This paper presents some of the results of a 2 year project on Improving Coaching. It was a participative, partnership project in which teachers played a considerable part in collecting data, making suggestions and validating outcomes. The paper asks two simple but important questions: 1. What are routine coaching sessions like (where they are not part of a specific intervention or programme)? 2. What are the common variations in coaching? Thirty coaching conversations were recorded, transcribed and analysed and the results are presented in the paper as part of answering these questions. Coaching has a significant popular reputation as a form of professional development. However in the research and professional discussion there is sometimes a lack of precision because all coaching is not the same. There is a shortage of language and criteria through which teachers and researchers can characterise episodes of coaching. The evidence base for the effect of coaching on a number of criteria is reasonable (see Costa & Garmston, 1994 for a summary at that date). However this evidence is predominantly from the US. In Europe academic research on coaching is most evident in Holland. Veenman & Dennessen (2001) provide strong evidence from a number of studies which included control groups, that there is a 'treatment' effect for trained coaches in terms of empowering teachers, good feedback, business-like approach, value placed on coaching by coached teachers, planning observations and coaching skills. In the UK (Roberts & Henderson, 2005, Leat et al., 2006) there is evidence that coaching is evaluated very positively as a face to face process, with teachers particularly valuing the opportunity to review their teaching in detail in a supportive environment. Less positively, Hargreaves & Dawe (1990) reporting from North America suggested that many teachers resist peer coaching as they see it as a management tool to impose collegiality. Simkins et al. (2006) suggested that in leadership coaching there was often a lack of skills and awareness, so that coaching drifts into more general supervision. Research in Holland (Bergen et al. 2000) which used content analysis of tape recorded coaching sessions, showed that coaches spent an inordinate amount of time clarifying their interpretation of what they had observed and what the teacher had intended. The coaches thus dominated the conversation more than intended. Where coaches make suggestions for improvement, many of these are not followed up. In a study by Timperley & Parr (2008) in New Zealand on a project on improving children’s writing, baseline data showed that coached teachers did not act on the indirect suggestions of teacher coaches. It took two further phases of research to help teachers improve their coaching. Through describing coaching we are seeking to offer tools to teachers so that they can assume analysis of their own practice thus allowing them to adopt a pragmatic approach to enquiry and practice (Dewey, 1920)  

This was a participatory project which drew upon the structure of research experiments in which interventions and designs are improved over time (Cobb et al,. 2003). Schools were recruited in 4 regions and volunteer teachers who were already coaching identified. The sampling was necessarily opportunistic. Teachers recorded their own practice and sent us DVDs of their coaching. 30 coaching episodes were recorded, transcribed, divided into over 1000 meaning segments and analysed. The analysis used 8 dimensions, most of which had several sub-categories. The dimensions were: 1. Initiation - who started a line of discussion; 2. Stimulus - what stimulus was used; 3. Topic; 4. Tone (of each segment); 5. Scale (ditto); 6, Time frame - looking forward or back; (of each segment); 7. Interaction function - with 15 sub-categories; 8. Co-construction - which reflects quality of dialogue; Various comparative analyses have been made, for example between pre and post lesson sessions and between different coaches.

The results have provided a fascinating insight into 'everyday' coaching practice. Some of the findings include: 1. The dominant pattern is that coaches ask most questions and coached teachers spend most time explaining or clarifying their teaching/decisions and also evaluating it. 2. Although most lessons have been filmed and both parties have watched it, surprisingly little use is made of it in the post-lesson sessions; 3. The general tone of interaction is neutral - but coached teachers do most negative evaluation; 4. The most common scale of analysis in coaching sessions is the 'theme' (about the class, subject, teacher, teaching etc.) closely followed by lesson episodes and the whole lesson. 5. Little attention is paid to small incidents in lessons or particular pupils; 6. The most common topic is student engagement; 7. Few coaches challenge their partners in any way and few relate episodes or critical incidents to key principles of teaching or learning (theory); 8. The most skilled coaches (subjective judgement) use a much wider repertoire of interactions than those who are less skilled. On the basis of the results we have developed a model of 4 levels of coaching which has been validated by the project teachers.