A Teacher Expectation Intervention

Accelerat ing t he world's research.
A teacher expectation intervention:

Modelling the practices of high
expectation teachers
Christine Rubie-Davies
Contemporary Educational Psychology
Cite this paper Downloaded from Academia.edu 
Get the citation in MLA, APA, or Chicago styles
Related papers Download a PDF Pack of t he best relat ed papers 
Teaching high-expect at ion st rat egies t o t eachers t hrough an int ervent ion process
Christ ine Rubie-Davies, Annaline Flint , Lynda Garret t , Lyn McDonald, Elizabet h Pet erson
Teacher expect at ions, et hnicit y and t he achievement gap

Christ ine Rubie-Davies, Melinda Webber, Hana B Turner
Successive t eacher expect at ion effect s across t he early school years

Anne Gregory
Contemporary Educational Psychology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Contemporary Educational Psychology

journal homepage: www.elsevier.com/locate/cedpsych
A teacher expectation intervention: Modelling the practices

of high expectation teachers
Christine M. Rubie-Davies a,⇑, Elizabeth R. Peterson b, Chris G. Sibley b, Robert Rosenthal c
a
School of Learning, Development and Professional Practice, Faculty of Education, The University of Auckland, Private Bag 92601, Auckland 1150, New Zealand
b
School of Psychology, Faculty of Science, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand
c
Department of Psychology, University of California at Riverside, 900 University Ave, Riverside, CA 92521, United States
a r t i c l e i n f o a b s t r a c t
Article history: Since the original Pygmalion study, there have been very few interventions in the teacher expectation
Available online xxxx field and none that have been randomized control trials designed to change teacher practices to reflect
those of high expectation teachers. The current study was designed to address this gap in the literature.
Keywords: Teachers (N = 84) were randomly assigned to either intervention or control groups. Those in the interven-
Teacher expectations tion group attended four workshops at which were presented the instructional strategies and practices of
Mathematics achievement high expectation teachers. At each workshop, the intervention group planned changed practices to intro-
Reading achievement
duce to their classrooms modelled on the behaviours of high expectation teachers. The researchers visited
Experimental study
Intervention
the teachers on three further occasions to ensure fidelity of the implementation. Students in the classes of
the intervention group teachers significantly improved their mathematics achievement over one year,
showing a rate of improvement beyond that shown by the students of the control group teachers. Teach-
ers reported high levels of satisfaction with their changed practices and overall, there was a demonstrable
degree of integrity in the implementation of the intervention as measured by the researchers. Practical
guidelines in relation to the intervention and future directions of the project are included.
Ó 2014 Published by Elsevier Inc.
1. Introduction through descriptive studies. For example, researchers have studied

student characteristics that may lead teachers to form high or low
Teacher expectations for student success are important because expectations for particular students. In a study of this kind,
they are deemed to have a self-fulfilling prophecy effect such that, Tenenbaum and Ruck (2007) have shown that student ethnicity
when teachers have high expectations for student achievement can influence teacher expectations with teachers having higher
they interact with their students in ways that cause their expecta- expectations for white and Asian students when compared with
tions to become realized (Good & Nichols, 2001). Indeed several non-Asian ethnic minorities.
studies have shown that students for whom teachers have high Studies have also identified teacher behaviors that transmit
expectations tend to achieve at higher levels than those for whom expectations. For example, Brophy (1985) showed that teachers
they have low expectations, even when achievement is controlled. tend to wait less time for low expectation students (lows) to
For example, Madon, Jussim, and Eccles (1997) controlled for prior answer a question than they do for high expectation students
mathematics ability and showed that in mathematics for both high (highs), they criticize lows more often for failure than highs, but
and low achievers, teachers’ under- as well as over-estimates of praise them less frequently for success than highs, and they call
achievement produced self-fulfilling prophecy effects. on lows less frequently than highs to answer questions.
Since the initial experimental study of Rosenthal and Jacobson Weinstein and her colleagues (Weinstein, 1989, 1993, 2002;
(1968), teacher expectations have mostly been investigated Weinstein, Marshall, Brattesani, & Middlestadt, 1982; Weinstein
& Middlestadt, 1979) have investigated how students ascertain
teachers’ expectations for them. They have shown that teacher
⇑ Corresponding author. Address: School of Teaching, Learning and Development, differentiation in the curriculum delivered to high and low expec-
Faculty of Education, The University of Auckland, Private Bag 92601, Auckland
tation students, feedback from the teacher, public acknowledge-
1150, New Zealand. Fax: +64 9 623 8827.
E-mail addresses: c.rubie@auckland.ac.nz (C.M. Rubie-Davies), e.peterson@
ment of achievement, the ways that students are grouped, and
auckland.ac.nz (E.R. Peterson), c.sibley@auckland.ac.nz (C.G. Sibley), robert. the verbal and non-verbal behaviors of teachers enable students
rosenthal@ucr.edu (R. Rosenthal). to interpret their teachers’ expectations for their achievement.
http://dx.doi.org/10.1016/j.cedpsych.2014.03.003
0361-476X/Ó 2014 Published by Elsevier Inc.
Please cite this article in press as: Rubie-Davies, C. M., et al. A teacher expectation intervention: Modelling the practices of high expectation teachers. Con-
temporary Educational Psychology (2014), http://dx.doi.org/10.1016/j.cedpsych.2014.03.003
2 C.M. Rubie-Davies et al. / Contemporary Educational Psychology xxx (2014) xxx–xxx
Researchers have also explored how teacher beliefs can moder- would ultimately lead to students achieving at higher or lower lev-
ate expectation effects. For example, Babad (2009) has reported els, dependent on their teachers’ expectations.
that teachers who hold biased beliefs about their students tend Following the initial study, there were several other replication
to be more dogmatic in their views and more authoritarian in their attempts. Raudenbush (1984) identified 18 such studies. He
interactions with students than those who are less biased and showed that the timing of the false information given to teachers
hence expectations are more salient in the classes of high bias regarding students likely to suddenly make great learning gains
teachers and expectation effects are greater. Such teachers are mattered. If the teachers knew the students for more than two
more inclined to judge students based on stereotypical information weeks then they were not influenced by researcher manipulation
rather than on objective results and so are also more likely to form but in the five studies where the experimental manipulation
biased expectations (Babad, 2009). occurred before teachers had met their students, there was an
Hence there is a large body of literature in specific areas of the effect of the intervention on student achievement.
teacher expectation field that has described teacher and student Kerman (1979) trained teachers to distribute their interactions
behavior, characteristics, and beliefs that are associated with high with students more equitably. He argued that frequent differentia-
and low teacher expectations. However, while the very first tea- tion among teachers in calling on high versus low expectation stu-
cher expectation study (Rosenthal & Jacobson, 1968) was experi- dents led to the lows disengaging from class discussions and
mental, there have been few other randomized control trials in therefore decreased their learning. The 742 teachers who volun-
the expectation field, particularly when compared with the large teered for the project were divided into an experimental and con-
number of descriptive studies. Further, no studies have been pub- trol group. Kerman reported that training teachers to interact
lished which have attempted to positively influence student equitably with students led to significant academic gains for the
achievement as a result of an intervention designed to change tea- low achievers as well as a reduction in absenteeism and discipline
cher practices and beliefs to replicate those of teachers who have referrals. Evaluations of this program (Gottfredson, Marciniak,
high expectations for all their students. The current study was Birdseye, & Gottfredson, 1995), however, were less enthusiastic
designed to address this gap in the literature; it is an experimental, about its positive effects for low achieving students. This is proba-
rather than a descriptive study. bly because teacher expectation effects are more complex than
simply a reduction to teacher behaviors.
1.1. Theorizing high expectations Through an aggregation of studies at the time, Brophy (1983,
1985) took a similar approach to changing specific teacher behav-
Policy makers and educational leaders appear to believe that iors by identifying mechanisms through which low teacher expec-
high expectations can translate into improved achievement for tations could be portrayed to students. For example, teachers were
students because there are consistent calls for teachers to have found to wait less time for lows to respond to questions, less was
high expectations for all their students. This implies that teachers demanded of lows and lows were praised less and criticized more
can change their expectations and that currently teacher expecta- than their high expectation peers. Brophy (1983) identified 17 tea-
tions are not sufficiently high for all students and they should be. cher behaviors that portrayed teachers’ expectations. His idea was
But whether or not teachers’ expectations can be increased for all that teachers should be made aware of these behaviors and that
students with corresponding improvements in students’ achieve- then the teachers would monitor their own interactions for equity
ment has not previously been empirically tested. Indeed, the con- in distribution to students. However, Brophy did not collect evi-
ception that some teachers do have correspondingly high or low dence to indicate that teachers did moderate their behavior. There
expectations for all their students has only been explored in the lit- was simply an assumption that making teachers aware of the
erature relatively recently (e.g., Rubie-Davies, 2007). interactions that portrayed expectations would lead to change.
Other researchers (Babad, 1990a; Good & Brophy, 1974) took a
1.2. Intervention studies in the teacher expectation field different pathway to addressing teacher differential behavior.
Good and Brophy (1974) observed teacher behavior towards low
The first ever study in the teacher expectation field (Rosenthal & participants (a group teacher rarely interacted with) and an exten-
Jacobson, 1968) was a randomized control trial in which teachers sion group (students that teachers did not persist with when they
in one school were told that randomly identified students would responded incorrectly to questions). This differentiation in teacher
suddenly blossom. By the end of the first year of the study, overall behavior towards these groups was considered inappropriate by
those who had been identified as ‘‘bloomers’’ had made greater the researchers. The researchers also recorded appropriate teacher
intellectual gains than students who were not randomly assigned. behavior towards two contrast groups. The teachers were then
However, this initial study caused much controversy, mostly individually interviewed, shown the observational data, and given
related to the methodology employed. For example, Snow (1969) feedback on their differentiating behaviors. The teachers were also
argued that the intelligence test used to track changes in the exper- encouraged to interact more frequently with the low participants
imentally identified students was not normed for the youngest stu- and to prompt the extension group more frequently so that the
dents in the study where the biggest gains were found and students had additional response opportunities. Subsequent obser-
therefore their results would have had to have been extrapolated. vations of the teachers did show changes in teachers’ behaviors
He believed this made the validity of the results questionable. towards the two target groups and there was some indication of
Elashoff and Snow (1971) raised the same issue but also added corresponding changes in student behavior. That is, both student
debate about the effects of multiple administrations of the test groups became more inclined than previously to respond to their
and pointed out that whereas there were experimental effects for teachers and to initiate interactions following the teacher
the younger students, the older students did not show statistically feedback.
significant intellectual gains. Nevertheless, while there was debate Babad (1998) claimed that teacher behavior can differ in the
related to whether IQ could be raised experimentally, none of the quantity and quality of learning support provided to students, in
critics at the time, and few researchers today, would deny the exis- how much pressure is put on students, and in the quality of the
tence of teacher expectation effects. In Rosenthal and Jacobson’s emotional support experienced by high and low expectation stu-
(1968) initial work, it was theorized that teachers must interact dents. Based on studies (Babad, 1995; Babad, Avni-Babad, &
differentially with students for whom they have correspondingly Rosenthal, 2003) in which he found that students were particularly
high or low expectations and that this differentiation is what resentful of teachers who gave more emotional support to some
C.M. Rubie-Davies et al. / Contemporary Educational Psychology xxx (2014) xxx–xxx 3
students than others, Babad (1998) argued that it was differential of their beliefs that high and low expectation students need quite
affect that was at the core of the teacher expectation issue. He different learning opportunities, high differentiating teachers group
(Babad, 1990b, 1995) found that teachers provided more emotional students by ability, tend to hold views that intelligence is fixed,
support to high expectation students than to lows although teach- motivate students extrinsically, limit student agency, create a cli-
ers reported the opposite. Similarly to the study by Good and mate of favored and non-favored students, have differential rela-
Brophy (1974) cited above, Babad gave teachers feedback on their tions with parents depending on whether students are high or low
differential emotional support. In the first study (Babad, 1990b), in achievers, and limit opportunities for recognition at the school level
some classrooms, but not all, students reported more equitable to only a few students. On the other hand, the very different beliefs of
emotional treatment following the teacher feedback. However, in low differentiating teachers lead them to create mixed ability group-
the second study (Babad, 1995), no effects were noted when teach- ings, consider that all students can learn, foster intrinsic motivation,
ers were informed about their differential emotional support. provide all students with choices, promote a classroom community,
Some researchers (Proctor, 1984; Weinstein, Soule, Collins, & develop relationships with all parents and offer opportunities for all
Cone, 1991) took a broader view than simply behavioral in endeav- students to be involved at the school level. Weinstein (2002)
oring to change teachers’ expectations. Proctor (1984) argued that reported that the beliefs of the teachers moderated the expectation
high teacher expectations were one of the major facets that differ- effects such that high differentiating teachers tended to exacerbate
entiated effective from less effective schools. He proposed a model differences in achievement between high and low achieving stu-
that was being used in Connecticut to drive the school improve- dents. In the classes of low differentiating teachers, although they
ment process, although no data were provided by Proctor (1984) still had high expectations for some students and low for others,
to indicate the effectiveness of the model in practice. The model because they treated all students similarly and encouraged peer
was designed to show how school climate influenced teacher support, the gap between high and low expectation student achieve-
expectations and how student achievement affected teachers’ atti- ment decreased over a year.
tudes right across the school. Similarly, student characteristics Babad has categorized teachers as high and low bias. High bias
were claimed to influence the learning climate of the school. Proc- teachers are those who react more negatively towards low expec-
tor also asserted that the model showed how teacher expectations, tation students than low bias teachers. In a series of studies, Babad
shaped by the school climate, were causally linked to student and his colleagues (Babad, Bernieri, & Rosenthal, 1989a; Babad,
achievement via teacher interactions. Bernieri, & Rosenthal, 1989b; Babad, Bernieri, & Rosenthal, 1991;
The most comprehensive teacher expectation intervention Babad & Taylor, 1992) showed that teacher education students,
research to date was by Weinstein and her colleagues (Weinstein secondary school students and elementary school students could
et al., 1991). This study was focused on improving the school and all identify from ten-second video clips whether teachers were
classroom environments for students who had been identified as talking to, or even just talking about, a high and low expectation
at-risk and who were transitioning to high school. The researchers student. In these studies, the sound was distorted such that judg-
worked alongside seven teachers whose 158 students were origi- ments needed to be made based on non-verbal behavior alone.
nally assigned to the lowest track in high school to implement All groups could do this accurately but elementary school students
positive expectancy practices. These included providing an were more accurate than secondary school students who in turn
enriched curriculum, implementing heterogeneous grouping such were more accurate than the adults (Babad, 2009). High bias teach-
that some low performing students were placed in the honors clas- ers were more easily swayed by stereotypical information about
ses, increasing the ways in which performance was evaluated, students thus creating biased expectations for their students.
implementing cooperative teaching strategies, encouraging more Low biased teachers judged each student in line with actual
active student participation in learning, developing warm stu- achievement (Babad, 2009).
dent–teacher relationships, encouraging positive relations with The conception that teachers might have high or low expecta-
parents, and fostering student participation in school events. tions for all their students, however, had not been investigated until
Results for the experimental group were compared with those of recently and yet Brophy (1983) proposed that, ‘‘differential teacher
two previous low-track groups at the school (N = 154). Teachers treatment of intact groups and classes may well be a much more
were so enthusiastic about the changes in student engagement widespread and powerful mediator of self-fulfilling prophecy
among the at-risk group that the school abolished tracking. Stu- effects on student achievement than differential teacher treatment
dents previously labeled low-track were reported by teachers to of individual students within the same group or class, even though it
be achieving at similar levels to honors students. Students also has received less attention in the expectation literature’’ (p. 312).
reported becoming more excited about learning than previously. Author has recently investigated this conception within the New
Quantitative measures showed statistically significant differences Zealand context and has been able to identify high and low expecta-
in the achievement of the experimental group when their results tion teachers. These are teachers who, contrastingly, have either
were compared with those of the matched students from previous high or low expectations for all their students relative to achieve-
cohorts during the year of the intervention although disciplinary ment. First (Rubie-Davies, 2006, 2007, 2008a) investigations sug-
referrals and student absences did not show positive trends com- gest that high and low expectation teachers appear to have quite
pared with the comparison groups. Unfortunately, however, one different pedagogical beliefs and instructional practices and these
year later when the experimental group moved to non-project differing beliefs and practices appear to account for the differences
teachers, the advantages in achievement were not maintained. in academic achievement found in the classes of high and low
expectation teachers. The mean achievement gain for students with
1.3. Teacher beliefs that moderate expectation effects high expectation teachers was d = 1.05 over one academic year,
whereas the gain for students with low expectation teachers was
A small group of researchers have identified particular teacher d = .05 (Rubie-Davies, 2007). Further, while the self-beliefs of stu-
beliefs which appear to moderate teacher expectancy effects. Wein- dents with high expectation teachers increased a little over the year
stein has distinguished high and low differentiating teachers of the study, those of students with low expectation teachers fell
(Weinstein, 2002; Weinstein et al., 1982). High differentiating substantially such that there was a statistically significant differ-
teachers believe that high and low expectation students need quite ence in the self-beliefs of the students with high versus low expec-
separate learning tasks whereas low differentiating teachers believe tation teachers after one year which was not evident at the
that all students should have similar learning opportunities. Because beginning of the year (Rubie-Davies, 2006). The differences in
beliefs and practices between high and low expectation teachers than lows (r = .20), and praising high expectation students while
appeared to relate to the way in which teachers grouped their stu- criticizing lows (r = .07).
dents and the consequent learning experiences they provided for A final area in which high expectation teachers have been
students; the classroom climate; and motivation, evaluation, tea- shown to differ markedly from low expectation teachers is the
cher feedback, and student autonomy. The ways in which high way that they motivated students, evaluated and provided them
expectation teachers managed and taught their students which dif- with feedback, and promoted student autonomy (Rubie-Davies,
fered from the beliefs and practices of low expectation teachers will Hattie, Townsend & Hamilton, 2007). Goal setting was at the core
be described below. of these differences. Low expectation teachers did not engage in
goal setting with their students. The high expectation teachers
1.4. Beliefs and practices of high expectation teachers set mastery goals with each of their students based on regular for-
mative evaluation of their learning needs. Teachers provided their
Commonly in New Zealand elementary schools, teachers group students with clear feedback about their progress towards their
their students by ability within their classrooms in core curriculum goals and frequently reviewed and updated students’ goals with
areas such as reading, mathematics, writing, and spelling. Indeed, them. They also promoted student autonomy and motivation, not
New Zealand has been reported as having the highest within-class just through the choices students made about the activities they
ability grouping rate of any OECD (Organization for Economic Coop- engaged in, but also by allowing students to choose the focus for
eration and Development) country (Wagemaker, 1993). In first their learning goals based on the results of their assessments.
author and colleagues’ study (Rubie-Davies, Hattie, Townsend & These behaviors of high expectation teachers are reinforced by
Hamilton, 2007), the low expectation teachers did ability group the literature. Clarke, Timperley, and Hattie (2003) have previously
their students for instruction and activities. However, surprisingly espoused the rich benefits for students when they are given spe-
considering the New Zealand tradition, none of the high expectation cific skills-based learning goals and provided with clear feedback
teachers grouped their students by ability for learning activities. about their learning. Indeed, Hattie (2009) has shown that forma-
Hence, students were ability grouped for instruction but not for tive feedback has a d = .73 effect on student achievement.
their learning experiences and in most classes students chose the Evidence within the motivation field (Brophy, 2004) reinforces
activities that they completed. This non-differentiation of learning high expectation teachers’ concern with fostering student engage-
experiences may relate to input as described by Harris and ment and motivation. Brophy (2004) has detailed the benefits to
Rosenthal (1985). All students in the classes of high expectation student learning when teachers can successfully motivate stu-
teachers were involved in challenging, exciting instructional activi- dents. When students are motivated, they become more cogni-
ties (Rubie-Davies, 2008b); there was not the differentiation in their tively engaged in their learning, eager to learn, excited about
learning experiences evident in classes of low expectation teachers. enhancing their skills, experience pride in their achievements,
Hattie (2009) has shown that within-class ability grouping has only and learn that they can achieve their goals with reasonable effort.
a small effect on student achievement (d = .16) yet it is favored in Further, the autonomy literature provides evidence for the posi-
New Zealand and in some other countries such as the UK. Further, tive effects on students when they are provided with choices, a fur-
while ability grouping in some form is relatively common in schools ther aspect of the way in which high expectation teachers catered
in many countries (e.g., tracking in the US), it has been shown to for student learning. In an experimental study, Patall, Cooper, and
have detrimental effects on student self-beliefs (Linchevski & Wynn (2010) randomly assigned students to have a choice of
Kutscher, 1998; Oakes, 1990). Overall, it could be argued that the instructional activity to complete or to have no choice. Later the
minimal positive effects on achievement of within-class ability conditions were reversed. The researchers found that providing
grouping are outweighed by the detrimental effects that such students with a choice resulted in greater intrinsic motivation to
grouping can have on student self-perception. complete the task, increased self-competence beliefs, and
High expectation teachers also appeared to create a warmer improved performance on a test when compared with those stu-
classroom climate than low expectation teachers (Rubie-Davies dents who were given no choice.
and Peterson, 2011). They were more supportive of students, The evidence above suggests that providing students with clear
engaged with them more constructively, appeared to develop war- feedback on their learning goals, fostering intrinsic motivation and
mer teacher–student relationships and managed student behavior providing students with choices in the tasks that they complete, all
more positively than their low expectation counterparts. There is appear to have marked effects on both student social-psychological
increasing evidence of the importance of a warm classroom climate and academic outcomes and these have all been found to be behav-
for enhancing student learning. For example, Hattie (2009) has iors of teachers with high expectations for all their students. How-
shown that a positive teacher–student relationship has a large effect ever, if teachers were to change their beliefs and instructional
on student learning (d = .72). The importance of the adult–child practices to reflect those of high expectation teachers, this would
relationship is exemplified in a study by Hamre and Pianta (2001) necessitate some form of professional development comprising an
in which 179 students were tracked to examine the degree to which intervention. It would involve working with regular teachers and
the quality of the teacher–student relationship at kindergarten pre- their students without any attempt to induce false expectations.
dicted academic and behavioral outcomes for students in eighth Hence, the focus would be on changing teacher practices in order
grade. Students whose relationship with their teacher was poor at to influence their beliefs and expectations about what was possible
kindergarten had lower achievement and more behavioral difficul- for students. To our knowledge, there has been no randomized con-
ties throughout elementary school and through to eighth grade, trol trial that has attempted to experimentally change teachers’
even accounting for gender, ethnicity, cognitive ability and initial expectations for all students through changing teacher beliefs and
behavior ratings. Further, a meta-analysis (Harris & Rosenthal, instructional practices. The current study was designed to address
1985) within the teacher expectation field suggested that teacher this gap in the literature and measured changes in student achieve-
behaviors that affected the whole class were more important than ment as a result of the high expectation teacher intervention.
the individual teacher–student interactions that had been fre-
quently investigated. For example, classroom climate (r = .37) and 1.5. Purpose
the provision of differential opportunities to learn in some classes
(r = .33) had larger effects on student achievement than did behav- The current project was a randomized control trial in which
iors such as more frequently calling on high expectation students 90 teachers were randomized at the school level into either an
expectancy or control condition, with the goal of assessing impact intervention teachers (52%) and 1167 (49%) with control group
on student achievement after one school year. Hence, the purpose teachers. Table 1 provides further demographic details for teachers
of the current study was to test the effects on student achievement and students.
when teachers changed their practices to those of high expectation
teachers. The research questions were:
2.2. Procedure and research design
(1) Does a teacher expectation intervention result in increased
achievement for students in mathematics? This study was a randomized controlled trial of a teacher expec-
(2) Does a teacher expectation intervention result in increased tation intervention. Schools in one area of Auckland, New Zealand’s
achievement for students in reading? largest city were invited to participate in the study based on their
socioeconomic status. One particular area of Auckland was chosen
We hypothesized that teacher participation in the intervention as it had the widest spread of socioeconomic areas and the project
would predict increases in student achievement. leaders wanted to ensure that schools in the study represented the
full socioeconomic range. One high socioeconomic school in a dif-
ferent area of Auckland agreed to be part of the project when one
2. Method school in a high socioeconomic part of the target area declined to
participate. The study involved randomly assigning teachers
2.1. Participants within each school to intervention and control conditions follow-
ing collection of baseline data. Because there were only small num-
2.1.1. Teachers bers of teachers at any one school, a simple random selection was
Participants were 90 teachers, 46 who were randomly assigned conducted whereby a research assistant not involved in the study
to the teacher expectation intervention, and 44 who formed the was blindfolded and drew names from a container for each school
control condition and undertook the regular professional develop- and the assignment was recorded by a further research assistant.
ment provided by their school (mostly within school and off-site When there were an uneven number of teachers participating in
workshops) but were not involved in the intervention activities. a school, more teachers were randomly assigned to the interven-
By the end of the year of the study that this paper is reporting, 6 tion group than to the control group. While it was recognized that
teachers had left the study, 3 for personal reasons, 2 due to retire- there could be contamination from intervention teachers (Bloom,
ment and 1 who left her school for promotion in another school. 2005), there were a number of factors that influenced this decision
Hence, the reporting for the remainder of this paper relates to in the design. First, the intervention was relatively complex and so
the 84 teachers who remained in their schools throughout the year. not easily explained to another teacher. Second, control group
Of those, 43 were from the intervention group and 41 were in the teachers were to receive the same information as the intervention
control group. Teachers were recruited from 12 schools across pre- teachers in the second year of the study and so only had to wait
dominantly suburban areas in Auckland, New Zealand. The number
of teachers per school ranged from 2 to 11 (M = 7.42). Approxi-
Table 1
mately 66% (n = 55) taught at elementary level (Years 1–6; kinder-
Teacher and student demographics.
garten-grade 5 in the US) and 33% taught at middle school level
(Years 7–8; grades 6–7 in the US). Most teachers were female Teacher demographics Intervention Control
(70%). In terms of ethnicity, 52 (62%) identified themselves as Teacher gender Female: 27 Female: 32
New Zealand Caucasian, 7 were Māori (8%), 9 were Pasifika Male: 17 Male: 8
(11%), 3 were Asian (4%) and 13 were from overseas but from a Teacher ethnicity NZ European: 28 NZ European: 24
Maori: 4 Maori: 3
Western country (e.g., England, South Africa, the US, Canada, Ger- Pasifika: 3 Pasifika: 6
many) (16%). Of these teachers, 73 were originally trained in Asian: 1 Asian: 2
New Zealand (87%). The teachers were teaching Years 3–8 (Grades Other: 8 Other: 5
2–7), with the median and mode Years 5–6. Teaching experience Years teaching experience 11.82 (11.11) 9.18 (7.63)
Country where trained to teach New Zealand: 36 New Zealand: 37
ranged from 1 to 41 years with the mean 10.6 years (SD = 9.60)
Other: 8 Other: 3
and the median 7 years. Most teachers were teaching in mid-socio- Teaching level Elementary: 26 Elementary: 29
economic schools (n = 50, 60%), with 16 (19%) teaching in high Middle School: 18 Middle School: 11
socioeconomic areas and 18 in low (21%). Student demographics Intervention Control
Student age 6 years: 2 6 years: 21
2.1.2. Students 7 years: 88 7 years: 132
8 years: 193 8 years: 232
A total of 2408 elementary school students in New Zealand in
9 years: 229 9 years: 201
the classes of the teachers above participated in this longitudinal 10 years: 216 10 years: 252
study. At the beginning of the study, student ages ranged from 6 11 years: 227 11 years: 205
to 13 with a mean of 9.7 years. There were slightly more boys than 12 years: 276 12 years: 123
girls (n = 1234; 51%). A range of ethnicities were represented; 1167 13 years: 10 13 years: 1
Student gender Girl: 597 Girl: 577
students were Caucasian (49%), 424 were Māori (18%), 398 were
Boy: 644 Boy: 590
Pasifika (17%), 331 were Asian (14%) and 88 were classified as Student ethnicity Caucasian: 606 Caucasian: 561
Other (4%). European students were defined as those who identi- Maori: 203 Maori: 221
fied as New Zealand European as well as those from countries such Pasifika: 203 Pasifika: 195
as Australia, the UK, or the US. Pasifika students were defined as Asian: 179 Asian: 152
Other: 50 Other: 38
students who identified their heritage as being from one of the Class level Grade 2: 56 Grade 2: 121
Pacific Island nations (e.g., Samoa, Cook Islands, Tonga, Fiji) and Grade 3: 202 Grade 3: 233
Asian students were those who identified themselves as coming Grade 4: 227 Grade 4: 210
from South East Asia and the Indian sub-continent. Any students Grade 5: 217 Grade 5: 240
Grade 6: 207 Grade 6: 225
who did not fall into one of the four major categories were classi-
Grade 7: 332 Grade 7: 138
fied as Other. There were 1241 students in the classes of
one year to be fully involved. Third, it was a condition of the inter- expectation teachers organized their classrooms in order to accom-
vention training that the experimental group did not discuss the modate these practices into their classes, and how they managed
intervention with other teachers. Fourth, the project leader students, were introduced. The intervention teachers then chose
believed that had the random assignment been by school, that one of the above practices and together with other intervention
principals would not have been so keen to be involved since there teachers planned how they would introduce the practice into their
would be no benefit to control group schools for a whole year. Ran- classrooms. This approach gave teachers ownership of the process
dom assignment within schools, from the perspective of principals, (Timperley, Parr, & Bertanees, 2009) and allowed them to begin
meant that provided the intervention was effective schools poten- experimenting with the practices in their own classes.
tially benefitted as soon as the intervention procedures began. All intervention teachers were videoed teaching whatever they
chose for 20 min in the 2 weeks preceding their attendance at the
2.2.1. Treatment condition first workshop. At this first workshop, the teachers were also
The treatment consisted of first, attendance by the intervention introduced to both the verbal and non-verbal behaviors that
teachers at four separate workshops spread over two months; sec- research has shown indicated teachers’ expectations of students
ond, self-analysis of video data; third, implementation in class- (Brophy, 1985; Brophy & Good, 1970; Harris & Rosenthal,
rooms of the practices learned in the workshops; and fourth 1985). Following this, teachers were asked to view their videos
follow-up meetings with the researchers as project partners. Each in order to examine their own behaviors (particularly non-verbal)
of these components of the intervention will be described below. in light of the evidence with which they had been presented.
Teachers in the experimental group attended four workshops Most teachers wanted to share their videos with other teachers
designed to introduce the intervention. These were held in the sec- at the workshop and this resulted in closer analysis of the videos
ond and fourth weeks of March (the academic year begins in Feb- than had been originally intended. From their own analysis of the
ruary in New Zealand), two weeks later in April and a final video and discussion with others at the workshop, teachers self-
workshop in mid-May following a two-week vacation for schools. identified areas where they wished to improve. Hence, the video-
The interval between each workshop enabled the teachers to pro- ing of teachers and subsequent analysis in relation to non-verbal
gressively introduce each component of the intervention. Because behavior was left entirely with the teachers. The idea was that
of the size of the intervention group and the desire of the research teachers were taught about non-verbal behavior, could examine
team to be able to work closely with the teachers, two parallel ses- this in the videos, and would then be more adept at monitoring
sions were held each week. This provided some flexibility for the their own non-verbal behavior, particularly as they were also
teachers such that if they were unable to attend one session, they filmed later in the year and so had a basis for comparison, reflec-
were able to attend the parallel session. The sessions were always tion, and future development.
delivered by the first author to ensure consistency. The workshops The second, third, and fourth workshops followed a specific pat-
were based on the literature and findings of Weinstein (2002) and tern. In the morning, the previous session was reviewed; teachers
first author (Rubie-Davies, 2007, 2008b; Rubie-Davies and were then introduced to the specific focus for that day, to the evi-
Peterson, 2011), in particular the types of teaching behaviors that dence within the teacher expectation field, and to research evi-
high expectation teachers engaged in. These became the focus of dence outside the field that might support the new practices
the intervention and were centered round grouping and learning being introduced. At each session they discussed and closely exam-
activities; class climate; and motivation, evaluation, feedback, ined the practices and beliefs of high expectation teachers in rela-
and enhancing student autonomy. tion to the workshop topic. Hence, the focus was always on the
At the first workshop, teachers were introduced to the relevant advantages of high teacher expectations for student outcomes.
research in the expectation area. Hence, the entire first day was The teachers then spent the afternoon cooperatively planning
spent introducing the teacher expectation field and providing how they would introduce the new practices into their classroom.
examples of the importance of high expectations for student learn- These ideas were recorded, later typed up by members of the
ing. The intervention teachers were specifically introduced to the research team and formed into a booklet for all intervention teach-
work of Babad (2009), Weinstein (2002) and Rubie-Davies (2007) ers to use. Thus, the teachers introduced each new idea to their
as these authors have studied expectation effects as being moder- classrooms over time and were able to choose when and how each
ated by a teacher factor or belief. This was followed by an introduc- was introduced. For example, some teachers chose to introduce a
tion to some fundamental differences in the practices and beliefs new practice into their mathematics program rather than their
associated with high and low expectation teachers. To reiterate, reading program whereas others introduced the practices into
first both high and low expectation teachers taught core subjects mathematics and reading. The grouping and learning activities ses-
such as reading and mathematics with students in ability-based sion (the second workshop) was designed to introduce the idea of
instructional groups but whereas the low expectation teachers allowing students to work with all their peers on reading or math-
kept their students in ability groups for learning activities, the high ematics activities and providing students with choice in their
expectation teachers did not. Instead, they gave all students choice learning activities. So, in high expectation teachers’ classes, all stu-
in the activities they could complete and allowed them to work dents completed the same reading or mathematics activities or, all
with a variety of peers (Rubie-Davies, 2008b). Second, high expec- students were given a choice about which activities they would
tation teachers created a much warmer socioemotional environ- complete. Teachers were taught how to organize their reading
ment for their students than did low expectation teachers. High and mathematics instruction to accommodate choice in activities
expectation teachers were very positive towards students, focused and how to accommodate having students working with mixed
on mastery goals both at the class and individual level, encouraged ability peers.
cooperation and collaboration among students, promoted student The third session focused on enhancing the class climate. Teach-
autonomy and used positive, preventive behavior management ers were introduced to ways to measure their class climate through
strategies (Rubie-Davies, 2007; Rubie-Davies and Peterson, 2011). being taught how to construct and interpret a sociogram and how
Third, high expectation teachers monitored student progress to use the My Class Inventory (MCI; Fraser, Anderson, & Walberg,
closely, set clear individual learning goals with students, provided 1982). Further, teachers were introduced to the research evidence
feedback on progress, and taught students how to set their from the positivity literature for enhanced well-being (Fredrickson,
own goals. Low expectation teachers did not engage in any of 2009) and were shown how the ideas presented in Fredrickson’s
these practices (Rubie-Davies, 2007, 2008b). Ways in which high book could be adapted for use in their classrooms.
The final session focused on motivation, evaluation, feedback, continual online support from the facilitators of the program. Over-
and promoting student autonomy. These aspects were subsumed all, the control group teachers were involved in a rigorous program
under the overarching construct of goal-setting. Teachers were of professional development most likely equivalent to the input
introduced to the literature on goal-setting and feedback both of received by the intervention teachers.
which have large effects on student achievement (d = .56 for goal We also assessed if any of the practices were already being used
setting and d = .73 for feedback; Hattie, 2009). Using SMART goals in intervention teachers’ classes before the intervention began. No
(Specific, Measurable, Attainable, Relevant, Time-bound; Doran, teachers reported using flexible grouping before the intervention.
1981) and the e-asTTle Individual Learning Pathways report (see Two teachers reported that they believed their class climate was
below in the Measures section for further details in relation to already very positive but following the workshop on class climate,
e-asTTle), teachers were shown how to teach students to set goals, they also both reported that they had learned new ideas that they
how to provide specific learning feedback that would enhance would implement into their classes. A further two teachers
motivation and learning, and how to enable students to focus on reported using goal setting in their classes. One indicated that
specific goals that they chose for their next learning steps. The she set goals with the students and parents at the beginning of
e-asTTle Individual Learning Pathways report provides information the year (so these were distal goals whereas the Teacher Expecta-
for each child in graphical and text form with a list of the next tion Project focused on proximal goals). The other teacher did
goals that each individual student should be focusing on. Clicking report using similar practices to those outlined in the goal-setting
on each goal takes the user to ‘‘Where to next?’’ webpages that workshop. Hence overall, use of the practices in the three key areas
provide resources for assisting students to learn and achieve the prior to the intervention was minimal.
next stages in their learning.
Following the workshops, a further component of the interven- 2.3. Measures
tion was that the researchers visited schools a further three times
during the year as project partners to meet with the intervention The measures in this study were of student achievement in
teachers, discuss their successes, brainstorm additional ideas, and reading and mathematics. Further, at the end of the year of the
provide support for the teachers in implementing the intervention. study, the intervention teachers completed a self-report on the
These meetings also provided a check of the fidelity of implemen- extent to which they had engaged with the intervention and chan-
tation of the intervention strategies. Each research team member ged their practice.
was assigned approximately two schools and, for the most part,
the teachers from the two schools met together at one of the 2.3.1. Student achievement
schools. At a meeting prior to visiting schools, the research team Students completed standardized tests in reading and mathe-
decided on the focus for each school visit and planned prompts matics at the beginning, middle and end of the 2011 academic year
to promote discussion. However, although the plans were used (February, June and November) using e-asTTle (Assessment Tools
as a guide to ensure there was some consistency in what was cov- for Teaching and Learning). e-asTTle is a school-based item bank
ered at each meeting, the intervention teachers could ask questions with an integrated, computerized system for creating tests in read-
and seek advice. An aspect the teachers found particularly helpful ing, writing, and mathematics and it analyzes student achievement.
was that at each session they shared what they had been doing in All items in the e-asTTle system were pre-calibrated using item
their classrooms in relation to each of the areas of the intervention. response theory (IRT) scoring procedures (Embretson & Reise,
Again, from a research perspective, this assisted in ensuring consis- 2000) in national norming trials between 2000 and 2004. Through
tency in the implementation of the intervention. At the end of the simulation studies of randomly created tests using the norming
year, intervention teachers, principals, and control group teachers data, the standard error of measurement for any e-asTTle test was
met and several of the intervention teachers presented what they estimated to be 15 points. Each e-asTTle reading or mathematics
had implemented within their class programs and highlighted test, in norming and in operational use, must have at least one
their successes. open-ended question; the proportion of multiple-choice items in
Once the videos had been taken of the intervention teachers at the e-asTTle item bank in mathematics is 57%, while about 83% of
mid-year, they were sent copies of their dvd. They viewed their reading items are multiple-choice. The curriculum validity of each
own dvds and then shared them with other intervention teachers item was established through curriculum-aligned item specifica-
in their school if they chose to do so. Again, this provided an oppor- tions, teacher-review panels, teacher-marking panels, and tea-
tunity for self-analysis and for analyzing the verbal and non-verbal cher-standard setting panels. Schools voluntarily use the e-asTTle
messages that were being portrayed to students. The researchers program to create their own standardized tests for their own in-
did not have input into this analysis apart from the initial teaching school use. Schools analyze student performance using the opera-
about non-verbal behavior. tional e-asTTle software and tests to generate individual and group
In New Zealand, it is compulsory for teachers to engage in some level reports against the pre-calibrated national norms and curricu-
form of professional development each year. Hence, while the con- lum standards. Because of the pre-calibrated IRT score values for
trol group was not engaged in the intervention, all were involved in each item, irrespective of which test is taken by the students, their
some form of professional training. Because the teachers came total scores can be compared across classes, years, and schools. e-
from 12 different schools, not all control group teachers were asTTle is designed to be used with students from Years 4 to 12.
involved in the same professional development because schools In this study, the achievement measure was student
have autonomy over the training they choose to be involved in performance on an e-asTTle test created by the first author in
each year. However, as an example of the kind of professional mathematics and reading. For the purposes of the current study,
development in which control group teachers were involved, dur- 40-min tests were created at each level of e-asTTle (i.e., Level 2,
ing the year of the intervention several of the schools (n = 7) partic- Level 2/3, Level 3 and so on up to Level 6). The levels relate to
ipated in a professional development program designed to enhance the New Zealand curriculum in reading and mathematics. In
teachers’ assessment practices and improve student achievement. New Zealand an average student spends 2 years on each curricu-
The control group teachers attended three full-day workshops in lum level so Level 2 is where average students in Years 3 and 4
relation to this program over a period of just over six months. In would be working, students in Years 5 and 6 would be at Level
between workshop sessions, the control group teachers imple- 3 and so on. However, students move at their own pace and tests
mented the practices into their classrooms and were able to access are assigned at students’ current level so it is not uncommon for
a high ability Year 6 student to be working at Level 4 or even 2.4. Overview of statistical model
Level 5 and for a struggling Year 8 student to be working at Level
2. Possible scores in the e-asTTle test range from approximately We conducted Bayesian Multilevel Latent Growth Models test-
1100–1900 in reading and in mathematics. In the current sample, ing the effect of our intervention at three time-points during the
the scores ranged from 1171 to 1672 in reading (M = 1395.80, school year on mathematics and reading test achievement. We
SD = 107.57) and from 1194 to 1754 in mathematics estimated our Latent Growth Curve models within a multilevel
(M = 1423.88, SD = 97.95) at the beginning of the year. Following grouping structure in which students were nested within schools.
the creation of the tests, they were then reviewed by a selection All effects were modelled as within-school effects assuming a
of deputy principals in schools involved in the study to ensure fixed-effects model for school-level variability. While this model
their appropriateness and acceptability to the schools. No prob- thus recognized that students in different schools might differ,
lems were identified. Teachers then assigned students to the given the complexity of the design, our model did not formally
appropriate test (i.e., each student was assigned a test at their account for classroom variability within this nested structure.
level), the tests were photocopied and delivered to schools, We opted not to estimate random effects at the school-level due
returned to the researchers once completed, and marked on-line to practical limitations in reliably modelling such effects across
by the research team. So that tests were comparative across age the limited number of schools in our sample. Similarly, we opted
groups, in reading, the areas chosen for each test were processes to model the effects of our intervention on the rate of change in
and strategies, ideas, and language features; in mathematics the mathematics and reading performance in separate models,
areas selected were number knowledge, number sense and alge- rather than in one model with parallel processes given that our
bra. This process was followed at each of the three data collection model was already exceedingly complex. We estimated the latent
points and the areas to be tested remained consistent. The overall growth curves for all models used Bayesian estimation with diffuse
total score for mathematics and for reading generated by the e- priors.
asTTle computer system was used in this study. Bayesian analysis is less well known than the standard Frequen-
tist (or Maximum Likelihood) analytic methods. Our purpose here
is not to review Bayesian methods, but a few brief comments are
2.3.2. Teacher engagement with the intervention warranted (for a review see Kruschke, Aguinis, & Joo, 2012). Yuan
At the end of the year, teachers self-reported how much they and MacKinnon (2009) when discussing the difference between
had engaged with the three major components of the interven- Bayesian credible intervals relative to more well-known Frequen-
tion. They were first asked if they had included the component tist Confidence Intervals (CIs) commented that ‘‘Bayesian credible
in their teaching (Yes/No); if they responded, yes, they were then intervals have more natural probability interpretations than CIs.
asked whether each aspect of the intervention had enhanced stu- A 95% credible interval means that there is a 95% chance that the
dent learning outcomes, enhanced classroom relationships, the credible interval contains the true value of the parameter on the
degree to which they had included the intervention component basis of the observed data’’ (Yuan & MacKinnon, 2009, p. 304).
into their program, and whether or not they were satisfied with Because of this, credibility intervals in Bayesian analysis do not
how they had integrated each specific aspect of the intervention have to be symmetric. Rather, intervals are determined as the
into their program. Teachers were also asked if they would rec- specific percentile values around the distribution of each parame-
ommend the intervention to others; if they believed they were ter (known as the posterior distribution). For example, the 95%
now a better teacher, having taken part in the program; and if credible interval for the posterior distribution of a regression slope,
they believed their class climate had improved. All items other such as those we estimate here, would take the 2.5 and 97.5
than the ‘Yes/No’ options cited above were responded to on a percentile values of the posterior distribution, thus allowing for
1–5 Likert scale. The anchors varied in relation to the items. For skew. The p-values we report in our analysis thus reflect the
example, for items related to evaluation of aspects of the proportion posterior distribution for a given parameter (or latent
intervention the options ranged from ‘strongly disagree’ to growth curve) that is above or below zero. Our models were
‘strongly agree’; for satisfaction they ranged from ‘not satisfied’ conducted in Mplus 7.0.
to ‘satisfied’, and for items measuring the extent of implementa- We first conducted baseline models which estimated the rate of
tion the options ranged from ‘not at all’ to ‘fully’. Space was also change in mathematics and reading achievement over the three
provided where teachers could comment on aspects of the testing phases. These models estimated an intercept representing
intervention although the qualitative data will not be reported the mean level of test performance at Time 0 (during the first test-
in the current paper. ing phase before the intervention took place), and a slope repre-
The degree of teacher engagement with the intervention was senting the rate of linear change in performance across the three
assessed by the researchers. A mark out of 85 was calculated for assessment periods (assessment periods were specified as three
each teacher based on attendance and participation at the work- equidistant time scores fixed at 0 (the first testing phase at the
shops (out of 40). Teachers were also assessed on the degree to beginning of the year before the intervention took place), 0.5
which each teacher had incorporated into their classrooms the (mid-way through the year) and 1 (the testing phase conducted
strategies they had learned and planned in the workshops related at the end of the year). This model thus provides information on
to flexible grouping (out of 10), class climate (out of 10), and goal- the initial mean level of performance (the intercept), the linear rate
setting (out of 10). Teachers were given an overall score (out of 15) of change in performance (the slope), and the residual variance in
for the degree to which the strategies implemented into their both the intercept and slope. In our baseline models, these esti-
classroom programs reflected what was presented at the work- mates of residual variance reflect the extent to which initial perfor-
shops, that is, the degree of fidelity to the original intervention mance and the rate of change in performance differed across
strategies. These marks for individual strategies and overall imple- students.
mentation were calculated based on evidence collected by the Pro- We then extended these models to assess the effect of the inter-
ject Partners. A mark for each teacher was allocated by two vention on the rate of change in mathematics and reading perfor-
researchers who assessed the evidence of implementation mance. We did this by regressing the intercept and slope in each
obtained from Project Partner meetings and then allocated a mark. model on a variable representing experimental condition (0 con-
These marks were then scrutinized by a further two researchers trol, 1 intervention). If the experimental condition variable signifi-
before a final mark was agreed by all four researchers. cantly predicted the slope, then this would indicate that the linear
rate of change in performance over the year differed across the mathematics performance at the beginning of the year (b = 2.964,
experimental and control conditions. We then integrated these Post. SD = 4.286, p = .245, 95% Cred. Interval = 5.255, 11.530).
model parameters to derive the predicted average score at the More generally, the intercept for mathematics achievement in
end of the year for students randomly assigned to the control our latent growth model was significant (b = 1422.300, Post.
and intervention condition, and the difference between them. SD = 3.122, p < .001, 95% Cred. Interval = 1416.211, 1428.453). This
coefficient represents the predicted level of mathematics achieve-
3. Results ment at the beginning of the year for students in the control con-
dition. The slope for mathematics achievement was also significant
3.1. Descriptive statistics (b = 32.472, Post. SD = 2.720, p < .001, 95% Cred. Interval = 27.129,
37.797). This indicates that, across the year, students in the control
The students’ mean achievement scores for reading and mathe- condition (coded as 0 in our model) typically increased by around
matics at each of the three time points are given in Table 2 below 32 points in their mathematics test scores. In contrast, students in
for the intervention and control groups. These results show that for the experimental condition typically increased by around 41 points
both reading and mathematics student achievement increased in their mathematics test scores (the addition of the b for main
over the course of the year. effect = 32.472 and the effect of receiving the intervention
b = 8.781).
3.2. Analysis of baseline mathematics and reading models In terms of predicted scores, our model indicates that at the
final testing phase conducted at the end of the year, the average
The baseline model for mathematics also indicated that perfor- mathematics score of students in the control condition was
mance increased significantly over the year by around 37 points 1454.769 (Post. SD = 2.986, p < .001, 95% Cred. Interval = 1448.883,
(b = 37.129, Post. SD = 1.883, p < .001, 95% Cred. Interval = 33.449, 1460.592). Conversely, the average mathematics score of students
40.819). The residual variance for this slope also varied signifi- in the intervention condition should be around 11–12 points
cantly, indicating that the rate of change in mathematics perfor- higher, at 1466.514 (Post. SD = 2.821, p < .001, 95% Cred. Inter-
mance differed across students (r2 = 2197.723, Post. val = 1460, 1460.949, 1471.982). Moreover, and in line with the
SD = 268.462, p < .001, 95% Cred. Interval = 1667.501, 2719.526). significant slope for the growth curve, this difference in predicted
Thus, while student performance increased over the year, the sig- points was significant (b = 11.748, Post. SD = 4.111, p < .001, 95%
nificant residual variance indicated that there was diversity across Cred. Interval = 3.740, 19.867).
students in the rate of improvement, with some students increas- In plain language, these results indicated that students in the
ing at a quicker rate than others. control and intervention condition had similar levels of mathemat-
Similarly, the baseline model for reading indicated that perfor- ics achievement at the beginning of the year. As would be
mance increased significantly over time by an average of around 35 expected, students generally increased their mathematics achieve-
points (b = 34.738, Post. SD = 1.714, p < .001, 95% Cred. Inter- ment over the year. However, students in the intervention condi-
val = 31.405, 38.113). This residual variance for this slope also var- tion performed better throughout the year, and made additional
ied significantly, indicating that the rate of change in reading gains of around 9 points in their performance beyond that of the
achievement differed across students (r2 = 2409.305, Post. base improvement rate shown by students in the control condition.
SD = 580.279, p < .001, 95% Cred. Interval = 1240.076, 3502.710). This represents an improvement above the control group of
approximately 28%. In simpler statistical terms, the improvement
3.3. Effect of intervention on mathematics achievement in mathematics scores for the intervention group amounted to an
effect size of d = 0.49. The improvement over and above that of
A schematic overview of the Latent Growth Model testing the the control group was an effect size of d = 0.24 (t = 4.86,
effect of the intervention on mathematics achievement across the df = 1637, r = 0.12). The intervention significantly improved stu-
year is presented in Fig. 1. We used this model to test our primary dents’ mathematics achievement.
hypothesis that students in classrooms randomly allocated to the
teacher expectation intervention condition would show greater 3.4. Effect of intervention on reading achievement
gains in mathematics performance over the year than students in
classrooms randomly allocated to the control condition. This We tested an identical model, as outlined in Fig. 1, to assess the
hypothesis was supported, as our critical test indicated that exper- effect of the intervention on reading achievement. Experimental
imental condition (0 control, 1 intervention) significantly predicted condition (0 control, 1 intervention) did not significantly predict
the latent slope representing the rate of change in mathematics the rate of change in reading performance over the year
achievement over the year (b = 8.781, Post. SD = 3.765, p < .001, (b = 3.969, Post. SD = 3.398, p = .121, 95% Cred. Inter-
95% Cred. Interval = 1.291, 16.035). Importantly, and as one would val = 10.535, 2.774) or the intercept representing initial rates of
expect given random assignment to conditions, experimental con- reading performance at the beginning of the year (b = 4.037, Post.
dition did not significantly predict students’ initial rates of SD = 4.698, p = .195, 95% Cred. Interval = 5.222, 13.213). Rather,
Table 2
Descriptives for e-asTTle reading and mathematics scores at the beginning (T1), middle (T2) and end (T3) of the academic year for the total sample and the intervention and
control groups.
Total sample Intervention group Control groups

N Mean SD Min Max N Mean SD Min Max N Mean SD Min Max
Mathematics Score, T1 1907 1423.88 97.95 1194 1754 1017 1426 97.51 1194 1754 1018 1421 98.44 1194 1695
Reading Score, T1 1850 1395.80 107.57 1171 1672 985 1399 109.80 1172 1643 865 1392 104.91 1171 1643
Reading Score, T2 1809 1415.51 109.36 1165 1780 981 1417 107.69 1165 1780 828 1413 111.33 1166 1780
Reading Score, T3 1560 1425.29 106.60 1174 1709 852 1427 106.86 1174 1709 707 1423 106.29 1176 1709
Fig. 1. Schematic overview of Latent Growth Model testing the effect of the intervention on test performance.
and as described in the baseline model, reading performance relationship between student mathematics achievement at the
increased significantly by around 37 points over the year beginning of the year and teacher engagement with the interven-
(b = 36.822, Post. SD = 2.473, p < .001, 95% Cred. Interval = 31.924, tion (r = 0.26) but by the end of the year, this correlation with
41.610), but this rate of increase did not differ across the control student achievement was positive (r = 0.38). In contrast, in reading
and intervention conditions. These results indicate that the inter- the correlation between teacher engagement with the intervention
vention did not affect reading achievement. This was confirmed and student achievement in reading at the beginning of the year
when simpler effect size estimates were calculated. The overall was r = 0.26 and remained negative at the end of the year
change in achievement in reading for the intervention group was (r = 0.12).
d = 0.26 across the year while the advancement of the intervention
group over the control group was d = 0.04, a very small effect.
4. Discussion
3.5. Teacher engagement with the intervention
We found evidence that the students of teachers who were
As outlined in Section 2, the teachers self-reported the degree to trained in the practices of high expectation teachers had increases
which they had incorporated each aspect of the intervention into in their mathematics scores over one year. The gain in comparison
their classrooms by the end of the year. They further reported on with the control group was the equivalent of students increasing
whether they would recommend the intervention to colleagues their scores by almost an additional three months over the course
and its overall effectiveness. However, not all intervention teachers of an academic year (28%). It is to be remembered that this is in
completed the end of year evaluation (n = 36). The results of the contrast to the classes of control group teachers many of whom
teachers’ self-report data showed that overall the teachers were undergoing equally intensive professional development. The
appeared to be very positive about the intervention and its effec- mean learning gain across New Zealand in e-asTTle scores for kin-
tiveness in leading to improved student achievement and dergarten to grade 7 students is 30 points each year. The mean for
improved teaching (see Table 3). All means were above the mid- the control group was 32 points, a gain slightly above what is nor-
point of 3.0 on the 1–5 Likert scale and some were above 4.0. Those mally anticipated, and perhaps related to the professional develop-
that were above 4.0 related to teacher reports that classroom rela- ment that they underwent during the same year. However, the
tionships had improved as a result of the intervention, that goal students with intervention teachers gained 41 points, an increase
setting had resulted in improved student achievement, and that beyond what was achieved by the control group. Nevertheless, in
they would recommend the Teacher Expectation Project to other terms of traditional measures of effect sizes, the gain of the inter-
colleagues. The lowest reported means were for items related to vention group over the control group was small (d = 0.24) and in
the extent of implementation of goal setting and changes to the terms of the Binomial Effect Size Display (Rosenthal & Rubin,
class climate. However, for items related to whether or not the 1982), the benefit to the intervention group would be equivalent
teachers had been working on each aspect of the intervention, to 56% of those students showing above average gains compared
positive responses for each key aspect were above 90% (see to only 44% of the control group students showing such gains.
Table 3). However, there were no equivalent gains in reading.
Scores for the degree to which the researchers judged that One explanation for greater gains in mathematics is that in end
teachers had effectively implemented the intervention ranged of year surveys of the degree to which teachers had implemented
from 47.5 to 81.5 of a possible score of 85. The mean score was the three aspects of the intervention, whereas 95% reported imple-
71.26 (SD = 6.82). This suggests that overall there was a high level menting flexible grouping to some extent, 63% reported fully inte-
of implementation and adherence to the integrity of the interven- grating flexible grouping into their mathematics program whereas
tion. At a correlation level, there was a small but negative only 22% reported full integration into their reading programs.
Table 3
Descriptives and proportions for teacher self-report of engagement with the intervention.
Survey item M SD Proportion

agreement
Flexible grouping
1. I have been using flexible grouping in my classroom 95%
2. I believe that what I did in relation to flexible grouping made a positive difference to my students’ outcomes 3.94 .58
3. I believe that what I did in relation to creating flexible grouping made a positive difference to my classroom relationships 4.19 .67
4. To what extent have you integrated flexible groups into your class program? 3.68 .87
5. I am satisfied with how I have integrated flexible grouping into my class program 3.65 1.13
Classroom climate
6. I have been working on the classroom climate in my classroom 97%
7. I believe that what I did in relation to changing the classroom climate made a positive difference to my students’ academic outcomes 3.97 .70
8. I believe that what I did in relation to changing the classroom climate made a positive difference to my classroom relationships 4.19 .58
9. To what extent have you made changes in your classroom climate? 3.56 .94
10. I am satisfied with the changes I have made with respect to my classroom climate 3.97 .84
Motivation, evaluation, and student autonomy (Goal Setting)
11. I have been using goal setting to enhance student motivation and autonomy 92%
12. I believe that my use of goal setting and feedback to increase student motivation and autonomy made a positive difference to my 4.09 .58
students’ academic outcomes
13. I believe that what I did in relation to goal setting and feedback made a positive difference to my classroom relationships 3.97 .64
14. To what extent have you made changes around increasing goal setting and feedback in your classroom? 3.49 .94
15. I am satisfied with the changes I have made with respect to goal setting and feedback in my classroom 3.59 1.10
Overall evaluation
16. I would recommend to my colleagues that they take part in the Teacher Expectation Project 4.03 .88
17. I believe I am a better teacher now than I was before I started the project 3.86 .76
18. I believe I have a better class climate now than I did before I started the project 3.79 .71
Similarly, overall 92% had implemented goal-setting by the end of opportunity to do so (Author, a). Fourth, the incorporation of flex-
the year, but, again, teachers reported that they found it easier to ible grouping is likely to lead to improved self-perceptions for stu-
do this in mathematics than in reading (75% for mathematics ver- dents (Author, b). There is evidence that the self-perceptions of
sus 25% for reading). Some teachers had not implemented goal set- both the very high and the very low achievers are damaged by abil-
ting until near the end of the year and intended to make it more of ity grouping (Liem, Marsh, Martin, McInerney, & Yeung, 2013;
a focus the following year (26%). Overall, 97% had been working on Oakes, 1988). Finally, an interesting observation that relates clo-
the class climate, however, the changes in teacher practices related sely to ability grouping is that While New Zealand has the highest
to class climate were more holistic and not so specifically related to within class ability grouping rate of all OECD countries
a curriculum area. The class climate changes were likely to affect (Wagemaker, 1993), it also has the largest disparity between the
all aspects of classroom life. Hence, it is possible that wider inclu- highest and lowest achievers (Tunmer, Chapman, & Prochnow,
sion of flexible grouping and goal setting in mathematics than in 2004). On the other hand, Finland has a policy of only utilizing het-
reading may have led to increased gains in mathematics for stu- erogeneous grouping throughout schooling and has one of the
dents whose teachers were in the intervention group because smallest gaps between the highest and lowest achievers (OECD,
those components of the intervention were more specific to a cur- 2007).
riculum area. It was interesting to note that there was a positive though small
Flexible grouping is likely to lead to enhanced achievement for correlation between teacher engagement with the intervention (as
students for a variety of possible reasons. First, flexible grouping, assessed by the researchers) and student achievement in mathe-
where students choose the activities they complete and with matics but that this relationship was very small and negative in
whom, means that all students work together. This creates a class- reading. In the original Pygmalion study (Rosenthal & Jacobson,
room community in which students form positive relationships 1968), the Reasoning IQ benefits of favorable teacher expectations
across the classroom rather than just with those in their group were significantly greater than the benefits found for Verbal IQ,
(Weinstein, 2002). Studies have shown that students learn better although similarly to the current study, the latter were in the pre-
in classroom environments that are warm and emotionally sup- dicted direction but not significantly so. It remains for future
portive (Babad, 2009; Pianta, Belsky, Vandergrift, Houts, & research to disentangle why the effects of a teacher expectation
Morrison, 2008; Weinstein, 2002; Wentzel, 2009). Second, when intervention appear to have greater effects on student mathemat-
students are working with their peers of varying abilities they ics gains than they do on reading gains. It may be that the parental
are likely to be exposed to positive peer modelling for both task relationship has more influence in reading than in mathematics
completion and behavior which in turn leads to enhanced learning because many parents read to and with their children during ele-
(Fuligni, Eccles, & Barber, 1995; Hornby, Witte, & Mitchell, 2011). mentary schooling but do not so frequently engage in mathematics
Third, flexible grouping diminishes perceptions of top group and support. This could result in the teacher contributing more to stu-
bottom group, where those in the top group are perceived of as dent achievement gains in mathematics than in reading where per-
somehow better than the others (Gamoran, 1992; Ireson, Hallam, haps parents play more of a role. This may then be reflected in
& Hurley, 2005); all students are exposed to challenging learning intervention attempts across both curriculum areas where the
opportunities and differentiation between high and low achievers influence may be greater in mathematics. However, currently this
is far less salient. Fourth, and a related possibility, is that with flex- is purely speculative.
ible grouping all students have similar opportunity to learn. When We acknowledge that we are unable to unravel which aspects of
students are ability grouped this can lead to those in high ability the intervention most contributed to the gains in mathematics
groups learning more, simply because they are given more achievement of the students with intervention group teachers. This
could be regarded as a limitation of the current paper. However, fidelity by the researchers suggested that overall teachers did
the intervention was considered a package and was not designed conform to the integrity of the implementation.
to be disentangled. The core aim of the intervention was to teach In our study the percentages of self-reported implementation
regular teachers the pedagogical beliefs and instructional practices for the three areas were high by the end of the year (all above
of high expectation teachers. It was hoped that when they adopted 90%), however many teachers chose to implement the changes
the practices and understood the underlying pedagogical beliefs gradually. For example, no teachers were using flexible grouping
of high expectation teachers that they would notice improved at the beginning of the year yet 95% were doing so by the end of
achievement for their students and therefore raise their the year. This was a substantial innovation for teachers and so
expectations of what students could achieve. Changes in teachers’ many concentrated on this instructional change throughout the
expectations will be measured in future years of the project. Hence, second term of school (May to mid-July) following the training.
at this stage in the project, we cannot be certain that teacher They then introduced changes to class climate in the third term
expectations among intervention teachers have increased. It may (July to end September) and for about one-third of teachers
be simply that the changed practices as a result of the intervention (32%), goal setting was not introduced until the fourth term
contributed to student achievement gains in mathematics but (mid-October to mid-December) (McDonald (submitted for
that there was no effect on teacher expectations. However, all publication)). The effects of the intervention for mathematics indi-
aspects of the intervention were core practices of high expectation cate the importance of the intervention for increasing student
teachers. It is of note that the improvement in mathematics was at achievement. While the exact component of the intervention that
a level that Hattie (2009) describes as one which would be is vital for its effectiveness is not known, flexible grouping is likely
visible to teachers (anything d = 0.40 and above) and, indeed, to be a key element. However, we did not measure whether there
many of the teachers did talk about noticeable improvements in was a relationship between the speed at which teachers intro-
their students’ achievement. They would be aware of the changes duced all aspects of the intervention and student achievement out-
within their own classes rather than any between class differences comes. This is a limitation of the current paper.
and so it is hoped that this will lead to increased teacher It is possible that by the end of the second year of the interven-
expectations. tion, when almost all teachers would be fully implementing the
The three major areas in which the earlier studies (Rubie- intervention across the entire year that larger effects may be found
Davies, 2006, 2007, 2008b) had noted differences between high in mathematics. Possibly, too, as teachers recognize the gains made
and low expectation teachers formed the basis of the intervention by their students in mathematics and become more confident in
and created an interconnected framework. The disestablishment of using flexible grouping and goal setting in that curriculum area
ability groupings has consequences for students as has already that even larger gains will be seen in mathematics. The teachers
been discussed. Flexible grouping has many benefits for student then may also transfer the learning to reading.
learning outlined above and also has favorable repercussions for While not reported in the current paper, teacher expectations
creating a positive class climate. Students are encouraged to work were also measured at the beginning of the year of the project
together. Moreover, because seating and work groups change before the intervention began. It will be interesting to measure
regularly, this serves to create a classroom community and con- the expectations of the same teachers in the second year of the
tributes to the class climate as does the positive atmosphere and study to see whether the changes in instructional practices and
warm teacher–student relationships created by high expectation noticeable gains in mathematics will lead to teachers reporting
teachers. Finally, the setting of specific, achievable goals for stu- higher expectations for their students in the following year. This
dents, monitoring progress closely, and providing students with is a limitation of the current paper; change in expectations over
clear feedback about their learning provide students with clear the first year of the project has not been included. This is mainly
directions for their learning and the focus is on mastering skills because, as outlined above, teachers implemented the intervention
and learning. Together these practices probably contributed to strategies gradually over one year and so it would not be until the
the enhanced achievement of the intervention students in very end of the first year that they would be likely to have been
mathematics. Of course, we cannot be certain. There could be other aware of the gains students were making in mathematics. The sec-
explanations for the improved achievement of students in ond year of the project will enable the measurement of class level
mathematics when compared with the gains of the control group expectations, controlling for student prior achievement in order to
teachers’ classes. see if the expectations of the intervention group of teachers have
The intervention was very complex involving changes in three changed.
major areas of teachers’ classroom practice. Because of the multi- At the beginning of the year of the project, we assessed the
faceted nature of the intervention, teachers were given the oppor- degree to which any of the intervention teachers were already
tunity to implement the changes as they felt comfortable and implementing the practices that were part of the professional
confident to do so. This is also is in line with the self-governing nat- development workshops. As reported earlier, before the interven-
ure of New Zealand schools where teachers act relatively autono- tion, there was very minimal use of the practices (one teacher
mously (Fergusson, McNaughton, Hayne, & Cunningham, 2011). reported using goal setting similarly to the practices in the inter-
The less scripted nature of the intervention also capitalizes on vention). However, we were unable to assess the control group’s
the capabilities of the teachers to make their own instructional use of any of the strategies or practices without letting them know
decisions. While program integrity is critical, adaption of programs what comprised the intervention. Therefore we cannot be certain
will always happen during delivery, as those implementing the that some control group teachers were not using the practices. This
program seek to take into account the context in which they are is a limitation of the study. However, given that among interven-
operating (Fergusson et al., 2011). This was one reason we enabled tion group teachers use of the practices and strategies was very
teacher planning during the workshops as that provided an oppor- minimal, there is no reason to suppose that control group teachers
tunity for the researchers to gauge teacher understanding of the would be any different. Further, if control group teachers were
specific component of the intervention and also to assess the using some of the practices and having positive effects on student
degree to which planning reflected fidelity to the original design. achievement, then it would make it more difficult for us to find dif-
At that stage, the researchers were more readily able to guide ferences between the groups.
teachers towards integrity of the intervention than they were once A further limitation of the current paper is that the intervention
teachers were back in their own classrooms. The measure of reported here is only over the course of one academic year and has
been researcher led. Creating programs that are sustainable References

within school contexts is not easy. As with most school-based
interventions there is a need to build and sustain on site capabili- Babad, E. (1990a). Calling on students: How a teacher’s behavior can acquire
disparate meanings in students’ minds. Journal of Classroom Interaction, 25, 1–4.
ties (e.g., staff training) and ensure sufficient resources are present Babad, E. (1990b). Measuring and changing teachers’ differential behavior as
for program implementation and ongoing effectiveness evaluation. perceived by students and teachers. Journal of Educational Psychology, 82,
As Fergusson et al. 2011 reported, ‘‘Success of any policy, program 683–690.
Babad, E. (1995). The ‘‘teacher’s pet’’ phenomenon, teachers’ differential behavior,
or intervention will depend critically on how well the program is and students’ morale. Journal of Educational Psychology, 87, 361–374.
implemented as well as the intrinsic efficacy of the program Babad, E. (1998). Preferential affect: The crux of the teacher expectancy issue. In J.
per se’’ (p. 290). Brophy (Ed.). Advances in research on teaching: Expectations in the classroom
(Vol. 7, pp. 183–214). Greenwich, CT: JAI Press.
In future work with this cohort we will be able to track to some
Babad, E. (2009). The social psychology of the classroom. New York: Routledge.
extent the sustainability of this intervention. Student achievement Babad, E., Avni-Babad, D., & Rosenthal, R. (2003). Teachers’ brief nonverbal
will be tracked when students move from intervention group behaviors can predict certain aspects of students’ evaluations. Journal of
teachers to other intervention teachers, to control teachers or to Educational Psychology, 95, 553–562. http://dx.doi.org/10.1037/0022-
0663.95.3.553.
those not in the project, to determine the extent to which the gains Babad, E., Bernieri, F., & Rosenthal, R. (1989a). Nonverbal communication and
in mathematics of students in the intervention group are leakage in the behavior of biased and unbiased teachers. Journal of Personality
sustained. Similarly, it will be interesting to track students with and Social Psychology, 56, 89–94.
Babad, E., Bernieri, F., & Rosenthal, R. (1989b). When less information is more
control teachers who move to intervention teachers, other control informative: Diagnosing teacher expectations from brief samples of behaviour.
teachers or to teachers not in the project. The second year of the British Journal of Educational Psychology, 59, 281–295.
project involves intervention group teachers training the control Babad, E., Bernieri, F., & Rosenthal, R. (1991). Students as judges of teachers’ verbal
and nonverbal behavior. American Educational Research Journal, 28, 211–234.
group in the three areas of instructional change that form the Babad, E., & Taylor, P. B. (1992). Transparency of teacher expectancies across
intervention and hence we will be able to see the extent to which language, cultural boundaries. Journal of Educational Research, 86, 120–125.
the intervention strategies can be passed on without high level Bloom, H. S. (Ed.). (2005). Learning more from social experiments: Evolving analytic
approaches. New York, NY: Russell Sage Foundation.
researcher involvement. Brophy, J. E. (1983). Research on the self-fulfilling prophecy and teacher
Finally, there is a need for the replication of this study in expectations. Journal of Educational Psychology, 75, 631–661.
another context, and independently of the intervention developers. Brophy, J. E. (2004). Motivating students to learn. Mahwah, NJ: Lawrence Erlbaum.
Brophy, J. E. (1985). Teacher-student interaction. In J. B. Dusek (Ed.), Teacher
This might also help identify the vital aspects of the study that
expectancies (pp. 303–328). Hillsdale, NJ: Lawrence Erlbaum.
make it effective. These aspects, along with availability of Brophy, J. E., & Good, T. L. (1970). Teachers’ communication of differential
resources and funding, cultural acceptability, and feasibility of expectations for children’s classroom performance: Some behavioral data.
implementing the program would all be necessary precursors Journal of Educational Psychology, 61, 365–374.
Clarke, S., Timperley, H. S., & Hattie, J. (2003). Unlocking formative assessment:
before a large scale roll out of this intervention (Fergusson et al., Practical strategies for enhancing students’ learning in the primary and
2011). intermediate classroom. Auckland, New Zealand: Hodder Moa Beckett.
This study has measured whether a teacher expectation inter- Doran, G. T. (1981). There’s a S.M.A.R.T. way to write mangement’s goals and
objectives. Management Review, 70, 35–36.
vention that targeted teacher behaviors could increase student Elashoff, J. D., & Snow, R. E. (1971). Pygmalion reconsidered. Worthington, Ohio:
achievement. The gains of students in mathematics when com- Jones.
pared with the control group are exciting. It is often difficult to Embretson, S. E., & Reise, S. (2000). Item response theory for psychologists. Mahwah,
NJ: Erlbaum.
achieve gains with a randomized control trial in schools due to Fergusson, D., McNaughton, S., Hayne, H., & Cunningham, C. (2011). From evidence
school constraints such as management or policy guidelines. to policy, programmes and interventions. In P. D. Gluckman (Ed.), Improving the
Working with many teachers across multiple classrooms makes transition: Reducing social and psychological morbidity during adolescence
(pp. 287–300). Auckland, New Zealand: Office of the Prime Minister’s Science
measuring the fidelity of the intervention difficult. Further, recruit- Advisory Committee.
ing schools to be actively involved in a research project for three Fraser, B. J., Anderson, G. J., & Walberg, H. J. (1982). Assessment of learning
years is not easy. The positive results achieved in this initial year environments: Manual for learning environment inventory (LEI) and my class
inventory (MCI) (3rd vers.). Perth, Australia: Western Australian Institute of
augur well for the two further years of the project.
Technology.
Fredrickson, B. L. (2009). Positivity: Top-notch research reveals the 3-to-1 ratio that
will change your life. New York: Three Rivers Press.
Acknowledgments Fuligni, A. J., Eccles, J. S., & Barber, B. L. (1995). The long-term effects of seventh-
grade ability grouping in mathematics. Journal of Early Adolescence, 15, 58–89.
This study was supported by the Marsden Fund Council from Gamoran, A. (1992). Is ability grouping equitable? Educational Leadership, 50, 11–17.
Good, T. L., & Brophy, J. E. (1974). Changing teacher and student behavior: An
Government funding administered by the Royal Society of New empirical investigation. Journal of Educational Psychology, 66, 390–405.
Zealand. The workshops that were a major part of the intervention Good, T. L., & Nichols, S. L. (2001). Expectancy effects in the classroom: A special
were funded by the Cognition Trust. These providers of financial focus on improving the reading performance of minority students in first-grade
classrooms. Educational Psychologist, 36, 113–126.
support played no role in the study design, in the collection, anal-
Gottfredson, D. C., Marciniak, E. M., Birdseye, A. T., & Gottfredson, G. D. (1995).
ysis and interpretation of data; in the writing of the report; and in Increasing teacher expectations for student achievement. Journal of Educational
the decision to submit the article for publication. Research, 88, 155–163.
Hamre, B. K., & Pianta, R. C. (2001). Early teacher-child relationships and the
The authors would also like to acknowledge the senior
trajectory of children’s school outcomes through eighth grade. Child
researchers and research assistants who have provided valuable Development, 72, 625–638. http://dx.doi.org/10.1111/1467-8624.00301.
assistance on this project and continue to do so. They are core to Harris, M. J., & Rosenthal, R. (1985). Mediation of interpersonal expectancy effects:
the success of the project and we are very appreciative of their 31 meta-analyses. Psychological Bulletin, 97, 363–386.
Hattie, J. (2009). Visible learning: A synthesis of over 800 meta-analyses relating to
involvement. achievement. London, England: Routledge.
We are very grateful to the principals, teachers and students Hornby, Garry., Witte, Chrystal., & Mitchell, David. (2011). Policies and practices of
who are participating in this project. We thank them for devoting ability grouping in new zealand intermediate schools. Support for Learning, 26,
92–96.
valuable time to this project. Ireson, J., Hallam, S., & Hurley, C. (2005). What are the effects of ability grouping in
The first author would particularly like to acknowledge the GCSE attainment? British Educational Research Journal, 31, 443–458.
exceptional support and guidance she received from Rhona Wein- Kerman, S. (1979). ‘‘Why did you call on me? I didn’t have my hand up!’’ Teacher
expectations and student achievement. Phi Delta Kappan, 60, 716–718.
stein during the planning and initial stages of writing this article. Kruschke, J. K., Aguinis, H., & Joo, H. (2012). The time has come: Bayesian methods
Rhona’s feedback and careful critique were, as always, much for data analysis in the organisational sciences. Organizatonal Research Methods,
appreciated. 15, 722–752. http://dx.doi.org/10.1177/1094428112457829.
Liem, G. A. D., Marsh, H. W., Martin, A. J., McInerney, D. M., & Yeung, A. S. (2013). The Rubie-Davies, C. M., Hattie, J. A. C., Townsend, M. A. R., & Hamilton, R. J. (2007).
big-fish-little-pond effect and a national policy of within-school ability Aiming high: Teachers and their students. In N. Galwye (Ed.). Progress in
streaming: Alternative frames of reference. American Educational Research Educational Psychology Research (pp. 65–91). Hauppauge, NY: Nova Publishers.
Journal, 50, 326–370. http://dx.doi.org/10.3102/0002831212464511. Rubie-Davies, C. M., & Peterson, E. R. (2011). Teacher expectations and beliefs:
Linchevski, L., & Kutscher, B. (1998). Tell me with whom you’re learning, and I’ll tell Influences on the socioemotional climate. In C. M. Rubie-Davies (Ed.).
you how much you’ve learned: Mixed ability versus same-ability grouping in Educational Psychology: Concepts, research and challenges (pp. 134–149).
mathematics. Journal for Research in Mathematics Education, 29, 533–554. London: Routledge.
Madon, S., Jussim, L., & Eccles, J. S. (1997). In search of the powerful self-fulfilling Snow, R. E. (1969). Unfinished pygmalion [Review of Pygmalion in the classroom].
prophecy. Journal of Personality and Social Psychology, 72, 791–809. Contemporary Psychology, 14, 197–199.
McDonald, L., Flint, A., Rubie-Davies, C., Peterson, E., Watson, P., & Garrett, L. Tenenbaum, H. R., & Ruck, M. D. (2007). Are teachers’ expectations different for
Teaching high expectation strategies for teachers through an intervention racial minority than for European American students? A meta-analysis. Journal
process. (submitted for publication). of Educational Psychology, 99, 253–273.
Oakes, J. (1990). Opportunities, achievement, and choice: Women and minority Timperley, H. S., Parr, J. M., & Bertanees, C. (2009). Promoting professional inquiry
students in science and mathematics. Review of Research in Education, 16, for improved outcomes for students in New Zealand. Professional Development
153–222. in Education, 35, 227–245. http://dx.doi.org/10.1080/13674580802550094.
Oakes, J. (1988). Tracking in mathematics and science education: A structural Tunmer, W. E., Chapman, J. W., & Prochnow, J. E. (2004). Why the reading
contribution to unequal schooling. In L. Weiss (Ed.), Class, race and gender in achievement gap in New Zealand won’t go away: Evidence from the PIRLS 2001
american education (pp. 106–125). Albany, New York: State University of New international study of reading achievement. New Zealand Journal of Educational
York Press. Studies, 39, 127–145.
OECD (2006). PISA 2006 science competencies for tomorrow’s world, Volume 1- Wagemaker, H. (Ed.). (1993). Achievement in reading literacy: New Zealand’s
analysis. Paris, France: OECD. performance in a national and international context. Wellington, New Zealand:
Patall, E. A., Cooper, H., & Wynn, S. R. (2010). The effectiveness and relative Ministry of Education.
importance of choice in the classroom. Journal of Educational Psychology, 102, Weinstein, R. S. (1989). Perceptions of classroom processes and student motivation:
896–915. Children’s views of self-fulfilling prophecies. In R. Ames & C. Ames (Eds.).
Pianta, R. C., Belsky, J., Vandergrift, N., Houts, R., & Morrison, F. J. (2008). Classroom Research on motivation in education (Vol. 3, pp. 187–221). New York: Academic
effects on children’s achievement trajectories in elementary school. American Press.
Educational Research Journal, 45, 365–397. Weinstein, R. S. (2002). Reaching higher: The power of expectations in schooling.
Proctor, C. P. (1984). Teacher expectations: A model for school improvement. The Cambridge, MA: Harvard University Press.
Elementary School Journal, 84, 469–481. Weinstein, R. S., Marshall, H. H., Brattesani, K. A., & Middlestadt, S. E. (1982). Student
Raudenbush, S. W. (1984). Magnitude of teacher expectancy effects on pupil IQ as a perceptions of differential teacher treatment in open and traditional
function of the credibility of expectancy induction: A synthesis of findings from classrooms. Journal of Educational Psychology, 74, 678–692.
18 experiments. Journal of Educational Psychology, 76, 85–97. Weinstein, R. S., & Middlestadt, S. E. (1979). Student perceptions of teacher
Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom: Teacher expectation interactions with male high and low achievers. Journal of Educational Psychology,
and pupils’ intellectual development. New York: Holt, Rinehart & Winston. 71, 421–431.
Rosenthal, R., & Rubin, D. B. (1982). A simple, general purpose display of magnitude Weinstein, R. S., Soule, C. R., Collins, F., & Cone, J. (1991). Expectations and high
of experimental effect. Journal of Educational Psychology, 74, 166–169. school change: Teacher-researcher collaboration to prevent school failure.
Rubie-Davies, C. M. (2006). Teacher expectations and student self-perceptions: American Journal of Community Psychology, 19, 333–363.
Exploring relationships. Psychology in the Schools, 43, 537–552. Weinstein, R. S. (1993). Children’s knowledge of differential treatment in school:
Rubie-Davies, C. M. (2007). Classroom interactions: Exploring the practices of high Implications for motivation. In T. M. Tomlinson (Ed.), Motivating students to
and low expectation teachers. British Journal of Educational Psychology, 77, learn: Overcoming barriers to high achievement (pp. 197–224). Berkeley, CA:
289–306. McCutchan.
Rubie-Davies, C. M. (2008a). Teacher beliefs and expectations: Relationships with Wentzel, K. R. (2009). Students’ relationships with teachers as motivational
student learning. In C. M. Rubie-Davies & C. Rawlinson (Eds.). Challenging contexts. In K. R. Wentzel & A. Wigfield (Eds.), Handbook of motivation at
Thinking about Teaching and Learning (pp. 25–39). Hauppauge, NY: Nova school (pp. 301–322). New York: Routledge.
Publishers. Yuan, Y., & MacKinnon, D. P. (2009). Bayesian mediation analysis. Psychological
Rubie-Davies, C. M. (2008b). Teacher expectations. In T. Good (Ed.). 21st Century Methods, 14, 301–322.
Education: A reference handbook (pp. 254–262). Thousand Oaks, CA: Sage
Publications.

A Teacher Expectation Intervention

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Teacher Expectation Intervention

Uploaded by

Copyright:

Available Formats

Accelerat ing t he world's research.

A teacher expectation intervention:

Contemporary Educational Psychology

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Teacher expect at ions, et hnicit y and t he achievement gap

Successive t eacher expect at ion effect s across t he early school years

Contents lists available at ScienceDirect

Contemporary Educational Psychology

A teacher expectation intervention: Modelling the practices

1. Introduction through descriptive studies. For example, researchers have studied

Total sample Intervention group Control groups

Survey item M SD Proportion

been researcher led. Creating programs that are sustainable References

You might also like