AQA Psychology: Research Methods Extended
Concepts
Observational Methods and Techniques
Observational studies involve the observation and recording of behaviour. Observation is often
used within an experiment in order to measure the DV. Therefore, it is best to view observation as
a technique used in conjunction with other research methods rather than a separate technique. If
observation is used in an experiment, then it is a research technique; the method is still
experimental. Observations can be naturalistic, in which nothing is controlled, or controlled in
which certain variables are controlled by the researcher, either as independent variables or as
environmental stimuli. Generally speaking, this is more likely to be lab based and more likely to
have obtained informed consent (overt over covert).
Another small addition here is the difference between participant and non-participant
observations. Most of the time non-participant observation is used, in which the observer does
not interact with the observed. Participant observation occurs when the observer is part of and
interacts with the observed.
+Naturalistic observations have high ecological validity
+Controlled observations mean less chance of extraneous variables impacted the EV and lowering
internal validity
+Covert observation increases ecological validity
+Participant observation allows for rich information
-Naturalistic observations lack control
-Controlled observations lack ecological validity
-Covert observation may breach ethical standards
-Participant observation is less objective
Bias and Reliability
Whilst these methods can capture spontaneous behaviour, there is a serious issue in observer
bias; where an observers’ expectations affect what they perceive, affecting the validity of the
study. Using more than one observer is therefore important to diminish this. Inter-rater reliability
can then be used to see the extent to which observers agree. 80% plus is usually considered high.
Additionally, despite the access to spontaneous behaviour observations can only show us what
people do but not what they think.
Observational Design
Unstructured observations involve recording all relevant behaviour but with no system, as
behaviour may be too unpredictable. There may be a problem with the sheer amount to record, or
that recorded behaviours may be the most reaction eliciting rather than most important or
relevant.
Structured observations involve systems used to organise observations, and are more common in
research. These systems include:
-Behavioural Categories
This involves reducing behaviour into a set of components that are objective, cover all possible
component behaviours and are mutually exclusive. Operationalisation is linked to achieving this.
, -Sampling Procedures
When conducting a continuous observation, the researcher should record every example of the
relevant behaviour in as much detail as possible. This is useful when behaviour is infrequent.
However, in many situations this is difficult due to the frequency of behaviour, therefore
systematic methods such as event sampling and time sampling can be used. The former involves
counting the frequency of behaviour, the latter involves recording behaviour within a given time
frame.
Self-report Techniques
Another non-experimental method is to acquire data directly from participants via self-report.
Questionnaires are a set of questions designed to collect information, with the great strengths of
ease of large sample size and access to people’s thoughts above that of observation or
experimental techniques, assuming what they answer truly reflects what they think. A good
questionnaire should include questions that have:
Clarity-Certain terms should be operationalised and there should be no ambiguity for the
reader. Double negatives and double-barrelled questions should be avoided.
Bias-Leading questions and social desirability bias are problems that need to be mitigated.
The latter is near impossible to mitigate.
Analysis-Closed questions as opposed to open questions can make results easier to analyse
due to being quantitative (pattern-spotting is easier) but may not adequately reflect
behaviour or cognition. Open questions produce rich qualitative data (but may be affected
by apathy)
Other considerations include:
Filler questions: Irrelevant questions designed to reduce demand characteristics
Sequence for the questions: Start with easier questions leaving any anxiety inducing or
difficult questions till later
Sampling technique: Important given the need to reduce sample bias. Questionnaires
often use stratified sampling.
Pilot study: The questionnaire can be tested on a small group of people beforehand,
leading to refinement and refocus of the questions if needed.
Interviews are essentially real-time questionnaires, and can be structured or unstructured. In the
former, questions are pre-determined. In the latter, questions are developed on the spot. A semi-
structured approach, or clinical interview, can also be used. A good interview, as well as also
having the same approach to questions as above, could have:
Recording: Helps the interviewer focus on listening and avoids participants thinking some
answers are not valuable
Effect of the interviewer: An interviewer should remain aware of how non-verbal
communication and listening skills can facilitate a good interview and rich information, or
bias answers
Unstructured interview skills: It is important here to avoid repeating questions, avoid too
many follow-up questions and probing too much
Concepts
Observational Methods and Techniques
Observational studies involve the observation and recording of behaviour. Observation is often
used within an experiment in order to measure the DV. Therefore, it is best to view observation as
a technique used in conjunction with other research methods rather than a separate technique. If
observation is used in an experiment, then it is a research technique; the method is still
experimental. Observations can be naturalistic, in which nothing is controlled, or controlled in
which certain variables are controlled by the researcher, either as independent variables or as
environmental stimuli. Generally speaking, this is more likely to be lab based and more likely to
have obtained informed consent (overt over covert).
Another small addition here is the difference between participant and non-participant
observations. Most of the time non-participant observation is used, in which the observer does
not interact with the observed. Participant observation occurs when the observer is part of and
interacts with the observed.
+Naturalistic observations have high ecological validity
+Controlled observations mean less chance of extraneous variables impacted the EV and lowering
internal validity
+Covert observation increases ecological validity
+Participant observation allows for rich information
-Naturalistic observations lack control
-Controlled observations lack ecological validity
-Covert observation may breach ethical standards
-Participant observation is less objective
Bias and Reliability
Whilst these methods can capture spontaneous behaviour, there is a serious issue in observer
bias; where an observers’ expectations affect what they perceive, affecting the validity of the
study. Using more than one observer is therefore important to diminish this. Inter-rater reliability
can then be used to see the extent to which observers agree. 80% plus is usually considered high.
Additionally, despite the access to spontaneous behaviour observations can only show us what
people do but not what they think.
Observational Design
Unstructured observations involve recording all relevant behaviour but with no system, as
behaviour may be too unpredictable. There may be a problem with the sheer amount to record, or
that recorded behaviours may be the most reaction eliciting rather than most important or
relevant.
Structured observations involve systems used to organise observations, and are more common in
research. These systems include:
-Behavioural Categories
This involves reducing behaviour into a set of components that are objective, cover all possible
component behaviours and are mutually exclusive. Operationalisation is linked to achieving this.
, -Sampling Procedures
When conducting a continuous observation, the researcher should record every example of the
relevant behaviour in as much detail as possible. This is useful when behaviour is infrequent.
However, in many situations this is difficult due to the frequency of behaviour, therefore
systematic methods such as event sampling and time sampling can be used. The former involves
counting the frequency of behaviour, the latter involves recording behaviour within a given time
frame.
Self-report Techniques
Another non-experimental method is to acquire data directly from participants via self-report.
Questionnaires are a set of questions designed to collect information, with the great strengths of
ease of large sample size and access to people’s thoughts above that of observation or
experimental techniques, assuming what they answer truly reflects what they think. A good
questionnaire should include questions that have:
Clarity-Certain terms should be operationalised and there should be no ambiguity for the
reader. Double negatives and double-barrelled questions should be avoided.
Bias-Leading questions and social desirability bias are problems that need to be mitigated.
The latter is near impossible to mitigate.
Analysis-Closed questions as opposed to open questions can make results easier to analyse
due to being quantitative (pattern-spotting is easier) but may not adequately reflect
behaviour or cognition. Open questions produce rich qualitative data (but may be affected
by apathy)
Other considerations include:
Filler questions: Irrelevant questions designed to reduce demand characteristics
Sequence for the questions: Start with easier questions leaving any anxiety inducing or
difficult questions till later
Sampling technique: Important given the need to reduce sample bias. Questionnaires
often use stratified sampling.
Pilot study: The questionnaire can be tested on a small group of people beforehand,
leading to refinement and refocus of the questions if needed.
Interviews are essentially real-time questionnaires, and can be structured or unstructured. In the
former, questions are pre-determined. In the latter, questions are developed on the spot. A semi-
structured approach, or clinical interview, can also be used. A good interview, as well as also
having the same approach to questions as above, could have:
Recording: Helps the interviewer focus on listening and avoids participants thinking some
answers are not valuable
Effect of the interviewer: An interviewer should remain aware of how non-verbal
communication and listening skills can facilitate a good interview and rich information, or
bias answers
Unstructured interview skills: It is important here to avoid repeating questions, avoid too
many follow-up questions and probing too much
