lable at ScienceDirect

Teaching and Teacher Education 27 (2011) 751e761

Contents lists avai

Teaching and Teacher Education

journal homepage: www.elsevier .com/locate/ tate

Exploring literacy teachers’ self-efficacy beliefs: Potential sources at play

Megan Tschannen-Moran*, Denise JohnsonCollege of William and Mary, School of Education, PO Box 8795, Williamsburg, VA 23187-8795, United States

a r t i c l e i n f o

Article history:Received 27 March 2010Received in revised form4 December 2010Accepted 7 December 2010

Keywords:Teacher self-efficacyLiteracy instructionAntecedentsContextual factors

* Corresponding author. Tel.: þ1 757 221 2187; faxE-mail addresses: mxtsch@wm.edu (M. Tschann

(D. Johnson).

0742-051X/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.tate.2010.12.005

a b s t r a c t

This study explored the antecedents of self-efficacy beliefs for literacy instruction and the relationship ofthese beliefs to self-efficacy for teaching in general. Factor analysis demonstrated construct validity of themeasure of TSELI developed. Moderate correlations between TSELI and the more general TSES suggestthat while there is some overlap, they are not the same thing. In a regression analysis, the quality ofuniversity preparation, highest level of education, participation in a book club, school level, resourcesavailable for classroom books, teachers’ sense of efficacy for instructional strategies and for studentengagement all explain variance in TSELI.

� 2010 Elsevier Ltd. All rights reserved.

The study of teachers’ sense of efficacy began in the mid 1970swith the RAND studies of reading instruction among low-incomeand minority students in an urban setting (Armor et al., 1976). TheRAND researchers, in search of variables that would explaindifferences in the effectiveness of certain teachers and methods,examined the extent to which teachers believed that they couldcontrol the reinforcement of their actions. They assessed the extentto which teachers believed they could control student motivationand performance and whether teachers believed environmentalfactors overwhelmed any power they could exert in schools. Theresearchers found that teacher self-efficacy was positively relatedto variations in reading achievement among minority students.Students taught by teachers who believed that they could signifi-cantly influence students’ motivation and learning tended to havehigher reading achievement than students whose teachers believedthat there was little they could do in light of the impediments tolearning posed by the environment.

The results of the RAND studies piqued interest in the constructof teachers’ self-efficacy beliefs and over the last three decadesresearchers have sought to expand and refine the notion of teacherself-efficacy and how it is measured. Teachers’ sense of efficacy hasproved to be a powerful construct, related to teachers’ motivationand behavior in the classroom as well as contributing to importantstudent outcomes (Ross, 1998; Tschannen-Moran, Woolfolk Hoy, &Hoy, 1998). Research has examined the effects of efficacy beliefs onteaching and learning in general as well as in selected subject areas.Yet, little research into teachers’ self-efficacy beliefs for literacy

: þ1 757 221 2988.en-Moran), cdjohn@wm.edu

All rights reserved.

instruction has been published since the RAND study. Moreover,there is little empirical evidence about how to cultivate strongerself-efficacy beliefs for literacy instruction. The purpose of thisstudy was to explore potential antecedents of self-efficacy beliefs inthe realm of literacy instruction and the relationship of thesebeliefs to self-efficacy for teaching in general.

1. The development of self-efficacy beliefs

Bandura (1977) introduced the concept of self-efficacy beliefs asan assessment of one’s capabilities to attain a desired level ofperformance in a given endeavor. He proposed that belief in one’sabilities was a powerful drive influencing one’s motivation to act,the effort one puts forth in the endeavor, the persistence of thateffort, and resilience in the face of setbacks. Bandura (1997) assertedthat these beliefs were more powerful than one’s actual abilities forthe task at hand in influencing people’s level ofmotivation, affectivestates, and actions (p. 2). Consequently, a teacher who did notexpect to be successful in literacy instruction for certain studentswould likely put forth less effort in the preparation and delivery ofinstruction and would likely give up more readily as the studentsstruggled, even if he or she actually possessed teaching strategiesthat would likely assist these students if they were applied. Self-efficacy beliefs can therefore become self-fulfilling prophesies,validating either beliefs of capability or of incompetence.

1.1. Sources of self-efficacy

Bandura (1997) proposed four major influences on self-efficacybeliefs e vicarious experiences, verbal persuasion, physiological

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761752

arousal, and mastery experiences. Vicarious experiences are thosein which the target activity, such as teaching a particular subject orgroup of students, is modeled by someone else. Verbal persuasionhas to do with verbal interactions that a teacher receives about hisor her performance and prospects for success. Psychological andemotional arousal also adds to a feeling of capability or incompe-tence, depending upon whether it is experienced as a sense ofanxiety or of excitement about a performance. The most powerfulsource of efficacy-relevant information for literacy teachers ismastery experiences or actual teaching accomplishments withstudents. Self-efficacy beliefs are raised when a teacher witnessesimprovement in student performances as a result of her or histeaching, which then contributes to optimism that future perfor-mances likewise will be proficient. This increase in self-efficacyresults in greater effort and persistence over time. Repeated fail-ures, on the other hand, lower self-efficacy beliefs resulting indecreased motivation and resilience (Guskey, 1988; Ross, 1998;Tschannen-Moran et al., 1998).

Researchers have tried to document the influence of thesevarious sources of self-efficacy. In a study in Hong Kong, prospec-tive teachers’ beliefs about their capability were found to beinfluenced by the quality of supervision they received (verbalpersuasion) as well as experiences during practice teaching(mastery experiences) (Yeung & Watkins, 2000). In a study ofprospective teachers in the US, it was found that vicarious experi-ences in the form of modeling by the teacher educator and masterteachers positively influenced pre-service teachers’ self-efficacy foreffective literacy instruction (Johnson, 2010). Tschannen-Moranand McMaster (2009) tested four different models of professionaldevelopment in the introduction of a pedagogical strategy to assistbeginning readers to bridge the gap between sounds and lettersymbols through the use of hand cues. The four models were eachconsidered to provide increasingly rich sources of self-efficacy,starting with verbal persuasion (telling alone), then the addition ofvicarious experiences (telling plus modeling), to the introduction ofa limited mastery experience (practice with colleagues), and finallyto a full mastery experience (coaching in the teachers’ own class-rooms). The model that included a full mastery experience showedthe largest gains in self-efficacy beliefs. This study demonstratedthat the development of self-efficacy beliefs is not a linearprogressionwith new sources of self-efficacy adding to incrementalgains in self-efficacy.

1.2. The role of context

The context inwhich the sources of efficacy are experienced alsoplays an important role in the development of teachers’ self-effi-cacy beliefs. In assessing their beliefs about their teaching capa-bility, it has been proposed that teachers make two interrelatedjudgments: an assessment of their personal teaching competencein light of the assumed requirements of an anticipated teaching task(Tschannen-Moran et al., 1998). Judgments of personal competenceare those a teacher makes about his or her capabilities based on anassessment of internal strengths and deficits. The assessment of theteaching task may include the instructional resources available aswell as the quality of the curriculum; student factors such as theirperceived ability, motivation, and socioeconomic status; andcontextual factors such as school climate, collegial support, andleadership. Support for this theory was found in a study conductedin Singapore. Teachers in high-track middle schools were found tobe have stronger self-efficacy beliefs compared to their fellowteachers who taught in regular schools where there was a greaterrange of student ability groupings (Chong, Klassen, Huan, Wong, &Kates, 2010). Furthermore, teachers’ beliefs about their ability tomeet the challenge of teaching high-track students were shaped in

part by the attitudes of other teachers, the availability of specificresources, as well as organizational expectations and goals.

Career stage seems to influence the role played by contextualvariables that matter as sources of self-efficacy. Contextual factorssuch as the availability of teaching resources were found tocontribute more to the self-efficacy beliefs of novice teachers thancareer teachers who had a wealth of mastery experiences onwhichto base their self-perceptions (Tschannen-Moran & Woolfolk Hoy,2007). Verbal persuasion, assessed as the interpersonal support ofadministrators, colleagues, parents, and members of the commu-nity, made significant contributions to explaining variance innovice teachers’ self-efficacy beliefs, but made little contributionfor career teachers. Bandura (1997) asserted that self-efficacybeliefs are most in flux early in learning and tend to become fairlystable and resistant to change once set. For novice teachers, whohave few mastery experiences to draw upon, other sources of self-efficacy seem to bemore salient in their self-assessments of efficacy(Tschannen-Moran et al., 1998).

While there is a large and growing body of literature on theeffects of teachers’ self-efficacy beliefs, little is known about theeffects of teachers’ self-efficacy beliefs in the complex realm ofliteracy instruction and the antecedents of these beliefs.

1.3. Self-efficacy and literacy instruction

Effective literacy instruction requires teachers to make complexand instantaneous teaching decisions to meet the diverse needs oftheir students (Block & Mangieri, 2003; Pinnell, 2002). Manyteachers view teaching students with awide range of reading levelsas one of the greatest challenges that they face (Baumann, Hoffman,Duffy-Hester, & Moon, 2000). Effective action depends, in part, onone’s perceived self-efficacy that the knowledge and skills neededto perform the task can be mobilized successfully under varied andunpredictable circumstances. In a situation where a teacher isattempting to enact a particular instructional strategy with a groupof students who are struggling to decode or comprehend a text, forexample, a teacher with a high sense of efficacy would be morelikely to try different instructional approaches, texts, or groupingstrategies until the students are successful (Allinder, 1994; Guskey,1988). A teacher with low self-efficacy is more likely to blame orcriticize the students for their lack of success, to persist with inef-fective instruction, to give up on the students, or to refer thestudents to special education (Ashton & Webb, 1986; Gibson &Dembo, 1984; Meijer & Foster, 1988; Podell & Soodak, 1993;Soodak & Podell, 1993).

Research on the impact of self-efficacy beliefs on literacyinstruction is quite limited, however the studies that have beendone provide promising evidence that self-efficacy beliefs matter inthe realm of literacy instruction. In a study of teachers workingwith disadvantaged students in New Zealand, Timperley andPhillips (2003) found generally low expectations and a low levelof self-efficacy at the beginning of their work with the teachers.After a six-month intervention in which teachers learned new andmore powerful literacy teaching strategies andwitnessed improvedstudent outcomes as a result, their self-efficacy beliefs were foundto have risen significantly. The teachers’ expectations of boththemselves and their students increased in spite of their students’disadvantaged backgrounds. In another study, it was found that thetype of training in reading instruction that pre-service teachersreceived based on the type of preparation program they attendedinfluenced their teaching in terms of differences in understandings,beliefs, and decision making. Beginning teachers who graduatedfrom reading specialist and reading-embedded programs weremore willing to experiment with teaching methods whereasbeginning teachers who graduated from general education

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761 753

programs tended to make decisions about teaching and learning inrelation to external factors, such as available materials, mandates,or the wishes of administrators (Maloch et al., 2003). Finally, self-efficacy beliefs have been found to influence the teaching ofwriting, in that the classroom practices in writing instruction ofhigh self-efficacy teachers differed substantially from that of lowself-efficacy teachers (Graham, Harris, Fink, & MacArthur, 2001).Teachers in the highest quintile of self-efficacy reported spendingsignificantly more time on writing each week, taught more writingprocesses, and more grammar. Teachers with stronger self-efficacybeliefs were also more positive about the impact of natural orincidental learning methods.

1.4. Cultivating teacher self-efficacy for literacy instruction

If the self-efficacy beliefs of literacy instructors are related to suchpositive outcomes as greater teacher effort and persistence andhigher student performance as has been found in these studies, itwould be helpful to know how teachers can be supported intocultivating stronger self-efficacybeliefs related to literacy instruction.Gregoire (2003) has proposed a model that explicates the mecha-nisms through which teachers’ self-efficacy mediates their responseto instructional reform. This model proposes that the degree ofteacher conceptual change when presented with an instructionalreform initiative will depend upon whether they appraise it asa threat or a challenge. When faced with a reform, teachers will firstassess whether they are implicated in the changes being proposed.Teacherswho believe that they are already implementing the reformwill judge that they are not implicated and will process the newcontent superficially. Teachers who do feel implicated by the reformspresentedwill experience stress and discomfort. Thosewith low self-efficacy are predicted to respond to the reform initiative as a threat,leading to an avoidance intention and superficial belief change. Onthe other hand, teachers with high self-efficacy who perceive thatthey have the resources, time, and support necessary to implementthe proposed changes, would likely interpret the reform as a chal-lenge and consequently engage in more systematic (and thuseffortful) processing of the information presented.

Two interrelated puzzles associated with teachers’ self-efficacyresearch concern the accuracy of teachers’ self-perceptions inrelation to external standards of knowledge and skills, and whetherit is universally productive to try to increase their self-efficacybeliefs. In a study of teachers’ knowledge of phonics, it was foundthat teachers’ assessment of the level of their expertise was notespecially accurate, with teachers tending to overestimate theirknowledge and skills (Cunningham, Perry, Stanovich, & Stanovich,2004). It was noted that novice teachers were able to calibratetheir reading-related content knowledge more accurately thanexperienced teachers which supports the assertion stated earlierthat the analysis of the teaching task may be most explicit fornovice teachers. Bandura (1997) proposed that slightly over-estimating one’s actual abilities may be useful if it leads to greatereffort, persistence, and resilience, and because it is difficult fora person to invest fully in an activity while fighting self-doubt.Wheatley (2002), on the other hand, proposed that doubts aboutone’s efficacy can sometimes be beneficial, arguing that uncertaintyor doubt is crucial for the teacher reflection that leads to newinsights. He claimed that it is difficult for teachers to learn andimprove without experiencing the disequilibrium and uncertaintythat stem from a challenge to teachers’ existing practices. Heworried that an overestimation of one’s skills may lead tocomplacency and a failure to pursue professional developmentopportunities to overcome deficiencies.

Adding to the argument of whether efforts should be made tospare teachers from experiences that would decrease their self-

efficacy, a number of researchers have found an “implementationdip” in self-efficacy to be a common experience as teachers begin toimplement a change initiative (Ross, 1994; Stein & Wang, 1988;Woolfolk Hoy & Burke-Spero, 2005). In a study of the imple-mentation of a new decoding strategy among primary-gradeteachers, Tschannen-Moran and McMaster (2009) found a substan-tial drop on the self-efficacy for literacy instruction for about halfof the teachers who witnessed the successful demonstration ofthe strategy with struggling beginning readers selected from thehost school. Both self-efficacy for literacy instruction and imple-mentation of the new instructional strategy were enhanced whenfollow-up coaching was offered to the teachers. Thus, the detri-mental effect on self-efficacy beliefs as teachers attempt toimplement a new teaching strategy may be ameliorated by addi-tional support during the crucial early stages of change (Guskey,1988; Joyce & Showers, 2002; Stein & Wang, 1988; Wheatley,2002). Self-efficacy beliefs have been found to rebound forteacherswho are able to successfully implement a new instructionalpractice and witness gains in student achievement. An interventionamong science educators that extended over two years wassuccessful in raising participating teachers’ self-efficacy and theirimplementation of the problem-based instructional practices thatwere the focus of the initiative (Haney, Wang, Keil, & Zoffel, 2007).

In considering mechanisms that may support the self-efficacybeliefs of literacy instructors, ongoing professional developmenthas been found to be important to the effectiveness of literacyinstruction, including participation in teachers-as-readers groupsand book clubs (Anders, Hoffman, & Duffy, 2000; Nathanson,Pruslow, & Levitt, 2008; Vardell & Jacobson, 1997; Wilson &Berne, 1999). Participants in teachers-as-readers groups meet ona regular basis to read and discuss the same literature; this providesopportunities for teachers to explore their own literacy throughsharing, reflecting, and learning from quality literature withcolleagues. Teachers-as-readers groups allow teachers to gainexperience and confidence with book discussions that in turnenhance teaching and learning. Research has verified the teacher’srole in using literature in the classroom to impact children’slanguage development, fostering interest in reading and literature,reading achievement, content learning, and writing (Arya et al.,2005; Galda & Cullinan, 2003; Hickman, 1981; Hoffman, Sailors,Duffy, & Beretvas, 2004; International Reading Association, 1999).Yet, in many preparation programs children’s literature courseshave been eliminated or integrated into existing methods courses(Hoewisch, 2000). In addition, having resources such as well-stocked classroom and school libraries is likely to play a critical rolein children’s reading achievement, and consequently on teachers’self-efficacy for literacy instruction.

Research on teachers’ self-efficacy for literacy instruction hasbeen hampered by the lack of a well-conceived and well-testedmeasure of the construct. Only one instrument was found in theliterature that measured literacy teachers’ self-efficacy beliefs, theReading Teaching Efficacy Instrument (RTEI) developed by Szaboand Mokhtari (2004). This instrument was designed based on theGibson and Dembo (1984) measure and consequently shares manyof the same psychometric problems of that measure (Tschannen-Moran & Woolfolk Hoy, 2001). As has been common for measuresbased on the Teacher Efficacy Scale, Szabo and Mokhtari found twofactors, one reflecting an assessment of personal capability forteaching and a second that assesses external forces beyond thecontrol of the teacher. This measure, then, is grounded in Rotter’s(1966) theory that “behavior is influenced by generalized expec-tancies that outcomes are determined either by one’s actions or byexternal forces beyond one’s control” (Bandura, 1997, p. 19). Thelabeling of the second factor outcome expectancy, a component ofsocial cognitive theory, perpetuates an unfortunate misconception.

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761754

While self-efficacy examines the extent to which a teacher believeshe or she has the capacity to orchestrate the necessary actions toperform a given task, outcome expectancy examines the teacher’sestimate of the likely consequences of performing that task at theexpected level of competence (Bandura, 1997, p. 21). These conse-quences are not the same as external forces beyond one’s control.Bandura asserted that because outcome expectancy stems from theprojected level of competence a person expects to bring to a givensituation, it adds little to the predictive power of efficacy measures.

The purpose of this study is to explore antecedents that mightbe important sources of self-efficacy beliefs for literacy teachers.The researchers sought to develop and test a new measure ofteachers’ sense of efficacy for literacy instruction and then toexplore factors that may be related to these beliefs. These includeexperiences during pre-service and inservice training, contextualfactors in the teaching setting, more general teacher self-efficacybeliefs, and demographic factors.

2. Method

This section describes the participants of this study, themeasures that were used, as well as the data analyses that wereconducted.

2.1. Participants

The participants were 648 teachers from 20 elementary schoolsand 6middle schools in Virginia, Kansas, and Arkansas. Participantswere recruited through a convenience sample of teachers ina diverse selection of schools in which the principal was willing toallow teachers to participate in the research. Surveys weredistributed in schools and in one case a district professionaldevelopment workshop for teachers. Teachers were asked tocomplete the surveys and return them to a designated volunteerwho mailed them to the researchers.

The obtained sample was predominantly female (92%) andWhite (80%), demographics that are reflective of the teachingpopulation as a whole, where 75% are female and 90% are White(National Education Association, 2003). African-American respon-dents comprised 13% of the sample (compared to 6% nationally),and 3% self-identified as “Other.” Three quarters (76%) of theparticipants taught at the elementary level, while a quarter (24%)taught in middle schools. Of those who indicated the subjectstaught, 63.2% taught all core subjects (n ¼ 366), 8.3% taughtlanguage arts (n ¼ 48), 3.3% taught math (n ¼ 19), 4.3% taughtscience (n ¼ 25), 1.9% taught social studies (n ¼ 11), 3.9% taught acombination of two or three subjects (n ¼ 25), and 14.7% indicatedthey taught “other” subjects (n ¼ 85). The context in which theytaught were varied, with 51% teaching in rural settings, 26% insuburban schools, and 23% in urban contexts. While only 8% of theteachers sampled reported proportion of free and reduced-pricedmeals (FaRM) of 20% or fewer, the remaining teachers were fairlyevenly distributed across the socioeconomic spectrum (31%:21e40%; 28%: 41e60%; 19%: 61e80% and 13%: 81e100% FaRM). Themean number of years of teaching experience was 14.4, witha range from .5 to 41, and an SD of 9.7. Asked about their highesteducational attainment, 55% reported having attained a bachelor’sdegree, 25% a master’s degree, 19% a specialists degree, and 2%a doctorate.

2.2. Measures

2.2.1. Teachers’ sense of efficacy for literacy instruction (TSELI)One of the greatest challenges in the measurement of teachers’

self-efficacy beliefs involves establishing the optimal level of

specificity for measurement. While researchers and theorists agreethat teachers’ self-efficacy beliefs are situation specific, the appro-priate level of specificity is often not clear. Discerning the mostuseful level of specificity depends on the purposes of the research,but either extreme of highly general or highly specific may poseproblems for researchers. Bandura (2001) cautioned that, “Scales ofperceived self-efficacymust be tailored to the particular domains offunctioning that are the object of interest.” (p.1). On the other hand,there is a danger of developing measures that are so specific theylose their predictive power for anything beyond the specific skillsand contexts being measured (Pajares, 1996). Support for thecontext-specific nature of teachers’ sense of efficacy is evident inthe finding that a teacher can hold divergent efficacy beliefs fordifferent teaching contexts. Efficacy beliefs vary within teachersdepending upon the subject area, characteristics of students, thenumber of course preparations they face, and whether they areteaching outside of their field of expertise (Raudenbush, Rowan, &Cheong, 1992; Ross, Cousins, & Gadalla, 1996; Ross, Cousins,Gadalla, & Hannay, 1999; Tournaki & Podell, 2005).

The development of a new instrument to measure teachers’sense of efficacy for literacy instruction was undertaken to providea means for researchers to examine this subject-specific aspect ofteachers’ self-efficacy beliefs. A pool of thirty-three items specific tovarious aspects of literacy instruction were constructed by theresearchers drawing on the NCTE/IRA (1996) Standards for theEnglish Language Arts and the IRA (2004) Standards for ReadingProfessionals. Items tapped such aspects of literacy instruction asthe ability to use word study, decoding, and comprehension strat-egies, modeling effective strategies, integrating instruction acrossthe language arts, grouping practices, use of a wide variety ofgenres, meeting the needs of both high ability and strugglingreaders, and the ability to motivate students to value reading.

The 33-item survey was submitted to a panel of four experts inthe field of reading and literacy instructionwho reviewed the itemsfor content validity. The instrument was then field tested witheleven graduate students in literacy instruction to assess the clarityof wording of the items and instructions, appropriateness of theresponse scale, and ease of administration. In keeping withBandura’s (1997) admonition that self-efficacy instruments shouldcapture assessments of current capability rather than futurepotential or future intention, and to make the instrument context-specific, the instructions directed the participants to, “Pleaserespond to each of the questions by considering the combination ofyour current ability, resources, and opportunity to do each of thefollowing in your present position.” This scale uses a unipolarresponse scale on a 9-point continuum with anchors at 1 e Not atAll, 3 e Very Little, 5 e Some Influence, 7 e Quite A Bit, and 9 e AGreat Deal. Sample items include:

� To what extent can you help your students figure out unknownwords when they are reading?

� To what extent can you adjust writing strategies based onongoing informal assessments of your students?

2.2.2. Teacher Sense of Efficacy Scale (TSES)Teachers’ efficacy beliefs for more general aspects of teaching

were measured using the Teacher Sense of Efficacy Scale (previ-ously called the Ohio State Teacher Efficacy Scale, Tschannen-Moran & Woolfolk Hoy, 2001). This measure consists of 12 items,including three 4-item subscales: self-efficacy for instructionalstrategies, self-efficacy for student engagement, and self-efficacyfor classroom management. Previous research has established thatin second-order factor analysis, these three subscales form a singlefactor; consequently, they are often combined as a single measure

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761 755

while in other instances the subscales are examined separately. Thedirections and the response scale were the same as for the TSELI.In the current sample, Cronbach’s alpha for the full 12 item scalewas a ¼ .82. The reliability for the 4-item self-efficacy for instruc-tional strategies subscale was a ¼ .75; for the self-efficacy forstudent engagement scale a ¼ .80; and for the self-efficacy forclassroom management a ¼ .84. Sample items include:

Efficacy for instructional strategies

� To what extent can you provide an alternative explanation orexample when students are confused?

Efficacy for Student Engagement

� How much can you do to motivate students who show lowinterest in schoolwork?

Efficacy for Classroom Management

� How much can you do to control disruptive behavior in theclassroom?

2.2.3. DemographicsRespondents were asked to provide information about their

gender as well as their race. Race was reduced to three commoncategories based on the characteristics of the anticipated samplepopulation: African-American, White, or Other. Respondents werealso asked to indicate their years of teaching experience as a freeresponse item.

2.2.4. Teaching contextIn the model of Teacher Self-efficacy, Tschannen-Moran et al.

(1998) proposed that teachers would make their self-efficacyjudgments in light of an analysis of the teaching task and context. Ininstances where they believed the teaching task to bemore difficultdue to constraints posed by the context, such as a higher proportionof students of lower socioeconomic status, teachers’ sense of effi-cacy might be lower; while contexts they believed to be moreconducivemight enhance their sense of efficacy. Consequently, a setof context variableswere examined in this study. These included thelevel of the schoolwhere the teacher taught (elementary ormiddle),the proportion of low socioeconomic students in the school, asmeasured by the proportion of students who qualify for free andreduced-priced meals (FaRM), and the setting of the school (urban,suburban or rural). Two items measured the level of literacy-rele-vant resources, including ratings of the availability of support for thepurchase of books for the classroom and the quality of the schoollibrary, both assessed on a four point scale as Poor, Adequate, Good,and Excellent. Resource support has been found to be related to self-efficacy beliefs in previous studies (Chester & Beaudin, 1996,Tschannen-Moran & Woolfolk Hoy, 2007).

2.2.5. Preparation and experience variablesSix variables assess the preparation and training experiences of

participants. These were considered to capture verbal persuasionand vicarious experiences. They included the highest educationaldegree attained (Bachelors, Masters, Specialist, or Doctorate),ratings of the quality of university preparation, having takena children’s literature course at either the graduate or undergrad-uate level, and ratings of the quality of professional developmentexperiences. Teachers were also asked whether they had beena member of a book club or had participated in a teachers-as-readers group, as these have been found to be related toteachers’ literacy instruction. Sample items include:

� How would you rate the quality of your university preparationfor reading instruction?

� How would you rate the quality of your professional develop-ment experiences (both within and outside your schoolsystem)?

� Have you ever been a member of a book club?

2.3. Data analysis

An exploratory factor analysis was used to pare down the pool ofitems to those that demonstrated a coherent factor structure and totest the construct validity of the new Teachers’ Sense of Efficacy forLiteracy Instruction Scale (TSELI). This was followed by a confir-matory factor analysis in which the new items were analyzed withthe more general teachers self-efficacy items assessed in the TSESin order to explore concurrent validity. t-Tests were then performedto assess whether there were significant differences in the self-efficacy for literacy instruction between different demographicgroups, and dichotomous context and experience variables. Simi-larly, ANOVA was used to examine the relationship betweenteachers’ self-efficacy for literacy instruction and categoricalcontext and experience variables. Pearson’s r was used to conductbivariate correlational analyses between Teachers’ Self-efficacy forLiteracy Instruction (TSELI) and themore general Teachers’ Sense ofEfficacy Scale (TSES), as well as the relationship of the TSELI to thecontext and experience variables that were continuous in nature.Finally, a hierarchical multiple regression was conducted, enteredin steps by demographics, preparation and professional develop-ment, context, and TSES, to assess whether variance in TSELI couldbe explained by the set of variables tested in this study and to seewhich, if any, made an independent contribution to the predictivepower of the set as a whole.

3. Results

The findings of this study confirmed many of the anticipatedrelationships between TSELI and the proposed antecedents.However, some surprises emerged regarding factors that wereunrelated to TSELI.

3.1. Factor analyses and reliability

Exploratory factor analysis, using principal axis factor analysis,was conducted to pare down and refine the Teacher Efficacy inLiteracy Instruction Scale to those items with a solid and coherentfactor structure. Six items with low communalities (below .50)were removed, along with five eliminated for low factor coeffi-cients. These items consisted primarily of statements related to theability to foster productive use of independent instructional time orliteracy attitudes and behaviors in the home, as well as the use ofparticular instruction strategies such as graphic organizers orliterature circles.

The criteria used to determine the initial factor structureincluded eigenvalues over 1.0, scree analysis, clarity of factorstructure, and interpretability of factors. Two factors with eigen-values over 1 emerged, and together these two factors explained62% of the variance in TSELI. Ten items loaded on the second factor.Of these, six assessed self-efficacy for writing instruction, three hadto do with oral reading, and the final item concerned the ability tomake use of students’ prior knowledge in reading tasks. The twofactors were strongly correlated (r ¼ .81, p < .01), however, andwhen the data were rotated using Varimax rotation, ten items hadcoefficients above .4 on both factors (When the elementary andmiddle school samples were run separately, the items that assessedwriting instruction and oral reading tended to form separate factors

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761756

as well, with many of these items sharing dual loadings of .4 orhigher with other factors.). Therefore, a second-order factor anal-ysis was conducted and the two factors converged into one strongfactor. Consequently, we ran the factor analysis calling for a singlefactor. All 22 items loaded on this single factor and all demonstratedstrong factor coefficients, ranging from .83 to .63. This factor had aneigenvalue of 12.17 and explained 55% of the variance in TSELI,which provided evidence of construct validity. The resulting22-item measure had a Cronbach’s alpha reliability of .96, whichprovided further justification for retaining a single factor. SeeTable 1 for the results of the factor analysis.

Next, confirmatory factor analyses (CFA) using LISREL 8.7(Joreskog & Sorbom, 1993) were conducted using the TSELI itemsand the TSES items to determine howwell the theoretical models ofthese as separate constructs fit the data (Table 2). The fit of themodels was evaluated using the following indicators: (a) the rootmean square error of approximation (RMSEA); (b) the standardizedroot mean square residual (SRMR); (c) the Goodness of Fit Index(GFI); (d) the Adjusted Goodness of Fit Index (AGFI); and (e) theComparative Fit Index (CFI). Hu and Bentler (1999) have suggestedthat an excellent model is one with RMSEA � .05, SRMR � .08,GFI � .95, AGFI � .95, and CFI � .95, and joint criteria of CFI � .96and SRMR � .08.

Five models were tested. In the first, all 34 items were tested asa single one-factor model. In the secondmodel, two latent variableswere tested, with the TSELI items forming one factor and the TSEScomprising the second. In the third model, these two latent vari-ables were retained and selected errors variances of closely wordeditems were allowed to correlate. In the fourth model, four latentvariables were formed, consisting of the TSE for literacy instruction(TSELI) and the three subscales of the TSES: TSE for instructionalstrategies, TSE for student engagement, and TSE for classroommanagement. In the fifth and final model, the four latent variableswere retained and the error variances of selected items with closewording were allowed to correlate. The fit statistics generally

Table 1Factor structure for the Teachers’ Sense of Efficacy for Literacy Instruction Scale.

Teachers’ sense of efficacy for literacy instruction

To what extent can you use a student’s oral reading mistakes as an opportunity to teaeffective reading strategies?

To what extent can you use a variety of informal and formal reading assessment stratTo what extent can you adjust reading strategies based on ongoing informal assessme

of your students?To what extent can you provide specific, targeted feedback to students’ during oral reTo what extent can you adjust writing strategies based on ongoing informal assessme

of your students?How much can you do to meet the needs of struggling readers?To what extent can you help your students monitor their own use of reading strategieTo what extent can you provide your students with opportunities to apply their prior

knowledge to reading tasks?To what extent can you get students to read fluently during oral reading?To what extent can you model effective reading strategies?To what extent can you implement effective reading strategies in your classroom?To what extent can you help your students figure out unknown words when they areTo what extent can you implement word study strategies to teach spelling?To what extent can you use students’ writing to teach grammar and spelling strategieTo what extent can you model effective writing strategies?To what extent can you use flexible grouping to meet individual student needs for reaTo what extent can you integrate the components of language arts?To what extent can you get children to talk with each other in class about books theyTo what extent can you recommend a variety of quality children’s literature to your sTo what extent can you provide children with writing opportunities in response to reHow much can you do to adjust your reading materials to the proper level for individHow much can you motivate students who show low interest in reading?

Eigenvalue% of variance explained

improved with each subsequent model (see Table 2). All of themodels except the first met the joint criteria for the CFI and SRMR.The RMSEA, GFI and AGFI indicators did not meet Hu and Bentler(1999) criteria for an excellent fit.

These findings suggest that teacher self-efficacy is a multi-faceted construct based on sets of subskills. Further, they demon-strate that teacher self-efficacy for literacy instruction can beconsidered an important set of skills that contributes to an overallset of self-efficacy beliefs among elementary and middle schoolteachers, as do more general instructional strategies, capacity forstudent engagement, and skill at classroom management. Thesefindings provide evidence of concurrent validity of the TSELI.

3.2. Comparison of means

To test whether the demographic variables, the context (analysisof the teaching task), and experience (verbal persuasion andvicarious experience) variables would make a contribution toexplaining teacher self-efficacy for literacy instruction, a series oft-tests and ANOVAs were performed on the dichotomous andcategorical variables. t-Tests revealed significant differences in self-efficacy for literacy instruction by gender (t ¼ 4.62, p < .01) withfemale teachers having higher TSELI. The effect size calculated asCohen’s d was .65. No differences, however, were evident by race.Elementary teachers evidenced significantly higher self-efficacy forliteracy instruction than middle schools teachers (t ¼ 11.33,p < .01), with an effect size of .99. Participation in various literacy-based experiences were related to differences in TSELI as well,including having taken a children’s literature course (t ¼ 4.97,p < .01, d ¼ .46), participation in a Teachers-As-Readers group(t ¼ 3.60, p< .01, d¼ .41), and participation in a book club (t ¼ 3.12,p < .01, d ¼ .26). See Table 3 for the results.

ANOVAwas used to examine the relationship between teachers’self-efficacy for literacy instruction and the context and experiencequestions with multiple response categories. The results revealed

ch .83

egies? .82nts .81

ading? .80nts .77

.77s? .75

.75

.74

.73

.73reading? .72

.71s? .70

.70ding instruction? .69

.69are reading? .69tudents? .67ading? .66ual students? .65

.63

12.1755.31

Table 2Confirmatory factor analysis TSELI model comparisons.

Competing models c2 df Dc2 Ddf RMSEA SRMR AGFI GFI CFI

1 One-factor model 6748.00* 527 e e .13 .087 .57 .62 .942 Two-factor model 3467.32* 520 3280.68 7 .093 .071 .73 .76 .963 Two-factor model with correlated errors 2739.80* 514 4008.20 13 .082 .066 .77 .80 .974 Four factor model 2446.11* 515 4301.89 12 .076 .061 .79 .82 .975 Four factor model with correlated errors 2299.21* 509 4448.79 18 .074 .056 .80 .83 .98

*p < .001.

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761 757

that school setting was significantly related to differences in TSELI(F ¼ 6.71, p < .01), although the effect size was quite small (partialeta2 ¼ .025). A Tukey Post Hoc analysis revealed significant differ-ences in TSELI between rural and suburban teachers, with suburbanteachers demonstrating higher mean self-efficacy. The SES of thestudent body, as measured by the proportion of students receivingFree and Reduced-Priced Meals (FaRM), had a significant relation-ship to TSELI (F ¼ 2.69, p < .05). A Tukey Post Hoc analysis revealedsignificant differences between teachers in schools with 21e40%FaRM and those in schools with 61e80%, with those teaching inschools with the higher proportion of low SES students reportinghigher self-efficacy for literacy instruction (partial eta2 ¼ .18).Finally, no significant differences on the TSELI were found based onthe highest educational degree attained. See Table 4.

3.3. Correlational analysis

An examination of the means of the continuous variables to beincluded in the correlational analysis revealed sufficient variabilityand an absence of ceiling effects among this set of variables.The mean score for the total sample on the TSELI was 6.93 ona nine-point scale, with a standard deviation of 1.24. See Table 5.A comparison of means using one-sample t-tests demonstratedsignificant differences in the mean of the TSELI and the mean scoreof the TSES and each of the subscales (t ranged from 174.2 to 217.8,p < .01). The effect sizes, using Cohen’s d, ranged from�.10 for self-efficacy for student engagement to .57 for self-efficacy forinstructional strategies.

In making comparisons between the new measure of teacher’sself-efficacy for literacy instruction and the previously establishedmeasure of overall teacher self-efficacy, correlations were calcu-lated for both the total measure and for each of the three subscales.

Table 3t-Test comparison of mean differences in teachers’ self-efficacy for literacy instruction.

Teachers’ self-efficac

N

GenderMale 51 (8%)Female 587 (92%)

RaceWhite 518 (84%)African-American 81 (13%)

LevelElementary 478 (76%)Middle school 150 (24%)

Course in children’s literatureYes 506 (79%)No 130 (21%)

Participation in teachers-as-readers groupYes 98 (15%)No 544 (85%)

Participation in a book clubYes 318 (50%)No 322 (50%)

*p < .05; **p < .01.

As predicted, teachers’ self-efficacy for literacy instruction (TSELI)was moderately related to teachers’ overall sense of efficacy forteaching (TSES) (r¼ .61, p< .01). Teachers’ scores on the TSELI wererelated to each of the subscales of the TSES: self-efficacy forinstructional strategies (r ¼ .55, p < .01), self-efficacy for studentengagement (r ¼ .56, p < .01), and self-efficacy for classroommanagement (r ¼ .40, p < .01). See Table 6 for the correlationmatrix.

Correlations between teachers’ self-efficacy for literacyinstruction and preparation, experience, and resource variableswere not strong and consequently did not explain a large propor-tion of the variance. Teachers’ ratings of the quality of theiruniversity preparation for literacy instruction were related to theirself-efficacy beliefs for literacy instruction (r¼ .23, p< .01), as weretheir ratings of the quality of their professional developmentexperiences (r ¼ .21, p < .01). Pre-service preparation for literacyinstructionwas weakly related to the more general teacher sense ofefficacy for instructional strategies (r ¼ .17, p < .01) and studentengagement (r ¼ .14, p < .01), and was unrelated to self-efficacy forclassroom management. Years of teaching experience was unre-lated to teachers’ self-efficacy beliefs for literacy instruction, as wellas for overall teachers’ self-efficacy beliefs and the subscales of theTSES. Teachers’ ratings of the level of resource support forpurchasing books for their classrooms was also related to teachers’sense of efficacy for literacy instruction (r ¼ .23, p < .01), however,ratings of the quality of the school library were unrelated to self-efficacy beliefs for literacy instruction.

3.3.1. Multiple regressionThe set of potential antecedents of self-efficacy for literacy

instruction (TSELI) were entered in a hierarchical multiple regres-sion analysis. Because years of teaching experience and the rating of

y for literacy instruction

Mean S.D. Cohen’s d

t ¼ 4.53, p < .001** .656.18 1.317.00 1.21t ¼ .48, p < .63 .016.95 1.256.94 1.26t ¼ 11.33, p < .001** .997.23 1.036.04 1.36t ¼ 4.91, p < .001** .467.04 1.166.45 1.41t ¼ 3.65, p < .001** .417.35 1.136.85 1.24t ¼ 3.33, p < .001** .267.10 1.166.75 1.31

Table 4ANOVA comparisons of TSELI by context and experience.

Teachers’ self-efficacy forliteracy instruction (TSELI)

N Mean S.D. Partial eta2

Setting F(2, 613) ¼ 7.90, p ¼ .001** .025Urban 142 6.85 1.39Suburban 160 7.14a 1.22Rural 314 6.65a 1.44

FaRM (% Low SES) F(4, 575) ¼ 2.69, p ¼ .03 .180e20% 45 6.72 1.4021e40% 178 6.66b 1.4941e60% 164 6.79 1.3761e80% 111 7.11b 1.3381e100% 82 6.96 1.16

Highest education level F(3, 493) ¼ .170, p ¼ .92 .001BA/BS 272 6.82 1.23MA 122 6.90 1.33EdS 94 6.73 1.38EdD/PhD 9 6.99 .70

N ¼ 648; *p < .05; **p < .01.a Indicates a significant difference based on setting.b Indicates a significant difference between 21e40% and 61e80%.

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761758

the quality of the school library were unrelated to TSELI in previousanalyses, these variables were excluded from the regression anal-ysis. In the first step, the demographic variables of gender and racewere entered into the analysis and these explained 2% of the vari-ance in TSELI. Gender was the only variable to make an indepen-dent contribution to the explanation of variance (b¼ .12). Althoughthe equation was statistically significant, the amount of varianceexplained by these variables was quite small, indicating thatpersonal characteristics did not play a strong role in the formationof teacher self-efficacy for literacy instruction.

Next, the preparation and experience variables were added tothe regression. These included ratings of the quality of theiruniversity preparation, highest level of education attained, andwhether their preparation had included a course in children’sliterature. In addition, experiences in the field including ratings ofthe quality of professional development experiences, havingparticipated in a teachers-as-readers group and/or a book clubwereincluded. This set of experiences was considered to representverbal persuasion and vicarious experiences as sources of teacherself-efficacy for literacy instruction. This set of variables explained13% of the variance in TSELI. Six variables made an independentcontribution to the variance explained. These included gender(b ¼ .10), the quality of university preparation (b ¼ .16), the highestlevel of education attained (b ¼ .10), having taken a children’sliterature course (b ¼ .15), the ratings of the quality of professionaldevelopment experiences (b ¼ .16) and participation in a book club(b ¼ .13).

In the third step, context variables representing the analysis ofthe teaching task and context were entered into the equation. These

Table 5Descriptive statistics for self-efficacy quality variables.

Variable Responseset

M

Teacher self-efficacy for literacy instruction 9 2Teacher self-efficacy scale 9 3Teacher self-efficacy for instructional strategies 9 4Teacher self-efficacy for student engagement 9 2Teacher self-efficacy for classroom management 9 2Quality of university preparation 4 1Quality of professional development 4 1Resource support for purchasing books for classroom use 4 1Quality of school library 4 1

N ¼ 648.

included school setting (urban, suburban, or rural), proportion offree and reduced price meals, school level (elementary or middleschool), and the ratings of resource support to purchase classroombooks were included in the regression. With the addition of thesevariables, the proportion of variance explained increased to 30%. Atthis stage, the quality of university preparation (b ¼ .18), highestlevel of education (b ¼ .11), participation in a book club (b ¼ .13),school level (b ¼ .39), and resource support (b ¼ .10) each madea significant independent contribution to the explanation ofvariance.

In the final analysis, the three subscales of the Teacher Sense ofEfficacy scale that assessed self-efficacy for instructional strategies,student engagement, and classroom management were added tothe regression. With this addition, the set of variables explained56% of the variance in TSELI. Seven variables made an independentcontribution to the explanation of TSELI, including school level(b ¼ .33), teachers’ sense of efficacy for instructional strategies(b ¼ .33), and for student engagement (b ¼ .26), the quality ofuniversity preparation (b ¼ .10), the highest level of educationattained (b ¼ .10), resource support (b ¼ .08), and participation ina book club (b ¼ .07). Teacher self-efficacy for classroom manage-ment did not make an independent contribution to explainingTSELI. See Table 7 for the results of the full model.

4. Discussion

In the following sections, the findings of this study are discussedin light of the antecedents of teachers’ self-efficacy proposed byprevious scholars (Bandura, 1977, 1997; Tschannen-Moran et al.,1998). These include demographics, verbal persuasion, vicariousexperiences, the analysis of the teaching task and context, andmore general teacher self-efficacy beliefs.

4.1. Demographics

Demographic factors proved to be weak predictors of Teachers’Sense of Efficacy for Literacy Instruction. A teacher’s race had nosignificant impact on his or her self-efficacy for literacy instruction.And although gender was related to TSELI, it explained only a tinyproportion of the variance. The sample was predominantly femaleand predominantly elementary, so it may be that the self-efficacy ofthese teachers was influenced by the availability of role modelswho were similar to themselves who provided self-efficacyenhancing vicarious experiences.

The number of years a teacher had been teaching was unrelatedto self-efficacy beliefs. This finding is consistent with Bandura’s(1997) assertion that self-efficacy beliefs tend to be establishedearly in learning and that once set, they are stable and resistant tochange without some kind of shock to provoke a reassessment. Thefinding by Tschannen-Moran and Woolfolk Hoy’s (2007) that the

in. Max. Mean S.D. t Cohen’s d

.23 9.00 6.93 1.24

.67 9.00 7.23 .84 217.8 .28

.50 9.00 7.54 .88 216.9 .57

.50 9.00 6.81 1.13 152.8 -.10

.25 9.00 7.34 1.07 174.2 .35

.00 4.00 2.40 .97

.00 4.00 2.58 .98

.00 4.00 2.22 .85

.00 4.00 2.21 .87

Table 6Correlations between teachers’ sense of efficacy for literacy instruction, TSES and sources of self-efficacy.

2 3 4 5 6 7 8 9 10

1. TSELI .61** .55* .56** .40** .06 .23** .21** .23** .032. Teacher sense of efficacy scale .77** .85** .83** .04 .14** .19** .14** .013. TSE for instructional strategies .49** .46** .04 .17** .12** .13** .074. TSE for student engagement .55** .03 .14** .21** .13** .035. TSE for classroom management .03 .03 .14** .08** .006. Years of teaching experience .07* 13** .12** .13**7. Quality of university preparation .19** .20** .048. Quality of professional development .34** .29**9. Resource support for purchasing books for classroom use .24**10. Quality of school library

N ¼ 648; *p < .05; **p < .01.

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761 759

self-efficacy beliefs of novice teachers are more readily influencedby contextual factors than those of career teachers providedevidence to support Bandura’s proposition. Teachers who begintheir careers with strong self-efficacy beliefs tend to build upon themotivation and persistence that those beliefs foster, fuelingcontinued strong self-efficacy beliefs. Conversely, teachers whobegin with weak self-efficacy are likely to engage in self-defeatingactions that reinforce those beliefs and thus persist in their beliefsas well.

4.2. Verbal persuasion and vicarious experiences

The verbal persuasion and vicarious experiences examined inthis study included the highest level of education attained, ratingsof the quality of university preparation for literacy instruction, andratings of professional development experiences. These experi-ences were characterized this way because during preparation andprofessional development, teachers are likely to be told about thepotential of teachers in general and themselves in particular, andperhaps to see other teachers in action, but are a step removed fromactual teaching experiences inwhich teachers judge for themselveswhether or not they are teaching successfully. These experiencescontributed to TSELI in both bivariate and multivariate analyses.Teachers’ ratings of the quality of their teacher preparation wascorrelated to TSELI and made an independent contribution toteachers’ self-efficacy beliefs for literacy instruction in the regres-sion analysis. This is an indication that teacher preparation has thepotential to provide powerful learning experiences which have animpact on teachers’ sense of their ability to successfully teach

Table 7Regression analysis of teacher self-efficacy for literacy instruction with demographic, con

Step 1 Demographics (gender, race)Step 2 Preparation (rating of quality of university preparation, highest education

level, participation in a children’s literature course, rating of quality of professionaldevelopment, participation in a teachers-as-readers group, and participation in a bo

Step 3 Context (school setting, FaRM, school level, resource support for classroom booStep 4 Other teacher self-efficacy beliefs (TSE for instructional strategies, TSE for stude

engagement, TSE for classroom management)

Rating of the quality of university preparationHighest level of education attainedParticipation in a book clubSchool levelResource support for classroom booksTSE for instructional strategies (Step 4)TSE for student engagement (Step 4)

N ¼ 648; *p < .05; **p < .01.

children literacy. In contrast, having attained a higher degree ofeducation was unrelated to self-efficacy for literacy instruction inthe ANOVA, although it did make an independent contribution tothe explanation of TSELI when combinedwith other variables in theregression analysis. This may suggest that it is the quality of what islearned and not the number of credit hours acquired that influ-enced teachers’ beliefs about their own capability to provide qualityliteracy instruction. Participation in quality professional develop-ment experiences and collaborative study with colleagues inTeachers-As-Readers groups or book clubs were also related toteachers’ self-efficacy beliefs for literacy instruction. Whether theconversations during these experiences related directly to thesharing of instructional strategies or not is unknown, but some-thing about these experiences seems to be related to teachers’conviction that they have the capacity to support their students’literacy learning. It may also be, however, that teachers with higherself-efficacy are more likely to seek out these experiences and totake advantage of them when they are offered.

4.3. Analysis of the teaching task and context

Teachers in this study appeared to be taking stock of elements ofthe teaching task and context in relation to the TSELI. Of thecontextual variables examined, having resources available topurchase books for classroom use was related to higher TSELI. Wemight interpret from this that they assessed the teaching task asmore manageable when these resources were more readily avail-able. School level was also an important contextual variable, withelementary teachers, in general, evidencing stronger self-efficacy

text, and source variables.

Teachers’ sense of efficacy for literacy instruction

R2 Adj R2 S.E. F Sig.

.017 .012 1.25 3.37 .035*

ok club)

.129 .011 1.87 6.87 .000**

ks) .304 .281 1.07 13.42 .000**nt .563 .545 .85 31.44 .000**

Beta t Sig..10 2.56 .011*.10 2.85 .005**.07 1.98 .049*.33 8.56 .000**.08 2.05 .041*.33 7.64 .000**.26 5.47 .000**

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761760

beliefs for literacy instruction than those teaching in middleschools. The setting of the school and the proportion of studentsreceiving free and reduced-priced lunches in their school playedonly a small role in relation to teachers’ self-efficacy beliefs forliteracy instruction, as evidenced by their failure to make anindependent contribution in the regression analysis.

4.4. General self-efficacy beliefs

The results of this study indicated that while the more generalTSE (Teachers’ sense of efficacy for instructional strategies, studentengagement, and classroom management) and the more specificTSELI are related, they are distinct constructs that function some-what independently. The CFA model that demonstrated the best fitwas the one that distinguished each of the subscales of the TSESand the TSELI. While there is some overlap between a strong senseof efficacy for general teaching tasks and a sense of efficacy forliteracy instruction, the constructs are not identical. The moderatecorrelations between these variables suggest that teachers withstronger self-efficacy for instructional strategies, student engage-ment, and classroommanagement also tend to feel more capable ofdelivering literacy instruction. Thus, a teacher may feel efficaciousfor instructional strategies, in a general sense, but perhaps notparticularly for those strategies involved in literacy instruction. Ora teacher may feel efficacious for the teaching of reading strategies,but not necessarily for student engagement or classroommanagement. The interplay of these important dimensions ofteaching needs to be better understood.

4.5. Limitations

This study has a number of limitations that should be kept inmind as the results are interpreted. One is that the results maysuffer from the effects of common-source bias, because bothdependent and independent variables were from a single surveyand set of respondents. In addition, because of the voluntary natureof the data gathering method, it may be that the teachers whoresponded differed in some systematic way related to the variablesunder study than teachers who did not. Finally, many of thedependent variables were single item measures and capturingassessments of pre-service preparation and professional develop-ment experiences as well as resource support might be enhancedwith richer measures in future studies.

5. Implications

Greater knowledge of the antecedents of self-efficacy beliefs ofteachers can expand scholars’ understanding of this construct aswell as assist teacher educators, principals, and other practitionersin fostering teachers’ sense of efficacy for literacy instruction. Thisstudy examined empirically two of the sources of self-efficacybeliefs proposed by Bandura (1977, 1997), vicarious experiencesand verbal persuasion, in the context of teachers’ self-efficacy forliteracy instruction, as well as the analysis for the teaching context(Tschannen-Moran et al., 1998). The findings identify factors thatbear further exploration.

The verbal persuasion and vicarious experiences embedded inuniversity preparation emerged as important elements in estab-lishing self-efficacy for literacy instruction. It would be helpful tobegin to explore which elements of their preparation studentsperceived asmost salient to their developing self-efficacy beliefs forthese various aspects of teaching. Evidence emerged that partici-pation in a children’s literature course was a salient part of teacherpreparation, but beyond that, this study did not explore whatdimensions of their preparation led them to rate it as higher or

lower in quality. Moreover, ratings of the quality professionaldevelopment was a salient factor in fostering teachers self-efficacybeliefs, but it remains to be explored which features of thoseexperiences teachers found to be relevant to bolstering theirself-efficacy beliefs. In addition, experiences with book clubsand teachers-as-readers groups each provided experiences thatcontributed to teachers’ self-efficacy for literacy instruction, whichsuggests that keeping teachers engaged with their own literaturelearning is somehow useful to their feelings of capability forengaging their students in learning.

Of the variables assessing the analysis of the teaching context,school level was the strongest factor, with elementary teachersclearly demonstrating higher self-efficacy for literacy instructionthan those in middle schools. Although greater attention has beenpaid recently to preparing middle school teachers for literacyinstruction, more work is evidently needed. Interestingly, theproportion of socioeconomically disadvantaged students in theirschools did not emerge as an independent predictor of TSELI in theregression analysis, suggesting that other factors are more prom-inent to teachers’ self-efficacy for literacy instruction than theirstudents’ circumstances outside of school.

Because our evidence supported the contention that efficacybeliefs are unrelated to years of teaching experience (as it ispresumed to be relatively stable once set), it seems prudent forteacher educators and school leaders to pay close attention to thosefactors that contribute to initial self-efficacy judgments for literacyinstruction. How efficacy beliefs are established, when they aremost malleable, and what factors may lead to their improvementare important issues that this study has begun to address, but moreresearch is needed. Longitudinal studies following pre-serviceteachers through their training and first years in the field with aneye toward their developing self-efficacy for literacy instructionwould be useful as there is much that we do not yet fully under-stand about the factors influencing these beliefs. School leadersmight examine the class assignment practices at their schools aswell as the professional development, coaching and mentoringsupports available to novice teachers in order to do all they can toassure the development of strong self-efficacy beliefs during thefirst years in the field.

The study of teachers’ sense of efficacy began with the RANDstudy’s search for variables that could explain differentialoutcomes in the reading instruction of low-income and minoritystudents in an urban setting (Armor et al., 1976). From that startingpoint, teachers’ sense of efficacy has been shown to be powerfullyrelated to teachers’ motivation and behavior in the classroom aswell as to important student outcomes (Tschannen-Moran et al.,1998). Given the importance of a strong sense of efficacy foroptimal motivation in teaching, scholars, teacher educators andschool leaders would do well to examine how efficacy beliefs forliteracy instruction are developed and fostered over the course ofa career. In this era of accountability, it is apparent that thedevelopment of teachers’ sense of efficacy for literacy instructionwarrants attention.

References

Allinder, R. M. (1994). The relationship between efficacy and the instructionalpractices of special education teachers and consultants. Teacher Education andSpecial Education, 17, 86e95.

Anders, P., Hoffman, J., & Duffy, G. (2000). Teaching teachers to teach reading: para-digm shifts, persistent problems, and challenges. In M. L. Kamil, P. B. Mosenthal,P. D. Pearson, & R. Barr (Eds.), Handbook of reading research, Vol. 3 (pp. 719e742).Mahwah, NJ: Erlbaum.

Armor, D., Conroy-Oseguera, P., Cox, M., King, N., McDonell, L., Pascal, A., et al.(1976). Analysis of the school preferred reading programs in selected Los Angelesminority schools. Report no. R2007-LAUSD. Santa Monica, CA: Rand Corporation.(ERIC document reproduction service no. 130 243).

M. Tschannen-Moran, D. Johnson / Teaching and Teacher Education 27 (2011) 751e761 761

Arya, P., Martens, P., Wilson, P., Altwerger, B., Jin, L., Laster, B., et al. (2005).Reclaiming literacy instruction: evidence in support of literature-basedprograms. Language Arts, 83(1), 63e72.

Ashton, P., & Webb, R. (1986). Making a difference: Teachers sense of efficacy andstudent achievement. New York: Longman.

Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change.Psychological Review, 84, 191e215.

Bandura, A. (1997). Self efficacy: The exercise of control. New York: W. H. Freeman &Co.

Bandura, A. (2001). Guide for constructing self-efficacy scale (Monograph). Stanford,California: Stanford University.

Baumann, J., Hoffman, J., Duffy-Hester, A., & Moon, J. (2000). Teaching teachers toteach reading: paradigm shifts, persistent problems, and challenges. InM. Kamil, P. Mosenthal, P. Pearson, & R. Barr (Eds.), Handbook of readingresearch, Vol. 3 (pp. 719e742). Mahwah, NJ: Erlbaum.

Block, C., & Mangieri, J. (2003). Exemplary literacy teachers: Promoting success for allchildren in grades K-5. New York, NY: Guilford.

Chester, M., & Beaudin, B. Q. (1996). Efficacy beliefs of newly hired teachers in urbanschools. American Educational Research Journal, 33, 233e257.

Chong, W., Klassen, R., Huan, V., Wong, I., & Kates, A. (2010). The relationshipsamong school types, teacher efficacy beliefs, and academic climate: perspectivefrom Asian middle schools. The Journal of Educational Research, 103, 183e190.

Cunningham, A. E., Perry, K. E., Stanovich, K. E., & Stanovich, P. J. (2004). Disciplinaryknowledge of K-3 teachers and their knowledge calibration in the domain ofearly literacy. Annals of Dyslexia, 54, 139e167.

Galda, B., & Cullinan, B. (2003). Literature for literacy: what research says about thebenefits of using trade books in the classroom. In J. Flood, D. Lapp, J. Squire, &J. Jensen (Eds.), Handbook for research on teaching the English language arts (pp.640e647). Mahwah, NJ: Lawrence Erlbaum.

Gibson, S., & Dembo, M. (1984). Teacher efficacy: a construct validation. Journal ofEducational Psychology, 76, 569e582.

Graham, S., Harris, K., Fink, B., & MacArthur, C. A. (2001). Teacher efficacy in writing:a construct validation with primary grade teachers. Scientific Studies of Reading,5(2), 177e202.

Gregoire, M. (2003). Is it a challenge or a threat? A dual process model of teacher’scognition and appraisal processes during conceptual change. EducationalPsychology Review, 15, 147e179.

Guskey, T. R. (1988). Teacher efficacy, self-concept, and attitudes toward theimplementation of instructional innovation. Teaching and Teacher Education, 4,63e69.

Haney, J. J., Wang, J., Keil, C., & Zoffel, J. (2007). Enhancing teachers’ beliefs andpractices through problem-based learning focused on pertinent issues ofenvironmental health science. The Journal of Environmental Education, 38(4),25e33.

Hickman, J. (1981). A new perspective on response to literature: research in anelementary school setting. Research in the Teaching of English, 15(4), 343e354.

Hoewisch, A. (2000). Children’s literature in teacher-preparation programs. ReadingOnline, Available: http://readingonlineorg/critical/hoewisch/index.html.

Hoffman, J., Sailors, M., Duffy, G., & Beretvas, S. (2004). The effective elementaryliteracy environment: examining the validity of the TEX-IN3 observationsystem. Journal of Literacy Research, 36(3), 303e334.

Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structureanalysis: conventional criteria versus new alternatives. Structural EquationModeling, 6(1), 1e55.

International Reading Association. (1999). Providing books other print materials forclassroom school libraries a position statement of the international reading asso-ciation. Newark, DE: International Reading Association.

International Reading Association. (2004). Standards for reading professionals.Newark, DE: International Reading Association.

Johnson, D. (2010). Learning to teach: the influence of a university-school part-nership project on pre-service elementary teacher’s efficacy for literacyinstruction. Reading Horizons, 50(1), 23e48.

Joreskog, K. G., & Sorbom, D. (1993). LISREL 8.72. Lincolnwood, IL: Scientific SoftwareInternational.

Joyce, B., & Showers, B. (2002). Student achievement through staff development (3rded.). Alexandria, VA: ASCD.

Maloch, B., Flint, A., Eldridge, D., Harmon, J., Loven, R., Fine, J., et al. (2003).Understanding, beliefs, and reported decision making of first-year teachersfrom different reading teacher preparation programs. The Elementary SchoolJournal, 103(5), 431e459.

Meijer, C., & Foster, S. (1988). The effect of teacher self-efficacy on referral chance.Journal of Special Education, 22, 378e385.

Nathanson, S., Pruslow, J., & Levitt, R. (2008). The reading habits and literacy atti-tudes of inservice and prospective teachers. Journal of Teacher Education, 59,313e321.

National Council of Teachers of English and the International Reading Association.(1996). Standards for the English language arts. Urbana, IL: NCTE.

National Education Association. (2003). Status of the American public school teacher2000e2001: Highlights. Washington, D. C.: NEA. Available online at: http://www.nea.org/edstats/images/statushighlights.pdf Downloaded 5/6/06.

Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of EducationalResearch, 66, 533e578.

Pinnell, G. (2002). Acquiring conceptual understandings and knowledge. InE. Rodgers, & G. Pinnell (Eds.), Learning from teaching in literacy education.Portsmouth, NH: Heinemann.

Podell, D., & Soodak, L. (1993). Teacher efficacy and bias in special educationreferrals. Journal of Educational Research, 86, 247e253.

Raudenbush, S. W., Rowan, B., & Cheong, Y. F. (1992). Contextual effects on the self-perceived efficacy of high school teachers. Sociology of Education, 65, 150e167.

Ross, J. A. (1994). The impact of an inservice to promote cooperative learning on thestability of teacher efficacy. Teaching & Teacher Education, 10, 381e394.

Ross, J. A. (1998). Antecedents and consequences of teacher efficacy. In J. Brophy(Ed.), Advances in research on teaching, Vol. 7 (pp. 49e74). Greenwich, CT: JAIPress.

Ross, J. A., Cousins, J. B., & Gadalla, T. (1996). Within-teacher predictors of teacherefficacy. Teaching and Teacher Education, 12, 385e400.

Ross, J. A., Cousins, J. B., Gadalla, T., & Hannay, L. (1999). Administrative assignmentof teachers in restructuring secondary schools: the effect of out-of-field courseresponsibility on teacher efficacy. Educational Administration Quarterly, 35,782e804.

Rotter, J. B. (1966). Generalized expectancies for internal versus external control ofreinforcement. Psychological Monographs, 80, 1e28.

Soodak, L., & Podell, D. (1993). Teacher efficacy and student problems as factors inspecial education referral. Journal of Special Education, 27, 66e81.

Stein, M. K., & Wang, M. C. (1988). Teacher development and school improvement:the process of teacher change. Teaching and Teacher Education, 4, 171e187.

Szabo, S., & Mokhtari, K. (2004). Developing a reading teaching efficacy instrumentfor teacher candidates: a validation study. Action in Teacher Education, 26(3),59e72.

Timperley, H. S., & Phillips, G. (2003). Changing and sustaining teachers’ expecta-tions through professional development in literacy. Teaching and TeacherEducation, 19, 627e641.

Tournaki, N., & Podell, D. M. (2005). The impact of student characteristics andteacher efficacy on teachers’ predictions of student success. Teaching andTeacher Education, 21, 299e314.

Tschannen-Moran, M., & McMaster, P. (2009). Sources of self-efficacy: fourprofessional development formats and their relationship to self-efficacy andimplementation of a new teaching strategy. Elementary School Journal, 110(2),228e248.

Tschannen-Moran, M., & Woolfolk Hoy, A. (2001). Teacher efficacy: capturing anelusive construct. Teaching and Teacher Education, 17, 783e805.

Tschannen-Moran, M., & Woolfolk Hoy, A. (2007). The differential antecedents ofself-efficacy beliefs of novice and experienced teachers. Teaching and TeacherEducation, 23, 944e956.

Tschannen-Moran, M., Woolfolk Hoy, A., & Hoy, W. K. (1998). Teacher efficacy: itsmeaning and measure. Review of Educational Research, 68, 202e248.

Vardell, S., & Jacobson, J. (1997). Teachers as readers: a status report. Journal ofChildren’s Literature, 23(1), 16e25.

Wheatley, K. (2002). The potential benefits of teacher efficacy doubts for educa-tional reform. Teaching and Teacher Education, 18(1), 5e22.

Wilson, S., & Berne, J. (1999). Teacher learning and the acquisition of professionalknowledge: an examination of research on contemporary professional devel-opment. In A. Iran-Nejad, & P. D. Pearson (Eds.), Review of research in education,Vol. 24 (pp. 173e209). Washington, DC: American Educational ResearchAssociation.

Woolfolk Hoy, A., & Burke-Spero, R. (2005). Changes in teacher efficacy during theearly years of teaching: a comparison of four measures. Teaching and TeacherEducation, 21, 343e356.

Yeung, K. W., & Watkins, D. (2000). Hong Kong student teachers’ personalconstruction of teaching efficacy. Educational Psychology, 20(2), 213e235.