3 4 5 6 7 Rating scales
1 2 3 4 5 6 7 Rating scales
teacher smiles at child
In this method the researcher is asked to child smiles at teacher
make some judgement about the events being teacher smiles at parent /
observed, and to enter responses into a rating parent smiles at teacher
scale (see http://www.routledge.com/textbooks/ In this scene the researcher notes down what is
9780415368780 – Chapter 18, file 18.7.ppt). For happening on the thirty-second point and notices
example, Wragg (1994) suggests that observed from these precise moments that the teacher
teaching behaviour might be entered onto initiates the smiling but that all parties seem to
rating scales by placing the observed behaviour
be doing quite a lot of smiling, with the parent
onto a continuum:
and the child doing the same amount of smiling 1 2 3 4 5 each (see http://www.routledge.com/textbooks/
Aloof 9780415368780 – Chapter 18, file 18.6.ppt).
Warm
Dull Instantaneous sampling involves recording what
Stimulating
Slipshod is happening on the instant and entering it on the
Businesslike
appropriate category. The chronology of events is An observer might wish to enter a rating according preserved.
to a 5-point scale of observed behaviour, for example:
Interval recording
1 = not at all 2 = very little 3 = a little 4 = a lot This method charts the chronology of events to
5 = a very great deal
some extent and, like instantaneous sampling, 1 2 3 4 5 requires the data to be entered in the appropriate
Child seeks teacher’s
category at fixed intervals. However, instead of
attention
charting what is happening on the instant, it Teacher praises the child charts what has happened during the preceding
Teacher intervenes to stop interval. So, for example, if recording were to take
misbehaviour
place every thirty seconds, then the researcher would note down in the appropriate category what
What is required here is for the researcher to had happened during the preceding thirty seconds.
move from low inference (simply reporting obser- While this enables frequencies to be calculated,
vations) to a higher degree of inference (making simple patterns to be observed and an approximate
judgements about events observed). This might sequence of events to be noted, because it charts
introduce a degree of unreliability into the obser- what has taken place in the preceding interval of
vation, for example through the halo effect, the time, some elements of the chronology might be
central tendency wherein observers will avoid ex- lost. For example, if three events took place in the
treme categories, or recency – where observers are preceding thirty seconds of the example, then the
influenced by more recent events than less recent order of the three events would be lost; we would
events. That said, this might be a helpful summary know simply that they had occurred.
way of gathering observational data. Wilkinson (2000: 236) distinguishes between whole interval recording and partial interval
recording. In the former, behaviour is recorded Duration recording only if it lasts for the whole of the interval; in the
So far we have concerned ourselves with single latter, behaviour is recorded if it occupies only a
events and their recording. This is very suitable part of the interval in question. In the case of the
for single and usually short-lived behaviours. partial interval recording, the researcher will need
However, sometimes certain behaviours last a long to specify how to record this.
time and would over-run the interval categories
STRUCTURED OBSERVATION 403
Chapter
or event categories described above, i.e. it is researchers to satisfy themselves that it is valid continuous behaviour rather than a single event.
to infer that a particular behaviour indicates a For example, a child may remove her shoes only
particular state of mind or particular intention or once, but she may continue to be without her
motivation. The desire to operationalize concepts shoes for a twenty-minute period; a child may
and constructs can easily lead researchers to
delay starting to do any writing for ten minutes, provide simple indicators of complex concepts. again a single behaviour but which continues for
Further, structured observation neglects the longer than each of the intervals in interval or
significance of contexts – temporal and spatial – instantaneous recording; a child may have a single
thereby overlooking the fact that behaviours may tantrum which continues for twenty minutes, and
be context specific. In their concern for the overt so on. What we need is an indication of the
and the observable, researchers may overlook un- duration of a particular behaviour. The observation
intended outcomes which may have significance; is driven by the event, not the frequency of the
they may be unable to show how significant are observation. This means that the observer needs
the behaviours of the participants being observed to structure the recording schedule to indicate the
in their own terms. If we accept that behaviour is total duration of a single continuous behaviour.
developmental, that interactions evolve over time For all the kinds of schedules discussed above,
and, therefore, are, by definition, fluid, then the
a decision will have to have been agreed in three methods of structured observation outlined advance on how to enter data. Consistency of
above appear to take a series of ‘freeze-frame’ snap- entering by a single and multiple observers will
shots of behaviour, thereby violating the principle need to be found on what counts as evidence,
of fluidity of action. Captured for an instant in when, where and how to observe, and how many
time, it is difficult to infer a particular meaning to people on whom to focus. For example, how
one or more events (Stubbs and Delamont 1976), will the observation schedule distinguish between
just as it is impossible to say with any certainty one person being observed demonstrating the
what is taking place when we study a single pho- same behaviour twelve times (1 person × 12) and
tograph or a set of photographs of a particular many people demonstrating the same behaviour
event. Put simply, if structured observation is to fewer times (e.g. 2 people × 6 times each, or
hold water, then the researcher may need to gather
4 people × 3 times each), i.e. is the focus to be additional data from other sources to inform the on people or on behaviour?
interpretation of observational data. While structured observation can provide
This latter point is a matter not only for struc- useful numerical data (e.g. Galton and Simon
tured observation but, equally, for unstructured 1980; Bennett et al. 1984), there are several
observation, for what is being suggested here is the concerns which must be addressed in this form
notion that triangulation (of methods, of observers, of observation, for example:
of time and space) can assist the researcher to
generate reliable evidence. There is a risk that The method is behaviourist, excluding any observations will be selective, and the effects of mention of the intentions or motivations of this can be attenuated by triangulation. One way the people being observed.
of gathering more reliable data (for example about The individual’s subjectivity is lost to an
a particular student or group of students) is by aggregated score.
them through the course of a day or a There is an assumption that the observed week, following them from place to place, event behaviour provides evidence of underlying to event. It is part of teaching folklore that stu- feelings, i.e. that concepts or constructs can dents will behave very differently for one teacher
tracking
be crudely measured in observed occurrences. than for another, and a full picture of students’ This last point is important, for it goes to the very
behaviour might require the observer to see the heart of the notion of validity, since it requires
students in different contexts.
404 OBSERVATION