I would suggest that the distinctive feature of observational techniques is their ability to record the flow of interaction, i.e. the dynamics of behaviour. Experiments have the disadvantage of being conducted in artificial situations where the results may be a product of the contrived circumstance and where all the relevant factors have to be identified beforehand and attempts made to control for all other variables. Pre- and post- tests, with all the refinements of control groups and proper statistical methods of multivariate analysis, etc, do not in themselves tell us anything about what went on in between. Too often the education process itself is treated as a black box, where we only measure the inputs and outputs. Surveys may be impressive in terms of the number of cases included but what they gain in quantity they often lose in quality - we learn a little about a large number of cases. Interviews and questionnaires are always one stage from the real action; they are recollections or interpretations rather than records of what actually happened.