13.1 Experiments

🔥 Definition

An experiment is a research method used to investigate, discover, or prove a hypothesis or theory.

Its primary purpose is to show a cause-and-effect relationship between the independent and dependent variables.

Features of experiments

The Independent variable ( $IV$ ) is manipulated and the dependent variable ( $DV$ ) is measured.
Involve making comparisons between data obtained in two or more groups/conditions.
- Data is compared from the experimental group/condition to the control group/condition to draw conclusions about cause-and-effect.
Cause and effect can only be surmised if all other variables (that could affect the dependent variable) are kept constant.
They have a standardized procedure and standardized instructions so that all participants have the same experience; the study can be replicated to test for reliability.

🔥 Standardisation

Standardisation involves making the procedure the same for every participant to eliminate biases.

Laboratory experiments

Features of lab experiments

Conducted in an artificial setting where the experimenter has high levels of control.
Control variables are kept the same to ensure that only the independent variable is causing any effect on the dependent variable.
High control over many variables allows for confidence in the cause-and-effect relationship.

Evaluating lab experiments

Reliability:
- The procedure is highly standardized.
- Easily replicable to test the findings for reliability and consistency.
Validity:
- Can infer cause-and-effect because of high levels of control.
- Low ecological validity because the tasks participants are asked to do may not reflect real life.
- Results should only be generalised to everyday life with caution, as people may behave very differently in a real-life setting.
- Participants might respond to demand characteristics or social bias and behave the way they think the researcher expects, instead of what feels natural.
Ethics:
- More ethical because participants know they are taking part in the research.
- Consent is given.
- Researchers are able to debrief the participants (following any deception).
- This ensures participants are given respect and leave in the same psychological state.

Field experiments

Features of field experiment

Research is conducted in real-world settings in natural environments.
The independent variable is manipulated to measure the dependent variable.
However, there may be uncontrolled variables within and between groups/conditions.
- These are factors that are not being measured or controlled by the researcher and can have an unwanted effect on the dependent variable.
Researchers try to maintain control over variables that could affect the dependent variable, but it’s difficult.
These uncontrolled, situational variables make it hard to draw firm conclusions.

Evaluating field experiments

Reliability:
- Much less control over variables.
- Field experiments tend to be more difficult to replicate due to the limited ability to standardise the procedure.
  - This makes it hard to test for reliability.
Validity:
- High ecological validity, meaning participants’ behavior is more likely to be natural.
- Results are generalizable to similar situations.
- Participants are less likely to show demand characteristics if they do not know they are taking part in the research.
Ethics:
- Participants often do not know that they are taking part in research.
- Consent cannot be sought.
- This removes the right to withdraw.
- When the deception is revealed, it could lead participants to doubt psychologists in the future.
- Deception should only be used when the research aims cannot be met without it and only if the participants are unlikely to feel discomfort. The research design should protect the dignity of the participants.

Comparing lab & field experiments

	Field experiment	Lab experiment
Advantages	Representative: normal situation, participants behave naturally. Less demand characteristics.	Validity: good control of variables. Reliability: Standard procedures.
Disadvantages	Less reliability/replicability: less control of variables. Ethical issues: unaware of study.	The artificial environment could make participants' behaviour unauthentic. Participants could respond to demand characteristics.

13.2 Experiment Designs

Independent measures design

An independent measures design uses different participants for each condition of the experiment.
Different groups participate in different conditions (different levels of the independent variable).
Results are compared between groups.
Researchers are able to do random allocation, assigning participants to either the experimental or the control group.
- This decreases the possibility of participant variables.
- This increases the validity of the results.

Repeated measures design

Both groups (experiment and control) perform both conditions.
The same group of participants provides data relating to both levels of the independent variable.
Results are compared within the groups.
This means participant variables cannot affect the validity of the data.
Because the same participants are in both groups, there are downsides regarding validity: demand characteristics and order effects.
It overcomes problems with individual differences, as each participant acts as their own baseline.

Matched pairs design

Participants are paired based on similar characteristics.
Each pair consists of two participants who share key traits (e.g., age, gender, IQ, or skill level).
One participant from each pair is assigned to the experimental group, and the other to the control group.
This reduces participant variables because their inherent traits are balanced across conditions and won’t affect the $DV$ .

Concepts relating to experimental design

🔥 Demand characteristics

Participants sometimes, consciously or unconsciously, change their behaviour in response to cues about the aim and hypothesis. This reduces the spontaneity of their behaviour, meaning they may not behave as they usually would. This may be because they want to please the experimenter or behave in ways that they think are expected of them.

🔥 Order effects

Order effects are consequences of doing tests/tasks more than once.

When participants have to perform a task twice under different conditions, behaviour in the second condition may be different from the first because they have performed the task once already.

Order effects are a threat to validity as it becomes unclear whether the $IV$ is affecting the $DV$ or whether the participants have brought about any changes observed.

One way researchers can reduce the impact of order effects is to leave a significant time period between the first and second conditions.

Practice effect: Participants may perform better the second time due to experience gained or increased motivation.

Fatigue effect: Participants may perform worse the second time due to boredom, tiredness, or strain.

🔥 Counterbalancing

Counterbalancing is a technique used in a repeated measures design to ensure order effects do not compromise the validity of the findings. This is done by having different groups of participants complete the conditions in different orders.

Example: Group A does the experimental condition first, then the control condition; group B does the control condition first, then the experimental condition.

This way, although practice effects will still occur, their impact on the findings is reduced.

🔥 Random allocation

A way to divide participants between levels of the independent variable such that any participant has an equal chance of being in any condition. This is often done by assigning each participant a number and then using a random number generator to select participants for the groups.

This decreases the possibility of participant variables.

This increases the validity of the experiment.

13.3 Self-reports

Self-reports are a method of research where the participant themselves provide information by responding to questions.
Data may be gathered about their beliefs and opinions as well as how they believe they would behave in certain situations.
Can collect both quantitative and qualitative data.

Questionnaire

A questionnaire is a research tool consisting of a set of written or digital questions used to collect data from participants.
Printed questionnaires:
- Completed by hand.
- Can be done by anyone (doesn’t require internet access).
- More convenient for many types of research.
Digital questionnaires:
- Saves printing and postage costs, and data can be collated much more quickly and easily.
- Can only be completed by participants who have access to the internet and understand the technology.

Closed questions

Participants answer from fixed options and produce quantitative data.
They may use some form of scale such as a Likert scale (e.g., rating 1 to 5).
Yes/no questions may not provide enough options to reflect the participants’ range of experiences. A rating scale may allow participants to provide more valid responses.
Fixed choice: ‘unsure/undecided/don’t know/neutral’
Forced choice: ‘yes/no’ or ‘agree/disagree’

Open questions

Allow participants to answer freely, give answers in their own words, and provide detail. It produces qualitative data.
No fixed answer choices.
Allow the participants to express their opinion more freely and tend to start with words such as ‘why’, ‘describe’, ‘how’.
Encourage participants to give detailed, qualitative answers.
Data is rich because it allows the researcher to understand the complexity of the issues they are asking about.
Harder to analyse because they are non-numerical.

Examples of open and closed questions in a questionnaire

What to Avoid in Questionnaires

Double-barrelled questions: These are questions where two things are asked in one question.
Ambiguous (unclear) questions: Participants should understand what the question is actually asking. If not, it will affect the validity of their responses.
- Pilot studies are useful to ‘trial’ drafts of newly developed questionnaires.
Technical terms: Avoid using technical language unless your target population will understand it.
Emotive language: You should avoid using words that are likely to offend or upset your participants for ethical reasons. Emotive language may also be persuasive and lead participants to give a certain type of answer.

Evaluating questionnaires

Table evaluating the strengths and weaknesses of questionnaires

Interview

Interviews differ from questionnaires in that the researcher speaks directly to individual participants, asks them questions, and records their responses. Interviews are generally done on smaller numbers of participants as they are far more time-consuming to conduct.

Can be done by telephone or in-person.
Likely to have more open questions as people are often willing to speak in greater depth about a topic than they are to write about it in a questionnaire.

Structured interview

The interviewer has a pre-determined, fixed list of questions; all participants are asked the same questions, in the same order.

Semi-structured interview

The interviewer has a number of key questions that guide the interview, but they can be asked in any order and the interviewer may introduce new questions based on what the participant says.
The interviewer is also able to ask follow-up questions to clarify anything that the participant has said and to probe more deeply into interesting or unexpected areas.
Some of the questions will be the same for each participant but not all.

Unstructured interview

The main topics are pre-determined.
There will be a research question that steers the direction; the interviewer does not follow a set list of questions.
Questions are based on whatever the participant chooses to talk about.

13.4 Case studies

Visual representation of a case study

A research method in which a single person, family, or institution is studied in detail.
They are particularly useful for looking at rare cases or disorders where a detailed understanding helps track the progress of the participant.

🔥 Triangulation

Triangulation is when three different techniques (e.g., observations, interviews, and psychometric tests) are used to study the same phenomenon. If they all produce similar findings, then the findings are considered valid.

Evaluating case studies

Validity (Risks):
- As there is a single participant and only one or a few researchers, it may make objectivity difficult.
- The researcher might grow too close to the participant. This may interfere with the researcher’s objectivity, which will affect the research’s validity.
Validity (How to improve):
- Improve by using triangulation, where different methods are used to obtain similar results.
- Use inter-observer reliability, where multiple researchers observe the same thing and then compare findings to check for similarity.
Ethics:
- The level of detail could be an ethical threat as it may intrude into the participant’s private life, and they may feel unable to answer certain questions.
- The details of the participant might make it hard to disguise their true identity, breaking confidentiality.
Reliability:
- Findings are specific to the individual being studied and may only be applicable to a few other instances, limiting generalisability.

13.5 Observations

Features of observations

Overt vs. Covert:
- Overt: The role of the observer is obvious/known to the participants.
  - Example: The researcher wears a staff badge and openly records how students behave during lunchtime.
- Covert: The role of the observer is not obvious/known (because they are hidden or disguised).
  - Example: The researcher disguises themselves as a cleaner and secretly observes students' lunchtime behavior.
Participant vs. Non-participant:
- Participant: A researcher who watches from the perspective of being part of the social setting.
  - Example: The researcher joins a yoga class and interacts with participants while observing their behavior.
- Non-participant: A researcher who does not become involved in the situation being studied.
  - Example: The researcher sits quietly at the back of the room, taking notes without participating in the yoga class.
Structured vs. Unstructured:
- Structured: A study in which the observer only records a limited range of behaviours.
  - Example: The researcher uses a checklist to record specific behaviors, like sharing toys or aggressive actions.
- Unstructured: A study in which the observer records the whole range of possible behaviours (usually confined to a pilot stage to refine behavioural categories).
  - Example: The researcher freely notes down any and all interesting behaviors without a set checklist.
Naturalistic vs. Controlled:
- Naturalistic: A study conducted by watching participants’ behavior in their normal environment without interference.
  - Example: Participants are observed solving puzzles in their usual classroom without any interference.
- Controlled: A study conducted by watching participants’ behavior in a situation where the social or physical environment has been manipulated by the researchers.
  - Example: Participants are invited to a lab room where the researcher changes the environment (like temperature or noise) while they solve puzzles.

Examples from core studies

Scenario (Core Study)	Feature Type	Example
Bandura et al. (Aggression)	Overt	Children knew they were being observed by adult researchers in the room.
Bandura et al. (Aggression)	Covert	(Hypothetical) If researchers had watched the children through a one-way mirror without their awareness.
Piliavin et al. (Subway Samaritans)	Participant	(Hypothetical) If observers had acted as passengers while observing helping behavior.
Piliavin et al. (Subway Samaritans)	Non-participant	Actual study: Observers sat separately and simply recorded helping/no-helping without intervening.
Milgram (Obedience)	Structured	Every participant received identical scripted prompts ("Please continue," "The experiment requires that you continue").
Fagen et al. (Elephant Learning)	Unstructured	(Hypothetical) If researchers had observed elephants freely interacting with materials without specific training.
Hassett et al. (Monkey Toy Preferences)	Naturalistic	Monkeys chose toys freely without human interference, resembling natural play behavior.
Hassett et al. (Monkey Toy Preferences)	Controlled	Actual study: Researchers carefully selected and presented specific toys (wheeled vs plush) under controlled conditions.

Evaluating observations

Controls

Use clear behavioral checklists (operationalize behaviors clearly).
Train observers thoroughly to increase consistency.
Conduct inter-rater reliability checks (compare multiple observers’ ratings).
Use multiple observers and compare their results.
- Example: In Andrade’s doodling study, observers used clear definitions of doodles to improve reliability.

Ethics

Obtain informed consent if the observation is overt.
Always debrief participants after the study.
Respect participants' privacy, especially in covert observations.
Balance achieving ecological validity with protecting ethical rights.
- Example: In Piliavin’s subway study, passengers were unaware they were part of a study, raising ethical issues around consent and deception. This could have been prevented by making a debriefing announcement after the observations.

	Key Points	Extra Tips
Strengths	- High ecological validity (if naturalistic) - Can capture spontaneous, genuine behavior - Useful when unethical/impractical to manipulate variables - Often rich, qualitative data (especially unstructured)	Use examples: Piliavin had high ecological validity because it was on a real subway.
Weaknesses	- Risk of observer bias (subjective interpretation) - Risk of demand characteristics (if overt) - Hard to replicate (especially naturalistic and unstructured) - Ethical issues (privacy, consent, deception) if covert	Use examples: Bandura’s study minimized observer bias by having multiple raters and checking inter-rater reliability.

13.6 Correlations

🔥 Correlations

A correlation is a relationship or link between two measured variables.

Important: Correlation is NOT causation.

Causation is when a change in one variable causes a change in the other variable.

Positive correlations: As one variable increases, the other also increases. Has a coefficient above 0 up to $+$ 1.
Negative correlations: As one variable increases, the other decreases. Has a coefficient below 0 down to $-$ 1.
Zero correlation: No relation between the variables.

Strength of correlations

Strong: When points are close to the line of best fit.
Weak: When points are scattered and not close to the line of best fit.

Evaluation of correlations

Correlations can guide ideas for future experimental work.
Can study variables that cannot practically be manipulated (such as activity levels in dreams).
Can study variables that cannot ethically be manipulated (such as phobia strength).
Correlations cannot determine causal relationships between the variables being tested.

13.7 Longitudinal studies

Follows a group of participants over a long period of time, studying one or more variables at intervals.
Used to detect certain changes in participants or to detect the effect of a specific incident.
Tracks patterns of changing beliefs, emotions, or intelligence.
Investigates possible correlations by recording longitudinal change in two or more variables.
Different from cross-sectional studies, which compare people at different stages of life by comparing them at one particular point in time.

🔥 Key Terms

Cohort: A group of participants who are all at the same age or stage.

Sample attrition: The loss of participants from a sample over time (due to losing contact, the desire to discontinue, boredom, unavailability, or death).

Issues

Sample attrition is likely to occur; a smaller sample size is less representative, threatening the generalisability and validity of the study.
Over time, the method to measure variables may be improved, leading to the need to introduce different measures. This causes data collection to be less standard and reliable.
Researchers working on the study might move on (e.g., those administering interviews may not be constant), affecting reliability.

Ethical concerns

Getting informed consent from children and explaining it to them in a way that they understand is important, as is getting consent from parents.
It is harder to decide whether a child wants to withdraw than an adult (who will speak up for themselves).
Participants must be repeatedly asked for consent throughout the duration of the study and must be re-contacted each time. This requires maintenance of contact details, which is a confidentiality issue.