Research Methods

1.1 Experiments

An experiment is a type of research used to investigate cause-and-effect relationships. It tests whether one factor (called the
independent variable or IV) causes a change in another factor (called the dependent variable or DV).

To do this, the researcher manipulates the IV to create different conditions or “levels”—for example, using either bright or dim
lighting. Then, they measure how these changes affect the DV. For instance, they might test if brightness (the IV) affects
attention (the DV). If people pay more attention in bright light than in dim light, this suggests that brightness has an effect on
attention.

KEY WORDS
• Experiment: A study that looks for a cause-and-effect relationship by changing an IV and measuring its effect on a DV.
• Independent Variable (IV): The factor that the researcher changes (manipulates) to create different conditions.
• Dependent Variable (DV): The factor that is measured to see if it changes as a result of the IV.

To be more confident that the IV is what’s causing changes in the DV, researchers try to control other factors that might
influence the DV. These variables should be kept the same in all conditions so they don’t interfere with the results.

Experiments can have:

• Two or more experimental conditions (e.g. bright vs. dull lighting), or
• One or more experimental conditions and a control condition, where the IV is absent. For example, comparing eating one
bar of chocolate to eating none.

KEY WORDS
• Uncontrolled Variable: A variable that isn’t controlled and might affect the DV. If it only affects one condition, it’s a
confounding variable, because it makes it hard to tell if the IV is truly causing the effect.
• Experimental Condition: One version of the IV being tested.
• Control Condition: A condition where the IV is not present; used for comparison.
Experimental Design

Experimental design refers to how participants are assigned to different conditions (levels) of the IV. The three main types are:
1. Independent Measures Design
2. Repeated Measures Design
3. Matched Pairs Design

KEY WORDS
• Experimental Design: The method used to assign participants to levels of the IV.
• Independent Measures Design: Each condition has a different group of participants.

1. Independent Measures Design

In this design, different people take part in each condition. So the data from one condition is completely separate
(“independent”) from the other.

Strengths:
• Each participant only takes part once, so they’re less likely to guess the purpose of the study (fewer demand
characteristics).

Weaknesses:
• Participant variables (differences between people, like memory ability or personality) might affect results.

(Individual differences refer to all the ways people naturally differ. Participant variables are a type of individual difference
specifically relevant in research settings. Participant variables are those traits could influence the results.)

To reduce this problem, researchers use random allocation. Each person is randomly assigned to a group (e.g. using a number
generator or drawing names). This spreads participant differences evenly across conditions.

KEY WORDS
• Demand Characteristics: When participants guess the purpose of the study and change their behaviour.
• Random Allocation: Each participant has an equal chance of being in any condition. This helps balance out participant
variables.
2. Repeated Measures Design
Here, the same people take part in every condition. For example, if we’re studying the effect of doodling on memory, the same
group would do the task with and without doodling.

Strengths:
• Each person acts as their own comparison, so participant variables are controlled.
• Example: If someone learns quickly, they’ll do so in both conditions, reducing individual differences.

KEY WORDS
• Repeated Measures Design: Each participant does every level of the IV.
• Participant Variables: Personal differences (e.g. age, intelligence) that might affect performance.
• Confounding Variable: A variable that affects only one condition and confuses the results.

Weaknesses:
Doing the same task more than once can lead to order effects:
• Practice effect: Performance improves due to repetition.
• Fatigue effect: Performance worsens due to boredom or tiredness.
• Participants are also more likely to figure out the study’s purpose (demand characteristics).

Solutions to Order Effects:

• Randomisation: Participants are randomly assigned to do the conditions in different orders (e.g. some do “music then no
music”, others the opposite).
• Counterbalancing: The group is split in half. One half does condition A then B; the other half does B then A.

KEY WORDS
• Order Effects: Changes in performance caused by repeating tasks.
• Practice Effect: Improved performance due to familiarity.
• Fatigue Effect: Worsened performance due to tiredness or boredom.
• Randomisation: Assigning orders randomly to reduce order effects.
• Counterbalancing: Dividing participants so all order combinations are tested evenly.

3. Matched Pairs Design

This design tries to reduce the problems found in both other designs. Each participant is paired with someone similar in
important ways (like age or aggression level). One person from the pair goes into one condition, the other into the second.
• This reduces participant variables (because people are matched).
• It also avoids order effects (because each participant only does one condition).

Example: In a study on violent video games, children could be matched based on aggression level. Each pair is split between the
violent and non-violent game conditions.

Twins are ideal matched pairs because they are very similar in many ways.

KEY WORD
• Matched Pairs Design: Each participant is paired with someone similar, and each is placed in a different condition.
Types of Experiment

1. Laboratory Experiments
These are experiments done in controlled, artificial settings, not in the participant’s normal environment. Example: To test how
lighting affects attention, children might do a computer task in a quiet university lab with controlled lighting and noise.

KEY WORD
• Laboratory Experiment: A study in a controlled setting, with an IV, a DV, and attempts to control all other variables.

Evaluating Laboratory Experiments

Strengths:
• High control of variables, which increases validity (you’re more sure the IV caused the effect).
• Standardisation (keeping procedures the same) increases reliability (results can be trusted and repeated).
• Pilot studies can be done beforehand to test and improve the procedure.
• Lab experiments are easier to replicate because everything is standardised. This improves replicability and helps confirm
that results are accurate.

KEY WORDS
• Controls: Keeping unwanted variables the same across all conditions.
• Standardisation: Making sure all participants are tested in the same way.
• Reliability: Whether the study gives consistent results.
• Validity: Whether the study is really testing what it claims to.
• Pilot Study: A small-scale test to check if the study design works well.

2. Field Experiments
Field experiments have both strengths and weaknesses, just like laboratory experiments.

One disadvantage is that it’s harder to control variables and make sure the procedure is exactly the same for all participants,
compared to a laboratory setting. Because of this, reliability (how consistently the experiment produces results) and validity
(how well the results reflect what is being studied) may be lower in field experiments.

However, validity can sometimes be higher in field experiments. This is because participants are in a familiar setting, doing tasks
that feel normal. For example, school students might behave differently if taken to a university lab — they could focus more
because they’re nervous or curious. This behaviour might hide the actual effects of the experiment. As a result, lab findings
might not apply well to real-life settings. This issue is called ecological validity — field experiments usually have better
ecological validity than lab experiments, but not always.

Another strength of field experiments is that participants might not know they are in a study. If they’re unaware, they are
less likely to show demand characteristics (where they change their behaviour to match what they think the researcher wants).
But if participants don’t know they’re in a study, they haven’t agreed to take part, which raises ethical concerns.

KEY WORDS
• Field experiment: An experiment that tests for a cause-and-effect relationship (by changing the IV and measuring the
DV) in the participants’ usual environment. Some control over variables is still possible.
• Generalise: Applying results from a study to other people, places, or situations.
• Ecological validity: How well results from a study apply to real-world situations. It depends on whether the setting and
tasks are realistic (also called mundane realism).
Hypotheses in Experimental Studies

To do this properly, Dr Huang needs to write a hypothesis. A hypothesis is a testable statement. It gives more detail than just
stating the aim, and describes what’s being tested.

A good hypothesis must be falsifiable — meaning it must be possible to prove it wrong. This is essential because if a hypothesis
cannot be proven false, it’s not scientific. There are different ways to write a hypothesis, depending on the type of prediction
you are making.

KEY WORDS
Hypothesis (plural: hypotheses): A testable statement based on the aim of the research.
Alternative hypothesis: A hypothesis that predicts a difference (in experiments) or a relationship (in correlations). It should
clearly define (operationalise) the variables.

Example:
Poorly written hypothesis: “Students using revision apps will learn better than students using mind maps.” (It’s unclear what
“better” means and which apps are being used.)
Operationalised hypothesis: “Students using the Gojimo revision app will score higher in a biology test than students using mind
maps.”

1. Non-Directional Hypotheses
A non-directional hypothesis (also called a two-tailed hypothesis) predicts that the independent variable will affect the
dependent variable, but does not say in which direction.

This is used when you don’t have enough previous research to suggest what will happen.

2. Directional Hypotheses
A directional hypothesis (one-tailed) predicts the direction of the effect. It’s used when previous research or theory suggests
what the result will be.

KEY WORDS
Non-directional (two-tailed) hypothesis: Predicts a difference or relationship but not the direction.
Directional (one-tailed) hypothesis: Predicts both the outcome and the direction (increase or decrease, positive or negative).

3. Null Hypotheses
The null hypothesis is the opposite of the alternative hypothesis. It says that any difference or relationship in the results is not
meaningful — it happened by chance.

This concept is important in inferential statistics, which are mathematical methods used to see whether a result is likely to be
real or just random.

If the results are statistically significant, the researcher can reject the null hypothesis and accept the alternative hypothesis.
If not, they must accept the null hypothesis — meaning there’s no real effect.

KEY WORDS
Null hypothesis: A testable statement that says any difference or correlation is due to chance and not the variables being
studied.
Ethics in Experiments

In a lab experiment, participants usually know they’re in a study and can give informed consent (agreeing to take part after
being told what the study involves).

However, researchers sometimes deceive participants (don’t tell them the true purpose) to avoid demand characteristics. This
helps keep the study valid but raises ethical issues. There’s always a need to balance good ethics with good science.

In field experiments, participants often don’t know they’re in a study, so they can’t give consent or choose to withdraw. This is
ethically problematic. If deception is used, researchers should debrief participants afterwards, though this can be difficult in
field settings.

It’s important that all participants are protected from harm, whether the experiment is in a lab or a field.

KEY WORDS
• Informed consent: Participants know what the study involves and agree to take part.
• Right to withdraw: Participants can leave the study and have their data removed at any time.
• Protection from harm: Participants should not face more risk than in everyday life.
• Deception: Misleading participants about the aim or procedure. If used, it must be justified and followed by a debrief.
• Privacy: Respecting participant’s space and not observing them in situations where they expect to be alone.
• Confidentiality: Keeping data safe and anonymous. If participants are unaware of the study, care must be taken not to
reveal their identity (e.g. by mentioning their workplace).

Applying Your Knowledge of Experiments

You should be able to:
• Identify an experiment and recognise the IV (independent variable) and DV (dependent variable).
• Operationalise the IV and DV (define them clearly and in detail).
• Decide if a study is a lab or field experiment.
• Evaluate experiments in terms of controls, standardisation, ethics, reliability, and validity.
• Plan an experiment by:
• Choosing between lab and field design,
• Defining the IV and DV clearly,
• Selecting a suitable experimental design,
• Applying proper controls.
 1.2 Self-reports

In a self-report, the participant gives information about themselves directly to the researcher. This is different from methods
like experiments or observations, where the researcher collects data about the participant’s behaviour or responses. There are
two main ways to carry out a self-report:
• Questionnaires
• Interviews

Both involve asking the participant questions.

Questionnaires
A questionnaire presents written questions to the participant. It can be done on a paper or online.

There are different types of questions, but the two most important formats are:
1. Closed questions – these have a set list of possible answers.
2. Open questions – these allow participants to respond in their own words, giving detailed personal and subjective answers.

Closed questions
Closed questions ask for fixed responses. These can be:
• Yes/No questions
• Multiple choice – choosing one or more options from a list
• Rating scales – selecting a number (e.g., from 0 to 5) to show how much someone agrees or feels a certain way
• Likert scales – where the participant shows how strongly they agree or disagree with a statement

Likert scale example: exercise is a necessity using options like:

Strongly agree / Agree / Neither agree nor disagree / Disagree / Strongly disagree

Examples of closed questions:

• What is your gender? (Boy / Girl)
• How do you travel to school? (Walk / Bicycle / Bus / Train / Car)
• Which animals scare you? (Dog, Spider, Cat, Rat, Fish, Rabbit, Bird – tick all that apply)
• On a scale of 0 to 4, how much do you like psychology? (0 = not at all, 4 = very much)

Open questions
Open questions ask participants to give detailed, descriptive answers. These are often used to explore why people think or
behave a certain way. They provide more depth than closed questions.

Examples of open questions:

• Why do you think it’s important to help people with phobias?
• Describe your views on how social media encourages helping behaviour.
• Explain how you would react if told to hurt someone.
Evaluating Questionnaires

Advantages:
1. Easy to analyse: Closed questions give straightforward results that are easy to count and summarise. You can calculate
averages to see overall patterns.
2. In-depth data: Open questions give richer information, helping researchers understand behaviours, feelings, and reasoning
better.

Disadvantages:
1. Open questions are harder to interpret: Researchers must decide what the answers mean, which can lead to inconsistency.
If different researchers interpret answers differently, it lowers inter-rater reliability. Inter-rater reliability: How much
agreement there is between two researchers who are interpreting the same qualitative (open) data.
2. Low response rates: Participants may not complete or return questionnaires, especially if done by post or online. This
reduces the sample size.
3. Unrepresentative samples: People who do reply might be similar in certain ways, such as being retired or having more free
time. This makes it hard to apply the results to the general population – a problem with generalisability. Generalisability:
How well the findings apply to other people or settings.
4. Dishonest responses: Participants might lie to give a better impression of themselves – this is called social desirability bias.
Social desirability bias: When someone tries to give answers that they think will be more acceptable or impressive to
others, not necessarily truthful.

To reduce this, researchers often include filler questions. These are questions not related to the real aim of the study. They’re
included to hide the purpose of the questionnaire so participants won’t try to give specific answers. Filler questions: Irrelevant
questions used to distract participants from the true purpose of the study.
Interviews
An interview is when the researcher asks questions directly, often face-to-face, though it can also be done over the phone or
via chat.

Interviews can use both open and closed questions, but they tend to use more open questions.

Types of interviews:
1. Structured interview
• All participants are asked the same questions in the same order.
• The researcher may also follow instructions on how to behave (e.g. posture, tone of voice) to keep things consistent.
• This makes the interview standardised.

2. Unstructured interview
• The questions depend on what the participant says.
• The conversation is more flexible and informal.
• Different participants may be asked different questions, which makes it harder to compare data across people.

3. Semi-structured interview
• A combination of both styles.
• Some fixed questions are asked to every participant, allowing comparisons.
• The researcher can also ask extra questions based on the participant’s answers to explore their ideas in more depth.

Evaluating Interviews

Problems:
• Dishonest answers: As with questionnaires, participants may lie to seem more acceptable (social desirability bias) or
because they’ve guessed the study’s purpose.
• Time-consuming: Interviews take longer to conduct and analyse. This might limit who takes part, which could make the
results less representative.
• Subjectivity: Researchers might interpret answers based on their own beliefs or feelings, rather than staying neutral.
This reduces objectivity.

Subjectivity: When a researcher’s personal views affect how they interpret data.
Objectivity: When data is interpreted fairly, without being influenced by personal views or bias.

To improve objectivity, researchers may ask other trained researchers (who don’t know the aim of the study) to interpret the
responses.
Applying Your Knowledge of Self-Reports

You should be able to:

• Recognise whether a study is using a questionnaire or an interview.
• Choose the most suitable method for a new research situation.
• Identify and create open and closed questions.
• Recognise and evaluate structured, unstructured, and semi-structured interviews, as well as different techniques (e.g.
face-to-face or telephone).

Important considerations:
• Think about whether the method allows a wide range of people to take part.
• Consider how honest participants are likely to be.
• Remember that open questions give qualitative data, which is detailed and often more valid. Closed questions give
quantitative data, which is easier to summarise but may not fully capture someone’s true feelings or opinions.

Reliability is also important:

1. Questionnaires and structured interviews are often more reliable because they are carried out in the same way every time.
2. Open questions are less reliable because researchers have to interpret the answers, which can be influenced by personal
opinions.

Finally, you should know how to:

• Plan a self-report study using either questionnaires or interviews.
• Write open and closed questions for either method.
• Plan the format (paper or online) for a questionnaire.
• Plan and conduct structured, semi-structured, or unstructured interviews, either face-to-face or by phone.
 1.3 Case Studies

A case study is a research method where one specific example is investigated in detail. This is usually one individual, but it can
also be a small group such as a family or an institution. The goal is to collect in-depth information, often using different
research techniques.

For example, a case study might include:

• Interviews with the participant
• Observations of their behaviour
• Psychological tests
• Questionnaires

This method is especially helpful in unusual or rare cases, where detailed insight is needed. It’s also useful for studying
changes over time, such as how a child develops or how a person with a mental disorder improves or declines. Although some
case studies are used during therapy, when we talk about case studies as a research method, the focus is on gaining
knowledge—not on helping the person directly.

Different sources of information can also be used, such as:

• Interviews with the participant
• Interviews with family, friends, or co-workers
• Reviewing existing documents, like medical or school records

Evaluating Case Studies

Strengths (validity):
• Case studies are high in validity because they explore the person in great detail and within real-life settings like home,
school, or work.
• A method called triangulation can improve validity further. This is when the same findings are confirmed by using
different techniques. For example, if interviews, observations, and questionnaires all lead to the same conclusion, the result is
more likely to be valid.

Case studies often include many aspects of the person’s life, such as:
• Their past and present situation
• Their emotions, thoughts, social life, and behaviours

Case study – A research method that studies one person, family, or institution in depth.
Triangulation – Using multiple research techniques to study the same topic. If they agree, it strengthens the study’s validity.

This detailed information is helpful, but it also comes with risks:

1. Objectivity risk: The researcher may become too emotionally involved with the participant. This could lead to subjective
interpretations, meaning they are biased by personal feelings or opinions, which lowers validity.
2. Ethical concerns: The questions might become too personal or make the participant feel pressured to answer. Even if the
person’s name isn’t used, the amount of detail might make them recognisable, which could break confidentiality.
3. Reliability issues: Often there is just one participant and one researcher (or a small team). Another researcher might
interpret the same data differently, especially when the results are not straightforward. The findings may only apply to this
one unique case, meaning generalising to others is difficult.

Subjectivity – When results are influenced by the researcher’s personal views or feelings.
Objectivity – When results are unbiased and independent of the researcher’s personal views.
Generalisability – How well the findings apply to other people or situations.
Applying Case Studies to New Research Situations

You should be able to:

• Identify when a case study is being used
• Know when it is suitable to use one
• Suggest methods that can be used within a case study

When deciding these things, always consider:

• How valid and reliable the findings will be
• Whether triangulation can be used (for example, collecting data through observation, interview, and questionnaire, and
comparing the results)

Ethical planning is also important:

• Make sure the participant understands what they are agreeing to and gives informed consent
• Take care to protect their privacy and maintain confidentiality

Planning a Case Study

You should also be able to plan a case study by:

• Choosing a suitable participant (or group like a family or institution)
• Selecting a range of methods (like interviews, tests, and observations) that allow for triangulation
• Explaining how you will analyse and interpret the results once the data is collected
 1.4 Observations

Observations involve watching people or animals to study their behaviour. There are several ways to carry out an observation,
and it’s important to understand these when designing or evaluating a study.

Setting of the Observation

One key decision is where the observation happens:
• Naturalistic observation: This takes place in the participant’s usual environment. The researcher does not interfere in any
way with the setting or the people involved.
• Controlled observation: This takes place in an environment that has been changed or set up by the researcher. Changes
could involve:
• The social environment (e.g. changing group size or adding a person to the group),
• The physical environment (e.g. adding toys, giving different foods, or changing the location).

Controlled observations can happen either in a natural environment or in a lab.

Key Terms
• Naturalistic observation: Watching participants in their normal environment without interfering.
• Controlled observation: Watching participants in an environment that has been altered in some way by the researcher.

Structured vs Unstructured Observations

At first, a researcher might watch all types of behaviour without a clear focus. This is often used in pilot studies to decide
which behaviours to track later.

1. Unstructured observation: The researcher records all behaviours seen. This is usually only done at the start of a study to
help identify what behaviours to focus on later.
2. Structured observation: The researcher focuses only on a set list of specific behaviours. These are called behavioural
categories.

Behavioural categories are:

• Clearly defined actions that can be observed directly (operationalised).
• Used to split a continuous flow of activity into separate events that can be recorded.
• Must not include assumptions about thoughts or emotions—only visible behaviour is recorded.

Structured observations improve inter-observer reliability (i.e. how consistent different observers are with each other), as
everyone is looking for the same things.

• Unstructured observation: Recording all behaviours without focusing on specific ones.

• Structured observation: Recording only pre-defined behaviours.
• Behavioural categories: Clearly defined, observable actions recorded during a structured observation.
• Inter-observer reliability: How similar different observers’ records are when they watch the same thing.
Role of the Observer

Observers can either be part of the group being studied or stay separate:
• Participant observer: The researcher joins the social setting, e.g. talks with adults or plays with children.
• Non-participant observer: The researcher watches without joining in, e.g. sitting apart or using one-way glass.

The observer’s presence may also be:

• Overt: Participants know they’re being watched.
• Covert: Participants do not know they’re being observed. This can involve:
• A participant observer pretending to be part of the group,
• A non-participant observer watching from a hidden place or through CCTV.

Key Terms
• Participant observer: Researcher joins in with the group.
• Non-participant observer: Researcher does not get involved in the group.
• Overt observer: Participants are aware they are being observed.
• Covert observer: Participants do not know they are being observed.

Strengths and Weaknesses of Observations

• Naturalistic observations show realistic behaviour, as people are in their normal environment. But there’s no guarantee the
behaviours researchers want to study will happen naturally, which is why controlled observations may be needed.
• Unstructured observations allow for unexpected behaviours to be noticed and recorded. But they can be difficult to
manage, and a lot of the behaviours recorded may not be useful. That’s why structured observations, which focus on specific
behaviours, often produce more reliable data.
• The observer’s role and whether participants know they’re being observed affect validity:
• If participants don’t know they’re being observed (covert), their behaviour is likely to be natural and valid.
• But covert observation can raise ethical issues, such as lack of informed consent and possible distress if the participants
discover they were being studied.
• If the observer is overt, participants may act differently because they know they are being watched. This reduces
validity.

Using Observations in Different Research Methods

Observations can be:
1. Used as a research method on their own – the study involves only watching and recording behaviour.
2. Used as a technique within another method:
• In experiments, observation may be used to measure the dependent variable.
• In correlations, it may be used to measure one or both variables.
• In case studies, observations can be combined with other methods like interviews or tests.

You should be able to:

• Decide when to use observations as a method or technique.
• Choose between naturalistic and controlled settings.
• Decide whether the observation should be structured or unstructured.
• Choose if the observer should be participant or non-participant, covert or overt.
• Suggest practical ways to set up each type of observation (e.g. how to join a group as a participant observer, or how to
observe covertly).

You also need to plan:

• Where the observation will take place (naturalistic or controlled setting),
• What behaviours will be recorded (structured observation),
• How the observer will interact with the group (participant or non-participant),
• How to stay hidden if the observation is covert.
 1.5 Correlations

A correlational analysis is a research method used to look for a relationship between two co-variables (measured variables). This
method is useful when you can only measure variables but cannot manipulate them – meaning you can’t do an experiment. This
could be because changing the variables is either not practical or unethical.

For example, it’s not ethical or practical to run an experiment where you control children’s long-term exposure to television or
purposely increase their exposure to violent content. But you can measure how much violent TV a group of children watches and
see if that relates to how aggressive they are at school. This would be a correlation.
• Co-variables: the two variables being measured in a correlation.
• Correlation: a research method that finds a relationship between two measured variables.
• Causal relationship: when one variable directly causes a change in the other (only possible to establish through an
experiment).

To carry out a correlation, each variable must be measured across a range (called continuous data) and must be expressed in
numbers. Suitable data could include:
• time durations
• total counts (tallies)
• scores from rating scales or tests

You can collect correlational data using methods like self-reports, observations, or tests/tasks.

It’s important to understand that a correlation does not prove causation. Even if two variables change together, we can’t assume
that one caused the other. For example:

If a study shows a correlation between paying attention in class and test scores, it might seem like attention causes better
scores. But a third factor, like being a hard-working student, could be influencing both. The student might naturally pay more
attention and also study more.

Because of this, we don’t use terms like independent variable (IV) and dependent variable (DV) in correlations. Instead, we refer
to measured variables or co-variables. To establish a cause-and-effect relationship, an experiment must be done – where you
manipulate the IV to see the effect on the DV.

If a correlation shows no link between two variables, we can conclude that there’s no causal relationship either.
Types of Correlation:
• Positive correlation: As one variable increases, the other increases too.
• Negative correlation: As one variable increases, the other decreases.

Hypotheses in Correlational Studies

Like in experiments, correlational studies need hypotheses that clearly define the two co-variables (operationalisation).
• Vague: “There will be a correlation between amount of sleep and emotional reactivity.”
• Better: “There will be a correlation between the number of hours a person sleeps and how loudly they cry during a sad
film.”

Let’s apply this to an idea from earlier in the chapter – whether playing computer games affects A-Level performance.
1. Non-directional hypothesis: “There will be a correlation between the number of computer games played and final A-Level
grade.” (This doesn’t say which way the correlation will go.)
2. Directional hypothesis: “There will be a negative correlation between the number of computer games played and A-Level
grades.” (Assumes more games = worse grades.) or As the number of games played increases, A-Level grades increase.”
(Assumes game-playing might improve learning via tech engagement.)

Note: In correlational hypotheses, you must not claim that one variable causes the other.

Evaluating Correlations
For a correlation to be valid, both variables must be clearly defined and must accurately measure what they are supposed to.
• Validity depends on how well the variables match the relationship being studied.
• Reliability depends on whether the measurements are consistent.
• Using scientific measures (like time or volume) often gives high reliability.
• Using self-reports or observations can lower reliability, since they’re more subjective.

The main limitation of correlations is that they cannot show causation – even if two variables are related, we can’t say one
causes the other.

When to Use Correlations

Correlations are a useful starting point in research. They help identify relationships that could be studied further with other
methods, like experiments.

They are also helpful when it is not possible or ethical to manipulate variables.

You need to be able to:

• Tell the difference between a correlation (two measured variables) and an experiment (one manipulated variable – IV –
and one measured – DV).
• Recognise positive and negative correlations.
• Decide when using a correlation is the most suitable method.
• Choose appropriate ways to measure the variables.
• Predict whether the study would show a positive or negative correlation.
• Know how to display the results using a scatter graph (covered in Section 1.9).

You should also be able to plan a correlational study. This means:

• Creating two co-variables that can be measured on a numerical scale.
• Selecting suitable methods of measurement for each co-variable, such as questionnaires, interviews, observations, or tests
that produce quantitative data.
 1.6 Longitudinal Studies

A longitudinal study follows the same group of participants over time, measuring one or more variables at different intervals—
weeks, months, years, or even decades. This type of study is used to track how individuals change over time, not just due to
age but also life experiences.

Longitudinal studies can also examine how certain experiences—like treatments or major life events—affect development.

This method is an alternative to cross-sectional studies, which compare different groups at a single point in time. For example, a
cross-sectional study on mindfulness might compare people aged 10, 20, 30, 40, and 50. While this shows age-related
differences, it can’t always separate effects of aging from generational differences (e.g. how people were raised or changes in
society).

Longitudinal studies avoid this problem by testing the same cohort—a group of participants selected at the same age or stage—
multiple times. A cohort might include people born in a certain year, those starting a treatment, or those experiencing a major
life event like pregnancy.

Types of Longitudinal Study Designs

• Some studies track how a single variable changes over time (e.g. changes in emotion or beliefs).
• Others track two or more variables to see if there’s a correlation (e.g. exposure to multiculturalism and levels of
prejudice).
• Some use an experimental design: researchers deliberately manipulate one variable and measure the effects over time.
In this case, the variable being changed is the independent variable, and what’s being measured is the dependent variable.

Usually, researchers measure the variable at the start (the baseline) and then again after certain time periods. This is like a
long-term repeated measures design, often called a quasi-experimental design. It’s useful for evaluating the impact of things
like education, health, or therapy interventions.

In this setup:
• The baseline is the pre-intervention measure.
• Later testing points are post-intervention, and there may be follow-up testing long after the intervention.

Key Terms
• Longitudinal study: follows the same group over time to track changes.
• Cross-sectional study: compares different groups at one point in time.
• Cohort: group of participants selected at the same age or stage.
• Longitudinal design: tests the same participants on two or more occasions over time.
• Situational variable: an environmental factor that might affect results (e.g. light or noise).
Evaluating Longitudinal Studies

Strengths:
1. They provide more valid results than cross-sectional studies, since they measure real changes in the same people over time.
2. Like repeated measures designs, they control for participant variables (differences between individuals).
3. They help eliminate confounding variables caused by comparing different groups.

A cross-sectional study looks at different groups of people at one point in time.

A longitudinal study follows the same group of people over a period of time — could be months, years, or even decades.

Sample attrition is when participants drop out of a study before it’s finished. Reasons for sample attrition include:
• Participants choosing to leave (e.g. older children withdrawing consent once given by parents)
• Boredom from repeated testing
• Moving and losing contact
• Life events (illness, prison, lack of time, death)

Sample attrition affects how representative the sample is. Over time, only certain types of people may stay—those who are
motivated, stable and healthy. This can reduce generalisability and affect validity. Also, participants might act differently just
because they’re part of the study.

Issues with Reliability

Reliability depends on consistent procedures at each testing point. But over time:
• New, improved measurement tools may be developed, making earlier data harder to compare.
• Researchers may leave and be replaced, so data collection may vary.
• If the same researchers stay, they may build relationships with participants, which could affect how participants respond
—a confounding variable.

Ethical Considerations
• If children are involved, both the child and parent/guardian must give informed consent.
• It’s harder to tell if a child wants to withdraw than it is with adults.
• Participants must be reminded regularly about their right to withdraw.
• Researchers must keep up-to-date contact details to reach participants over time—this creates a confidentiality issue.
Applying Longitudinal Studies to New Situations

Longitudinal studies are best for investigating development, ageing and other long-term life changes. These could include:
• Natural life events (e.g. illness recovery, divorce)
• Planned interventions (e.g. therapy or exposure to new groups)

You need to identify:

• What variables are being measured
• Whether variables are just being tracked over time or if there’s a manipulation causing change

In studies with a manipulation, time points usually include:

• Before, during, after, and follow-up

This makes it a longitudinal experimental design.

You should also be able to:

• Decide when a longitudinal study is more suitable than other methods
• Choose the right type of longitudinal study
• Suggest useful variables and ways to measure them (e.g. tests, interviews, observations)
• Plan how to collect, analyse, and interpret results
• Plan testing intervals, standardised procedures, and participant follow-up
 1.7 The Definition, Manipulation, Measurement and Control of Variables

Variables are factors that can change or be changed. In experiments, these include:
• The independent variable (IV) – what the researcher changes.
• The dependent variable (DV) – what is measured.
• Other variables – which may or may not be controlled.

In correlational studies, there are two co-variables that are measured to see if they are related.

In an experiment, researchers look for changes in the DV between two or more conditions (or levels) of the IV. These conditions
are created by the experimenter. For the study to be valid, the IV must be operationalised, meaning it’s defined in a clear and
measurable way. This ensures that the differences created in each condition represent what the researcher intends to study.

The aim of a study is the researcher’s intention—what they are trying to find out. The aim mentions the variables, but not
always clearly. Giving operational definitions to variables helps clarify how the aim will be tested.

• The DV also needs to be operationalised. We could measure it by counting how many false details participants remember,
or how convinced they are that those details are true.

Aim = the overall purpose or question the study is trying to answer.

Controlling Variables and Standardising Procedures

Controlling Variables
Controlling variables helps ensure that the study results are valid. In experiments, it is especially important to control
confounding variables—these are unwanted variables that can affect the DV in one condition of the IV but not others. They
confuse the results.

Confounding variables can cause two problems:

1. They might reduce or cancel out the effect of the IV, hiding a real result.
2. They might exaggerate the effect of the IV, suggesting a result that isn’t real.

These variables must be controlled. Other uncontrolled variables that affect all conditions randomly are less serious, but still
need attention. It can be hard to know in advance which variables might become a problem.

One way to find and control these variables is by running a pilot study—a small-scale version of the main study. This helps
identify and fix problems with uncontrolled variables before the real study begins.
Standardisation

Standardisation ensures that all participants are treated the same way, helping to maintain fairness and accuracy in the study.

One way to standardise a study is to give all participants the same standardised instructions—either written or spoken—before
and during the study. This helps ensure that differences in behaviour are due to the IV, not how the study was explained.

Example:
In a questionnaire on attitudes to helping, all participants should receive the same instructions about how to fill it out. This
ensures that any effects from social desirability—the pressure to give socially acceptable answers—are the same for everyone.

Standardised instructions = the same written or spoken directions for all participants to make the study consistent and fair.

The procedure of the study also needs to be standardised. This includes using the same equipment or tests in the same way
across all participants.

In the helping behaviour questionnaire, all questions should focus on helping—not on unrelated traits like friendliness. In lab
studies, this is easier because the equipment (e.g. stopwatches or brain scanners) is consistent. However, some results (like brain
scan images) may still need interpretation, and that should also be done in a standardised way.

Applying Your Knowledge

When designing experiments, you must:
• Decide how to operationalise the IV (to ensure the conditions are valid).
• Decide how to measure the DV consistently (for reliability).
• Choose appropriate controls and describe how to implement them.
 1.8 Sampling of Participants

A population is a group of people (or animals) who share one or more characteristics. For example, a country’s population
includes all its residents. Other populations could include all internet users, all football fans, or all left-handed people.

A sample is the group of people chosen from that population to take part in a study. A good sample should represent the
population well, so the results can be generalised (applied to the wider group).

Important sample features like age, ethnicity and gender are often recorded because they can affect behaviour. Other useful
characteristics might include socio-economic status, education level, employment, location, or occupation.

Sample size also matters. Small samples are usually less reliable and less representative because they may not include all the
variety found in the whole population.

Different sampling techniques can affect how representative the sample is. The more representative the sample, the more
confidently we can apply the results to the larger population.

Opportunity Sampling
Opportunity sampling means choosing people who are available at the time, like students at a university where the research is
happening.
• This method is quick and easy, which is why it’s often used.
• However, it usually doesn’t represent the population well, because people who are easily available are often similar in
some way.

Example: Using university students gives a sample that is mostly young and better educated than average, which might affect
the results. But in some studies, age and education may not matter much, so this method is still useful.

Key Terms
• Population: The larger group from which a sample is drawn.
• Sample: The participants selected for the study.
• Sampling technique: The method used to choose participants.
• Opportunity sample: Participants chosen because they are easily available.

Volunteer (Self-Selected) Sampling

In volunteer sampling, the researcher invites people to take part, for example by posting an advert or online message. Those
who respond become the sample.
• Volunteer (self-selected) sample: Participants respond to an invitation to take part in the study.
• Volunteers choose to join the study, so this method may also produce biased samples.
• Volunteers often share characteristics such as having more free time or being better educated.
• Still, this method is useful when you need participants with specific traits, such as in studies on rare conditions.
Random Sampling
Random sampling gives every person in the population an equal chance of being chosen. This makes it more likely that the
sample will be representative.
• Opportunity and volunteer samples often include similar types of people, which makes them biased.
• Random sampling helps avoid this problem.

Example:
If you only advertise for volunteers in the library, your sample may favour hard-working students. If you sample from a
common room, you might get more relaxed students. To avoid bias, a better method is to list all students, give them numbers,
and select numbers at random.

If the population is small (like one class), you can draw numbers from a hat to choose your sample.

Key Term
• Random sample: Everyone in the population has an equal chance of being selected, usually through a random method like
a number generator.

Applying Sampling Knowledge to Research

Choosing the right sample is an important early step in research. How well the findings can be generalised depends partly on
how representative the sample is.

In reality, practical issues (like time or access) often make random sampling difficult. Still, for many studies, some bias is
acceptable if we believe the behaviour being studied is generally similar across people. But it’s wrong to assume all groups
behave the same way.

Some areas of psychology—like cross-cultural research, individual differences, and developmental psychology—specifically study
how behaviour varies across populations. This is why recognising the limits of a sampling technique is important.

You should be able to spot when differences between participants might matter in a study.

Example:
Two researchers at different universities want to study obedience. One samples people from a nearby police college. The other
samples from a nearby hospital. Both use opportunity samples with the same age and gender mix. Even so, the difference in
jobs could affect results—police officers might be more obedient than nurses.

You need to:

• Explain how each sampling technique works.
• Justify why you might choose (or avoid) a certain method.
• Identify when a sample is or isn’t likely to be generalisable.
• Recognise when differences between participants might affect the findings.
 1.9 Data and Data Analysis

Psychologists often collect numerical results from their studies, called raw data. Since large sets of numbers can be hard to
understand, the data is often simplified and shown visually using graphs. This makes it easier to interpret the findings. In this
section, we’ll look at techniques used to analyse data. You should be confident in counting scores, finding the mode or range,
making comparisons, and interpreting data from tables or graphs.

Types of Data

Different research methods in psychology produce different types of data. The two main types are:
• Quantitative data – numerical results that show the amount or quantity of something (e.g., pulse rate, test scores).
• Qualitative data – descriptive and detailed information about psychological characteristics (e.g., answers to open-ended
questions or case study observations).

Key Definitions:
• Quantitative data: Numerical results about the amount of a psychological measure, like a test score or heart rate.
• Qualitative data: Descriptive, in-depth information that shows the quality of a psychological characteristic, like responses
to open questions or detailed observations.

Quantitative Data
Quantitative data shows the amount of something being measured—often as totals, frequencies, or scores. These are usually
measured on scales, like time or ratings, or as test scores for things like intelligence or personality.

Quantitative data is common in experiments and correlational studies but can also be collected from observations,
questionnaires, or interviews. For example:
• Counting how often a behaviour occurs.
• Adding up the number of responses to a closed question.

Because this kind of data uses fixed scales or clear categories, it’s usually objective and easy to interpret. This makes it high in
validity and reliability.

Qualitative Data
Qualitative data gives more detailed, descriptive information. It includes:
• Observer notes,
• Open-ended questionnaire or interview responses,
• Case studies.

While this type of data can be harder to analyse (since it’s more open to interpretation), it allows participants to express
themselves more fully. This can increase validity, even though the analysis might be more subjective.
Descriptive Statistics

Descriptive statistics help summarise and understand data collected in psychology studies. Raw data (the original, unprocessed
results) needs to be simplified, especially when there’s a lot of it. Here we focus on basic descriptive statistics—not more
advanced methods like inferential statistics.

The first step is usually to create a summary table. This could show:
• Totals from an observation tally,
• Percentages from a questionnaire,
• Averages and data spread.

Tips for tables:

• Always include a title.
• Label all rows and columns clearly.
• For experiments, show DV measures and IV levels.
• If units are used (like seconds or volts), include them once in the heading—not in every table cell.

You can then compare:

• Across columns
• Across rows

Measures of Central Tendency

To summarise a set of quantitative results, we use measures of central tendency—a way of finding a “typical” or average score.
These include:
• Mode – the most frequent score.
• Median – the middle score when data is arranged in order.
• Mean – the average (sum of all scores divided by the number of scores).

Key Definitions:
• Measure of central tendency: A mathematical way to find a typical score in a dataset.
• Mode: The most frequent score.
• Median: The middle score when values are ranked.
• Mean: The average value.

The Mode
The mode is the score that appears most often. It can be used with numerical data or categories (like subject preferences). It’s
the only average that works with categorical data.

However, the mode doesn’t consider the value of scores—just their frequency—so it’s less informative than the median or mean.
• If two or more values occur equally often, there can be multiple modes.

Example:
A survey asked students for their favourite subject:
• Maths: 4
• English: 6
• Psychology: 10
So the mode is Psychology.

Another question asked: “On which day do you do the most homework?” If:
• Boys mostly answered “Sunday”,
• Girls were split between “Friday” and “Sunday” (6 each),
Then Sunday is the mode for boys, while girls have two modes.

Overall, Sunday is the mode when combining both groups (17 students total chose it).
The Median
The median is used for numerical data on a scale. To find it:
1. List all values in order.
2. Find the middle one.
• If there’s an even number of values, add the two middle ones and divide by 2.

The median reflects the order of values, so it’s more informative than the mode. It’s also not affected by outliers (extreme
scores), but that means it may not fully represent the whole dataset.

Example:
Students rated how hard they work (1 to 10).

1, 2, 2, 3, 3, 3, 5, 5, 5, 6, 6, 6, 6, 7, 8, 8, 8, 9, 9, 10
111111IIIIIIIIIIII (As students)
Median = (6 + 6)/2 = 6

or

4, 6, 6, 7, 7, 7, 7, 7, 7, 7, 9, 9, 9, 9, 9, 10, 10, 10, 10, 10

111111111111111/ (A level students)
Median = (7 + 9)/2 = 8

This shows A-Level students believe they work harder than AS students.

The Mean
The mean (average) is the total of all scores divided by the number of scores. It can only be used for numerical data on a
scale.

The mean considers every score’s value, making it more detailed than the median or mode—but it can be affected by extreme
values.

Using the same data:

• AS scores added = 112
Mean = 112 / 20 = 5.6
• A Level scores added = 160
Mean = 160 / 20 = 8

This also shows that A-Level students believe they work harder than AS students.
Measures of Spread

Measures of spread show how much the scores vary—are they close together or far apart? Two sets can have the same average
but differ in how spread out the scores are. The two main types are:
• Range – the difference between the largest and smallest score, plus 1.
• Standard deviation – the average difference between each score and the mean.

Key Definitions:
• Measure of spread: A way to describe how varied the scores are in a dataset.
• Range: Largest score minus smallest score, plus one.
• Standard deviation: The average distance of each score from the mean. A higher value means greater variation.

The Range
To calculate the range:
1. Find the largest and smallest scores.
2. Subtract the smallest from the largest, then add 1.

Psychologists add 1 to account for the scale between points. For example, a score of 3 might mean a range from 2.5 to 3.5, so
the actual range is slightly wider than just 3 to 6.

Example:

AS students (smallest = 1, largest = 10):

Range = 10 - 1 + 1 = 10

A Level students (smallest = 4, largest = 10):

Range = 10 - 4 + 1 = 7

This means AS students have more varied opinions about how hard they work, while A-Level students are more consistent in
their responses.

One limitation of the range is that it can be affected by outliers. For example, if one A-Level student had scored 1 instead of 4,
the range would become 10, but the mean would hardly change. This shows the range may not reflect how typical those
extreme values are.
Graphs and Charts – Histograms and Bar Charts

Histograms and bar charts are both used to show data, but they are used for different types of data.

Bar Charts
• Used for: Discrete data (data that fits into separate categories, like “red,” “blue,” “yes,” “no,” or “1,” “2,” “3”).
• Bars are separated: Each bar has space between it. This shows that the categories are different and not continuous.
• Each bar has a label: Labels are written underneath each bar to show what category it represents.
• Height of the bar: Shows how many times (frequency) that category appeared.
• Example: A bar chart showing how many people prefer different fruits — “apple,” “banana,” “orange,” etc.

Histograms
• Used for: Continuous data (data that is measured and can take any value in a range, like height, weight, or time).
• No gaps between bars: The bars are joined together because the data values are continuous and flow into each other.
• X-axis shows ranges (classes): Instead of labels like “apple” or “banana,” the x-axis shows intervals like “150–155 cm”,
“155–160 cm”, etc.
• Y-axis shows frequency: This shows how many values fall into each range.
• Bar width: It represents the size of the class interval.
• Bar height (or area): In most histograms, the height of the bar shows the frequency, but if the class widths are not
equal, then the area of the bar shows the frequency.
• Example: A histogram showing the number of students in different height ranges.
Scatter Graphs

Scatter graphs are used to show the relationship between two numerical variables. They help identify if changes in one variable
are related to changes in another.

Key Features of Scatter Graphs:

• Each point shows one pair of values: For example, a person’s height and their shoe size.
• X-axis and Y-axis: Each axis represents a different variable.
• X-axis: The independent variable (the one you control or change).
• Y-axis: The dependent variable (the one you measure or observe).
• Points are plotted but not joined: Unlike line graphs, the dots are not connected.

Used for:
• Showing correlation (how one variable changes as the other changes).
• Useful in experiments or surveys where you want to see if two things are linked.

Types of Correlation:
1. Positive correlation:
• As one variable increases, the other also increases.
• Example: The more hours studied, the higher the test score.
2. Negative correlation:
• As one variable increases, the other decreases.
• Example: The more time spent on social media, the lower the exam marks.
3. No correlation:
• No clear pattern; the variables don’t seem related.
 1.10 Ethical Considerations

Psychologists must follow ethical guidelines when doing research. These rules help make sure that participants are treated
properly and that psychology is respected in society.

Ethical Issues
Research with humans or animals can raise concerns about their well-being. These are called ethical issues.

For example:
• A study about stress might cause emotional discomfort.
• The procedure might involve hiding the true aim of the study.
• The results could lead to negative effects on a group in society.

Ethical issues: problems in research that could harm participants or have a wider harmful effect on society.

To manage these issues, countries often have organisations that create codes of conduct. At universities, an ethical committee
usually checks whether a study is acceptable before it happens.

These codes provide ethical guidelines — clear advice on how to protect participants and the reputation of psychology. If
participants are treated badly or misled, they may not take part again, may distrust psychology, and may discourage others too.

Ethical guidelines: advice that helps psychologists protect participants and respect society’s values.

Guidelines for Research with Human Participants

Research should never expose participants to more risk than they face in everyday life. Harm could be:
• Psychological (e.g. embarrassment, stress, self-doubt).
• Physical (e.g. injections, risky behaviours).

Risks should be reduced during planning. Use experienced researchers, screen participants, and stop the study if unexpected
risks arise.

Valid Consent
People must give informed consent before participating — meaning they understand what they’re agreeing to.

Sometimes the study’s aim is hidden to avoid demand characteristics (when people change their behaviour to match what they
think is expected). In such cases:
• Participants should still be told what will happen and any risks.
• This lets them decide whether to take part, even if the full aim isn’t revealed.

Consent must be:

• Informed: participants are given enough details.
• Freely given: they are not pressured.
• By competent individuals: they must understand and be capable of deciding.

Extra care is needed with people who:

• Have mental health problems or learning difficulties.
• Struggle with reading, speaking the research language, or have memory issues.
• Are children (consent should also come from parents/guardians, and the child should agree in a child-friendly way).
• Might be under pressure (e.g. prisoners).
In naturalistic observations or field experiments, getting consent may not be possible. In these cases, presumptive consent is
used — people similar to the real participants are asked if they would find the study acceptable.

Presumptive consent: asking a similar group if they would agree to the study, and assuming real participants would feel the
same.

If consent wasn’t fully informed, a debrief must happen afterward. Debriefing: explaining the study in full after it ends so
participants leave in a good state.

Right to Withdraw
Participants must be told they can leave the study at any time. This is their right to withdraw.
• Any rewards promised for joining must still be given even if they leave.
• Researchers must not pressure participants to stay.
• Participants should be reminded of this right, especially if the researcher is in a position of authority.

Lack of Deception
Deliberately misleading participants (deception) should be avoided. If deception is necessary (e.g. to avoid demand
characteristics):
• Participants must be told the true aim as soon as possible.
• They must be given the option to withdraw their data.
• A proper debrief should follow.

Confidentiality
Personal information and data must be kept private:
• Names should not be linked to data unless consent is given.
• Data must be stored securely.
• Participants can be given numbers to match results across conditions in repeated measures.

If a case study or field experiment involves schools, hospitals, or famous individuals, their identity must also be protected (e.g.
by using initials — though this may not always be enough).

Information may only be revealed if:

• Participants give permission.
• Someone’s safety is at risk.

Privacy
Participants must not be put in situations that feel like an invasion of their personal or emotional space.

For example:
• In questionnaires or interviews, they should be allowed to skip questions.
• In lab settings, they should have private space.
• Observations should only happen in public places where people expect to be seen.

Debriefing
Anyone who knows they were in a study must be:
• Thanked,
• Given a chance to ask questions,
• Told the real aim of the study.

Participants must also have the chance to withdraw their data. If the study upset them in any way, the researcher must return
them to their original condition.

Debriefing is helpful but should not be used as a substitute for designing an ethical study in the first place.
Guidelines for Research with Animals
Animals are often used in psychology because:
• They can help us understand processes like learning.
• They can be used in procedures that would be unethical on humans.
• Their behaviour can be interesting in its own right.

Psychologists must follow the law and make sure that:

• Animal suffering is kept to a minimum.
• Veterinary advice is sought when in doubt.
• The benefits of the research must outweigh the harm.

Bateson’s cube helps weigh up:

• Level of animal suffering,
• Certainty of benefit,
• Quality of research.

When benefits are high and suffering is low, the research may be justified.

Replacement
Researchers should consider alternatives to using animals, like:
• Previous study recordings,
• Computer simulations.

Species
Choose species that are:
• Least likely to suffer.
• Bred in captivity (not taken from the wild).
• Not highly sentient (able to think and feel).

Number of Animals
Use the smallest number of animals needed for valid and reliable results. Use:
• Pilot studies,
• Good design,
• Accurate measurement,
• Proper data analysis.

Procedures
Animals will be affected by research. Still, their experiences should be as normal and positive as possible.
Pain, Suffering, and Distress
Avoid anything that causes death, pain, disease, or stress. Try to:
• Use methods that improve animals’ experience,
• Study natural situations (e.g. stress from environment).
Daily care and vet needs must be met. The suffering must be justified by scientific benefit (see Bateson’s cube).
Housing
Animals should be housed in a way that:
• Matches their social behaviour (social vs. solitary species),
• Prevents overcrowding, which can cause stress or aggression,
• Takes into account age and sex (which affect stress levels),
• Allows freedom to move, eat, and drink.

Only the important parts of the natural environment need to be recreated. Clean cages carefully — cleanliness shouldn’t cause
stress (e.g. from unfamiliar smells).

Reward, Deprivation, and Aversive Stimuli

Deprivation means taking away something important (e.g. food or water). If used:
• It must fit the animal’s natural habits (e.g. young animals need more food).
• Preferred food should be used instead, where possible.
• Aversive stimuli (unpleasant experiences) should be avoided if alternatives exist.

Applying and Evaluating Ethical Guidelines

All human and animal research must be ethical.

But sometimes, small risks are accepted if they’re outweighed by the benefit. Researchers must always:
• Justify the risk,
• Reduce the risk as much as possible,
• Consult others and follow ethical committee advice.

You must be able to:

• Identify ethical issues in real or hypothetical studies,
• Explain if and how guidelines like consent, protection, right to withdraw, deception, confidentiality, privacy, and debriefing
were followed or broken.

For animal research, a balance must be struck between potential suffering and possible scientific gain. This should be judged by
the importance of the research and the quality of the methods used.
 1.11 Evaluating Research: Methodological Issues

When evaluating research, it’s not just about ethics. You also need to assess whether the research is scientifically sound—this
means looking at methodological issues. The main concepts to focus on are reliability, validity, replicability, and generalisability.

Reliability
Reliability is about how consistent the results are. If the method of collecting data isn’t the same each time, the results might
vary—not because of what’s being studied, but because of the inconsistencies. This inconsistency is a reliability problem.
• Tools used to collect data affect reliability. For example, machines that record reaction times or pulse rates usually give
consistent results, so they are reliable.
• One way to test reliability is the test-retest method:
• The same test is given to the same group of participants on two separate occasions.
• Conditions (like time of day or day of the week) should be the same both times.
• If the results are similar (i.e. highly correlated), the test is reliable.
• If they’re not, the test may need to be redesigned.

Test-retest: Checks consistency over time. If scores from the two occasions match well, the test is reliable.

Replicability: Means the procedure can be repeated exactly. This is important when repeating a study to check if the results are
the same. It also allows researchers to test other variables using the same setup.

Another reliability issue comes from subjective interpretation, which happens when:
• Researchers interpret open-ended questionnaire or interview answers differently.
• Different interpretations = low inter-rater reliability.

If observers interpret behaviours in different ways, this leads to low inter-observer reliability.

To improve reliability in both cases, researchers can:

• Agree on clear definitions of what they are measuring (operational definitions).
• Go over examples together to ensure consistency.

Using standardised procedures also helps improve reliability. This means keeping the method the same across all participants:
• Same instructions, materials, and equipment.
• Researcher behaviour (like tone of voice or posture) should also be consistent across conditions.

Replicability
When researchers publish studies, they include detailed methods. This is so other psychologists can repeat the study—this is
called replication.
• Replication helps confirm whether the findings are true.
• If results can’t be repeated, people may lose trust in that research.
• It also affects the reputation of psychology overall.

To ensure replicability, the study must be planned carefully. The researchers should record details like:
• Variables (how they are measured and manipulated),
• Sample (who took part and how they were selected),
• Procedure (e.g. use of controls, group assignment, counterbalancing),
• Materials (what was used in the study).
Validity
Validity is about whether the research measures what it’s supposed to measure.
• If a test isn’t reliable, it can’t be valid. You need consistency to measure something accurately.
• Objectivity matters too. If a researcher interprets data subjectively, it lowers validity.

Face validity: A test has face validity if it seems to measure what it claims to.
Example: If you design a helping behaviour test where people are asked to help someone stuck in a bath full of spiders or
worms, people might not help—not because they’re unhelpful, but because they’re scared. This means the test lacks face
validity.

Demand characteristics can also reduce validity:

• If participants guess the aim of the study, their behaviour might change.
• For instance, in a study on false memories, if participants know what the study is about, they might try to guess which
memory is fake and either focus more on it or avoid mentioning it at all.
• To improve validity, researchers should hide the aim and remove any clues.

Ecological validity: This is about whether the results apply outside the specific research setting.
• A test done in a lab may not reflect real-life situations.
• But even research done at home may not apply to other real-world settings like work or hospitals.

Mundane realism: This is about how realistic the task is.

• A task with high mundane realism is similar to everyday activities.
• For example, measuring emotional responses to polar bears is less realistic than using common insects—more people
encounter insects, so the task is more valid.

Generalisability
Generalisability means how far the results apply to other people or situations.
• High ecological validity helps improve generalisability.
• The sample is another key factor:
• If the sample is too small or not diverse, it may not represent the population.
• This is more likely in opportunity or volunteer samples than in random samples.
Evaluating and Applying Methodological Concepts
When analysing or designing studies, ask yourself:
1. Are the measures reliable?
• Is the tool consistent?
• Is it being used the same way by everyone?
• Is the interpretation objective?
2. Is the study valid?
• Does it actually test what it claims to?
• Does the task seem realistic (face validity)?
• Does the setting reflect real life (ecological validity)?
• Could demand characteristics have influenced the results?
3. Are the findings generalisable?
• Do they apply to other settings or people?
• How broad is the sample?
• Is the task or setting realistic?
4. Could the research be replicated?
• What details would be needed to do so?
• Why is it important for others to be able to repeat it?

Improving Methodology
To improve research design, you can:
• Change the method
(e.g. use a field study instead of a lab study, or a questionnaire instead of an interview)
• Change the design
(e.g. use independent measures to reduce order effects, or repeated measures to avoid participant differences)
• Improve the sample
(e.g. opportunity sampling might give a bigger group, volunteer sampling can target specific people, and random sampling
improves generalisability)
• Improve the tool
(e.g. test for inter-rater or test-retest reliability, adjust the procedure to increase consistency)
• Refine the procedure
(e.g. reduce demand characteristics, make the task more realistic to raise validity)