While research on teacher development has shown changes over time in self-reported competence beliefs during initial teacher education (Malmberg & Hagger, 2009), and during professional practice (Hoy, & Spero, 2005), less is known about the development of the observed quality of teachers’ classroom practice. Since the promotion of adaptive non-academic and academic student outcomes are emphasized by educational systems worldwide it is important to know more about how classroom quality develops; that is how teachers’ competence to engage in warm, well-organized and cognitively stimulating interactions with students changes over time. In order to expand our knowledge of the development of teachers’ classroom quality, the aims of this study were to investigate variability and change, and the effects of personal characteristics and contextual factors on observed classroom quality over three years (one year of pre-service teacher education and two years in practice). Six research questions drove the study: (1) How variable is classroom quality, within lessons, across time and between teachers? (2) Does classroom quality change during the year of pre-service teacher education year and the first two years of professional practice? (3) Are there differences between individual teachers with regard to change over time? (4) How do teacher characteristics (age, gender and teaching subject) predict classroom quality, and change in classroom quality? (5) How do features of the classroom context (number of students, ability grouping and year-group) predict classroom quality? (6) How is student engagement related to average and variability of emotional support, classroom organization and instruction support at the lesson level? Analysis was framed by four hypotheses about the shape (i.e., linear and/or quadratic) of the trajectories of change for beginning teachers over time, drawn from the literature related to teachers’ self-reported competence beliefs, stages of teachers’ concerns, and the supported transition into working life. 1. Based on international research suggesting that teacher concerns during the earlier survival stage are related to the self and tasks, and concerns during the later stages relate to impact on students (Watzke, 2007) the first hypothesis assumed an early increase in observed classroom organization, and a later increase in instructional support. (i.e.impact related concerns). 2. Some studies of supported in-service and beginning teachers studies have found that teachers entering the profession became less democratic and more custodial over time (Hoy & Woolfolk, 1990), but also point to increases in organized, systematic, affective and stimulating teaching behavior (Adams, 1982). A significant inverted U-shaped effect in addition to a significant linear effect would depict an earlier steeper increase over time that would later plateau. 3. The third hypothesis suggested a dip during a “reality shock” (Veenman, 1984). A U-shaped effect, that is a dip in the trajectory during entry into the profession, could be indicative of such a reality shock. 4. Student engagement will be positively and strongly related to the emotional support, classroom organization and instructional support provided by the beginning teacher. (Malmberg & Hagger, 2009; Skinner & Belmont, 1993).

The data was drawn from a longitudinal study tracking 17 secondary teachers of English, Mathematics or Science through their initial postgraduate teacher preparation courses (at two English universities) and the first two years of teaching. The database comprised ten video recordings of lessons for each participant (spread across the three years) although not all recordings proved to be of sufficient quality for analysis. The lessons were analyzed using the adolescent version of the Classroom Assessment Scoring System for secondary school (CLASS-S; Pianta, La Paro & Hamre, 2006), which specifies four constructs in relation to observed behaviour: emotional support, classroom organization, instructional support and student engagement. Between one and four 20-minute segments were coded from each lesson (376 segments in total). Three researchers carried out the coding, following rigorous training and the cross-validation of three coded lessons and their justification scripts by the original instrument authors.

Confirmatory Factor Analyses with a reduced number of indicators per a priori construct fitted the data well. Using factor scores in 3-level multilevel model for change (lesson segments nested within time-points, within teachers), we found more variability over time than within lessons, and least variability between teachers. Emotional and instructional support showed an inverted U-shaped change over time, while classroom organization increased linearly over time. Emotional support was lower for older students and students were more engaged in larger classes. Older teachers provided more instructional support. A higher level of classroom organization during lessons was related to less variability in student engagement, but variability of instructional support was unrelated to variability of student engagement within each lesson. These findings are important for our interpretation of the possible effects of pre-service teacher education and support for newly qualified teachers. The challenge of ensuring that the pre-service education and induction of secondary teachers effectively promotes higher levels of emotional and instructional support, rather than merely strengthening teachers’ classroom organization, is clearly a critical one.