The most commonly used form of observation in education are lesson observations carried out as part of OFSTED inspections – technically these are a form of quantitative non-participant structured observation: OFSTED inspectors have half a dozen criteria to look out for and grade each criteria 1-4, with 1 being outstanding and 4 meaning unsatisfactory; observers will also add in some qualitative notes.
If a researcher is using previously gained records of lesson observations from OFSTED, this of course would count as a form of secondary data, but such a method is relatively easy (compared to participant-observation) for researchers to carry out as a part of their own primary research into schools.
One example of a structured observational schedule which has been used by education researchers is the Flanders System of Interaction Analysis (FIAC) which has been used to measure pupil and teacher interaction quantitatively. The researcher uses a standard chart to record interactions at three second intervals, placing each observation in one of ten pre-defined behaviour categories:
Teacher accepts pupils’ feelings
Teacher praises or encourages pupils
Teacher accepts or uses ideas of pupils
Teacher asks questions
Teacher gives directions
Terrace criticises pupils or justifies authority
Pupils talk in response to teacher
Pupils initiate talk
Silence or confusion.
Flanders used this form of quantitative behavioural analysis to discover than the typical American classroom is taken up by teacher talk 68% of the time, pupil talk 20% of the time with 12% spent in silence or confusion.
The advantages and disadvantages of OFSTED style non-participant observations applied to education
A practical problem is gaining access to observe lessons – although this is easier than with participant observation, it would still be relatively difficult to get schools and teachers to agree to this
Structured observations are relatively quick to carry out and don’t required much training on the part of the researcher.
Funding would be more likely than with more unstructured forms of observation.
Validity might be an issue – You can only observe with Non Participant Observation, you have little opportunity to get people to explain why they are doing what they are doing.
The Hawthorne Effect can be an issue – students and teachers act differently because they know they are being observed.
Reliability is good if the observation is structured because someone else can repeat the research looking for the same things.
Representativeness is easier than with unstructured observations because they are quicker to do thus larger samples can be achieved. HOWEVER, it is likely that you’ll end up with a self-selecting sample because better schools and teachers are more likely to give their consent to being observed than bad ones.
Dis-empowering for teachers and pupils – The observer is detached and acts as an expert.
Schools might give permission for observers to come in without getting the consent of the pupils.