Materials

Materials for this study included the (a)BEST in CLASS teacher manual,(b)BEST in CLASS coaching manual,and(c)Teacher–Child Interaction Direct Observation System(TCIDOS).The BEST in CLASS teacher manual included readings for teachers to complete related to the BEST in CLASS intervention practices,teacher-learning activities,and other supportive materials such as home–school communication forms.The BEST in CLASS coaching manual provided standardized coaching procedures and included all required forms and materials to be used during the coaching component of the intervention.The TCIDOS was used to collect observational data for the purposes of providing graphical feedback during coaching and of examining child and teacher outcomes and interactions across the course of the intervention.

Behavioral Definitions and Observation

Procedures

Behavioral definitions.The following teacher behaviors were observed and coded:rules,precorrection,opportunity to respond (OTR), behavior-specific praise (BSP),instructive feedback,corrective feedback, active supervision, and reprimands. Focal children’ s behaviors observed and coded included disruption, aggression, defiance(DAD) and engagement.In addition,teacher–child positive and negative interactions were coded.See Table 1 for behavioral definitions and codes.

Observation procedures.All observation sessions were conducted by trained research staff in each early childhood classroom during small and large group teacher-directed instructional activities. The TCIDOS was used for recording teacher and child behaviors during these times.The TCIDOS is a researcher-developed interval recording observation system (10-s observation and 10-s recording window) that utilized a paper/pencil format and audio cue. Partial interval recording was used to code teachers’ display of rules,precorrection,OTR,BSP,instructive feedback,corrective feedback, and reprimands,as well as children’ s DAD.Momentary time sampling(MTS)was used to code teachers’ active supervision,child engagement,and teacher–child interactions,both positive and negative.Observation periods lasted approximately 20 min and were conducted twice per week.During each 20-min observation period,10 min of observation and 10 min of recording time occurred.MTS occurred at each point in time when observers were cued to record.Thus,MTS occurred at 60 momentary time points across each of the 20-min observation periods.

Data were collected every week during two 20-min observation sessions across two teacher-directed instructional formats(large and small group)during the three phases of the study:baseline,coaching phase(which lasted 14 weeks),and maintenance,which occurred approximately 1 month after completion of the full intervention.During baseline,data were collected on all codes across two 20-min observations.During the coaching phase,two 20-min observations were conducted every week on each targeted BEST in CLASS instructional strategy and other responses observed(e.g.,child engagement,teacher reprimands).For example,the 2nd and 3rd weeks of the coaching phase targets teachers’ use of rules,expectations,and routines;therefore,two 20-min observations were conducted on the teachers’ implementation of rules,expectations,and routines in weeks 2 and 3 of the coaching phase.Data on each BEST in CLASS instructional strategy were averaged at the end of the time period when coaching was completed on that instructional strategy.For example,the observations of teachers’ use of rules,routines,and expectations following coaching were averaged across Weeks 3 and 4.Throughout each week of the coaching phase,data on other responses(e.g.,child behaviors [engagement,disruptive behavior]and positive and negative teacher–child interactions)were collected and the occurrence of these responses was averaged across the entire 14 weeks of the coaching phase.Similar to baseline,maintenance data were collected across two 20-min observations.

Interobserver reliability estimates.On 24.68% of the observation sessions (across phases),interobserver reliability estimates were assessed for the occurrence or nonoccurrence of the responses coded,represented by intervals coded. Interobserver reliability estimates were collected by having the secondary observer collect data at the same time as the primary observer.Interobserver reliability estimates were calculated by computing agreements/agreements + disagreements×100.Overall,interobserver reliability estimates.Percentage agreement by code can be found in Table 2.