You should be familiar with observation being included as part of the data collection for experiments. But in an experiment where the investigator manipulates the IV and controls as many extraneous variables as possible, the observation itself is not a research method. For observation to be a research method, it must be the main method of gathering data, and this is done in a natural environment where nothing is manipulated or controlled by the experimenter – behaviour is simply observed, and recorded as natural.
Sometimes, an observation can be in the form of a structured observation, where the same situation is repeated with different groups of participants and researchers observe what happens to each different participant in that situation. The IV is not manipulated by the researcher, but the setting and environment are well-controlled, making it structured. However, structured observations are very rarely used in Psychology. They may be used in child psychology, but otherwise a naturalistic observation is used.
There are four types of naturalistic observation, which fall under two conditions:
- Participant observation or non-participant observation – either the observer takes part in the study as one of the participants, or they remain a sole observer to record information only
- Covert observation or overt observation – either the observation takes place secretly without the participants knowing they are being studied, or the observer tells them fully about the study to gain informed consent
Whilst one of the main weaknesses of observations in general is that they tend to be hard to replicate, and therefore more often than not lack much reliability (as it cannot be tested for), an observation can have high inter-observer reliability. This occurs when there is more than one observer allocated to the study, and they each record their own data separately. After the data collection has taken place, the findings from each are compared and if there is a clear correlation in the data then the observation is said to have inter-observer reliability, which is a strength.
Types of Observations
Data collection in observations
An observation does not only collect qualitative data, but also quantitative data. Whilst less in-depth and rich with interpretable information, quantitative data (numerical) are useful when it comes to analysing the results obtained from the observation as a whole. There are two methods explained here for collating quantifiable data:
Tallying: This involves making a mark each time a specific behaviour is observed. For successful tallying there should be an initial observation, preferably with more than one observer, in which categories of behaviour are recorded so that all the researchers know what behaviour should be tallied.
Time-tallying: This involves using a tally table to shown behaviour being observed but, rather than giving one tally for every time a behaviour is observed, it means putting down a tally mark for each interval of time (set by the observer) that the behaviour remains to be done. For example, if you are observing the types of toys a child plays with, and they play with some play-dough, are you going to wait until he’s finished playing with the play-dough until your next tally? The child could play with it for a long time. Instead, mark off one tally for every minute (or other period of time) he continues playing with the play-dough. When he’s finished, the tallying stops and the next toy goes up.