27
This article was downloaded by: [Umeå University Library] On: 15 November 2014, At: 18:20 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Teacher Development: An international journal of teachers' professional development Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rtde20 Evaluating the impact of collaborative action research on teachers: a quantitative approach John A. Ross a & Catherine D. Bruce b a Ontario Institute for Studies in Education of the University of Toronto , Peterborough , ON , Canada b School of Education & Professional Learning , Trent University , Peterborough , ON , Canada Published online: 13 Nov 2012. To cite this article: John A. Ross & Catherine D. Bruce (2012) Evaluating the impact of collaborative action research on teachers: a quantitative approach, Teacher Development: An international journal of teachers' professional development, 16:4, 537-561, DOI: 10.1080/13664530.2012.734746 To link to this article: http://dx.doi.org/10.1080/13664530.2012.734746 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 &

Evaluating the impact of collaborative action research on teachers: a quantitative approach

Embed Size (px)

Citation preview

Page 1: Evaluating the impact of collaborative action research on teachers: a quantitative approach

This article was downloaded by: [Umeå University Library]On: 15 November 2014, At: 18:20Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Teacher Development: An internationaljournal of teachers' professionaldevelopmentPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rtde20

Evaluating the impact of collaborativeaction research on teachers: aquantitative approachJohn A. Ross a & Catherine D. Bruce ba Ontario Institute for Studies in Education of the University ofToronto , Peterborough , ON , Canadab School of Education & Professional Learning , Trent University ,Peterborough , ON , CanadaPublished online: 13 Nov 2012.

To cite this article: John A. Ross & Catherine D. Bruce (2012) Evaluating the impact ofcollaborative action research on teachers: a quantitative approach, Teacher Development:An international journal of teachers' professional development, 16:4, 537-561, DOI:10.1080/13664530.2012.734746

To link to this article: http://dx.doi.org/10.1080/13664530.2012.734746

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 tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand 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 Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Page 2: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 3: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Evaluating the impact of collaborative action research onteachers: a quantitative approach

John A. Rossa* and Catherine D. Bruceb

aOntario Institute for Studies in Education of the University of Toronto, Peterborough, ON,Canada; bSchool of Education & Professional Learning, Trent University, Peterborough,ON, Canada

(Received 20 April 2011; final version received 19 March 2012)

The authors extend findings from qualitative research on the effects of actionresearch by reporting two linked quantitative studies (N = 80 and 105). Theyfound that teachers who participated in collaborative action research experiencedstatistically significant improvements in attitudes to educational research andteacher efficacy. The pre–post changes were robust across conditions of teachergender, career stage, and qualifications. Teachers benefited more if they (i) rec-ognized the importance of the data analysis and process reflection stages ofaction research; (ii) participated in action research that was rigorous and/or ledto changes in their conceptual understanding; (iii) worked in schools thatfostered professional learning; and (iv) had participated in research activitiesprior to these action research studies.

Keywords: action research; teacher efficacy; attitudes to research; in-service;collaborative action research

Action research is an empowering approach to teachers’ professional learning thatis associated with the transformation of instruction (Goodnough 2008), the integra-tion of research and practice (Postholm 2009) and the enhancement of teachers’self-efficacy beliefs (Moran 2007). To illustrate the potential contribution of quanti-tative investigation to the corpus of findings on the effects of action research, wereport two linked studies of collaborative action research that generated statisticallysignificant benefits for teachers. The focus of our investigation was on two claimsreported by qualitative researchers: participation in action research improves teacherattitudes to research (Atay 2008) and increases confidence in their professional abil-ities (Henson 2001). Our first goal was to triangulate the findings from our quantita-tive inquiry with the results from qualitative studies in order to increase thegeneralizability of claims previously reported. Our intent was to provide between-study triangulation of qualitative findings (comparing the results of our quantitativestudy to conclusions drawn by qualitative research) rather than to conduct within-study triangulation (in which quantitative and qualitative data from a single studyare juxtaposed). Our second goal was to identify potential moderators of actionresearch impact on teachers.

*Corresponding author. Email: [email protected]

Teacher DevelopmentVol. 16, No. 4, November 2012, 537–561

ISSN 1366-4530 print/ISSN 1747-5120 online� 2012 Teacher Developmenthttp://dx.doi.org/10.1080/13664530.2012.734746http://www.tandfonline.com

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 4: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Knowledge to guide practice and policy emerges by aggregating the results ofmultiple studies. Although procedures for the rigorous synthesis of qualitativeresearch are available, see Au (2007) for an especially powerful example, meta-analysis requires quantitative data. Even when quantitative studies of action researchare conducted, the statistics required by meta-analysts are often omitted: for exam-ple, Edwards (2008) used General Linear Modeling to identify the effects of actionresearch but did not provide the statistics that would enable meta-analysts to esti-mate the effect size of the intervention. When meta-analysts review strategies fordeveloping teacher capacity and creating knowledge of teaching, action research islikely to be overlooked.

Action research poses multiple challenges for the quantitative researcher. First,action research is emergent. Pre-ordinate planning, for example, the identification ofpossible teacher outcomes prior to the implementation of pre–post designs, is frus-trated when teachers and their academic partners cannot predict where an actionresearch project will lead. Second, the democratic ideals and non-hierarchical rela-tionships among action research participants inhibit the czar-like control requiredfor the establishment of equivalent experimental conditions; particularly difficult toimplement is the randomization of professional learning strategies, the gold-standardof quantitative research. Third, teacher samples in the typical action research projectare too small to provide adequate statistical power. Fourth, action research is not ahomogenous treatment. Taxonomies of action research types abound (e.g., Kemmis2009; Rearick and Feldman 1999; Somekh and Zeichner 2009). In addition, thesame label is used for different types, e.g., participatory action research may or maynot require commitment to radical social change (Brydon-Miller and Maguire 2009)and some frequently encountered forms of action research are rejected as inadequateby leading theorists in the field (Kemmis 2006). Despite these challengesquantitative examination of the teacher outcomes of action research is feasible anddesirable.

Framework for quantitative study

Context

The context of our research was an action research initiative of the ElementaryTeachers Federation of Ontario (Canada) in 2007–08 and 2008–09. The federation(union) invited teams of four to six Kindergarten to grade 8 teachers to submit pro-posals to participate in job-embedded professional learning by investigating a ques-tion related to their teaching practice and their beliefs about student learning. Eachteam was assigned an academic researcher to support the team’s inquiry and to con-duct selected case studies. The second author was a member of the academic facili-tation group; the first author conducted the quantitative evaluation and was notinvolved in program design or delivery.

Each team participated in a summer training session, met face-to-face and elec-tronically with their academic facilitators through the school year, and participatedin a final debriefing. The action research teams (1) formulated an action researchquestion based on a problem the team identified in their teaching and/or studentlearning; (2) established that the question was worth exploring through reference toexternal data sources, such as research articles, and a personal desire to ‘learn moreabout this topic’; (3) developed a comprehensive action plan, including instructionalinterventions, to address the research question using the project’s Action Research

538 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 5: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Planner; (4) implemented the action plan and gathered data; (5) drew conclusionsabout the data and the action research project; and, (6) presented findings in aninterim and final report, and through conference presentations. For example, oneaction research team identified a concern with the limited level of patterning under-standing their primary students seemed to demonstrate, including an overemphasison simple AB color patterns. They wondered whether presenting more complexproblems and questions to their students would improve the depth and range of stu-dent thinking and understanding about repeating and linear growing patterns. Theseteachers developed a research question based on their concern: How does a focuson rich problems and teacher questioning improve student understanding of patternsand their relationships in the primary grades? The group then searched for and readresearch articles about promising strategies for teaching patterning. They also reliedon their mathematics research partners to support teacher content learning about pat-terning. With several days of release time, the team then designed a pre- and post-assessment for target students as well as a series of problem-based lessons toexplore patterning with their students (ages 4–8). Each teacher assessed their stu-dents and implemented the co-planned lessons, documenting student learningthrough photographs, observations, and video (with the support of researchers). Atthe end of two cycles of lesson co-planning and co-teaching, the teachers thenimplemented the post-assessment with students and analyzed the data. Over severalmeetings, the team finalized their analysis of student data (work samples, fieldnotes, video and photographs) as well as teacher journals. These analyses were thendocumented in the form of a full report, with the support of research partners. Thereport included a table describing teacher learning through collaborative actionresearch, included here as an example of the power of collaborative action researchas a model of professional learning (Table 1).

The results were shared with other teachers in the district at professional learn-ing sessions, with administrators in the district at a principal meeting, and with par-ticipants at a provincial mathematics conference. Additional research articles forpeer-reviewed journals and conference presentations were generated in collaborationwith researchers.

The projects instantiated collaborative action research (CAR) as defined byCapobianco’s (2007) criteria: equal participation by teachers and researchers indefining the classroom-based problem to be addressed; equal participation in mak-ing decisions at each CAR stage, including decisions about research methods; co-learning by teachers and researchers in the development of data collection, analysis,and reporting skills; mutual sharing by teachers and researchers of their personaland critical reflections on the CAR process; development of individual and collec-tive knowledge about teaching and learning.

Outcome variables

Figure 1 displays the framework that guided the data collection. Since the CARquestions were unique to each team and changed over time, we focused thepre- and post-surveys on the overall intent of action research, which is not theproduction of abstract knowledge but a combination of theoretical and practicalknowledge that contributes to school improvement (Cochran-Smith and Lytle 1996).There were two sets of outcomes:

Teacher Development 539

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 6: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Attitudes to research

Teachers tend to find educational research unhelpful because the issues addressedby researchers are not the core concerns of teachers and the culture of researchsupports values that differ from the practicality ethic of teachers (Ramaley 2004;Shkedi 1998). The goal of action research is not to re-create teachers as researchersbut to support them in developing an inquiry stance and a willingness to considerresearch-based instructional practices (Postholm 2009). There is qualitative evidencethat action research participation improves teacher attitudes toward research (Atay2008; McDonough 2006; Warren, Doorn, and Green 2008). Following Cousins andWalker (2000) we distinguished two sets of attitudes: perceived relevance ofresearch to teaching and usefulness of applied research.

Teacher efficacy

Teacher efficacy concerns teachers’ beliefs about their professional abilities, specifi-cally whether they believe that they will be able to bring about student learning.Teachers with high levels of confidence in their teaching ability set higher goals forthemselves and their students, persist through obstacles, are more willing to imple-ment challenging instructional strategies, and produce higher student achievement

Table 1. Teacher-reported professional learning and related shifts in practice (one casestudy team).

Shift from Shift to

Teaching problem solving in isolation Teaching math with problem solving as thecore component through teacher questions,rich tasks, having students make conjectures,facilitating student analysis of ideas.

Teacher expectations based on grade-specificcurriculum

Teacher expectations based on theunderstanding of a continuum of learninggenerated through listening to and observingstudents in order to see what they can do – afocus on higher order thinking as part of thecurriculum and going beyond the curriculumceiling.

Surface-level content knowledge of repeatingand growing patterns (leading to rotedelivery of instruction)

Deep understanding of repeating andgrowing patterns as the foundation tomathematics (leading to investigation-basedteaching and learning).

Limited approach to teaching patterning Multiple approaches/strategies to teachingpatterning (e.g., show the core in 3 differentways: hands, SmartBoard, yarn; use of arange of manipulatives, books, charts).

Teaching concepts and strands in isolation Teaching concepts with their natural inter-strand connections (patterning linked togeometry and geometric representations,number sense/near and far predictions, datamanagement/graphing).

Managing a math class where the teacher isthe consistent leader

Facilitating a community of learners wherestudents are listening to one another, buildingon one another’s ideas, and all parties arevaluing children’s thinking and learning.

540 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 7: Evaluating the impact of collaborative action research on teachers: a quantitative approach

than teachers with low teacher efficacy (Ross 1998; Tschannen-Moran, WoolfolkHoy, and Hoy 1998). Qualitative studies have found that participation in actionresearch has positive effects on teacher efficacy (Bruce and McKenzie 2008; Bonner2006; Caro-Bruce and McCreadie 1994; Henson 2001).

Moderators of action research outcomes

We investigated two dimensions of school context and several dimensions ofteachers’ prior experience as variables that might affect the impact of CAR.

Attitudes to collaboration

CAR is intensely collaborative within the team and between the team andtheir academic facilitators. Qualitative researchers report that action researchcontributes to teacher support for collaboration (Capobianco 2007; McDonough2006; Moran 2007). We anticipated that teachers who reported that they workin schools with collaborative cultures would be more likely to have the skillsneeded to establish constructive working relationships within their CAR teams.We predicted that teachers with higher commitment to collaboration wouldbenefit more from CAR than teachers with lower levels of collaborativeattitudes.

Figure 1. Framework for investigating the teacher effects of action research.

Teacher Development 541

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 8: Evaluating the impact of collaborative action research on teachers: a quantitative approach

School culture

Cochran-Smith and Lytle (1996) argued that organizational conditions such as reluc-tance to ask for help and norms of privacy are obstacles to action research. We pre-dicted that teachers who believed they were part of a learning community in theirschools (Louis and Marks 1998) would be more likely to benefit from CAR thanteachers in schools with weaker learning cultures.

Prior research experience

Teachers need to have sufficient knowledge of research issues to participate equallyin CAR decisions, especially those involving methodological choices. AlthoughCAR provides opportunities for teachers to develop research skills, we predictedthat teachers who had prior experience in research activities would benefit morefrom CAR than teachers with less experience.

Stages of CAR

We anticipated that individual teachers would find some stages of CAR moreengaging than others. We predicted that teachers who attached greater importanceto analyzing data and reflecting on CAR experiences, stages in which individualand collective reflection on teacher practice is especially strong, would benefitmore from CAR than teachers who attached greater importance to other CARstages.

Sharing control

Capobianco (2007) emphasized equality of participation by teachers and researchersas essential attributes of CAR. Qualitative researchers report that hijacking of actionresearch by elites (administrators, government officials) has occurred (Frankhamand Howes 2006; Salleh 2006). A division of labour may emerge in which teacherscede responsibility for key technical decisions (especially research design and dataanalysis) to academic partners (Ross and Ben Jaafar 2006; Hall 2009; Koutselini2008). Hierarchical relationships between teachers and academics are particularlylikely when action research is a degree assignment (Cain et al. 2007). Failure toparticipate in all decisions could reduce action research to unreflective, proceduralactivities (Ax, Ponte, and Brouwer 2008). We predicted that teachers who partici-pated in CAR projects in which control was equally shared would benefit morefrom CAR than teachers in projects with a hierarchy of roles.

Rigor of action research

Giving teachers who are relatively inexperienced in research an equal voice in tech-nical decisions may jeopardize the credibility of the research but denying equity indecision-making may limit teacher commitment to the process and its outcomes.The central dilemma is the conflict between ownership and expertise. Qualitativeresearchers provide a variety of criteria for judging CAR rigor (Anderson and Herr1999; El-Dib 2007; Elliott 2007; Oja and Smulyan 1989) with the strong recom-mendation that these be applied selectively and flexibly (McMahon and Jefford2009). We predicted that teachers who believed they participated in more rigorous

542 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 9: Evaluating the impact of collaborative action research on teachers: a quantitative approach

CAR would benefit more than teachers who believed they participated in lessrigorous CAR.

Conceptual understanding

Teachers’ understanding of their practice and their professional knowledge candeepen during action research because it provides a safe place in which reflectioncan occur (Goodnough 2008; Li 2008). We predicted that teachers whoexperienced a CAR that deepened their understanding of core elements of teachingwould benefit more than teachers who reported having a less reflective CAR pro-ject.

Demographic moderators

Teacher background variables likely predict the outcome variables. For example,females report higher teacher efficacy than males (e.g., Anderson, Greene, andLoewen 1988), possibly because teaching is viewed as a female occupation andfemale teachers tend to be more in tune with the dominant ideology of schools(Kalaian and Freeman 1994). However, our interest is not in these well-establishedrelationships but in whether teacher responses to CAR vary on the basis of teachercharacteristics like gender, career experience, and certification. Although there islimited qualitative evidence suggesting that younger teachers are more responsive toaction research (Vogrinc and Zuljan 2009) we can think of no plausible reason whythis would be the case. Other demographic factors, such as gender, certification,additional training, have not been investigated.

Research questions

Given the large number of potential moderators included in our framework, weaddressed a subset in each year: Study 1 (2007–08) and Study 2 (2008–09). In eachstudy we focused on the effects of CAR on teacher attitudes to educational researchand teacher efficacy. We were guided by three research questions:

(1) What are the effects of CAR on teacher attitudes to research?(2) What are the effects of CAR on teacher efficacy (beliefs about their profes-

sional ability)?(3) Are the effects of CAR on teachers’ attitudes and beliefs moderated by (i)

teachers’ backgrounds, especially prior research experiences, (ii) their schoolcontext, and/or (iii) the attributes of their CAR?

Study 1

Methodology

Teacher teams submitted proposals to their federation to conduct action researchprojects of their own design and received support from university-based facilitators.Pre- and post-surveys were completed by 80 of the 167 teachers, for a response rateof 48%.1

Teacher Development 543

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 10: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Instruments

Instruments used in Study 1 are shown in Appendix 1. Outcome measures wereadministered pre and post. Teacher attitudes to educational research consisted offive items measuring usefulness of research (items 13–17) and five measuringrelevance of research (items 18–22), selected from items developed by Cousins andWalker (2000). Teacher efficacy, i.e., teachers’ beliefs in their ability to bring aboutstudent learning, was measured with 12 items from Tschannen-Moran and WoolfolkHoy (2001), arranged in three scales measuring teacher efficacy for student engage-ment (items 2, 3, 4, 11), for developing instructional strategies (items 5, 9, 10, 12),and for classroom management (items 1, 6, 7, 8).

Potential moderators consisted of items measured at pre-test only. Professionalbackground consisted of items for highest academic degree (item 46), number ofAdvanced Qualifications courses taken (item 47), specialist certificates held (item48), number of years teaching (item 50), gender (item 51) and school or district role(item 49). Contextual factors consisted of three sets of pre-test items. We developedeight items to measure teachers’ collaboration in their school (items 23–30), eightitems for teachers’ prior research experience (items 31–38), and seven items forschool culture (items 39–45).

Data analysis

All analyses were conducted in PASW/SPSS 18.0. Missing values (one value oneight variables) were replaced with Estimation Maximization and outliers (one lowvalue on two variables) were reduced to three standard deviations beyond the mean.

To address the first research question about the impact of CAR on teacher atti-tudes to research, we conducted a within-subjects analysis of variance using GeneralLinear Modeling (GLM). The outcomes were the two sets of attitudes towardresearch, repeated at pre- and post-test occasions. To address the second researchquestion, we used the same procedure for the three dimensions of teacher efficacy.The third research question, which focused on moderators of action research impact,was addressed through correlation and hierarchical multiple regression.

Results

All but one of the scores generated by the instruments was reliable, i.e., Cronbach’salpha = .70+, while the exception (teacher efficacy for instructional strategies) wasslightly below criterion (alpha = .66). All scales were normally distributed: skew-ness and kurtosis scores < 1.0.

Effects of action research on outcomes

We conducted a within-subjects analysis in which the outcomes were perceived use-fulness and relevance of research, repeated over pre- and post-test occasions. Thetop panel of Table 2 displays the main effects on teacher attitudes to research. Therewas a statistically significant improvement on both measures; the effect sizes(Cohen’s d) were small. The bottom panel of Table 2 shows there was a small butstatistically significant improvement in teacher confidence about their ability toengage student attention. These results provide statistical evidence that action

544 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 11: Evaluating the impact of collaborative action research on teachers: a quantitative approach

research may have made a positive contribution to teacher attitudes toward educa-tional research and to their beliefs about their professional abilities.

Moderators of action research effects

We began by constructing a correlation matrix of the post-test scores for teacherefficacy and research attitudes with the demographic and teacher training variables.None was significantly correlated. We concluded that the positive impact of CARon teacher efficacy and research attitudes was relatively constant across all teachergroups.

We continued our investigation by examining whether contextual variables (in-school collaboration, prior research experience, and school culture) influenced theoutcomes of CAR. We searched for moderators using correlations, followed byHierarchical Multiple Regression in which the dependent variable was the post-testscore on an outcome variable, the first block of predictors was the pre-test score,and the second block consisted of variables that correlated with the outcome.2 Noneof the contextual variables significantly predicted post-test scores, after we con-trolled pre-test effects.

Finally, we repeated the regression using stage variables as moderators; i.e., theitems in which teachers indicated how important each of the five stages was to theirprofessional development. Teachers who indicated that analyzing data (beta = .246,t = 2.057, p = .043) and/or reflecting on their CAR (beta = .330, t = 3.048, p =.001) were important were more likely than other teachers to show a pre–postimprovement in their attitudes to the usefulness of educational research. None ofthe other stages – identifying your research question, writing a research plan or col-lecting data – influenced the outcomes.

Study 1 discussion

Study 1 provided statistical evidence that participation in action research had apositive effect on teachers’ perception of the value of educational research. Weinterpret these results to mean that by conducting their own research, data collec-

Table 2. Study 1 teacher attitudes to research and teacher efficacy by time and dimension,significance and effect sizes, N = 80.

Outcome variables Time Mean SD Within-subjects analysis ES

Useful research pre 4.96 .54 F(1,79) = 4.12, p = .046 0.24post 5.09 .52

Relevant research pre 4.68 .53 F(1,79) = 5.60, p = .020 0.27post 4.83 .63

TE instruction pre 4.08 .51 F(1,79) = 3.52, p = .064 0.23post 4.19 .42

TE management pre 4.36 .49 F(1,79) = 0, p = .984 0post 4.36 .52

TE engagement pre 3.99 .59 F(1,79) = 10.77, p = .002 0.33post 4.18 .53

Note: TE = teacher efficacy.

Teacher Development 545

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 12: Evaluating the impact of collaborative action research on teachers: a quantitative approach

tion and analysis skills were de-mystified and teachers acquired specific job-relatedresearch skills. Improving teacher attitudes to research matters because teachersneed to assess research-based claims about the value of particular instructionalstrategies and conduct ongoing research into their own practice. The use ofresearch-based instructional strategies contributes to positive school change (Muijset al. 2004).

The study also provided statistical evidence that participation in actionresearch contributed to teacher beliefs about their ability to engage students inclassroom activities. We interpret this result to mean that collaborative reflec-tions occurring during CAR affirmed teachers’ sense of their own competencethrough such mechanisms as clarification of professional standards, provision ofvicarious experiences and opportunities for enactive mastery (Ross and Bruce2007). Increased professional confidence is compatible with the finding thataction research helps teachers become more aware of their practices (Atay2008). In Study 1 we found that one dimension of teacher efficacy improved(confidence in engaging students in learning activities) while two othersremained unchanged (confidence in classroom management and confidence ininstructional strategies). CAR topics may have given more attention to studentengagement than to other themes.

In Study 1 the findings were robust, i.e., participation in action research benefit-ted all identifiable groups. None of the contextual variables was a significant moder-ator of CAR effects. Teacher outcomes were influenced by teachers’ engagement inparticular CAR stages: teachers who recognized the importance of data analysis andprocess reflection to their professional growth developed more positive attitudestoward educational research than other CAR participants.

The limitations of our findings are first, that the changes from pre to post maynot be attributable exclusively to the specific features of the CAR program but toother factors such as the experience of participating in a study or to some unob-served structural feature affecting teachers across the province. Included in this limi-tation is the possibility that teachers interpreted the surveys in different ways afterparticipating in CAR than they had before. Second, our claims about the influenceof teachers’ contexts are limited to the specific contextual features measured in thesurveys we administered.

Study 2

We replicated Study 1 using virtually the same instruments and procedures with anew sample. In the ‘searching for moderators’ section we replaced CAR stages withother measures represented in Figure 1.

Methodology

In 2008–09, teachers were recruited in the same way as in 2007–08, except theywere asked to focus their CAR on mathematics teaching. Pre- and post-test surveyswere received from 105 teachers for a 66% response rate. Table 3 summarizes thedemographic characteristics of the Study 1 and 2 samples. The typical participantwas a female classroom teacher who had been teaching for 12 years and had com-pleted a four-year undergraduate degree and four Advanced Qualifications coursesprior to participating in CAR.

546 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 13: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Instruments

The outcome measures were identical to those used in Study 1 (shown in Appendix1): three dimensions of teacher efficacy and two dimensions of research attitudesadministered at pre and post. Contextual factors consisted of the same pre-test mea-sures of school collaboration, school culture, and prior research experience. Teacherbackground items were also the same.

The post-test included features of teachers’ projects as potential moderators ofCAR outcomes (shown in Appendix 2). Seven items measured teachers’ perceptionof the rigor of their CAR (items 23–29). Nine items measured sharing control ofCAR activities; each item identified a particular CAR task with five ways in whichcontrol could be shared, ranging from ‘completed by our school administrator’ to‘shared equally by all team members’ (items 30–38). Thirteen items focused onfunctional and dysfunctional features of their projects (items 39–51). Ten itemsprobed the overall effects of CAR on teachers’ understanding of their teaching(items 52–61), generated from the categories of mathematics content knowledgeand pedagogical content knowledge in Borko and Putnam (2005).

Data analysis

We replicated Study 1’s analysis procedures.

Results

We refined two of the scales during the reliability analysis. The final version of thefunctional features scale involved four items (39, 40, 43, 50); the dysfunctional

Table 3. Study 1 and 2 sample characteristics.

Variable Values

Study 1proportions

Study 2proportions

N = 80 N = 105

Highest degree 3-year university 24% 26%4-year university 51% 57%

Masters 24% 17%

AQ courses completed 0–2 40% 33%3–5 33% 34%6–10 27% 32%

Number specialists 0 49% 45%1 31% 31%2–3 20% 24%

School role classroom teacher 68% 80%school support 27% 15%district role 4% 5%

Years teaching (mean = 12.56, SD = 7.22) 1–8 35% 36%9–17 34% 33%18–36 31% 31%

Gender Female 94% 94%

Teacher Development 547

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 14: Evaluating the impact of collaborative action research on teachers: a quantitative approach

features consisted of two items (39 and 49). All scales in the study reached the cri-terion for adequate internal consistency, Cronbach’s alpha = .70 or more. Wereplaced two missing values (both on School Culture) using Estimation Maximiza-tion and replaced outliers (1–2 low scores on six variables) with values three stan-dard deviations below the mean. All but one of the 18 scales was normallydistributed (i.e., skewness < 1.5 and kurtosis < 2.0).

Effects of action research on outcomes

We conducted a within-subjects GLM analysis in which the outcomes were the per-ceived usefulness and relevance of research, repeated over pre- and post-test occa-sions. In Study 2 the attitudes of teachers toward research grew more positive overthe course of their CAR. Table 4 shows that the improvements in teacher attitudesto research were statistically significant. The effect sizes were small to medium butlarger than in Study 1.

We repeated the within-subjects analysis for the second research question.Table 4 shows that there were statistically significant gains on all three dimensionsof teacher confidence in their professional abilities. The improvements were smallto medium. The effect sizes in Study 2 (shown in Table 4) exceeded those in Study1 (shown in Table 2) on two of the three teacher efficacy measures. The pre–postchange on teacher efficacy for student engagement was statistically significant inboth program years and had the same effect size (ES = .34).

Moderators of CAR outcomes

A correlation matrix of the post-test scores for teacher efficacy and research atti-tudes with the demographic and teacher training variables found no statistically sig-nificant correlations. We concluded that the positive impact of CAR on teacherefficacy and research attitudes was constant across all teacher groups.

Table 5 displays the means and standard deviations of potential moderators. Forreasons of reliability, we used the conceptual understanding scale rather than theindividual items in the analysis. However, the item means, not shown, were ofinterest. Teachers perceived that CAR had the greatest impact on aspects of

Table 4. Study 2 teacher attitudes to research and teacher efficacy by time and dimension,significance and effect size levels, N = 105.

Outcome variables Time Mean SD Within-subjects analysis ES

Useful research pre 4.97 .52 F(1,104) = 49.48, p < .001 0.44post 5.20 .53

Relevant research pre 4.74 .52 F(1,104) = 14.33, p < .001 0.28post 4.89 .55

TE instruction pre 4.12 .45 F(104) = 15.34, p < .001 0.39post 4.29 .42

TE management pre 4.50 .42 F(104) = 5.09, p = .026 0.23post 4.60 .43

TE engagement pre 4.28 .55 F(104) = 10.48, p = .002 0.34post 4.46 .49

Note: TE = teacher efficacy.

548 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 15: Evaluating the impact of collaborative action research on teachers: a quantitative approach

teaching associated with teaching for deep understanding such as ‘listening to stu-dents talk about math’ and ‘focusing on mathematical explanations’. CAR had les-ser impact on other dimensions of teaching such as ‘how to sequence [objectives]in the curriculum’ and ‘understanding mathematical procedures and algorithms’.

We searched for moderators using Hierarchical Multiple Regression. We con-structed linear regression models for each outcome separately. We entered the pre-test variable as the first block in the model and in the second block we entered thepotential moderators that correlated with the target outcome measure and their inter-actions.

Once we had controlled for pre-test attitudes, we were unable to find any statis-tically significant predictors of attitudes toward research use or research relevance.The result was the same for teacher efficacy beliefs about classroom management.We found moderators that influenced the impact of CAR on two of the teacherefficacy measures.

Table 6 shows the final model for teacher efficacy for instructional strategies;variables that were not statistically significant predictors have been removed. Thepre-test score was the strongest predictor of the post-test: teachers with higher tea-cher efficacy on the pre-test had higher efficacy on the post-test. Table 6 shows thatthere were two statistically significant predictors of post-test scores after the effectsof the pre-test had been removed. There was also an interaction, displayed in Fig-ure 2. (Figure 2 was created by dichotomizing the predictor variables, above andbelow their medians, and plotting the mean outcome scores for each value from preto post.) The top graph shows the pre–post changes for teachers with low researchexperience; the bottom graph shows the pre–post changes for teachers with highresearch experience. The thick dark lines in each graph show the trend for teachers

Table 5. Study 2 means and standard deviations for potential moderators (N = 105).

Potential moderators Mean SD

Collaboration attitudes 5.47 .57Prior research experience 3.80 1.21Supportive school culture 4.43 .70CAR 5.35 .65CAR shared control 4.63 .41CAR functional attributes 5.70 .50CAR dysfunctional attributes 4.29 .74CAR conceptual impact 5.17 .60

Table 6. Final multiple regression model for pre-test, research experience and CAR rigor,predicting teacher efficacy for instructional strategies (N = 105).

Model Beta⁄ t Sig.

(Constant) .299 .765Pre-test .391 4.579 <.001Research experience .201 2.309 .023CAR .656 2.803 .006Research experience X CAR -1.385 -2.159 .033

Note: ⁄ standardized coefficients; CAR = collaborative action research.

Teacher Development 549

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 16: Evaluating the impact of collaborative action research on teachers: a quantitative approach

who experienced CAR projects of high rigor; the thin lines in each graph depict thetrend for low rigor projects. Teachers who reported conducting a rigorous CAR pro-ject (using deliberate, considered and effective data collection and analysis strate-gies) were more likely than other teachers to experience greater benefits from CAR,regardless of their prior research experience. But teachers who reported being

Figure 2. Interaction of CAR rigor and research experience on pre–post changes in teacherefficacy for instructional strategies.

550 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 17: Evaluating the impact of collaborative action research on teachers: a quantitative approach

involved in a less rigorous CAR showed pre–post improvements only if they hadprior research experience (e.g., those who had previously worked with universityresearchers on a research project or had taken courses in research). If they did nothave prior research experience and were not involved in rigorous CAR, their confi-dence in their ability to develop instructional strategies declined from pre to post.

Table 7 identifies the factors that influenced the size of pre–post changes in tea-cher efficacy for student engagement. As in Table 6, there was a statistically signifi-cant effect for the pre-test: teachers who were confident on entry to the programwere more likely than other teachers to be confident about their professional abili-ties on the post-test. There were greater pre–post gains in teacher efficacy for teach-ers who reported being in a CAR that enhanced their understanding of mathematicsteaching and for teachers working in a school that supported teacher learning. Fig-ure 3, in the same format as Figure 2, shows the interaction. The relationshipbetween pre–post teacher efficacy gains and a CAR feature (promotion of concep-tual understanding) was stronger for teachers who worked in schools in which therewas less support for teacher learning than it was for teachers who worked in moresupportive school cultures.

Study 2 discussion

Study 2 confirmed Study 1 findings that there was a statistically significant pre–postimprovement on teacher attitudes toward educational research and on teacher effi-cacy. This finding is important because teachers who are confident about their pro-fessional abilities implement a broader range of instructional strategies, includingdifficult-to-implement alternatives, and elicit higher achievement from students(Ross 1998; Tschannen-Moran, Woolfolk Hoy, and Hoy 1998). Positive attitudestoward research contribute to teachers’ ability to use data to set goals, track studentperformance, and recalibrate their practice, features that characterize schools thatare successful in challenging circumstances (Harris et al. 2006).

The pre–post changes on teacher outcomes were stronger in 2008–09 than in2007–08. We interpret this to mean that CAR presenters and academic facilitatorslearned from their 2007–08 experiences and provided stronger support for teachers.A key change in strategy was to provide additional professional development, net-working, and research support opportunities for the teachers who led the CARteams. In addition, the topics of the CAR projects in 2008–09 were concentratedwithin a single subject, mathematics, enabling presenters to provide a more specificcontext for the in-service. The benefits of the single subject focus were particularly

Table 7. Final multiple regression model for pre-test, school culture and CAR effects,predicting teacher efficacy for student engagement.

Model Beta⁄ t Sig.

(Constant) -2.371 .020Pre-test .286 3.343 .001School culture 2.716 3.162 .002CAR conceptual impact 2.267 3.389 .001School culture X CAR conceptual impact -3.555 -3.018 .003

Note: ⁄ standardized coefficients; CAR = collaborative action research.

Teacher Development 551

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 18: Evaluating the impact of collaborative action research on teachers: a quantitative approach

visible in the areas of teacher knowledge in which teachers reported that CAR hadthe most impact. Teachers stated that their thinking was most affected with regardto listening to students’ mathematical talk and focusing on mathematical explana-tions. This result provides statistical evidence for the claim that CAR promotedteaching for deep understanding, in line with the curriculum priorities of the teach-ers’ federation that sponsored the program (Leithwood et al. 2004).

Figure 3. Interaction of CAR conceptual understanding and school culture on pre–postchanges in teacher efficacy for student engagement.

552 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 19: Evaluating the impact of collaborative action research on teachers: a quantitative approach

The findings of positive effects were robust. There were statistically significantpre–post changes regardless of teachers’ qualifications, gender, and teaching back-grounds. The size of pre–post changes in teacher efficacy was moderated by (i) fea-tures of teachers’ CAR projects (pre–post changes were greater if teachersparticipated in CAR that was rigorous and/or led to changes in their conceptualunderstanding); (ii) their school contexts (pre–post changes were greater for teachersin schools that fostered teacher learning); and (iii) prior experience (pre–postchanges were greater for teachers who had participated in research activities priorto engaging in this study).

The effect of the moderators was interactive. Teacher efficacy for instructionalstrategies improved from pre to post if teachers reported that they participated inrigorous CAR or had prior research experience. If neither condition prevailed,teachers’ confidence in their math teaching declined. The practical implication isthat program organizers should give special attention to the quality of the collabora-tive action research process for participants inexperienced in research.

Teacher efficacy for engaging student attention improved more for teachers whoreported that they experienced CAR that had a high conceptual focus and for teach-ers working in schools that supported professional learning than was the case forother teachers. These moderators interacted such that the gap between experiencinga low rather than a high conceptual CAR was reduced in schools with collaborativecultures. We interpreted this result to mean that school culture compensated for lowconceptual focus in the CAR.

We were surprised that the other variables included in our framework were notsignificant moderators of pre–post changes in teacher outcomes. In two instancesthe problem may have been ceiling effects. The means for collaboration and func-tional features of CAR projects were 5.47 and 5.70 on a scale with a maximumscore of 6.0, which would attenuate correlations. We thought that experiencing aCAR project with dysfunctional features would have a negative effect on outcomesbut this was not the case, perhaps because our final version of the scale was basedon only two items. Our biggest surprise was that sharing control was not a modifierof outcomes. Since sharing control is an essential attribute of CAR and participatoryevaluation, the result warrants further investigation.

Finally, we reiterate that both Study 1 and Study 2 were limited by two impor-tant features of our methodology: first, the use of a pre–post design is susceptibleto several internal validity threats (particularly history and testing and to a lesserextent in our study context, maturation, instrumentation, mortality and regression).Second, our claims about the robustness of pre–post changes is limited to the spe-cific contextual features measured in our surveys. It is plausible that by providingteachers with different prompts or with open-ended opportunities to describe theirteaching situations we may have found other factors that affect the size of pre–postchanges.

Lessons learned

The accumulation of qualitative inquiries into the processes and products of actionresearch is a core strength of the field. There is a natural affinity of action researchwith the qualitative paradigm and we anticipate that qualitative studies of the impactof action research on teachers will continue to dominate. What is problematic is theshortage of quantitative investigations: it restricts opportunities for methodological

Teacher Development 553

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 20: Evaluating the impact of collaborative action research on teachers: a quantitative approach

triangulation and limits meta-analysis. The two studies reported here demonstratethat quantitative study of action research is feasible if it addresses the actionresearch features that inhibit quantitative inquiry. Even though the design of individ-ual action research projects was emergent, there was sufficient structure (in terms ofa budget, timeline, goals and tools to improve teacher reflection and practice) toenable pre-ordinate planning. By measuring outcomes at a higher level of generalitythan the teacher outcomes for individual CAR projects, we were able to honour tea-cher decision-making in CAR and our inability to control participants’ actionsbecame moot. The problem of action research samples being too small to providesufficient statistical power was circumvented by studying a collection of actionresearch projects organized by a single agency. Finally, the problem of heterogene-ity of action research cases was addressed by focusing on a single type, collabora-tive action research.

The limitation of these two studies was our use of a single group, pre–postdesign to measure the impact of action research on teachers. Although a variety ofinternal and external validity threats to this design have been identified (Posavacand Carey 2007), the most critical in our study context were undetected changesfrom pre to post in teacher interpretations of the outcome scales (testing effect) andenvironmental changes occurring simultaneously with participation in actionresearch (history). These deficiencies weaken the claims we can make and highlightthe need to design randomized experiments in which the teacher impact of actionresearch is compared to the effects of other approaches to professional learning.

The contribution of our two studies to the corpus of action research literature istwofold. First, we confirmed two important benefits of action research participationreported by qualitative researchers, improved teacher attitudes to educationalresearch and confidence in teachers’ professional abilities (increased self-efficacy).Second, we found moderators of the impact of action research that help identify con-ditions in which action research is particularly likely to benefit teachers. The mostimportant finding from this section is that action research is very robust: it is equallysuited to early and later career teachers, to males as well as females, to the minimallyqualified and to those with additional training. The key limiting condition has to dowith implementation: we found evidence that teachers are more likely to benefit fromaction research if they experience a version that is rigorous and contributes to deepunderstanding of the subject they teach. This strengthens the case for teacher andresearcher collaboration, an essential feature of collaborative action research.

Future quantitative studies of action research might focus on important benefitsthat have been reported only in qualitative investigations. For example, actionresearch is a promising strategy for helping teachers in racially conflicted situationsrecognize how their instructional practices and interactions with minority groupmembers are influenced by teachers’ attitudes toward minorities (Foreman-Peck andMurray 2008; Magos 2007; Sowa 2009). Examining the impact of action researchon these and other worthy outcomes from quantitative as well as qualitative stanceswould add support for action research as a powerful technique for professionallearning.

AcknowledgmentsThe research was funded by the Elementary Teachers Federation of Ontario, through theOntario Ministry of Education. The views expressed in the article do not necessarilyrepresent the views of the Federation.

554 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 21: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Notes1. Teachers who did not complete the data collection did not drop out of the CAR; they

declined to complete the post-survey for various individual reasons. There were no sta-tistically significant pre-test differences between those who did and did not complete thepost-survey.

2. We did not use simple change scores (i.e., post minus pre) because they are less reliablethan the scores on which the change is based (see the review in Bowman 2010). Of theseveral solutions recommended by Campbell and Kenny (1999) we applied the residual-ized change score method; i.e., we examined the effect of contextual variables (block 2in the regression) after removing the effect of the pre-test score (block 1) on the post-test.

Notes on contributorsJohn A. Ross is Professor Emeritus of Curriculum, Teaching, and Learning at the OntarioInstitute for Studies in Education at the University of Toronto. The author of over 100refereed journal articles, his research interests are school change, mathematics education,student assessment, and program evaluation.

Catherine D. Bruce is an Associate Professor at Trent University where she teachesmathematics education courses. Her research program focuses on teacher efficacy, models ofprofessional learning and related student learning in mathematics.

ReferencesAnderson, G.L., and K. Herr. 1999. The new paradigm wars: Is there room for rigorous practi-

tioner knowledge in schools and universities? Educational Researcher 28, no. 5: 12–21.Anderson, R., M. Greene, and P. Loewen. 1988. Relationships among teachers’ and students’

thinking skills, sense of efficacy, and student achievement. Alberta Journal of Educa-tional Research 34, no. 2: 148–65.

Atay, D. 2008. Teacher research for professional development. ELT Journal 62, no. 2:139–47.

Au, W. 2007. High stakes testing and curricular control: A qualitative metasynthesis. Educa-tional Researcher 36, no. 5: 258–67.

Ax, J., P. Ponte, and N. Brouwer. 2008. Action research in initial teacher education: Anexplorative study. Educational Action Research 16, no. 1: 55–72.

Bonner, P.J. 2006. Transformation of teacher attitude and approach to math instructionthrough collaborative action research. Teacher Education Quarterly 33, no. 3: 27–44.

Borko, H., and R. Putnam. 1995. Expanding a teacher’s knowledge base: A cognitive psy-chological perspective on professional development. In Professional development in edu-cation: New paradigms and practices, ed. T. Guskey and M. Huberman, 35–65. NewYork: Teachers College Press.

Bowman, N.A. 2010. Can 1st-year college students accurately report their learning anddevelopment? American Educational Research Journal 47, no. 2: 466–95.

Bruce, C., and J. Mackenzie. 2008. Collaborative action research: A model for professionaldevelopment. ETFO Final Report. Peterborough, ON: Trent University.

Brydon-Miller, M., and P. Maguire. 2009. Participatory action research: Contributions to thedevelopment of practitioner inquiry in education. Educational Action Research 17, no. 1:79–93.

Cain, T., M. Holmes, A. Larrett, and J. Mattock. 2007. Literature-informed, one-turn actionresearch: Three cases and a commentary. British Educational Research Journal 33, no.1: 91–106.

Campbell, D.T., and D.A. Kenny. 1999. A primer on regression artifacts. New York: Guil-ford.

Capobianco, B.M. 2007. Science teachers’ attempts at integrating feminist pedagogy throughcollaborative action research. Journal of Research in Science Teaching 44, no. 1: 1–32.

Caro-Bruce, C., and J. McCreadie. 1994. Establishing action research in one school district.Elementary School Journal 95, no. 1: 33–40.

Teacher Development 555

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 22: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Cochran-Smith, M., and S. Lytle. 1996. Communities for teacher research: Fringe or fore-front. In Teacher learning: New policies, new practices, ed. M. McLaughlin and I. Ober-man, 92–112. New York: Teachers College Press.

Cousins, J., and C. Walker. 2000. Predictors of educators’ valuing of systematic inquiry inschools. Special issue. Canadian Journal of Program Evaluation 15: 25–52.

Edwards, K. 2008. Examining the impact of phonics intervention on secondary students’reading improvement. Educational Action Research 16, no. 4: 545–55.

El-Dib, M. 2007. Levels of reflection in action research. An overview and an assessmenttool. Teaching and Teacher Education 23, no. 1: 24–35.

Elliott, J. 2007. Assessing the quality of action research. Research Papers in Education 22,no. 2: 229–46.

Foreman-Peck, L., and J. Murray. 2008. Action research and policy. Journal of Philosophyof Education 42: 145–63.

Frankham, J., and A. Howes. 2006. Talk as action in ‘collaborative action research’: Makingand taking apart teacher/researcher relationships. British Educational Research Journal32, no. 4: 617–32.

Goodnough, K. 2008. Dealing with messiness and uncertainty in practitioner research: Thenature of participatory action research. Canadian Journal of Education 31, no. 2:431–58.

Hall, E. 2009. Engaging in and engaging with research: Teacher inquiry and development.Teachers and Teaching 15, no. 6: 669–81.

Harris, A., C. Chapman, D. Muijs, J. Russ, and L. Stoll. 2006. Improving schools in chal-lenging contexts: Exploring the possible. School Effectiveness and School Improvement17, no. 4: 409–25.

Henson, R.K. 2001. The effects of participation in teacher research on teacher efficacy.Teaching and Teacher Education 17: 819–36.

Kalaian, H., and D. Freeman. 1994. Gender differences in self-confidence and educationalbeliefs among secondary teacher candidates. Teaching and Teacher Education 10, no. 6:647–58.

Kemmis, S. 2006. Participatory action research and the public sphere. Educational ActionResearch 14, no. 4: 459–76.

Kemmis, S. 2009. Action research as a practice-based practice. Educational Action Research17, no. 3: 463–74.

Koutselini, M. 2008. Participatory teacher development at schools: Processes and issues.Action Research 6, no. 1: 29–48.

Leithwood, K.A., P. McAdie, N. Bascia, and A. Rodrigue. 2004. Teaching for deep under-standing: Towards the Ontario curriculum that we need. Toronto: Ontario Institute forStudies in Education of the University of Toronto & Elementary Teachers’ Federation ofOntario.

Li, Y.-L. 2008. Teachers in action research: Assumptions and potentials. Educational ActionResearch 16, no. 2: 251–60.

Louis, K.S., and H.M. Marks. 1998. Does professional community affect the classroom?Teachers’ work and student experience in restructuring schools. American Journal ofEducation 106, no. 4: 532–75.

Magos, K. 2007. The contribution of action-research to training teachers in intercultural edu-cation: A research in the field of Greek minority education. Teaching and Teacher Edu-cation 23, no. 7: 1102–12.

McDonough, K. 2006. Action research and the professional development of graduate teach-ing assistants. The Modern Languages Journal 90, no. 1: 33–47.

McMahon, T., and E. Jefford. 2009. Assessing action-research projects within formal aca-demic programmes: Using Elliott’s context-related criteria to resolve the rigour versusflexibility dilemma. Educational Action Research 17, no. 3: 359–71.

Moran, M.J. 2007. Collaborative action research and project work: Promising practices fordeveloping collaborative inquiry among early childhood preservice teachers. Teachingand Teacher Education 23, no. 4: 418–31.

Muijs, D., A. Harris, C. Chapman, L. Stoll, and J. Russ. 2004. Improving schools in socio-economically disadvantaged areas: A review of research evidence. School Effectivenessand School Improvement 15, no. 2: 149–75.

556 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 23: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Oja, S.N., and L. Smulyan. 1989. Collaborative action research: A developmental process.London: Falmer Press.

Posavac, E., and R. Carey. 2007. Program evaluation: Methods and case studies. 7th ed.Englewood Cliffs, NJ: Prentice Hall.

Postholm, M.B. 2009. Research and development work: Developing teachers as researchersor just teachers? Educational Action Research 17, no. 4: 551–65.

Ramaley, J.A. 2004. The challenges of making research relevant to practice. Paper presentedat the annual meeting of the American Educational Research Association, April, in SanDiego, CA.

Rearick, M.L., and A. Feldman. 1999. Orientations, purposes, and reflection: A frameworkfor understanding action research. Teaching and Teacher Education 15, no. 4: 333–49.

Ross, J.A. 1998. The antecedents and consequences of teacher efficacy. In Research onTeaching, Vol. 7, ed. J. Brophy, 49–74. Greenwich, CT: JAI Press.

Ross, J.A., and S. Ben Jaafar. 2006. Participatory needs assessment. Canadian Journal ofProgram Evaluation 21, no. 1: 131–54.

Ross, J.A., and C. Bruce. 2007. Self-assessment and professional growth: The case of agrade 8 mathematics teacher. Teaching and Teacher Education 23, no. 2: 146–59.

Salleh, H. 2006. Action research in Singapore education: Constraints and sustainability. Edu-cational Action Research 14, no. 4: 513–23.

Shkedi, A. 1998. Teachers’ attitudes towards research: A challenge for qualitative researchers.International Journal of Qualitative Studies in Education 11, no. 4: 559–77.

Somekh, B., and K. Zeichner. 2009. Action research for educational reform: Remodellingaction research theories and practices in local contexts. Educational Action Research 17,no. 1: 5–21.

Sowa, P.A. 2009. Understanding our learners and developing reflective practice. Conductingaction research with English language learners. Teaching and Teacher Education 25:1026–32.

Tschannen-Moran, M., and A. Woolfolk Hoy. 2001. Teacher efficacy: Capturing an elusiveconstruct. Teaching and Teacher Education 17: 783–805.

Tschannen-Moran, M., A. Woolfolk Hoy, and W.K. Hoy. 1998. Teacher efficacy: Its meaningand measure. Review of Educational Research 68, no. 2: 202–48.

Vogrinc, J., and M.V. Zuljan. 2009. Action research in schools – An important factor inteachers’ professional development. Educational Studies 35, no. 1: 53–63.

Warren, S., D. Doorn, and J. Green. 2008. Changes in vision: Teachers engaging in actionresearch. The Educational Forum 72, no. 3: 260–70.

Teacher Development 557

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 24: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Appendix 1.

558 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 25: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Teacher Development 559

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 26: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Appendix 2.

560 J.A. Ross and C.D. Bruce

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014

Page 27: Evaluating the impact of collaborative action research on teachers: a quantitative approach

Teacher Development 561

Dow

nloa

ded

by [

Um

eå U

nive

rsity

Lib

rary

] at

18:

20 1

5 N

ovem

ber

2014