Estimating effectiveness of school-based counselling: Using data from controlled trials to predict improvement over non-intervention change