The starting point for scientific enquiry is observation. We observe what is going on and then try to make sense of it. For example, people started to observe their surroundings thousands of years ago. They observed that the sun rose on one side of the landscape and then set on the other side and this happened at regular and predictable intervals. They observed and recorded and then made the best sense of this, which was the belief that there was a bloke in a fiery chariot chasing across the sky every day. The conclusions were wrong, but the observation was sound.
The principles and problems are there in psychological observations. We observe behaviour record it, look for patterns in the behaviour and then try to make sense of it.
The first task is to make good observations, and to devise ways of categorising it, recording it that helps our understanding. The tricky bit is to make sense of it and say what it means. What we are looking for is an explanation that tells us something that we did not know already about behaviour ... and does not involve flying blokes in flaming vehicles.
Some studies are entirely observational, but more frequently researchers use observational techniques as a way of recording behaviour as part of an experiment. The core study by Bandura, Ross and Ross is an example of this kind of research. The researchers set up a laboratory experiment but then used observational techniques to code the children’s behaviour.
The core study by Rosenhan, “On being sane in insane places”. is another example of observation. In this study, Rosenhan conducts a participant observation. This means that he and other researchers went “undercover”, posing as psychiatric patients in order to observe the interactions between hospital staff and patients. Research like this has a very high ECOLOGICAL VALIDITY as it is conducted in real life environments and participants were unaware that they were being observed. However, there is a lack of control in such research and serious ethical issues are often encountered.
Observational research is often very difficult to REPLICATE. You will never be able to observe the exact same set of circumstances in the real world, as you have little control over the CONFOUNDING VARIABLES. This is very different from experimental research, where it is possible to reconstruct the exact same environment within the laboratory. It is not always possible to be sure that you are drawing the right conclusions from your research.
|Strengths of observations||Limitations of observations |
Useful when it is unethical to manipulate variables
Now download this slideshow about the different types of observations used in Psychology.
Aim: To investigate how to make observations
Method: Work with your partner and take turns observing each other for five minutes. Person A should spend the time doing a piece of work, while Person B (the observer) should note down anything person A does.
Unstructured and Structured Observations
Aim: To consider individual differences in anxiety when speaking in public.
Method: Pilot study. Begin by conducting an unstructured observation. Ask several volunteers (male and female) to stand up in front of the class and deliver a one minute speech on any topic. The rest of the class should make a note of any non conventional behaviour, for example, scratching nose, licking lips, waving hands, saying “um “er” etc.
|Behaviour||Person 1 ||Person 2 ||Person 3 |
| Saying "um", "err", etc|
|Touching face with hand|
| Shifting feet|
Results: Summarise your findings in a table showing totals for each behavioural category for each speaker. You can also illustrate the behaviour categories using a bar chart.
You might think that making observations is easy, but after the last lesson you should now realise that it is difficult to
Observational research, like all research, aims to be objective and rigorous. For this reason it is necessary to use observational techniques. We will look at some of these techniques.
Sampling Observational data
An observer needs to decide when and how often to make observations. This may be continuous, where the observation records every instance of behaviour in as much detail as possible. This is useful if the behaviours of interest do not occur very often. Usually continuous observation is not possible because of there may be too much to record. Therefore observers use a systematic method such as:
The researcher records the relevant behaviour but has no system. The behaviour to be studied is largely predictable. Out of the Core Studies, Rosenham was one where unstructured observations were made. One problem with unstructured observations is that behaviour recorded will often be those which are most visible or eye-catching to the observer. However they may not necessarily be the most important or relevant.
One of the hardest aspects of using the observational method is deciding what you are going to record and how often different behaviours should be categorised. This is because our perception of behaviour is often seamless, when we watch somebody perform particular action we see a continuous stream of action rather than a series of separate behavioural components.
In order to conduct systematic observations we need to break up this stream of behaviour into different categories. In order to do this a researcher has to break down behaviour into a set of components, for example, when observing infants behaviour, having a list such as sucking, crying etc. The categories are sometimes called a behaviour checklist or a coding system (when a code is invented to represent each category of behaviour). You can see an example of a checklist in the core studies by Bandura et al. (aggression) and Griffiths (gambling).
A behaviour checklist or coding system may be adapted from previous research or may be developed after first making preliminary observations. It should:
Controlled and Naturalistic
In a naturalistic observation behaviour is studied in a natural situation where everything has been left as it normally. In a controlled observation some variables are controlled by the researcher, possibly in a lab, reducing the "naturalness" of behaviour being studied.
Participant and Non-Participant
In many cases the observer is merely watching the behaviour of others and acts as a non-participant. In some studies observers also participate, which may affect their objectivity. This is not so much an either/or, more a sliding scale of participation.
Disclosed and Undisclosed
One-way mirrors can be used to prevent participants being aware that they are being observed. This is called undisclosed or COVERT observation.. This method was used by in the study on aggression by Bandura et al.. Knowing that your behaviour is being observed is likely to alter your behaviour; in the study by Thigpen & Cleckley the young woman with multiple personalities knew the doctors were writing a book about her and may have "acted up" to give them something to write about.
Direct and Indirect
We have mainly been considering direct observation. In some studies observations of human behaviour are made indirectly of data that has already been collected, for example observing adverts on TV or ads in newspapers. This is called indirect observation and may use CONTENT ANALYSIS. Studies have looked at the way gender is portrayed by the media, magazines, books, TV, and so on. Such studies are described as content analysis because they make indirect observations of behaviour by looking at the content of communications provided by people. For example, Manstead & McCullough (1981) looked at ads over a one week period, ignoring those that only contained children and animals. In each ad they just looked at what the central adult figure was doing and recorded frequencies in a table like the one below. For each ad there might be no ticks, one tick, or a number of ticks.
| Credibility - product user|
| Credibility - authority on product|
| Role - dependent|
| Role - independent|
| Argument given - factual|
| Argument given - opinion|
| Type of product - food/drink|
| Type of product - alcohol|
| Type of product - household|
| Type of product - body|
In this study women were found to be more likely than men to be portrayed as product users, to be cast in a dependent role, to produce no arguments in favour of the product and to be shown at home.
Work in small groups of 4. In each group, two people work on a time sampling schedule and two on an event sampling observation. Your topic is going to be observing the socialisation and foraging behaviours of apes in a Zoo. Once the schedules are determined students will carry out a 10 minute observation using their schedules. Use http://sandiegozoo.org/videos/ . Then groups get together and compare their observations. What did some pairs miss? Did both observation schedules record the same information? Which was most effective in drawing conclusions; time sampling or event sampling?
The term reliability refers to how consistent any measurement is. If you use a ruler to measure the length of a table you expect the ruler to be reliable so that if you used it again it would give you the same measurement.
When making observations of a person or animal or event we require the observations to be something we can rely on. If they are reliable we would expect to end up with the same data even if the observations were made by a different person- two observers should produce the same record This is called inter rater reliability.
This can be checked by comparing the recordings nade by two or more observers and calculating agreement. Using time sampling a count can be kept for the number of times the observers agree with each other. As a general rule if the total agreements are more than 80% of the total observations made then the data has high inter rater reliability
Improving inter-rater reliability
Reliability can be improved by making sure that observers are trained in the use of the behavioural checklist. It might also be necessary to review the checklist and see if some categories are unclear or need sub-dividing to make for more accurate coding.
Validity is the extent to which the research has measured what it set out to measure. When making observations the main issue is observer bias, what someone observes is influenced by expectations. For example if you think that football fans tend to be aggressive this may lead you to see more aggression than an observer who believes the opposite. This reduces the objectivity and validity of observations.
Observer bias can be dealt with by using more than one observer and averaging data across observers to balance any biases. It can also be improved by keeping observers naive about the aims of the research in order to prevent their expectations biasing their observations
You will find it helpful to download this handout, summing up the strengths and limitations of different types of observation.