There is a newish report at http://www.york.ac.uk/res/herc/documents/wp/10_11.pdf
that includes analysis of coaching effects. Around p22 it concludes: "The coaching effect is large and statistically significant". This is pretty much in line with the last comment on the Nfer pages that coaching gains are substantial http://www.nfer.ac.uk/nfer/research/assessment/eleven-plus/#10
While I'm at it, some other conclusions on coaching that inexplicably haven't yet made it onto the Nfer pages:
"the moderately bright tend to gain most" (Jensen, 1980)
"The results clearly indicate that coaching has a massive effect on 11+ marks." (Egan & Bunting, 1991).
"Those who did not receive coaching, regardless of ability as assessed by the Raven's matrices, obtained very poor marks in the 11+, and in no case would one of these students have passed this selection process" (Egan & Bunting, 1991).