Observation
Observation for Psychological Investigations
All types of research involve some element of observation. It is not just observational
studies that use observation. For example, when we use self report measures we observe
the responses of the participants, when we carry out experiments we observe the behaviour
of our participants and so on.
Observational studies are investigations where the researcher observes a situation and
records what happens but does not manipulate an independent variable.
Observational studies therefore tend to be high in ecological validity as there is no
intervention and if the observer remains undetected the method avoids problems with
demand characteristics.
A main strength of observational studies is that they get to see how participants actually
behave rather than what they say they do.
A further strength of observational studies is that they offer ways of studying behaviour
when there are ethical problems with manipulating variables. For example there will be less
ethical issues with carry out a naturalistic observation of school children compared to
carrying out experiments on school children.
Observational studies are also useful as a starting point in research. For example the
researchers may be investigating a new area of research in order to produce hypotheses for
future investigations such as experiments.
On the other hand observational studies are difficult to replicate.
Observations do not provide information about what participants are thinking or feeling.
There is little or no control of extraneous variables in observational studies therefore we can
not make cause and effect statements.
There is also the problem of observer bias with observational studies. This occurs if the
observers ?see? what they expect to see.
A number of ethical issues can arise with observational studies including problems with a
lack of informed consent and invasion of privacy.
Observations can also be very time consuming, require careful preparation and possibly
training for the observers.
There are a number of different types of observational studies includingnonparticipant and participant observations, undisclosed
observations and structured and unstructured observations.
A non-participant observation is a type of observational study whereby the researcher
does not join in with the activity being observed.
A participant observation is a type of observational study where the observer is also a
participant in the activity being studied. This type of observation can be useful because it
provides more insights about behaviour but does have a problem that the observer may lose
some objectivity.
An undisclosed observation (or covert) is a type of observational study whereby the
participants are not fully aware that they are being studied. The researchers may use oneway mirrors. This ensures that the participants are not aware they are being studied.
A structured observation is where the researchers design a type of coding scheme to
record the participants' behaviour.
Structured observations generally provide quantitative data. Coding schemes are ways of
categorising behaviour so that you can code what you observe in terms of how often a type
of behaviour appears.
Below is an example of a coding scheme which is also known as a behavioural checklist. The
observer simply ticks the relevant category when one of the behaviours occurs. You may
notice that a problem does arise when a behaviour occurs which fits more than one category
such as ?carries on working? whilst at the same time ?listens to music?. Behavioural
categories should be mutually exclusive (not overlapping) but in reality this is often difficult
to do with a checklist.
Advantages of using a coding scheme are that they are fairly simple to carry out and that
they provide quantitative data which can be analysed statistically.
However observation using coding schemes has a main weakness. It gives a very restricted
view of what is actually happening. The researcher may miss important behaviour and the
data is not as in-depth as simply observing behaviour which is occurring.
An unstructured observation involves the researchers recording the behaviour they can
see. This can be difficult without the use of recording equipment (such as a video camera),
can be difficult to analyse but does provide rich qualitative data.
Furthermore with unstructured observations there is a tendency for observers to record the
most eye-catching or noticeable behaviours which might not be the most relevant or
important behaviours to record.
A controlled observation occurs when the researchers control some variables. These
observations may be carried out in laboratory situations or natural situations.
Sampling observational data
Sometimes observations are made continuously where the observers record everything that
happens in detail ? perhaps with a video camera. Sometimes researchers use a sampling
technique as it may be difficult to record everything. Two techniques are event sampling and
time sampling.
Event sampling
Event sampling consists of the researcher recoding an event every time it happens. For
example, ticking a box every time somebody picks their nose. Although behaviours should
not be missed as in time sampling, if too many observations happen at once it may be
difficult to record everything.
Time sampling
Time sampling occurs when the researcher decides on a time say 5 seconds and then
records what behaviour is occurring a at that time. A problem with time sampling is that
some behaviours will be missed and therefore the observation may not be representative.
Reliability and validity of observations.
Reliability refers to how consistent a measuring device is. A measurement is said to be
reliable or consistent if the measurement can produce similar results if used again in similar
circumstances. For example, if a speedometer gave the same readings at the same speed it
would be reliable. If it didn't it would be pretty useless and unreliable.
A common way of assessing the reliability of observations is to useinter-rater reliability.
This involves comparing the ratings of two or more observers and checking for agreement in
their measurements.
A way of improving the reliability of an observational study is to ensure that the categories
are clearer or that the observers are well trained in the use of the checklists.
Validity refers to whether a study measures or examines what it claims to measure or
examine. Observations could lack validity for a number of reasons. If participants are aware
they are being observed they may behave in the way they feel they should behave. Validity
could also be reduced by observer bias. That is the observer may be influenced by
expectations and not record objectively what happened
Validity could be improved in a number of ways. Perhaps some of the categories could have
been coded in a different or clearer way. Observers could be kept unaware of the aims of
the observation or more observers could be employed.
