Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Chapter 1 — Research Methods

Cambridge AS & A Level Psychology (9990) — Full Detail Study Notes

Complete syllabus content, in coursebook order — every definition, example, number and evaluation
point, written as memorisable bullets instead of paragraphs.

KEY TERMS (full glossary)

Term Definition
Aim The intention of a study — the idea being tested, problem to be solved, or question
being answered
Experiment Investigation allowing researchers to look for a cause-and-effect (causal) relation-
ship
IV Independent Variable — factor manipulated to create 2+ conditions, expected to
cause change in the DV
DV Dependent Variable — factor measured, expected to change due to the IV
Uncontrolled variable Variable with random OR systematic effect on the DV, not deliberately manipulated
Confounding variable Uncontrolled variable acting systematically on ONE level of the IV — con-
fuses/hides/exaggerates the real effect
Experimental / Control A level of the IV being tested / the level where the IV itself is absent
condition
Experimental design The way participants are allocated to levels of the IV
Independent measures A different group of participants is used for each level of the IV
design
Repeated measures The same participants perform in every level of the IV
design
Matched pairs design Participants paired on relevant traits; one member of each pair per condition
Random allocation Equal chance of being placed in any condition — spreads individual differences
evenly
Demand characteris- Features giving away the study’s aim; participants change behaviour to match ex-
tics pectations
Order effects Practice or fatigue effects from doing a task more than once (repeated measures)
Practice effect Performance improves due to familiarity/learning the task
Fatigue effect Performance declines due to tiredness or boredom
Randomisation Random order of conditions allocated to each participant — fixes order effects
Counterbalancing ABBA design — half do A then B, half do B then A — fixes order effects
Participant variables Individual differences (age, personality, intelligence) that could distort results
Laboratory experiment IV + DV + strict controls, in an artificial (not the participant’s usual) setting
Field experiment IV manipulated in the participant’s NORMAL environment for that behaviour
Controls Ways of keeping confounding variables constant, raising validity
Standardisation Keeping the procedure identical for every participant, raising reliability
Reliability Consistency of a procedure/measure — would it produce the same results again?
Validity Extent to which the researcher is testing what they claim to be testing
Pilot study Small-scale preliminary test of the procedure before the main study
Replication Keeping procedure/materials exactly the same between studies to verify results
Operational definition Clear description of a variable so it can be manipulated, measured and replicated
Placebo Fake pill/procedure with no active ingredient, used for comparison
Generalise To apply findings more widely, e.g. to other settings and populations
Ecological validity Extent to which findings from one situation would generalise to other (real-life) situ-
ations
Hypothesis A testable statement based on the aims of an investigation
Alternative hypothesis The main testable prediction of a difference (experiment) or relationship (correlation)
Non-directional hypoth- Predicts an effect but not its direction (two-tailed)
esis
Directional hypothesis Predicts the direction of the effect/relationship (one-tailed)
Null hypothesis States any difference/correlation is due to chance
Informed consent Knowing enough about a study to decide whether to agree to participate

Page 1 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Term Definition
Right to withdraw A participant can remove themselves, and their data, from a study at any time
Protection from harm No greater physical/psychological risk than everyday life
Deception Deliberately misinforming participants — should be avoided/minimised
Privacy Participants’ emotional/physical space should not be invaded
Confidentiality Results/personal info kept safe, not released outside the study
Self-report Method obtaining data by asking participants to give info about themselves
Questionnaire Self-report using written questions (paper or online)
Closed question Fixed alternative responses — produces quantitative data
Open question Full, descriptive answers in participant’s own words — produces qualitative data
Inter-rater reliability Consistency between 2+ researchers interpreting qualitative responses
Social desirability bias Trying to present oneself in the best light rather than answer honestly
Filler questions Irrelevant items added to a questionnaire to disguise its real aim
Structured interview Fixed, scripted questions in fixed order for every participant
Unstructured interview Questions depend on the respondent’s previous answers
Semi-structured inter- Fixed core of open/closed Qs + interviewer can add more if needed
view
Subjectivity Interpretation biased by the researcher’s own feelings/beliefs
Objectivity Unbiased, external, consistent viewpoint independent of the researcher
Case study Detailed investigation of a single instance — person, family or institution
Triangulation Using different techniques on the same case to check consistency (validity)
Naturalistic observation Watching behaviour in the normal environment, no researcher interference
Controlled observation Watching behaviour where the social/physical environment has been manipulated
Unstructured observa- Recording the whole range of possible behaviours
tion
Structured observation Recording only a limited, pre-defined range of behaviours
Behavioural categories Operationalised, observable, discrete activities recorded in an observation
Overt observer Observer’s role is obvious to participants
Covert observer Observer’s role is hidden or disguised
Participant observer Researcher watches from within the social setting
Non-participant ob- Researcher does not become involved in the situation studied
server
Inter-observer reliability Consistency between two observers watching the same event
Correlation Method looking for a relationship between two measured co-variables
Co-variables The two measured variables in a correlation (NOT IV/DV)
Causal relationship A change in one variable is responsible for (causes) a change in another
Positive correlation Both co-variables increase together
Negative correlation One co-variable increases as the other decreases
Longitudinal study Follows the same group (cohort) of participants over time
Cross-sectional study Compares different age/stage groups at one point in time
Cohort A group of participants selected at the same age or stage
Situational variable A confounding variable that is an aspect of the environment
Sample attrition The loss of participants from a sample over time
Standardised instruc- The same written/verbal information given to all participants
tions
Population The group, sharing characteristics, from which a sample is drawn
Sample The group selected to represent the population
Opportunity sample Participants chosen because they are available
Volunteer (self- Participants who respond to an advert/request to take part
selected) sample
Random sample Every population member has an equal chance of being chosen
Quantitative data Numerical results about the amount/quantity of a measure
Qualitative data Descriptive, in-depth results about the quality of a characteristic
Mode The most frequent score(s) in a data set
Median The middle score of a ranked data set
Mean Sum of all scores divided by the number of scores
Range (Highest value - lowest value) + 1
Standard deviation Average distance of every score from the mean; bigger = more spread out
Bar chart Graph for discrete categories/totals — has gaps between bars

Page 2 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Term Definition
Histogram Graph for continuous data — bars touch (no gaps unless empty category)
Scatter graph Displays correlational data — each dot = one participant’s two scores
Ethical issues Problems raising concerns about the welfare of participants or society
Ethical guidelines Advice guiding psychologists to protect participant/society welfare
Presumptive consent Asking a similar (not actual) group whether a study would be acceptable
Debriefing Full explanation of aims/consequences given to participants at the END of a study
Bateson’s Cube 3-factor model (benefit, research quality, suffering) justifying animal research
Test-retest Using a measure twice; a high correlation between scores = high reliability
Replicability The extent to which a study’s procedure can be kept the same when repeated
Generalisability How widely findings apply, e.g. to other settings and populations

1.1 Experiments

DEFINITION: An experiment looks for a cause-and-effect relationship. The researcher manipulates

the IV to produce 2+ conditions/"levels" and measures the effect on the DV.

MEMORY TRICK: IV = "I Vary it" (researcher controls it). DV = "Data" you collect (what changes
because of the IV).

• Example: IV = brightness of lighting (bright/dull). DV = how well people pay attention.

• To be certain the IV caused the DV difference, the researcher must control OTHER variables that
might affect the DV — e.g. whether people recently ate, exercised, or sat through a dull class.

Uncontrolled & Confounding Variables

Uncontrolled variable: acts RANDOMLY (affects DV in all levels of the IV) OR SYSTEMATICALLY
on just ONE level (= confounding variable).

DON’T CONFUSE: Random = affects all conditions roughly equally (less serious, just noise). Con-
founding = affects only ONE condition (serious — hides OR exaggerates the IV’s real effect, making
results impossible to interpret).

• Minimise uncontrolled variables by keeping ALL factors except the IV the same in each condition.

Experimental condition: a level of the IV being tested. Control condition: the level where the IV
itself is absent (a baseline).

EXAMPLE: Chocolate & attention study — one bar vs two bars = 2 experimental conditions. One bar
vs no chocolate = 1 experimental + 1 control condition.

Experimental Design (3 types)

MEMORY TRICK: "I R Match" = Independent measures / Repeated measures / Matched pairs

1. Independent Measures Design

DIFFERENT group of participants for EACH level of the IV. Data is "independent" — not related to
any other data, comes from different people.

• Random allocation reduces individual-differences bias: give each participant a number, then ran-
domly split numbers into groups (hat, or random number generator).

Page 3 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

• Good because participants only encounter the setting ONCE — less likely to notice demand charac-
teristics.
• Problem: individual differences between groups could bias findings (e.g. more natural dream-rememberers
ending up by chance in the ’no noise’ group of a sleep study).

2. Repeated Measures Design

SAME group of participants does EVERY level of the IV. Think "repeating" their performance under
different conditions.

• Main advantage: each person is their own baseline — participant variables affect BOTH levels equally,
so are unlikely to bias findings.
• Participant variables = individual differences (age, gender, personality, intelligence) that could hide/exaggerate
differences between IV levels — a type of confounding variable.

DON’T CONFUSE: Practice effect = performance IMPROVES (familiarity/learning). Fatigue effect

= performance DECLINES (tiredness/boredom). Both = "order effects".

• Order effects happen because participants do the same/similar task 2+ times.

• Participants tested SECOND may do better (practice) or worse (fatigue/boredom) than those tested
first.
• Seeing BOTH levels also gives more chance to guess the aim → more demand characteristics.

MEMORY TRICK — fixing order effects: "R and C save the day" = Randomisation OR Counter-
balancing.

HOW EACH WORKS:

• Randomisation: music (M) vs no music (N) learning task. Each participant RANDOMLY allocated
to do M-then-N or N-then-M. Advantage of doing one condition first roughly evens out across the
sample by chance.
• Counterbalancing: group split into TWO — half do M-then-N, half do N-then-M (an ABBA de-
sign). If accidentally learning items from test 1 affects test 2, this is a problem for EXACTLY HALF
of each condition — balances out precisely rather than relying on chance.

3. Matched Pairs Design

Participants arranged into PAIRS matched on relevant variables (age, gender, intelligence, personal-
ity). One member of each pair per condition. Overcomes BOTH independent measures and repeated
measures problems.

• Identical twins = ideal matched pairs (genetically same, similar experiences).

• Example: violent computer game study — match participants on EXISTING aggression level.

Table 1.1 — Strengths & Weaknesses of Experimental Designs

Design Strengths Weaknesses

Independent Measures No order effects (different peo- Participant variables can distort re-
ple per condition). Less de- sults. Needs MORE participants
mand characteristics (see one level than repeated measures — less
only). Random allocation reduces ethical/effective if hard to recruit.
individual-differences bias.

Page 4 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Design Strengths Weaknesses

Repeated Measures Participant variables unlikely to dis- Order effects could distort results.
tort effect (each person does all lev- Greater exposure to demand char-
els). Counterbalancing fixes order acteristics (task seen more than
effects. Fewer participants needed. once).
Matched Pairs Only one level seen → less de- Similarity between pairs limited by
mand characteristics. Participant matching criteria chosen. Availabil-
variables less likely to distort effect ity of matches may be limited →
(matched). No order effects. small sample.

Types of Experiment

Laboratory experiment: IV + DV + strict controls, in a setting that is NOT the participant’s usual
environment.
Field experiment: IV manipulated, conducted in the participant’s NORMAL environment for that
behaviour; some control still possible.

EXAMPLE: Testing light levels on attention. Lab: children on a computerised task in a univer-
sity room. Field: altering number of classroom lights, measured via topic-test score (still has
IV/DV/controls, but is FIELD because it’s their usual environment/behaviour).

Table 1.2 — Strengths & Weaknesses of Experimental Methods

Laboratory Experiment Field Experiment

Strengths Good control raises validity. Causal Natural behaviour → representative
relationships determinable (only IV results. Less demand characteris-
should affect DV). Standardised tics if participants unaware they’re
procedures raise reliability and al- in a study.
low replication.
Weaknesses Artificial situation → unrepresenta- Harder to control variables → lower
tive behaviour → low ecological va- reliability, harder replication. Less
lidity. Demand characteristics likely. certain IV caused DV change. Ethi-
cal issues if participants unaware.

WORKED EXAMPLE: Testing whether TV affects children’s helpfulness (measured by how often they
help a "crying" doll). Lab: each child shown a helpful or neutral cartoon, given the doll for 10 mins,
videoed. Field: parents show the cartoon at home; child videoed with the doll afterwards. Field
version has MORE ecological validity (familiar, less nervous) but risks home distractions preventing
attention to the cartoon.

Key Evaluation Vocabulary

Term Exam-ready meaning

Controls Keep confounding variables constant between IV lev-
els → raises validity
Standardisation Same procedure for every participant → raises relia-
bility
Reliability Would the procedure/measure give the same results
again?
Validity Are you testing what you claim to test?
Pilot study Small trial BEFORE the main study to catch problems
(NOT to test ethics)
Replication Repeat the exact procedure/materials to verify results
Operational definition Clear, precise description so a variable can be mea-
sured/replicated

Page 5 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Term Exam-ready meaning

Placebo Fake pill/procedure with no active ingredient, used for
comparison
Ecological validity Do findings apply to the real world? Depends on set-
ting realism + mundane realism of the task

IMPORTANT: Controls + standardisation together raise BOTH reliability AND validity. A pilot study is
NOT for testing ethics — that’s the researcher/ethics committee’s job.

Hypotheses in Experimental Studies

Hypothesis: testable statement based on the aim, giving more detail than the aim. Must be falsifi-
able (provable wrong) and have operationalised variables.

UNFALSIFIABLE EXAMPLE: Freud explained a woman being rude to a man as EITHER uncon-
scious hatred (avoiding being ruder) OR unconscious love (avoiding showing desire) — whatever she
does, the theory "explains" it, so it can never be proven wrong. Bad science.

MEMORY TRICK: Non-directional = "no direction stated". Directional = "direction stated" (needs
previous evidence). Null = "it’s just chance".

Type Says Used when Example

Non-directional (2-tailed) Effect WILL happen, direc- No previous research to "There is a difference be-
tion not stated predict direction tween mind maps and re-
vision apps in helping stu-
dents learn."
Directional (1-tailed) States WHICH direction Previous evidence sug- "Students using revision
(higher/lower; +/- correla- gests a direction apps will learn better
tion) than students using mind
maps." (Opposite direction
is also a valid directional
hypothesis with different
reasoning.)
Null Any difference/correlation Always the "backup" to the "There will be no differ-
is due to CHANCE alternative hypothesis ence..." OR "Any differ-
ence... is due to chance."

EXAM TIP: Inferential statistics work out how likely results arose by chance. A "significant" result →
REJECT null, ACCEPT alternative hypothesis. Non-significant → ACCEPT null.

Ethics in Experiments

• Lab: participants usually know they’re taking part → informed consent possible — but validity may
need deception (hiding aim) to avoid demand characteristics.
• Field: participants often unaware they’re in a study → can’t consent, can’t exercise right to withdraw
→ presumptive consent may be needed; participants must still be protected from harm.
• Deception should be avoided; explain the truth afterwards in a debrief where possible (harder in field
studies, sometimes impossible).
• Privacy: easier to respect in lab (pre-planned tasks); harder in field (risk of invading personal space).

Page 6 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

• Confidentiality: keep data secure/anonymous; vital that unaware field participants can’t be individually
identified (e.g. by workplace).

1.2 Self-Reports

Self-report: obtains data by asking participants to give info about THEMSELVES directly (different
from experiments/observations, where the researcher finds data out FROM the participant).

MEMORY TRICK: Closed = "countable" → quantitative. Open = "own words" → qualitative.

Questionnaires

Questionnaire: self-report using written questions (’paper and pencil’ or online).

Question type Data Examples

Closed Quantitative Yes/no; item lists; rating scales (0–
5); Likert scales (strongly agree →
strongly disagree)
Open Qualitative "Why do you believe it is important
to help people who suffer from pho-
bias?"

EXAM TIP — full weakness list:

• Open answers need interpretation → risk of low inter-rater reliability between researchers.
• Easy to ignore → low return rate → respondents may share characteristics (e.g. unem-
ployed/retired with spare time) → poor generalisability.
• Participants may lie: to look acceptable (social desirability bias), or if they’ve guessed the aim.

Filler questions = irrelevant items added to disguise a questionnaire’s real aim (not analysed).

Interviews

Interview: verbal questions (face-to-face, telephone, or real-time chat); often uses MORE open
questions than questionnaires.

MEMORY TRICK: "SUS" = Structured (fixed/scripted) / Unstructured (flows from answers) / Semi-
structured (fixed core + extra Qs — best compromise).

Format Definition
Structured Fixed order/scripted questions for everyone; may
standardise interviewer’s tone/posture too
Unstructured Most questions (after the first) depend on the respon-
dent’s answers — flexible, hard to compare between
participants
Semi-structured Fixed core of open + closed Qs, interviewer CAN add
more — comparable AND explores individual issues

DON’T CONFUSE: Subjectivity = interpretation biased by the researcher’s OWN feelings (bad). Ob-
jectivity = unbiased, consistent regardless of researcher (good).

• Interviewees may lie (social desirability, or trying to help/disrupt research if they guess the aim).
• Time-consuming → restricts type of participant who volunteers → narrow representation of views.

Page 7 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

• Researchers should aim for objectivity — may ask other aim-unaware researchers to independently
interpret findings.

EXAM TIP: Questionnaires/structured interviews tend to be MORE reliable (consistent, numerical,

no interpretation needed). Open questions give more VALID, in-depth info but need interpreting.

1.3 Case Studies

Case study: detailed investigation of ONE instance (usually a person, could be a family/institution).
Uses a VARIETY of techniques (interviews, observation, tests, questionnaires) AND sources (partic-
ipant, relatives, colleagues, existing records).

• Useful for: rare cases needing detail; tracking developmental change (child’s progress, decline/improvement
of a disorder).
• Often linked to therapy, BUT the therapeutic purpose is NOT the research aim.

MEMORY TRICK: Triangulation = using different techniques on the same case — if results agree,
this suggests validity.

Evaluating Case Studies

✓ Strengths ✗ Weaknesses
Highly valid — great depth, genuine context Close researcher-participant relationship → risk of
(work/family). Validity improved by triangulation. subjectivity → lowers validity.
Detail threatens confidentiality/privacy — hard to dis-
guise identity even without names.
Single participant/researcher → hard to be objective;
low reliability.
Findings specific to individual → very limited general-
isability.

WORKED EXAMPLE: Sleep clinic case study on patient "PMA" (nightmares). Used EEG to track
sleep cycles/dreaming, woke patient to ask about emotions, kept a dream diary, AND interviewed
family — this is triangulation. Referred to only by initials to protect confidentiality.

EXAM TIP: Ethically, participants should give informed consent about their commitment; particular
attention to privacy and confidentiality given the depth of detail involved.

1.4 Observations

MEMORY TRICK: 4 decisions for every observation — "S-F-R-O": Setting, Focus, Role,
Obviousness.

Decision Options & Definitions

Setting Naturalistic = normal environment, no researcher
interference (social or physical). Controlled = so-
cial/physical environment manipulated by researchers
(can be normal OR artificial/lab setting).

Page 8 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Decision Options & Definitions

Focus Unstructured = records the WHOLE range of be-
haviours (often a pilot stage). Structured = records
only a LIMITED, pre-defined range using opera-
tionalised behavioural categories (observable ac-
tions, not inferred states).
Role Participant observer = part of the social setting (may
talk/play with people watched). Non-participant =
stays uninvolved (one-way glass, sits apart).
Obviousness Overt = role obvious (e.g. holding a clipboard).
Covert = hidden/disguised (disguised as a group
member, or physically hidden via CCTV).

Evaluating Observations

• Naturalistic: ✓ true-to-life behaviour. ✗ no guarantee the target behaviour occurs.

• Unstructured: ✓ captures unexpected important behaviours. ✗ very hard to record ALL activity accu-
rately; much may be irrelevant.
• Structured: ✓ more reliable data (focus on few categories).
• Covert: ✓ more valid (no demand characteristics/social desirability). ✗ practical issues (must stay
hidden, harder data collection) AND ethical issues (no informed consent, distress if discovered).
• Overt: more ethical, but participants aware of being watched → may alter behaviour → lowers validity.

IMPORTANT: Covert + Participant observation = BEST validity (natural, undetected) but WORST
ethics (no consent, invades privacy, distress risk if discovered).

Inter-observer reliability = consistency between two researchers watching the SAME event.

WORKED EXAMPLE: A teacher tests whether students detect each other’s emotions. 3 students
act as confederates, taking turns to appear sad at lunchtime. The teacher + a colleague act as
non-participant, overt observers (students used to seeing them, behave normally). They use a
fixed checklist (structured observation) with operational definitions ("shows concern without action"
/ "verbalises concern" / "takes action") — raises inter-observer reliability since both use the same
definitions.

EXAM TIP: Observation can be a method in ITSELF, or a TECHNIQUE within another method (mea-
suring the DV in an experiment, a co-variable in a correlation, or data within a case study).

1.5 Correlations

Correlation: looks for a relationship between two measured co-variables. Used when manipulating
variables isn’t practical or ethical (i.e. can’t do an experiment). Both variables must exist over a range
and be measurable numerically (durations, tallies, ratings, scores).

THE #1 EXAM TRAP: Correlation ̸= Causation. NEVER say one co-variable

"causes"/"makes"/"leads to" the other — a THIRD variable may explain both. Always say "co-
variables", never "IV/DV".

CLASSIC EXAMPLE: Ice cream sales & murder rate correlate POSITIVELY — NOT because ice
cream causes murder; both likely linked to hot weather (a third variable). Similarly, attention in class
& test scores correlating doesn’t prove attention CAUSES good scores — both could stem from how

Page 9 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

hard-working the student generally is.

MEMORY TRICK: Positive = both go up the stairs together. Negative = a see-saw (one up, one
down).

Type Example
Positive More exposure to aggressive models → more violent
behaviour
Negative Fewer years in education → more obedient

WORKED EXAMPLE — operationalising a correlational hypothesis: Bad: "There will be a cor-

relation between sleep and emotional reactivity" (nothing operationalised). Good: "There will be a
correlation between the number of hours a person sleeps and their emotional reactivity, indicated
by how loudly they cry during a sad film." Directional version: "There will be a negative correlation
between number of computer games played and final A Level grade."

Evaluating Correlations

• Only valid if BOTH co-variables are measured in clearly-defined, effective ways.

• Reliability depends on consistent measurement — lower for self-reports/observations than scientific
scales (e.g. cm3 , seconds).
• Main issue: conclusions do NOT necessarily reflect a causal relationship.
• Good starting point for research — shows whether a relationship is worth investigating further (e.g.
with an experiment).

EXAM TIP: Strength shown by ’r’ value: +1 = strong positive; -1 = strong negative; 0 = no/weak
relationship. Displayed on a scatter graph.

1.6 Longitudinal Studies

Longitudinal study: follows a SINGLE group (a cohort) over time (weeks to decades), studying
variables at intervals.
Cross-sectional study: compares DIFFERENT groups at differing ages/stages at ONE point in time.

MEMORY TRICK: Longitudinal = "LONG" time, SAME cohort. Cross-sectional = "CROSS-

comparing" DIFFERENT groups, ONE time.

PROBLEM WITH CROSS-SECTIONAL: hard to separate age effects from differences due to grow-
ing up at DIFFERENT times (upbringing, changing societal expectations). Longitudinal avoids this —
one cohort, tested repeatedly.

• Different designs: tracking ONE variable’s change over time; recording 2+ variables to look for cor-
relations; OR an EXPERIMENTAL ("quasi-experimental") design — baseline (pre-intervention) then
post-intervention/follow-up measures.

Evaluating Longitudinal Studies

✓ Strengths ✗ Weaknesses
Changes due to TIME, not cohort differences → most Sample attrition — drop-outs (withdrawal, boredom,
valid test of development. moving, poor health, homelessness, death).

Page 10 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

✓ Strengths ✗ Weaknesses
Retests same people → no participant-variable con- Smaller/less representative sample; survivors may be
found. atypical/"special" (motivated to succeed).
Overcomes confounding situational variables (e.g. dif- Reliability risk — measures/researchers may change
ferent educational methods) that would affect a cross- over a long period. Ethical issues — ongoing consent
sectional study. (harder with children), confidentiality of long-term con-
tact details.

CLASSIC EXAMPLE: Terman’s "Life Cycle Study of Children with High Ability" (began 1922) —
"Terman’s Termites". Debunked the myth that high IQ automatically leads to success (though some,
e.g. psychologist Lee Cronbach, did become famous). Initial sample = 1528 children. Sample
attrition: comparing 1996 & 1999 data, only 119 of 162 returned questionnaires gave data on BOTH
occasions (Holahan & Velasquez, 2011). Participants’ "special" identity (being named "Termites",
knowing their IQs) may itself have affected their outcomes.

1.7 Definition, Manipulation, Measurement & Control of Variables

Aim: the intention of the study — idea tested, problem solved, or question answered.

MEMORY TRICK: Participant variable = about the PERSON (age, personality, intelligence). Situa-
tional variable = about the SITE/environment (noise, light, weather).

EXAMPLE: False-memory study: IV = age, operationalised as ’young’ (under 20), ’middle-aged’

(40–50), ’old’ (over 70). DV operationalised as number of details "remembered", or how convinced
participants were the memory was true.

• Confounding variables act on ONE level of the IV — can either (1) work AGAINST the IV’s effect,
masking a real effect, or (2) INCREASE the apparent effect, suggesting a fake one.
• A pilot study helps identify problem variables BEFORE the main study.

WORKED EXAMPLE (Dr Huang’s chair study): Comparing soft-chair class vs hard-chair class. If
the soft-chair room has better lighting → situational variable (environment). If soft-chair students
all happen to do arts subjects and hard-chair students all do maths/science → participant variable
(individual ability) — could make it look like soft chairs cause WORSE performance when really it’s
the subject-ability difference.

Standardisation

Standardised instructions: the same written/verbal info given to every participant, keeping the
experience as similar as possible regardless of IV level.

• Every participant must be treated the SAME way — includes standardised instructions AND a stan-
dardised procedure (consistent equipment/tests).
• Easier in lab experiments (consistent equipment e.g. stopwatches); some measures (brain scans)
need interpretation, which must ALSO be standardised.

1.8 Sampling of Participants

Population: the group sharing characteristics, from which a sample is drawn. Sample: the group
selected to represent the population. Sampling technique: the method used to obtain participants.

Page 11 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

MEMORY TRICK: "O-V-R" = Opportunity (available) / Volunteer (advert response) / Random (equal
chance, numbers from a hat).

• Samples should ideally be REPRESENTATIVE so findings generalise. Commonly reported details:

age, ethnicity, gender, socio-economic status, education, employment, location, occupation.
• Sample SIZE matters — small samples are less reliable/representative (less likely to contain the full
range of population variation).

Technique Definition ✓ Strength ✗ Weakness

Opportunity Participants chosen be- Quick/easy, larger sample Non-representative —
cause available (e.g. readily obtained limited/similar variety (e.g.
nearby uni students) young, above-average
education)
Volunteer (self-selected) Participants respond to an Easy; committed (willing to Non-representative —
advert/announcement return); good for finding respondents share traits
UNUSUAL participants (free time, motivation)
Random Every person has an Likely representative Not everyone truly ac-
EQUAL CHANCE (num- cessible if list incom-
bered list + random plete/biased, especially if
generator) sample is small

MILGRAM’S ACTUAL ADVERT (quotable): "WE WILL PAY YOU $4.00 FOR ONE HOUR OF YOUR
TIME — Persons Needed for a Study of Memory... at Yale University... All persons must be between
the ages of 20 and 50." A VOLUNTEER sample. Baron-Cohen et al. also used volunteer sampling
to find people with autism spectrum disorder.

CLASSIC TRAP: Advertising on a LIBRARY noticeboard biases toward hardest-working students

(only they visit); an opportunity sample from a COMMON ROOM biases toward the LEAST hard-
working (relaxing there). Neither is truly random even though it might feel unbiased.

EXAM TIP: Individual differences and developmental psychology are specifically devoted to studying
differences between people — recognising sample limitations is especially important there.

1.9 Data and Data Analysis

MEMORY TRICK: Quantitative = QUANTITY (numbers). Qualitative = QUALITY (words, depth).

Quantitative Qualitative
Strengths Typically objective; Often valid — participants express
scales/questions often very re- themselves exactly, not limited to
liable; easy to analyse (central fixed choices; unusual responses
tendency/spread) and compare not lost to averaging
Weaknesses Data collection often limits re- Often relatively subjective — find-
sponses → less valid if the wanted ings may be invalid if interpretation
answer isn’t available is biased by the researcher; may
not generalise if from only a few in-
dividuals

Measures of Central Tendency — "the 3 M’s"

Page 12 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Measure Definition Key facts

Mode Most frequent score(s) ONLY measure usable on
named/discrete categories. Can
have 2+ modes. Doesn’t consider
score VALUES, so less informative
than median/mean.
Median Middle score of RANKED data (2 Cannot be used on discrete cat-
middle numbers → add & ÷2) egories — only numerical/linear
scales. Unaffected by outliers (ben-
efit) but ignores their value (less
representative than mean).
Mean Sum of ALL scores ÷ number of Most informative — considers every
scores score’s VALUE. Can be swayed by
extreme outliers, but by taking them
into account is MORE representa-
tive than median/mode.

FULL WORKED EXAMPLE (AS vs A Level students rating how hard they work, scale 1–10):
• AS data ranked: 1,2,2,3,3,3,5,5,5,6,6,6,6,7,8,8,8,9,9,10 — two middle numbers both 6 → median
=6
• A Level data ranked: 4,6,6,7,7,7,7,7,7,9,9,9,9,9,10,10,10,10,10,10 — middle numbers 7 and 9 →
(7+9)/2 = median = 8
• A Level students believe they work HARDER (median 8 vs 6).
• Mean: AS total 112 ÷ 20 = 5.6. A Level total 160 ÷ 20 = 8. Same conclusion.
• Standard deviation: AS = 2.62, A Level = 1.72 → GREATER variation in AS students’ opinions
(bigger SD = more spread out).

Measures of Spread — "the 2 measures"

Measure Definition/Formula Key facts

Range (Largest - smallest value) + 1 +1 added because psychology
scales measure GAPS between
points, not the points themselves.
Problem: doesn’t accurately re-
flect outliers — one extreme score
barely changes the mean but can
drastically change the range.
Standard deviation Average difference (deviation) be- Bigger SD = more spread out;
tween each score and the mean smaller = more clustered. Ad-
vantage over range: uses EVERY
score, so not distorted by outliers
alone.

EXAM TIP: If the least hard-working A Level student rated themselves 1 instead of 4, the MEAN
barely changes (8 → 7.85) but the RANGE jumps from 7 to 10 — shows how sensitive range is to
single outliers vs mean/SD.

Graphs

MEMORY TRICK: "X is a-cross" → IV/categories go on the x-axis.

Page 13 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Graph Used for Key feature

Bar chart Discrete/named categories, totals GAPS between bars (categories
or averages (incl. all measures of not linearly related)
central tendency)
Histogram Continuous data (score distribution) NO gaps between bars (unless a
category is empty — leave a visible
gap)
Scatter graph Correlational data Each dot = 1 participant’s scores on
both co-variables; line of best fit; r
value shows strength/direction

IMPORTANT: A scatter graph only shows a RELATIONSHIP exists — NEVER that one co-variable
causes the other. An experiment would be needed to test causation.

1.10 Ethical Considerations

Ethical issues: problems raising concerns about participant welfare (or wider negative impact on
society) — may arise from the NATURE of the study, the PROCEDURE, or the IMPLICATIONS of
findings.
Ethical guidelines: advice (e.g. BPS Code of Ethics and Conduct, 2018) helping psychologists
MINIMISE HARM and MAXIMISE BENEFIT.

• Deceived/distressed participants may refuse future studies, view psychology badly, and distrust find-
ings — outcomes to avoid.
• University research usually needs approval from an ethical committee.

Human Participants — 7 Guidelines

MEMORY TRICK: "Concerned People Won’t Deceive, Care Properly, Debrief" = Consent, Protection
from harm, Withdraw (right to), Deception (avoid), Confidentiality, Privacy, Debrief.

Guideline Full exam-ready explanation

Informed consent Enough info to freely & knowingly agree. Must be
FREELY given by a COMPETENT individual — harder
for children, mental health/learning difficulties, low
literacy, non-native speakers. Children need par-
ent/guardian consent AND their own (child-friendly
way). If real consent is impossible (naturalistic obs,
field experiments), use presumptive consent — ask
a SIMILAR group instead.
Protection from harm No greater physical/psychological risk than everyday
life. Minimise via screening, experienced researchers,
stopping if unexpected risks arise.
Right to withdraw Can leave (and remove data) at ANY time, made clear
at the START. Incentives can be offered but not taken
away if they leave. Researchers must not use author-
ity to pressure continuation.
Lack of deception Don’t deliberately misinform. If essential (e.g. to avoid
demand characteristics), reveal the real aim ASAP
and allow removal of results.

Page 14 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Guideline Full exam-ready explanation

Confidentiality Data stored separately from names; never published
without specific agreement. Use ID NUMBERS in-
stead of names (e.g. to re-contact or pair repeated-
measures scores). Institutions (schools/hospitals)
also anonymised. EXCEPTION: identifiable info can
be shared with informed consent, or in exceptional
safety-risk circumstances.
Privacy Don’t invade personal/emotional space. Participants
can ignore questions they don’t want to answer. Ob-
servations only where people would expect to be on
public display.
Debriefing Full explanation of aims/consequences AFTER the
study; return participants to at least as positive a state
as they arrived. NOT a substitute for designing an eth-
ical study — minimise harm in advance, don’t rely on
debrief to "fix" it.

WORKED EXAMPLE: A relaxation-therapy ethics review. Researchers only ask consent for PART of
the procedure (listening to a recording), not the full aim. They plan deception — telling participants
some people see "disturbing flashing lights" rather than revealing the true aim (measuring pulse
rate falling). Participants told they can leave any time (right to withdraw), reassured they won’t be
asked about intimate thoughts shared (privacy), and given a number instead of using their name
(confidentiality).

Animal Research Guidelines

Based on BPS Guidelines for Psychologists Working with Animals (2020). Reasons animals are used
(Driscoll & Bateson, 1988): convenient models (e.g. learning); procedures unethical on humans (e.g.
brain surgery); or interesting in their own right (e.g. whale communication).

MEMORY TRICK: "Replace Sick Numbers, Protect Housing & Rewards" = Replacement, Species,
Number, Procedures, Pain/suffering/distress, Housing, Reward/deprivation/aversive stimuli.

Guideline Full explanation

Replacement Consider alternatives (videos of previous studies,
computer simulations) instead of live animals
Species Choose the one LEAST likely to suffer; consider cap-
tivity history, prior experimentation experience, and
sentience (ability to think/feel)
Number of animals MINIMUM needed for valid/reliable results — min-
imised via pilot studies, reliable DV measures, good
design, appropriate analysis
Procedures Legally controlled; animal’s experience should be
as normal/positive as possible within research con-
straints
Pain, suffering, distress Avoid where possible; use designs that IMPROVE
(not worsen) the animal’s experience (e.g. studying
enrichment vs normal, rather than deprivation); costs
justified by scientific benefit
Housing Match the species’ social needs (isolation worse
for social animals; overcrowding causes dis-
tress/aggression); enough space, food, water;
artificial environment only needs to recreate what’s
important to WELFARE/SURVIVAL (not what looks
nice to humans)

Page 15 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Guideline Full explanation

Reward/deprivation/aversive stimuli Consider each species’ natural feeding needs; prefer
PREFERRED food as reward over deprivation; avoid
aversive (unpleasant) stimuli where possible

EXAM TIP: Bateson’s Cube (1986) = 3-factor model: certainty of BENEFIT (high=good), quality
of RESEARCH (high=good), animal SUFFERING (low=good). Research only justified if benefit cer-
tainty is HIGH and suffering is LOW.

WORKED EXAMPLE: Professor Chao tests whether animals approach liked things and avoid disliked
things. Considered replacing live animals with videos but rejected this (couldn’t be sure what they
were reacting to). Chose rats as the species (bred for lab use, fine housed alone). Used a repeated
measures design to limit the number of animals needed. Used liked food (peanuts) vs disliked
food (lettuce) rather than aversive stimuli to reduce pain/distress; rats were NOT deprived of food
beforehand.

EXAM TIP — justifying an ethical decision, cover 3 points: (1) choices about animals/care
(species, housing, procedures); (2) WHY the study is done (benefits to humans/animals); (3)
STRENGTHS of the design (controls, objectivity, validity, reliability).

1.11 Evaluating Research: Methodological Issues

MEMORY TRICK — THE BIG 4, always in this order: Reliable? → Valid? → Generalisable? →
Replicable?

Concept Full definition The ONE question to ask

Reliability Consistency of a proce- Would I get the SAME result if I re-
dure/measure — produces the peated this?
same results with the same people
each time
Validity Testing what you claim to test. Am I measuring what I CLAIM to
Affected by reliability, objec- measure?
tivity/subjectivity, face validity,
demand characteristics, ecological
validity
Generalisability How widely findings apply — de- Do findings apply to OTHER peo-
pends on ecological validity AND ple/places?
sample size/representativeness
Replicability Extent the procedure can be Could someone repeat this EX-
kept the same when repeated, ACTLY using my write-up?
by same/different researchers, to
verify results

DON’T CONFUSE: Inter-rater reliability = consistency interpreting QUESTIONNAIRE/INTERVIEW

answers. Inter-observer reliability = consistency watching the SAME behaviour/event.

EXAM TIP — Test-retest: use the SAME test TWICE, same people, same conditions. HIGH corre-
lation between the two sets of scores = high reliability.

EXAM TIP — Face vs Ecological validity: Face validity = does the TASK itself seem to test the right
thing (e.g. a helping test using spiders lacks face validity for people scared of spiders)? Ecological
validity = do findings from this SITUATION apply to real life/other situations — improved by mundane
realism (a realistic task).

Page 16 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Ways to Improve Methodology

Area How
Method Switch field ↔ lab experiment, or questionnaire ↔ in-
terview
Design Independent measures = fewer order-effect problems;
repeated measures = fewer individual-difference
problems
Sample Opportunity = larger sample; volunteer = rare/specific
participants; random = better generalisability
Tool Check inter-rater/test-retest reliability; refine opera-
tional definitions
Procedure Reduce demand characteristics; increase task real-
ism (raises validity)

EXAM TIP: Always tie your improvement to the SPECIFIC scenario given — generic textbook an-
swers score fewer marks.

WORKED EXAMPLE: Dr Singh tests whether drivers are more obedient to a traffic warden in white
vs black clothing (same warden, different clothing = controls for validity). 4 observers use operational
definitions for behaviours ("slowing down" = visibly reducing speed; "stopping late" = halting past the
line) to raise reliability. Drivers unaware they’re in a study, so unlikely to show demand character-
istics. Drivers who stop past the line are given a debrief note with a phone number to call — this
respects their right to withdraw without Dr Singh needing to take their contact details first.

Page 17 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

QUICK REVISION CHEAT SHEET

Topic What to Remember

Experiment basics IV manipulated → DV measured → cause & effect. Control confounding variables
to raise validity.
3 Experimental designs "I R Match" — Independent (diff people, no order effects, participant-variable risk,
needs more people) / Repeated (same people, order-effect risk, no participant-
variable risk, fewer people) / Matched pairs (paired, best of both, hard to match,
small-sample risk)
Order effects fix Randomisation (random order per person) OR Counterbalancing (ABBA — half
A-then-B, half B-then-A)
Practice vs Fatigue Practice = improves (learned it) | Fatigue = declines (tired/bored) — both = "order
effects"
Lab vs Field Lab = control, causal certainty, standardised/replicable, but low ecological validity
& demand characteristics | Field = natural behaviour, less demand characteristics,
but harder to control, lower reliability, consent issues
Hypothesis types Non-directional = effect, no direction | Directional = states direction (needs evi-
dence) | Null = due to chance (always the "backup")
Closed vs Open Qs Closed = quantitative (countable, easy to analyse, less valid) | Open = qualitative
(in-depth/valid, needs interpreting, inter-rater risk)
3 Interview types "SUS" — Structured (fixed/scripted) / Unstructured (flows from answers, hard to
compare) / Semi-structured (fixed + extra Qs, best compromise)
Case study 1 instance, many methods+sources, triangulation = validity check, but n=1 = very
poor generalisability, researcher subjectivity risk
4 Observation decisions "S-F-R-O" — Setting (Naturalistic/Controlled), Focus (Structured/Unstructured),
Role (Participant/Non-participant), Obviousness (Overt/Covert)
Best validity/worst ethics Covert + Participant observer
combo
Correlation golden rule NEVER claim causation — only a relationship; ice cream & murder rate = classic
example (3rd variable: hot weather); say "co-variables" not IV/DV
Positive vs Negative cor- Positive = both increase together | Negative = one up, one down
relation
Longitudinal vs Cross- Longitudinal = same cohort over time (sample attrition risk) | Cross-sectional =
sectional different groups, one time point (cohort/generation confound risk)
Terman’s Termites 1922 gifted-children study; debunked "high IQ = automatic success"; sample fell
1528 → 119; participants’ "special" identity may have affected outcomes
Participant vs Situational Participant = the PERSON (age, personality) | Situational = the ENVIRONMENT
variable (lighting, noise)
3 Sampling techniques "O-V-R" — Opportunity (available, biased/limited) / Volunteer (advert, bi-
ased/motivated, good for rare participants) / Random (equal chance, best gen-
eralisability, list may be incomplete)
Milgram’s sample Volunteer — $4 newspaper ad, Yale University, "study of memory", ages 20-50
3 M’s (central tendency) Mode = most frequent (only usable on named categories) | Median = middle score,
ignores outlier values | Mean = average of all scores, most informative but swayed
by outliers
2 Spread measures Range = (highest-lowest)+1, simple but hides outlier detail | Standard deviation =
avg distance from mean, uses every score
3 Graphs Bar chart = gaps, discrete categories | Histogram = no gaps, continuous data |
Scatter graph = correlation dots + best-fit line, r value +1 to -1
7 Human ethics guide- "Concerned People Won’t Deceive, Care Properly, Debrief" — Consent, Protec-
lines tion, Withdraw, Deception, Confidentiality, Privacy, Debrief
Presumptive consent Ask a SIMILAR group — used when real consent is impossible (naturalistic obs,
field experiments)
7 Animal ethics guide- "Replace Sick Numbers, Protect Housing & Rewards" — Replacement, Species,
lines Number, Procedures, Pain/suffering, Housing, Reward/deprivation
Bateson’s Cube Benefit certainty (high) + Research quality (high) + Animal suffering (low) = justi-
fied
The Big 4 (evaluation) Reliable? → Valid? → Generalisable? → Replicable? — always in this order

Page 18 of 19
Chapter 1 — Research Methods Notes Cambridge AS & A Level Psychology (9990)

Topic What to Remember

Inter-rater vs inter- Rater = questionnaire/interview interpretation | Observer = watching same be-
observer haviour
Test–retest Same test, twice, same people — high correlation between scores = high reliability
Face vs Ecological valid- Face = does the TASK look right? | Ecological = REAL LIFE application (mundane
ity realism helps)
Improving methodology Method / Design / Sample / Tool / Procedure — always tie the fix to the SPECIFIC
scenario given

Page 19 of 19