Research Methods

• Experiments
• Extraneous variables are identified and controlled for
• Involves the manipulation of an IV and DV
• High levels of control
• Can draw cause and effect relationships

• Observations - Non-experimental but can be used to collect data
• Behavioural categories are often used to record behaviour

• Correlations
• Involves two co-variables 
• No manipulation of variables
• Scattergrams show the relationships between variables

Content analysis = Systematically summarising and describing any form of content - written, spoken or visual. 

• Decide on research Q, decide on media type and how to collect a sample
• Develop a list of coding categories
• Pilot the analysis and make any changes needed
• Take a sample and count the number of times the categories occur
• Check the reliability by correlating one researcher's scores with another

Coding system must be objective, cover all possibilities and mutually exclusive. 

AO3: Avoids many ethical issues as the material already exists in the public domain so consent is assumed. 

• Suffers from subjectivity - the researcher imposes their ideas on the communication in content analysis.  

• This is an indirect observation of communication so there is a danger that opinions and motivations are attributed incorrectly. 

A way of summarising qualitative data. 

The researcher looks for themes that recur and summarises the main themes using quotations from the evidence as support. 

How to: 

• Collect the data
• Code the data and put into categories
• Look for recurring themes and patterns in the data
• Give example quotes to illustrate themes in the final report

AO3: High external validity, ethical and easily replicated if original sources are available. 

Easy comparison over time, culture or source. 

Science = The practice of generalising from observations to form a rational explanation... a theory of a particular part of the world and then testing that theory by experiment, rejecting it if it does not match up and accepting it if it does. (Williams 2000)  

• It is empirical: Claims are tested systematically and are based on fact rather than beliefs 

• It is replicable: The procedures can be reproduced and repeated and results verified 

• It is rational and explained through theories 

• Science involves testing hypotheses and modifying theories 

The hypothetico-deductive method = By employing empirical methods we can develop hypotheses which we can go out to test, we are engaging in deductive reasoning. This means that we have a theory or hypothesis, and we assume it to be correct. We can then test it in various ways to see if it stands up.  

Objectivity - Scientific observations should not be affected by the expectations of the researcher 

Theory construction - Theories must be constructed to explain observations 

Hypothesis testing - A good theory generates testable hypotheses. If an experiment does not support the hypothesis then the theory must be changed. 

Empiricism - Information is gained through direct observation/experiment rather than by reasoned argument 

Replicable - Results should be verifiable through replication 

Falsification = The process of proving a hypothesis incorrect. This is the only way that Scientific knowledge can be built up.  The attempt to disprove a theory only strengthens it.  

Null hypothesis 

• To decide that a theory is correct we must attempt to disprove it. This is why a theory starts with a null hypothesis. We look for evidence and record it to see if the null hypothesis is disproved.  

• If a theory survives repeated attempts to disprove it, it can then be considered a good one.  

• We can then reject the null and accept the experimental hypothesis. 

Paradigm = Scientific disciplines have a shared set of assumptions and methods. (Kuhn) 

Paradigm shift = Sometimes in science we see major changes in the dominant way of thinking. 

Because psychology has no singular defining approach, it is, by Kuhn’s definition a ‘pre-science however it could be argued that today’s dominant paradigm is the biological approach.  

An example of a paradigm shift is when psychology moved away from Freud’s psychoanalytic ideas towards the more scientific ideas of behaviourism.  

Peer review = The process of assessing scientific work to decide whether it is worthy of publication in an academic journal.  

• Scientist writes their study, sent to 2 or 3 experts in the same field
• Peers review the quality of the study
• Reviewers make comments on the work and it is returned to the scientist who must make any corrections needed
• It may have to not be published if the results are invalid. Reviewers are anonymous. 

Why? - Ensure only quality research is published - validity of current knowledge is maintained

• Allocate funding - Universities are given more funding for future projects if they have better research as it is seen as less of a risk for donors. 
• Comments are anonymous so they can be honest but rival researchers may use criticism to promote their own work instead. This is to prevent publication and so they could do similar research. If scientists don't allow research to be published then scientific progression is slowed. 

Referencing = The process of informing the reader of your sources. 

Your claims can be checked for accuracy as the reader can find and read the original material.  

Items in the reference list at the end must be listed in a specific way:  

1. Surname 

1. Date published 

1. Book title 

1. City 

1. Publisher 

SDBCP - Sid doesn't bake crispy penguins. 

Aim - False aim as long as it is realistic. 

Procedure - What participants will be expected to do and how long it will take. 

Ethics - Tell P's study has received ethical approval. Confidential and anonymous. Right to withdraw. 

Signature - Print name and date. 

Thank you for taking part. 

You will be asked to. 

This is not a judgement of your IQ, it is testing X

There are no right or wrong answers

Do you understand? 

Do you have any questions? 

TAJNUQ

Take Any Jam Never Use Q-tips

Thank participants for taking part 
Tell them the true aim of the study 
Ask if they have any questions  
Remind them they have the right to withdraw and that their data is confidential 
Offer a contact number / email 

TAQRC

Take Any Q-Tips Really Cry

Reliability = The extent to which a test produces a consistent finding every time it is done 

Internal reliability = the consistency of items in one test or tool 

External reliability = Will I get the same results if I do my test on a different occasion?  

Inter-rater reliability where two observers watch the same participant at the same time and rate them using the same checklist, the scores are correlated. A high correlation means the checklist and observers are reliable.  

Assessing external reliability – Test-retest method 

Usually used for self-report measures or assessments such as diagnoses.  

1. Test the participant using the measurement tool 

1. Test them again after long enough so they cannot recall their answers 

1. Re-test them using the same measurement tool 

1. Correlate the scores between the two different times  

1. A high correlation indicates the measure is reliable 

Questionnaire - Test-retest has a low correlation coefficient then some questions may need to be re-written. 

Interview - Use the same interviewer every time. Trained interviewers. Structured interviews have better reliability as there are fixed questions. 

Observation - Clear categories which are operationalised and mutually exclusive. 

Experiment - Standardised instructions, same conditions and multiple measurements to make sure one is not anomalous. 

Validity = Does your study measure what it claims to be measuring? If so can it apply to other times, places and people?  

Internal validity = How much do the findings of the dependent variable have to do with the independent variable and not other factors?  

Ps guessing the aim and showing demand characteristics is the greatest threat to internal validity.  

Ecological validity = Can your results be generalised beyond the testing environment? E.g. a lab study can lack mundane realism if Ps are asked to do something that they would never do in the real world.  

Population validity = If the study findings can be generalised to the rest of the population then they have high population validity.  

Temporal validity = Your results may show a clear effect of the IV on the DV and be applicable to different environments and populations but if they are only true to this era then they lack temporal validity.  

Predictive validity = One test will predict the outcome of a test at another point. 

Concurrent validity = Results it gives are the same as existing measures. 

Face validity = Involves asking an expert in the field to look at your study/measurement tool and say whether on the ‘face of it’ / surface it seems to be measuring what you want to measure.  

Questionnaire - lie scale - a question to check if your P is lying because you know the answer. Anonymity ensures honesty. Avoid weighting all questions the same way to avoid Ps ticking all yes or no. 

Observation - Observer training and well-operationalised categories.  

Experiment validity - Use a control group to isolate the true effect of the IV on the DV.  

Deception reduces demand characteristics.  

Independent groups prevent order effects.  

Field and natural studies improve ecological validity over lab studies.  

Experimental hypothesis
A hypothesis that says there will be a difference in the DV between your conditions 
There will be a difference between...

Null hypothesis
A hypothesis that says there will be no difference 
There will be no difference between...

One tailed/ directional hypotheses
An experimental hypothesis predicts the direction in which the results are expected to go
This is because we have some previous research to base our prediction

Two tailed/ non-directional hypotheses
A hypothesis that does not state a direction, but simply states that there will be a difference between two sets of scores

• Include all levels of the IV (all the conditions of the study)

• Include the DV

• Operationalised all the variables 

• Predicted which group you think will score more highly (if it is directional)

IV and DV, Operationalise and Predict

Repeated measures - Everyone does everything

Less likely that participant variables will influence the results as the same participants are used in each condition. However order effects and demand characteristics are more likely. 

Repeated measures: Counterbalancing 

Participants are split in half and complete the conditions in different orders to reduce order effects. 

Independent groups - Different participants complete the different experimental conditions

Advantage in not having order effects however there is no control over participant variables but this can be improved by random allocation. 

Matched pairs - Recruit participants and find another group that matches them on participant variables. Compare results. Minimises participant variables and no order effects however can be time consuming and cannot match on everything. 

Ethical issue: Consent - Get participants to sign a consent letter or form. 

Ethical issue: Deception - In a debrief the participants should be informed of the deception and asked again for their consent to use their data. 
Limitations: In a field or quasi study it can be hard to catch up with participants after the study. 
 
Ethical issue: Confidentiality - Maintain anonymity by using initials or numbers. 
Limitations: If a case is infamous it can be hard for the participant to remain anon. Participants withdrawing their data can limit the data pool. 
 
Ethical issue: Debriefing - Ensure that all participants receive this information, allowed to ask questions and be given contact details.
 
Ethical issue: Withdrawal rights  - Remind participants that they may withdraw at any time. 
Limitations: Too much information may change behaviour. Too many participants may withdraw limiting your data pool. 
 
Ethical issue: Protection from harm  - Counselling. 
Limitations: It is harder to know the extent of mental damage.

Random Sampling

Every member of the target population has an equal chance of being selected. Free from investigator bias. Time consuming and often difficult in practice.

Systematic sampling: Target population placed in order. How many participants you need = N. Select every Nth person to be in your sample. Minimises researcher bias.

Stratified sampling: identify the different types of people that make up the target population and work out the proportion needed from each group for the sample to be representative. Time-consuming.