Research Methods (Edexcel A-Level Psychology): Revision Notes

Research Methods

Research methods form the foundation of psychological investigation. This topic encompasses the various approaches psychologists use to study behaviour, the procedures they follow, and the techniques they employ to analyse collected data. Understanding these methods allows you to evaluate research critically and design your own investigations effectively.

Types of data

Psychologists collect different types of data depending on their research aims. Understanding which type to use is essential for designing appropriate investigations.

Qualitative data consists of non-numerical information, typically in the form of words, descriptions, or images. This type of data provides rich, detailed insights into experiences, opinions, and behaviours. For example, interview transcripts or diary entries would constitute qualitative data. The depth of information obtained is a key strength, allowing researchers to explore complex phenomena and unexpected themes. However, qualitative data can be difficult to analyse objectively, and findings may be influenced by researcher interpretation. The time-consuming nature of analysis and challenges in generalising results to broader populations are notable limitations.

Quantitative data comprises numerical information that can be counted or measured. Examples include test scores, reaction times, or the number of words recalled in a memory experiment. This type of data can be easily compared and statistically analysed, making it simpler to identify patterns and relationships. Quantitative data also allows for objective measurement and easier replication of studies. Nevertheless, it may oversimplify complex behaviours and fail to capture the full richness of human experience.

Sampling techniques

Selecting participants for psychological research requires systematic approaches to ensure the sample is appropriate for the study's aims. Different sampling methods have distinct advantages and limitations.

Random sampling involves selecting participants in a way that gives every member of the target population an equal chance of being chosen. This might be done by assigning numbers to everyone in the population and using a random number generator or pulling names from a container. The main strength of random sampling is that it minimises bias, as the researcher has no influence over who is selected. This increases the likelihood that the sample will be representative of the population, improving the generalisability of findings. However, this method can be time-consuming and impractical, particularly when the target population is large or difficult to access. Additionally, there is no guarantee that the sample will actually be representative, especially with smaller sample sizes.

Stratified sampling involves dividing the target population into subgroups (strata) based on relevant characteristics, such as age, gender, or ethnicity, and then selecting participants from each subgroup in proportion to their presence in the population. This method ensures that important characteristics are represented proportionally, making it more representative than random sampling. However, stratified sampling requires detailed knowledge of the population composition and can be complex and time-consuming to implement.

Experimental and research designs

The design of a study determines how participants are allocated to different conditions and how data is collected. Each design has specific applications and considerations.

Independent groups design involves using different participants in each experimental condition. Each participant only experiences one level of the independent variable. This design eliminates order effects (such as practice or fatigue) because participants only complete the task once. It also reduces the risk of demand characteristics, as participants are unaware of other conditions. However, individual differences between participants in different groups may confound results. This design also requires more participants than repeated measures, and there may be differences between groups that affect the dependent variable.

Repeated measures design uses the same participants in all experimental conditions. Each participant experiences every level of the independent variable. The main advantage is that participant variables are controlled, as the same people are tested in each condition. This design also requires fewer participants overall. However, order effects can be problematic: participants may perform better on later tasks due to practice, or worse due to fatigue or boredom. Demand characteristics may also be more likely, as participants experience all conditions and may guess the study's aims.

Matched pairs design involves matching participants in different groups on relevant characteristics (such as age, IQ, or personality traits) so that each person in one condition has an equivalent person in the other condition. This design controls for participant variables while avoiding order effects. However, it is time-consuming and difficult to match participants on all relevant variables, and perfect matching is practically impossible. This design also requires more participants than repeated measures.

Hypotheses

A hypothesis is a testable prediction about the relationship between variables or the outcome of a study. Different types of hypotheses serve different purposes in psychological research.

The null hypothesis predicts that there will be no significant effect or relationship between variables. Any difference or correlation observed will be due to chance alone. In experimental research, the null hypothesis typically states that the independent variable will have no effect on the dependent variable. Researchers test the null hypothesis to determine whether it can be rejected in favour of an alternative hypothesis.

An experimental hypothesis (also called an alternative hypothesis) predicts that there will be a significant effect or relationship between variables. This is what the researcher actually expects to find. The experimental hypothesis is accepted if the null hypothesis is rejected following statistical testing.

A directional hypothesis specifies the expected direction of the relationship or difference. For example, "participants will recall significantly more words in the organised list condition than in the random list condition." Directional hypotheses are appropriate when previous research or theory provides a clear basis for predicting the direction of an effect. These hypotheses are tested using one-tailed statistical tests.

A non-directional hypothesis predicts that there will be a difference or relationship but does not specify the direction. For example, "there will be a significant difference in the number of words recalled between the organised list condition and the random list condition." Non-directional hypotheses are used when there is insufficient evidence to predict the direction of an effect. These hypotheses are tested using two-tailed statistical tests.

Operationalisation is the process of defining variables in a way that is specific, measurable, and testable. This ensures that research can be replicated and that results can be interpreted objectively.

Questionnaires and interviews

Questionnaires and interviews are self-report methods that gather information directly from participants about their experiences, attitudes, or behaviours.

Questionnaires consist of written questions that participants answer in their own time. They can include different types of questions:

Open questions allow participants to respond in their own words, providing detailed, qualitative information. These questions generate rich data and may reveal unexpected insights, but responses can be difficult to analyse and compare objectively.

Closed questions provide predetermined response options, such as "yes/no" or multiple choice answers. These produce quantitative data that is easy to analyse and compare statistically. However, they may restrict participants' responses and fail to capture the full complexity of their views.

Ranked scale questions (such as Likert scales) ask participants to rate their agreement or frequency on a numerical scale, such as 1-5. These questions produce quantitative data while allowing some degree of nuance in responses.

Questionnaires can be structured (with predetermined questions in a fixed order), semi-structured (with core questions but flexibility to add probes), or relatively unstructured (with general topic areas but flexibility in specific questions asked).

Interviews involve direct interaction between researcher and participant, with questions asked verbally.

Structured interviews follow a fixed set of predetermined questions in a specific order. This approach ensures consistency across participants and produces quantitative data that is easy to analyse. However, it lacks flexibility and may not allow exploration of interesting points raised.

Semi-structured interviews use predetermined questions but allow the interviewer to probe further or add follow-up questions based on responses. This balances consistency with flexibility.

Unstructured interviews are more like guided conversations, with general topic areas but flexibility in the specific questions asked. These generate rich, detailed qualitative data but may lack consistency across participants and can be influenced by interviewer bias.

Experiments

Experiments are a key research method in psychology, allowing researchers to establish cause-and-effect relationships by manipulating variables under controlled conditions.

Laboratory experiments take place in controlled, artificial settings. The researcher manipulates the independent variable (IV) – the factor that is systematically changed – and measures the dependent variable (DV) – the factor that is measured to see if it has been affected. Laboratory experiments allow high levels of control over extraneous variables, making it easier to establish causation. They can be replicated precisely to test the reliability of findings. However, the artificial setting may lack ecological validity, meaning findings may not generalise to real-world situations. Participants may also respond to demand characteristics, altering their behaviour based on their perception of the study's purpose.

Field experiments are conducted in natural, real-world environments while the researcher still manipulates the IV. These experiments have higher ecological validity than laboratory experiments, as behaviour is studied in a realistic context. Participants may be unaware they are being studied, reducing demand characteristics. However, field experiments offer less control over extraneous variables, making it harder to establish clear cause-and-effect relationships. They can also raise ethical concerns if participants are unaware of their involvement. Replication may be difficult due to the natural setting.

Observations

Observational methods involve systematically watching and recording behaviour as it occurs naturally or in controlled settings.

Naturalistic observations take place in the participant's natural environment without researcher interference. These observations provide high ecological validity and allow study of behaviours that cannot be manipulated ethically or practically. However, extraneous variables cannot be controlled, cause-and-effect relationships cannot be established, and observer bias may influence what is recorded.

Controlled observations occur in settings where the researcher has some control over the environment and possibly the situation. These offer more control than naturalistic observations while retaining some ecological validity.

Observations can vary in participant awareness:

Covert observations involve observing people without their knowledge. This reduces demand characteristics and observer effects (changes in behaviour due to being watched), but raises ethical concerns regarding consent and privacy.

Overt observations involve observing people with their knowledge and consent. This is ethically preferable but may lead to demand characteristics as participants may alter their behaviour when aware they are being watched.

The observer's role also varies:

Participant observation involves the observer joining the group being studied. This provides detailed insights and understanding of the group's perspective but risks observer bias and loss of objectivity.

Non-participant observation involves the observer remaining separate from the group. This maintains objectivity but may limit understanding of the group's perspective and experiences.

Additional research methods and techniques

Beyond experiments and observations, psychologists employ various specialised research methods.

Twin studies compare the similarity of monozygotic (identical) twins, who share 100% of their DNA, with dizygotic (non-identical) twins, who share approximately 50% of their DNA. If identical twins are more similar than non-identical twins for a particular trait, this suggests genetic influence. These studies help researchers understand the relative contributions of nature (genetics) and nurture (environment) to psychological characteristics and behaviours.

Adoption studies examine individuals who were adopted in infancy, comparing them to their biological parents (with whom they share genes but not environment) and adoptive parents (with whom they share environment but not genes). If adopted individuals resemble their biological parents more closely, this suggests genetic influence; if they resemble adoptive parents more closely, this suggests environmental influence.

Animal experiments use non-human animals as research subjects. These allow researchers to conduct studies that would be unethical with humans and to control variables precisely in ways not possible with human participants. However, findings may not generalise to humans due to physiological and psychological differences between species. Animal experiments raise ethical concerns about the welfare and rights of animal subjects.

Case studies involve in-depth investigation of a single individual, small group, or event. Multiple research methods are typically used to gather detailed information, including interviews, observations, tests, and examination of records. Case studies provide rich, detailed data and allow study of rare phenomena or unique cases. However, findings cannot be generalised to other people or situations, and researcher bias may influence interpretation. The uniqueness of the case means the study cannot be replicated.

Content analysis is a research method for systematically analysing qualitative material, such as media content, written documents, or interview transcripts. Researchers develop coding categories and count how often particular themes or items appear. This allows qualitative material to be converted into quantitative data for statistical analysis.

Correlational research examines the relationship between two or more variables without manipulation. The strength and direction of relationships are measured using correlation coefficients. While correlations can identify relationships, they cannot establish causation: a correlation between two variables does not mean that one causes the other.

Longitudinal research involves studying the same participants over an extended period, sometimes many years. This design is valuable for studying developmental changes and long-term effects. However, longitudinal studies are time-consuming, expensive, and suffer from participant attrition (dropout) over time.

Cross-sectional research compares different groups of people at the same point in time, often different age groups. This design is quicker and cheaper than longitudinal research but cannot show developmental changes within individuals, only differences between groups.

Cross-cultural research compares behaviour, attitudes, or psychological characteristics across different cultures. This helps identify universal aspects of human psychology and culture-specific variations.

Meta-analysis is a statistical technique that combines and analyses the results of multiple studies investigating the same research question. This produces more reliable conclusions than individual studies and can identify overall patterns across research. However, meta-analyses may include studies of varying quality, and the combination of different methodologies can be problematic.

Control issues

Controlling extraneous variables is essential for establishing clear cause-and-effect relationships in research. Various factors can threaten the validity of research findings if not properly managed.

Experimenter effects occur when the researcher's expectations, behaviour, or characteristics unintentionally influence participants' behaviour or the study's results. This can include giving subtle cues about expected responses or treating participants in different conditions differently. Double-blind procedures, where neither participants nor researchers know which condition participants are in, help reduce experimenter effects.

Social desirability bias occurs when participants respond in ways they believe are socially acceptable or that present them in a favourable light, rather than responding truthfully. This is particularly problematic in self-report measures. Researchers can reduce social desirability bias by ensuring anonymity, using indirect questions, or employing deception about the study's true purpose.

Demand characteristics are cues in the research situation that may reveal the study's aims or expected behaviour to participants. When participants identify these cues, they may alter their behaviour to either help the researcher (please-U effect) or deliberately perform differently (screw-U effect). Single-blind procedures, where participants are unaware of the condition they are in or the study's aims, help reduce demand characteristics.

Participant variables are individual differences between participants that may affect the dependent variable, such as age, intelligence, personality, or prior experience. These are particularly problematic in independent groups designs. Researchers can control for participant variables through random allocation, matched pairs designs, or using repeated measures designs.

Situational variables are features of the research environment that may affect participants' behaviour, such as temperature, noise level, time of day, or lighting. Standardised procedures help control situational variables by ensuring all participants experience the same environmental conditions.

Careful operationalisation of variables is necessary to ensure they are defined precisely, measurably, and unambiguously. This allows replication and reduces subjective interpretation.

Descriptive statistics

Descriptive statistics summarise and describe data, making it easier to identify patterns and communicate findings.

Measures of central tendency indicate the typical or average value in a dataset:

The mean is calculated by adding all values together and dividing by the number of values:

$\text{Mean} = \frac{\sum x}{n}$

where $\sum x$ represents the sum of all values and $n$ is the number of values.

The mean uses all data points and is the most sensitive measure of central tendency. However, it can be distorted by extreme values (outliers).

The median is the middle value when data is arranged in order. If there is an even number of values, the median is the average of the two middle values. The median is not affected by extreme values, making it useful when data includes outliers. However, it does not use all values in the dataset and is less sensitive to changes in the data.

The mode is the most frequently occurring value in a dataset. A dataset can have more than one mode (bimodal or multimodal) or no mode if all values occur with equal frequency. The mode is the only appropriate measure of central tendency for nominal data (categories). However, it does not use all values in the dataset and may not be representative if several values occur with similar frequency.

Measures of dispersion indicate the spread or variability of data:

The range is calculated by subtracting the lowest value from the highest value:

$\text{Range} = \text{Highest value} - \text{Lowest value}$

The range is easy to calculate and provides a quick indication of spread. However, it is heavily influenced by extreme values and does not indicate how data is distributed between the highest and lowest values.

The standard deviation measures the average distance of values from the mean. A small standard deviation indicates that data points cluster closely around the mean, while a large standard deviation indicates greater spread. Standard deviation uses all values in the dataset and provides precise information about variability. However, it is more complex to calculate than the range and, like the mean, can be affected by extreme values.

Frequency tables organise data by showing how often each value or category occurs. These make patterns in data easier to identify.

Graphs and charts provide visual representations of data:

Bar charts display categorical data using separate bars for each category. The height of each bar represents the frequency or value. Bars should not touch, as they represent discrete categories.

Histograms display continuous data. Unlike bar charts, bars touch because the data is continuous rather than categorical. The area of each bar represents frequency.

Scatter diagrams display the relationship between two continuous variables by plotting pairs of values as points on a graph. The pattern of points indicates the strength and direction of any relationship. A positive correlation shows an upward trend, a negative correlation shows a downward trend, and no correlation shows no clear pattern.

When interpreting descriptive statistics, researchers must select appropriate measures for their data type and research question. Comparing different datasets requires consideration of both central tendency and dispersion to draw meaningful conclusions.

Inferential statistics

Inferential statistics allow researchers to determine whether patterns observed in sample data are likely to reflect genuine effects in the population or whether they could have occurred by chance.

Levels of measurement classify data types and determine which statistical tests are appropriate:

Nominal data consists of categories with no inherent order, such as eye colour or gender. Frequency counts show how many observations fall into each category.

Ordinal data consists of ranked categories with a meaningful order, but the intervals between ranks are not equal. Examples include ranking preferences or satisfaction ratings. While we know one value is greater than another, we cannot quantify the difference precisely.

Interval data consists of measurements on a scale with equal intervals between values but no true zero point. Temperature in Celsius is an example: the difference between 10°C and 20°C is the same as between 20°C and 30°C, but 0°C does not represent the absence of temperature.

Ratio data has equal intervals and a true zero point, meaning zero represents the complete absence of the measured variable. Examples include reaction time, number of errors, or height.

Directional and non-directional testing relates to hypothesis type:

A one-tailed test is used when the hypothesis is directional, predicting the specific direction of a difference or relationship. This test only examines one end of the distribution of possible results.

A two-tailed test is used when the hypothesis is non-directional, predicting a difference or relationship but not its direction. This test examines both ends of the distribution of possible results.

Critical value tables are used to determine whether an observed result is statistically significant. The researcher compares their observed (calculated) value with the critical value for their chosen significance level and sample size. Whether the observed value needs to be equal to or greater than the critical value, or equal to or less than it, depends on the specific test used.

Type I errors occur when the null hypothesis is true but the researcher incorrectly rejects it, concluding there is an effect when there is not (false positive). This is more likely when a lenient significance level is used.

Type II errors occur when the null hypothesis is false but the researcher incorrectly retains it, concluding there is no effect when there actually is one (false negative). This is more likely when a stringent significance level is used.

Observed and critical values are compared to determine statistical significance. The observed value is calculated from the sample data using the chosen statistical test. The critical value is found in statistical tables based on the significance level, whether the test is one- or two-tailed, and the sample size or degrees of freedom. The specific decision rule (whether observed must be greater than, or less than, critical) depends on the test used.

Probability is expressed as a value between 0 and 1, or as a percentage. In statistical testing, symbols indicate probability levels:

• p ≤ .05 means probability of 5% or less
• p > .05 means probability greater than 5%
• p ≤ .01 means probability of 1% or less

Understanding probability is essential for interpreting statistical conclusions and evaluating research claims.

Methodological issues

Evaluating research quality requires understanding key methodological concepts that affect the trustworthiness and applicability of findings.

Validity refers to whether a study measures what it claims to measure or whether the conclusions are legitimate:

Internal validity concerns whether the study actually measures what it intends to measure within the research situation itself. Poor operationalisation of variables, confounding variables, and demand characteristics all threaten internal validity.

External validity concerns whether findings can be generalised beyond the specific research situation:

• Ecological validity refers to whether findings generalise to real-world situations beyond the research setting. Laboratory experiments often have low ecological validity due to artificial conditions.
• Population validity refers to whether findings generalise to other groups beyond the sample studied.
• Temporal validity refers to whether findings generalise across time, remaining relevant in different time periods.

Predictive validity concerns whether a test or measure successfully predicts future behaviour or performance.

Generalisability refers to the extent to which findings from a study can be applied to other situations, populations, or time periods. Research with high generalisability has broad applicability. Factors that limit generalisability include biased samples, artificial laboratory settings, and cultural specificity of findings.

Objectivity refers to research that is free from researcher bias and personal interpretation. Objective measurements are observable and quantifiable, with different researchers able to obtain the same results. Quantitative data is typically more objective than qualitative data.

Subjectivity refers to research influenced by personal opinions, interpretations, or bias. Subjective measurements involve interpretation and judgment. Qualitative data is typically more subjective, though systematic analysis methods can increase objectivity.

Researcher bias occurs when the researcher's expectations, beliefs, or preferences influence the research process or findings. This can affect which behaviours are observed and recorded, how data is interpreted, or how participants are treated. Standardised procedures, operational definitions, and blind techniques help reduce researcher bias.

Credibility refers to whether research findings are believable and trustworthy. Credibility is enhanced by transparent methodology, appropriate research design, acknowledgement of limitations, and consistency with other research in the area.

Analysis of qualitative data

Qualitative data requires systematic analysis approaches to identify patterns and themes while maintaining rigour.

Thematic analysis is a method for identifying, analysing, and reporting patterns (themes) within qualitative data. The process typically involves:

1. Familiarisation with the data through repeated reading
2. Generating initial codes to label interesting features
3. Searching for themes by grouping codes into potential themes
4. Reviewing themes to ensure they work in relation to coded data and the entire dataset
5. Defining and naming themes
6. Producing the final analysis with examples from the data

Thematic analysis provides rich, detailed accounts of data and is flexible in terms of theoretical approach. However, the analysis is subjective and requires careful researcher judgment. Different researchers may identify different themes from the same data.

When analysing qualitative data, researchers must evaluate findings in terms of:

• Credibility (whether findings are believable)
• Transferability (whether findings transfer to other contexts)
• Dependability (consistency and reliability)
• Confirmability (whether findings can be confirmed by others)

Providing rich descriptions, examples from data, and transparency about the analytical process all enhance the quality of qualitative research.

Conventions of published psychological research

Psychological research follows standardised conventions for reporting and publication, ensuring findings are communicated clearly and can be evaluated critically.

Publishing psychological reports involves a peer review process. Researchers submit their report to an academic journal, where it is reviewed by other experts in the field (peer reviewers). These reviewers evaluate the quality of the research, identify strengths and limitations, and recommend whether the report should be published, revised, or rejected. This process helps ensure published research meets quality standards and contributes meaningfully to knowledge. However, peer review can be time-consuming, and biases in the review process may occur. The peer review system strengthens the credibility of published findings but is not infallible.

Ethical issues in research using humans

Psychological research involving human participants must adhere to ethical guidelines that protect participants' welfare and rights.

The BPS Code of Ethics and Conduct (2009) provides ethical principles for psychological research in the UK. Key ethical principles include:

Deception - Deliberately misleading participants or withholding information should be avoided wherever possible. Deception may only be justified when the research could not be conducted without it and when the scientific or social value outweighs the ethical concerns. When deception is used, participants must be fully debriefed afterwards.

Debriefing - After participation, researchers should explain the full purpose of the study and address any questions or concerns. Debriefing is particularly important when deception has been used or when the study may have caused distress. Participants should leave the research in the same state they entered it.

Confidentiality - Participants' data and identity should be protected. Personal information should be stored securely and anonymised where possible. Published reports should not allow identification of individual participants unless they have explicitly consented to being identified.

Privacy - Researchers should respect participants' privacy, particularly in observational research. Observation in public spaces where people expect to be seen by strangers is generally acceptable, but observation in situations where privacy is expected raises ethical concerns.

Ethical issues in research using animals

Animal research in psychology is governed by legal and ethical frameworks designed to protect animal welfare.

The Scientific Procedures Act 1986 (UK legislation) regulates research using animals. This Act requires that:

• Animal research must be licensed by the Home Office
• Researchers must demonstrate that the research has potential benefits that outweigh animal suffering
• The research could not be conducted using non-animal alternatives
• The minimum number of animals necessary are used
• Suffering is minimised through appropriate housing, care, and procedures

Home Office regulations require detailed proposals demonstrating scientific justification, ethical justification (why animal use is necessary and appropriate), and welfare provisions. Regular inspections ensure compliance with regulations.

Ethical arguments regarding animal research involve a tension between potential scientific and medical benefits and concerns about animal welfare and rights. Those supporting animal research argue it has led to important discoveries and treatments that benefit humans and animals. Critics argue that animals have rights not to be harmed, that animals experience suffering similarly to humans, and that results may not generalise reliably from animals to humans.

When evaluating animal research, consider the species used (as different species have different capacities for suffering), the severity of procedures, the potential benefits of the research, and whether alternatives could have been used. The ethical acceptability of animal research remains a subject of debate.

Remember!