions,
cy as a predictor of writin g performance. Cogniti ve Therapy and and outcome expectancy mechanisms in read ing and writing ach ievement:
Research , 8, 107- 120. Grade-level and achievement-level differences. Journal of Educational
Multon , K. D. , Brown, S. D., & Lent, R. W. ( I99 1). Relation of self-effica- Psychology, 87, 386-398.
cy beliefs to academ ic outcomes: A meta-analytic investigation. Journal Shell , D. F., Murphy, C. C., & Bruning, R. H. ( 1989). Self-efficacy and out-
of Counseling Psycholog1·. 38, 30-38. come expectancy mechanisms in reading and writing achievement. Jour-
Pajares, F. ( I995, April). The role of perceived self-efficacy in se lf-regulation nal of Educational Psychology, 8 1, 9 1- 100.
and achievement across domains. In D. H. Schunk (Chair), Learner per- White, E. M. ( 1985). Teaching and assessing writing. San Franc isco, CA:
ceptions ofself-regulatOiy competence: From se/f-schemas to selreffica- Jossey- Bass.
cy. Paper presented at a symposi um conducted at the meeting of the Amer- Wigfield, A., & Eccles, J. ( I 992). The development of achievement task val-
ican Educational Researc h Assoc iation, San Franc isco. ues: A theoretical analysis. Developmental Review, 12, 265- 3 10.
Pajares, F. ( 1996). Self-efficacy beliefs in academic settings . Review of Edu- Wolcott, W. (I 989). Perspectives on holistic scoring: The impact of moni-
cational Research, 66, 543-578. toring on wrillen evaluation . Unpubli shed doctoral dissertation. Universi-
Pajares, F., & Johnson , M. J. ( I994). Confidence and compete nce in writing: ty of Florida.