★ Must-know
📌 A directional hypothesis predicts the direction of a difference or relationship, whereas a non-directional hypothesis predicts a difference or relationship without specifying its direction.
Further detail
Uncontrolled EVs weaken the establishment of an IV–DV cause-and-effect relationship and can reduce study reliability.
Participants may overperform to please the experimenter (the please-you effect) or deliberately underperform to sabotage results (the screw-you effect).
Uncontrolled influences blur the IV–DV relationship
📌 Independent groups avoid order effects but may have participant-variable differences between groups, which can weaken cause-and-effect conclusions; random allocation is used to address this imbalance.
Independent groups designs are less economical than repeated measures because each participant contributes once and twice as many participants are needed to produce the same amount of data.
Repeated measures control participant variables and require fewer participants, but order effects such as boredom or fatigue may worsen performance on the second task, and participants may infer the study’s aim.
📌 Matched pairs reduce participant-variable differences but cannot match participants exactly, and matching is time-consuming and expensive; because participants take part in only one condition, order effects and demand characteristics are less problematic.
★ Must-know
📌 Opportunity sampling selects people who are willing and available, whereas volunteer sampling uses people who select themselves to participate.
Further detail
Laboratory experiments provide control over EVs, support cause-and-effect conclusions, and allow replication, but may have low generalisability, external validity, and mundane realism because the setting or tasks can be artificial and behaviour may be unnatural.
Field experiments can produce more authentic behaviour and higher mundane realism and external validity, but reduced control over EVs makes cause and effect harder to establish and precise replication difficult.
Field experiments may raise ethical issues because participants may not give informed consent or know they are being studied, creating a risk of privacy invasion.
Natural experiments can enable research that is impractical or unethical to conduct otherwise and may have high external validity, but rare events limit research opportunities and generalisability, while non-random allocation can weaken cause-and-effect conclusions.
Quasi experiments can be conducted under controlled conditions, allow replication, and may have high external validity, but participants cannot be randomly allocated to conditions, so participant differences may confound results.
In natural experiments the IV occurs naturally; in quasi experiments it is an existing difference between people
★ Must-know
Participants have the right to withdraw from a study at any time if they wish.
Researchers must protect participants from physical or psychological harm and must not expose them to greater risk than they would face in daily life, including embarrassment, inadequacy, stress or pressure.
Privacy is participants’ right to control information about themselves, whereas confidentiality requires protecting personal data if privacy is invaded; anonymity can be maintained by recording no personal details and using numbers or initials.
Further detail
At the end of a study, debriefing reveals the true aims and withheld details, explains the use of participants’ data, offers the right to withhold data, reassures them about their behaviour, and may include counselling after extreme stress or embarrassment.
The BPS code of ethics provides guidelines researchers must observe, and ethics committees use a cost-benefit approach to assess whether research ideas are ethically acceptable.
📌 An investigation usually compares an experimental group or condition with a control group or condition, which provides a baseline for comparison; a greater behavioural change in the experimental group supports a cause-and-effect link between the IV and DV if confounding variables remain constant.
Single blind hides information from participants; double blind hides the aims from both participants and researcher.
★ Must-know
📌 Covert observation records people without their knowledge or consent, whereas overt observation records them with their knowledge and consent.
📌 Participant observation involves the researcher joining the group and producing a first-hand account, whereas non-participant observation keeps the researcher outside the group.
Further detail
Naturalistic observations are difficult to replicate because the researcher has little control over variables.
Uncontrolled extraneous variables make patterns of behaviour and cause-and-effect relationships difficult to establish in naturalistic observations.
Controlled observations are easier to replicate because extraneous variables may have less influence.
Controlled observations may be difficult to generalise to real life because an unnatural setting can produce demand characteristics.
📌 Covert observation must take place in public to be ethical.
Covert observation avoids participant reactivity because participants do not know they are being watched, so behaviour is natural and validity may increase.
Covert observation raises informed-consent concerns because people in public may not want their behaviour observed.
Participant observation can increase validity by giving researchers insight into the situation as participants experience it.
Participant observation risks researchers identifying too strongly with the group and losing objectivity, a process called going native.
Non-participant observation helps researchers maintain objective psychological distance and avoid going native, but may sacrifice insight gained through participation.
Naturalistic observation follows everyday settings; controlled observation structures the setting.
Structured observations use behavioural categories to make recording easier and more systematic, and produce quantitative data that is easier to analyse and compare.
Structured observations can produce reliable, replicable data.
Structured observations may lack contextual depth and can be susceptible to participant reactivity and demand characteristics.
Unstructured observations produce rich, detailed data but risk observer bias because objective behavioural categories are absent and researchers may record only conspicuous behaviours.
📌 Behavioural categories should be precisely defined, cover all ways the target behaviour may occur, avoid inference, and be mutually exclusive.
Clear behavioural categories make data collection more structured and objective and require no further interpretation, but ambiguous boundaries such as smiling versus grinning can reduce validity.
Event sampling is useful for infrequent behaviours that could be missed by time sampling, but complex events may lead observers to overlook important details.
Time sampling can reduce the chance of sampling observations that are unrepresentative of the observation as a whole, but may miss infrequent behaviours that occur outside the specified time frame.
Event sampling counts occurrences; time sampling observes within fixed intervals.
★ Must-know
Further detail
At least two researchers should conduct observations because a single observer may miss details or notice only events that confirm their opinions or hypothesis, introducing bias.
Designing an observational study involves deciding its aims and hypothesis, setting, observer status and involvement, sampling method, dependent-variable recording, observation structure, behavioural categories and checklist, ethical issues, and results presentation.
Train observers, pilot together, compare judgements, then calculate reliability.
★ Must-know
📌 Structured interviews are easy to replicate because their standardised format reduces differences between interviewers and the effect of interviewer bias, but their fixed questions limit elaboration and the detail obtained.
📌 Unstructured interviews allow follow-up questions and free elaboration, but their data are difficult to analyse, conclusions may be hard to establish, and respondents may give socially desirable answers; a skilled interviewer may build rapport to encourage truthful discussion of sensitive topics.
Further detail
Questionnaires are cost-effective because they can quickly gather data from large numbers of people.
Questionnaires reduce researcher bias and effort because the researcher need not be present while participants complete them.
Fixed-choice closed questions can produce quantitative data that is easy to analyse and can support statistical analysis and comparisons between groups using graphs and charts.
Questionnaire responses may be untruthful when respondents try to present themselves positively, producing social desirability bias as a demand characteristic.
Acquiescence bias occurs when respondents tend to answer in the same way, such as always ticking yes, possibly because they completed a questionnaire too quickly to read the questions properly.
Structured interviews standardise questions; unstructured interviews let the conversation flow.
★ Must-know
Further detail
Interviewers may take notes during an interview or record it for later analysis.
Interviews commonly involve one interviewer and one participant in a quiet room away from other people, while neutral opening questions can help the participant feel comfortable and build rapport, especially when topics are sensitive or personal.
Good interview questions avoid jargon unfamiliar to people outside a specific subject area.
Good interview questions replace emotive wording with neutral alternatives so the wording does not lead the respondent toward a particular answer.
★ Must-know
📌 A positive correlation means that as one co-variable increases, the other also increases; a negative correlation means that as one increases, the other decreases; and a zero correlation means there is no relationship between them.
📌 Experiments manipulate or control the IV to measure its effect on the DV, whereas correlations do not manipulate either co-variable and therefore cannot establish cause and effect, even when the correlation is strong.
Further detail
Correlations provide a precise, quantifiable measure of the strength and direction of a relationship, which may suggest further research when the variables are closely related.
Correlations can be quick and economical because they require neither a controlled environment nor variable manipulation, and use of data collected by others can make them less time-consuming than experiments.
Correlations cannot show why variables are related or which co-variable causes the other to change, and an untested third variable may explain their relationship.
Correlation shows association; an experiment manipulates an IV to test its effect on a DV.
★ Must-know
Further detail
📌 Quantitative data are simple to analyse, allow easy comparisons between groups, and are more objective and less open to bias, but their narrower scope may reduce ecological validity and make real-life generalisation difficult.
📌 Qualitative data provide rich detail and broader scope because participants can develop their views, and may offer greater external validity through meaningful insight, but they are difficult to analyse, cannot be summarised statistically, and rely on subjective interpretation that may be affected by researcher bias.
Primary data is authentic information obtained from participants specifically for an investigation, with questionnaires and interviews designed to target the researcher's required information.
Collecting primary data requires time and effort, and producing it through an experiment requires considerable planning, preparation and resources.
Secondary data can be found in:
Secondary data is quick, cheap and easy to access, requires minimal effort, and removes the need to conduct an experiment to collect primary data.
Quantitative data use numbers; qualitative data use words.
📐 Formula - The range is calculated as the highest score minus the lowest score, plus 1 as a mathematical correction: .
Standard deviation measures how much scores deviate from the mean; the variance is calculated from the differences between the mean and each score, and the standard deviation is the square root of the variance.
A bar chart represents the frequency of each variable by bar height and can display mean values across categories of discrete data.
📌 In a positive skew, the long tail is on the positive (right) side of the peak and most of the distribution is concentrated on the left; in a negative skew, the long tail is on the negative (left) side of the peak and most of the distribution is concentrated on the right.
★ Must-know
Further detail
📐 Formula - The percentage of participants is calculated as the number of participants divided by the total number of participants, multiplied by 100: .
To convert a percentage to a decimal, move the decimal point two places to the left, which is equivalent to dividing by 100.
Results can be estimated by calculating the mean or range of a data set or by rounding to a specified significant figure.
📌 In the Sign Test, if S is smaller than the critical value, the difference is significant, so the null hypothesis is rejected and the hypothesis is accepted.
★ Must-know
Peer review helps decide which research should receive funding, assess research quality and validity of methods, check that conclusions rest on solid evidence, recommend improvements and, in extreme cases, advise withdrawing unsuitable work from publication.
Publication bias occurs when journal editors favour significant, headline-grabbing positive findings, causing research without such results to be ignored and creating a misleading impression of findings.
Peer review can suppress research that challenges mainstream theories because reviewers may favour work supporting their own views, allowing established findings to pass more readily than innovative opposition and slowing change in a discipline.
Further detail
Peer review is described as supporting the validity and accuracy of research.
Reviewers remain anonymous to encourage honesty, but anonymity can be used to criticise rival researchers, particularly when researchers compete for limited funding.
Correlations and Experiments
| Feature | Correlation | Experiment |
|---|---|---|
| Variable handling | Measures co-variables without manipulation | Manipulates or controls the IV |
| Inference | Shows association, not causation | Measures the IV’s effect on the DV |
| Setting and effort | No controlled environment required; may use others’ data | Requires manipulation and control |
Take a 51-question practice quiz on Observational and Self-Report Methods, with every answer explained.
1. A researcher varies the amount of background noise and records participants’ reading accuracy. Which feature makes this an experimental method?
2. Which statement is a hypothesis for a study comparing conversation levels after caffeine or water?
Memorize the key concepts of Observational and Self-Report Methods with 98 flashcards.
What does the experimental method manipulate to measure an effect?
The independent variable.
What does a hypothesis predict about variables and study outcomes?
A testable relationship between variables and the expected outcome.
What does a directional hypothesis specify that a non-directional hypothesis does not?
The direction of a difference or relationship.
Upload your class notes and Revizly writes a study guide, a quiz and flashcards in about a minute.
Study guide maker