1
Classroom Interactions and Teacher Expectations
Classroom Interactions: Exploring the Practices of High and Low Expectation Teachers
Expectancy theory has been refined progressivelysince the initial study of Rosenthal and Jacobson (1963) which appeared to show that when teachers held expectations of particular students they interacted with their students in differing ways such that their initial, sometimes erroneous, expectations were fulfilled (the self-fulfilling prophecy effect). These findings led investigators to further explore the direct and indirect exchanges of teachers which might have implications for student achievement.
In the search for ways in which teacher expectations may be communicated to students much of the earlier literature concentrated on the direct exchanges that occurred in the classroom. Brophy and Good (1970) designed their classroom observation instrument to record dyadic teacher-student interactions and then conducted observations in four separate classrooms. Their observations in first-grade classrooms enabled them to identify 17 differing behaviours that teachers used with high and low expectation students. They found, for example, that teachers were more likely to praise correct answers from high expectation students than they were to praise such answers from low expectation students even though the latter occurred less frequently. They further reported that low expectation students were criticised more often when answers were incorrect and that teachers more often accepted poor performances from these students than they did from the high expectation students. In contrast teachers more often rephrased questions for high expectation students when they had answered incorrectly and provided support for them in reading that the low expectation students did not receive to the same degree. Brophy (1985) contended that these differential proximal behaviours may have affected the progress of students and therefore acted as self-fulfilling prophecies.
Cooper and Good (1983; Cooper, 1985) also identified similar behaviours as contributing to teacher expectation effects but added that teachers interacted more frequently with high expectation students in public and with low expectation students in private. They argued that teachers actually discouraged low expectation students from making public responses which served to inform the students of the teacher’s expectations for them.
This kind of research provided the impetus such that many researchers concentrated on further investigating the dyadic behaviours of teachers in their endeavours to unravel the consequences of teacher expectations for student learning. Such behaviours were readily observable and could be recorded relatively easily. Evidence allowed these teacher behaviours to be scrutinised and focused teacher attention on the possible behaviours that could be contributing to self-fulfilling prophecy effects. Neither Brophy and Good (1974) nor Cooper and Good (1983), however, provided effect sizes that would enable researchers to determine the relative significance of these behaviours as contributing to self-fulfilling prophecy effects.
This provision of effect sizes was conducted by Harris and Rosenthal (1985), who identified 31 different teacher behaviours that had been categorised by researchers in 136 investigations as contributing to teacher expectation effects. Their meta-analysis showed that many of the proximal behaviours identified in previous studies and which researchers and teachers had concentrated on altering actually had less impact on student outcomes than others not so frequently focused upon. For example, Harris and Rosenthal (1985) provided the following correlations for some of the behaviours identified: wait time, r = .18; praising high expectation students more, r = .12; smiling more at these students, r = .19. On the other hand behaviours that appeared to mediate teacher expectations to a greater extent included: creating a friendlier classroom climate, r = .32; teaching high expectation students more concepts and more difficult concepts, r = .29; creating a warm socioemotional climate, r = .29. It seems from this meta-analysis that the dyadic teacher-student interactions, while important, were of less significance in the mediation of teacher expectations than were the whole-class factors such as the classroom climate.
Brattesani, Weinstein and Marshall (1984) identified teachers who could be classified as high-differentiating or low-differentiating teachers. These were teachers who differentiated in their interactions with high and low expectation students to a greater or lesser degree. High differentiating teachers espoused a fixed view of ability, placed students in fairly inflexible ability groups, clearly differentiated between the instructional activities of high and low ability students, stressed performance goals and used largely negative behaviour management techniques. The low differentiating teachers on the other hand held incremental notions of intelligence, mainly used interest-based grouping, emphasised task-mastery goals and created positive relationships with students (Weinstein, 2002). The ways in which high and low differentiating teachers interact with their students in the classroom and the consequences for all students’ learning also point to whole class factors being more important in the mediation of teacher expectations than the previously identified dyadic teacher-student interactions.
Rubie-Davies (in review) recently identified teachers who had correspondingly high or low expectations for all the students in their respective classes. One month into the academic year Rubie-Davies asked the 21 teachers in her sample to rate their students’ expected achievement in reading at the end of the year from very much below average to very much above average on a 7-point scale. These ratings were compared with students’ beginning of year achievement based on running record data collected by the author. From the initial sample, when the data were aggregated for each teacher, 12 teachers could be identified who had expectations for their students’ achievement that were significantly above or below the students’ actual performance. In order to further clarify that the teachers’ expectations were class-centered rather than based on especially high or low expectations for particular ability groups the expectation and achievement data were then re-analyzed for above, below and average students in each classroom. The results confirmed that the expectations were indeed at the class level. When teachers had high expectations for their high ability students they had similarly high expectations for their average and below average students. Likewise the low expectations of some teachers were found to relate to all ability levels. Furthermore, the students placed with high expectation teachers made markedly more progress in reading (d > .5) than did those in the classes of low expectation teachers (d < .02), over the year of the research.
In a subsequent study Rubie (2004) showed that the pedagogical beliefs and self-reported practices of the high and low expectation teachers identified in her initial study differed in substantive ways. In contrast to the low expectation teachers, the high expectation teachers taught their students in homogeneous ability groups but then allowed their students to choose their learning activities. They believed that students could select appropriate activities and that students should work in mixed ability groupings with a range of peers. They did not provide discrete activities for high and low ability students as was the case with low expectation teachers. Moreover the high expectation teachers believed that while students should be given some ownership for their learning, teachers should monitor their progress closely, provide them with feedback about their learning and set clear learning goals with students. Furthermore they thought that teachers should ensure that the activities students completed were exciting and interesting, as the high expectation teachers believed that such activities were motivating for students. Again these findings point to the role of whole class factors as a medium for teacher expectation effects.
Moreover, the beliefs and self-reported practices of the high expectation teachers in the above study show parallels with the practices of effective classroom teachers identified in the literature. The fostering of self-regulation is one such approach (Bohn, Roehrig, & Pressley, 2004; Hall & Harding, 2003; Topping and Ferguson, 2005). The teachers in the above study reported providing their students with choices in the learning experiences they completed and encouraging them to set clear learning goals. Both these practices would have led to increased student self-regulation. They also reported monitoring their students’ progress closely (Berliner, 2004; Bohn et al., 2004; Hall & Harding, 2003; Pressley, Rankin & Yokoi, 1996) and providing their students with feedback related to their learning (Berliner, 2004; NICHD ECCRN, 2005; Topping & Ferguson, 2005), practices which also reflect those identified in the literature. Student motivation was considered important by these teachers, as it has been by researchers (e.g. Berliner, 2004; Block, Hurt & Oakar, 2002; Bohn et al., 2004) and one way in which the high expectation teachers endeavoured to enhance student motivation was to provide activities that students would find exciting and interesting. Bohn et al. (2004) and Pressley, Wharton-McDonald, Allington, Block, Morrow, Tracey, Baker, Brooks, Cronin, Nelson & Woo (2001) have described how effective teachers provide learning experiences for students that are meaningful and engaging. Furthermore, the high expectation teachers in the above study appeared to value the social climate of the classroom in that they reported encouraging their students to work with a variety of their peers. In recent studies both Bohn et al. (2004) and Topping and Ferguson (2005) identified this practice as common among effective teachers who appeared to encourage the development of community values in their classrooms (Bohn et al., 2004).
It was thought likely that a further mechanism for the teachers’ expectations could be the interactions that teachers had with all students in their classrooms, hence that high expectation teachers may interact with their students in different ways from low expectation teachers. While the expectancy literature has not explored teacher interactions with students at the class level, the literature on teaching styles and effective teaching practices identifies a range of practices that differentiate effective teachers from their less effective colleagues. These include focusing and re-focusing students’ attention on the topic (Topping & Ferguson, 2005; Wray, Medwell, Fox & Poulson, (2000); taking account of student prior learning (Berliner, 2004); providing high levels of instructional talk (Connor, Morrison & Petrella, 2002; Sylva, Hurry, Mirelman, Burrell & Riley, 1999); providing students with encouragement and feedback (Bohn et al., 2004; Topping & Ferguson, 2005); asking more higher level questions than less effective teachers (NICHD ECCRN, 2005; Taylor, Pearson, Clark, & Walpole, 2000;Topping & Ferguson, 2005); managing student behaviour positively (Bohn et al., 2004; Hall & Harding, 2003; NICHD ECCRN, 2005; Topping & Ferguson, 2005); and establishing self-directed classroom procedures (Berliner, 2004; Bohn et al., 2004; Topping & Ferguson, 2005). The aim of the current investigation, therefore, was to explore the classroom exchanges of high and low expectation teachers with their students as a mechanism for teachers’ expectations. In particular the teaching, feedback, questioning, procedural and classroom management strategies of the differing groups of teachers were explored.
Method
The participants in this study were 12 primary school teachers from eight different schools working in the Auckland area of New Zealand. As described above there were originally 21 teachers from 12 schools in an initial study. These schools were randomly chosen from among over 60 schools that worked with the author’s university in its pre-service teacher education programme. School principals in those schools were asked to nominate up to two teachers who might be interested in being part of the study and all teachers who were invited to participate agreed to be included. The teachers in the current study had been identified, as outlined above, fromthe earlier study (Rubie-Davies, in review) as having expectations for their students’ learning that were either significantly above or below the children’s level of achievement. The teachers formed three groups once the teacher expectation data had been collected at the beginning of the academic year and running record data had been gathered at both the beginning and end of the year by the author. The first group was made up of six teachers called the high expectation teachers who had expectations for their students’ learning that were significantly above their students’ performance and their students made statistically significant achievement gains over the year (HiEx Group). Mean effect size gains in reading for the students in these classes were d = .50, .73, .86, 1.27, 1.28, 1.44 respectively. The second group of three teachers were called the average progress teachers. These were teachers whose expectations were significantly above their students’ achievement but whose students did not make statistically significant gains over the year of the research (AvPr Group); mean effect size gains for the students in these classes were d = .27, .18, -.08 respectively. The final group also comprised three teachers who were called the low expectation teachers. Their expectations were significantly below their students’ level of achievement and their students made small or negative relative gains over the year (LoEx Group). The mean effect size gains for the classes in this group were d = .20, -.02, -.03 respectively.
Much of the research based on teacher expectations of individual students has shown that teachers have higher expectations for those students with more ability and lower expectations for students with lesser ability. Given the large reading gains of the students with the HiEx Group teachers and the smaller gains of the students with AvPr and LoEx teachers the role of students’ initial achievement in the formation of teachers’ class-level expectations needed to be established. The running record data for each student were converted to a 1-7 scale from very much below average to very much above average depending on chronological ageand in line with national curriculum expectations in New Zealand. It would therefore be anticipated that an average student across a national sample would score ‘4’. The mean for the students in the HiEx and AvPr Groups was a little below what might be expected(3.52 and 3.80 respectively) while that for the LoEx Group was above the national average (4.69). A one-way between groups analysis of variance (ANOVA) revealed a statistically significant difference between the groups (F (1,2) = 10.68, p < .001). A post-hoc Tukey test showed that the statistically significant differences at the beginning of the year were between the HiEx and LoEx Groups (p < .001) and between the AvPr and LoEx Groups (p < .007). Given that the mean for the LoEx Group was greater than that for the students of the HiEx and AvPr teachers it would seem that the teachers’ class-level expectations were not based on students’ beginning of year achievement.
The participants were observed on two separate occasions, once during the middle of the academic year (June or July) and once later in the spring (September or October) while they taught reading. Two people observed each lesson. In order to ensure that interactions were observed over a consistent time period in each lesson the first thirty minutes of each lesson was observed and recorded. Hence in total one hour of instruction was recorded and coded for each teacher.
The observation schedule was based on one previously developed by Bond, Smith, Baker and Hattie (2000) and required one observer to complete a running record of the lesson in progress recording as much as possible of what the teacher said and did during the lesson. This observer also audiotaped each lesson. At the same time a second observer completed the structured observation protocol. The author later coded the information gathered.
Four observers were recruited to conduct the classroom observations. All were student teachers in the final year of a teaching degree who were familiar with conducting classroom observations. They were told only that the author was interested in the interactions of teachers with their students in the classroom; they were not told the main focus of the research or the designation of the teacher (HiEx, LoEx or AvPr Group). The author used two videos of classroom lessons to train the observers in the use of the observation protocol. All four observers and the author coded ten-minute portions of the audio taped lessons and discussed any differences until all observers were comfortable in the use of the observation schedule and an agreement rate above 97% had been reached between all observers and the author. The observers worked in pairs throughout their observations with the same pairs observing the same teachers on both occasions. The same person completed the running record and audio taping, or the observation protocol on each occasion.