📚 Experimental Psychology Essentials

Brief Overview
This note covers experimental psychology methods and was created from a PDF source.
It gives a concise walkthrough of key concepts like experimental design, hypothesis
formulation, variable operationalisation, and statistical analysis—plus practical guidance
on sampling and reliability.

Key Points
Overview of the experimental method and variable control
Comparative strengths of laboratory, field, quasi‑experimental, and natural
designs
Sampling techniques and their impact on generalisability
Basic descriptive statistics and inferential tests for common data types

Experimental Method 🧪
Definition: The experimental method manipulates an independent variable (IV) to
observe its effect on a dependent variable (DV), which is then measured.

Settings: field, laboratory, quasi‑experimental, natural.

Key requirement: control all other variables so any change in the DV can be
attributed to the IV.

Aims 🎯
Definition: A general statement of what the researcher intends to investigate; derived
from theory and prior research.

Guides the overall purpose of the study.

Informs the formulation of hypotheses.

Hypotheses 📊
 Definition: A precise, testable statement about the expected relationship between
variables.

Directional hypothesis: predicts the direction of the effect (e.g., “More sleep →
better memory performance”).
Non‑directional hypothesis: predicts an effect but not its direction (e.g.,
“Sleep duration will affect memory performance”).
Use directional hypotheses when prior research suggests a clear outcome;
otherwise adopt a non‑directional hypothesis.

Variables 🔄
Independent Variable (IV)

The factor manipulated by the researcher (or that changes naturally) to test its impact
on the DV.

Dependent Variable (DV)

The factor measured as the outcome of the IV manipulation.

Experimental condition: participants experience the IV.

Control condition: participants do not receive the IV (baseline).

Operationalisation of Variables 📏
Definition: Translating abstract variables into concrete, measurable terms.

Example (directional hypothesis on sleep & memory):

IV: Number of hours of sleep (e.g., 4 h, 6 h, 8 h).
DV: Score on a memory test (e.g., number of correctly recalled words
out of 20).

Control of Variables ⚖️
Extraneous vs. Confounding Variables
 Extraneous variables: Any variable other than the IV that may influence the
DV but does not systematically vary with the IV (e.g., lab lighting, participant
age).
Confounding variables: Variables that systematically vary with the IV, making
it unclear whether the IV or the confounder caused the DV change (e.g., time of
day the memory test is taken).

Demand Characteristics & Investigator Effects

Definition: Cues that lead participants to guess the study’s purpose and alter their
behaviour (demand characteristics) or unconscious cues from the researcher that
influence participants (investigator effects).

Can lead to participant reactivity, placebo/sabotage effects, and reduced

validity.

Randomisation & Standardisation 🎲

Randomisation: Assign participants, order of conditions, or materials by
chance to minimise bias.
Standardisation: Use identical procedures and instructions for all participants
to eliminate procedural variability.

Types of Experiment (Strengths & Limitations) 📋

Design Description Strengths Limitations
Laboratory Highly controlled; High internal Low ecological
replicable validity; precise validity; artificial
control setting
Field Conducted in High ecological Reduced control;
natural settings validity; natural harder to replicate
behaviour
Quasi‑experimental No random Good external Threatened internal
allocation; IV validity; feasible in validity (confounds)
occurs naturally many contexts
Natural IV is a naturally Very high external Rare, cannot be
occurring event validity manipulated, hard
 to replicate

Sampling Methods 🎯
Method Description Strengths Limitations
Opportunity Participants are Quick, inexpensive, Poor
readily available low researcher bias generalisability; not
representative
Random sampling Every member of Minimises Requires complete
the population has researcher bias; sampling frame; can
equal chance representative be costly
Systematic Every n‑th member Simple to May introduce
selected after implement; fairly periodic bias if
random start unbiased pattern exists
Stratified Population divided Ensures Time‑consuming;
into strata; random representation of requires knowledge
sample from each sub‑groups of strata
Volunteer Participants Easy recruitment; Volunteer bias;
self‑select motivated limited
participants generalisability

Experimental Designs 📐
Design Description Strengths Limitations
Independent Different No order effects; Requires more
groups participants in each simple to analyse participants;
condition possible group
differences
Repeated Same participants Controls individual Order effects;
measures experience all differences; fewer fatigue/boredom
conditions participants possible
Counterbalanced Order of conditions Eliminates order Complex to
varied across effects organise; still needs
participants enough participants
 Matched pairs Participants paired Controls specific Difficult to find
on key variables confounds suitable matches;
then split limited variables

Pilot Studies 🧭
Definition: Small‑scale preliminary studies that test procedures, identify problems, and
allow refinements before the main investigation.

Saves time and resources.

Helps optimise measurement tools and recruitment strategies.

Blind Procedures 👁️
Single‑blind
Participants do not know whether they receive the experimental or control
treatment, reducing demand characteristics.
Double‑blind
Both participants and experimenters are unaware of condition allocations,
minimising investigator effects and placebo influences.

Observational Techniques 👀
Type Strengths Limitations
Naturalistic High ecological validity; Low control; potential
participants behave confounds; replication
naturally difficult
Controlled Greater control over May induce artificial
variables; easier replication behaviour; lower ecological
validity
Overt Ethical (informed consent) Participants may modify
behaviour (demand
characteristics)
 Covert Captures authentic Ethical concerns (no
behaviour consent); privacy issues
Participant Insider perspective; richer Risk of bias; loss of
insight objectivity
Non‑participant Greater objectivity; May miss nuanced
minimal influence interactions

Observational Designs & Inter‑Observer Reliability 📏

Inter‑observer reliability formula:
Number of agreements
​

× 100
Total observations

Score ≥ 80 % indicates high reliability.

Unstructured observations: rich qualitative data; higher risk of observer bias.
Structured observations: systematic coding of behaviours; easier
quantification and reliability testing.
Sampling within Observations
Method Description Strengths Limitations
Time sampling Observe at Reduces data load; May miss brief
predetermined systematic behaviours
intervals
Event sampling Record each Captures rare Time‑consuming;
occurrence of a events may overlook
target behaviour start/end

Correlation 🔗
Definition: A statistical technique that measures the strength and direction of the
relationship between two variables (co‑variables) without manipulating them.

Coefficient (r): Ranges from –1 (perfect negative) to +1 (perfect positive); 0

indicates no linear relationship.
Types of relationships:
 Positive: both variables increase together.
Negative: one increases while the other decreases.
Zero/No correlation: no systematic relationship.

Strengths & Limitations

Strengths Limitations
Identifies patterns for further study Cannot infer causation (third‑variable
problem)
Quick, economical; can use secondary Misinterpretation as causation; vulnerable
data to outliers

Data Types 📂
Type Description Strengths Limitations
Qualitative Words, images, Rich detail; Hard to analyse
narratives captures statistically;
participants’ subjective bias
perspectives
Quantitative Numerical values Enables statistical May oversimplify
(e.g., scores, analysis; easy complex
counts) comparison phenomena
Primary Collected directly Tailored to study Time‑consuming;
by the researcher aims; high potentially costly
relevance
Secondary Existing data Saves time; May be outdated;
sources (e.g., inexpensive; large less control over
archives) samples possible variables

Descriptive Statistics – Central Tendency 📊

Measure Calculation Strengths Limitations
Mean Sum of scores ÷ Uses all data; Sensitive to outliers
number of scores appropriate for
interval/ratio data
 Median Middle value when Robust to extreme Less informative for
ordered scores multimodal
distributions
Mode Most frequent Works for nominal May be non‑unique
value data; easy to or absent
identify

Descriptive Statistics – Dispersion 📏

Measure Calculation Strengths Limitations
Range Highest – lowest Simple to compute Ignores distribution
score of intermediate
scores
Standard ˉ )2
∑(xi −x Precise; uses all Affected by
data points outliers; more
​

Deviation (SD)
​ ​

complex to
calculate

Presentation & Display of Quantitative Data 📈

Bar Charts
Ideal for discrete categories (e.g., Condition A vs. Condition B).
Bars do not touch, indicating separate groups.
The bar chart shows Condition A (~40) lower than Condition B (~65), visualising a clear
difference between groups.
Histograms
Used for continuous data; bars touch to indicate interval continuity.
X‑axis: equal‑sized score ranges (e.g., 0‑10, 11‑20).
Y‑axis: frequency of scores within each range.

Line Graphs
Connect data points to display trends over time or ordered conditions.

The darker line shows a dip then rise, while the lighter line starts high, falls sharply, then
recovers, illustrating differing trajectories.
Scattergrams
Plot each participant’s scores on two variables (X vs. Y).
Reveal association patterns; the line of best fit indicates direction and
strength.

Distributions 📊
Right‑Skewed (Positive Skew)

Mean (yellow) is to the right of median (green) and mode (blue), showing a tail extending
toward higher values.
Left‑Skewed (Negative Skew) / Normal‑like illustration

When mean < median < mode, the distribution is left‑skewed; a symmetric bell curve
would have all three coinciding.
Peer Review 📝
Purpose: Ensure research quality, relevance, and methodological soundness before
publication or funding decisions.

Validates the rigor of the study.

Helps allocate resources to worthwhile projects.

Publication Bias & Fraudulent Research 📚

Publication bias: The tendency for journals to preferentially publish studies with
positive, “headline‑grabbing” results, while negative or null findings are left
unpublished (the file‑drawer problem).

Consequences
Distorts the scientific record, leading to over‑optimistic effect size
estimates.
Enables fraudulent research to persist (e.g., the Wakefield 1998
MMR‑autism study) because retractions are slow and the original
claim remains in public memory.
Anonymity of reviewers can reduce objectivity; some journals now
employ open peer review to mitigate this.
Remedies
Encourage pre‑registration of hypotheses and analysis plans.
Adopt open review and transparent reporting (e.g., sharing raw
data, analysis scripts).
Use registered reports, where the study’s methods are
peer‑reviewed before data collection.

Economic Implications of Psychological Research 💷

Psychological research informs policy, workplace productivity, and public health,
thereby influencing national economies.
 Economic Domain Psychological Example
Contribution
Workforce productivity CBT & Rational‑Emotive Reduction in sick‑leave
Behaviour Therapy reduce saves millions of pounds
depression‑related annually.
absenteeism.
Family dynamics Research on father Higher household income,
involvement (e.g., Field greater consumer
1978) shows flexible spending.
working improves
caregiving, increasing
labour‑force participation.
Public health campaigns Social influence studies Effective campaigns lower
guide anti‑vaccine or disease burden and
environmental messaging, healthcare expenditure.
shaping health‑related
costs.
Legal system Cognitive‑interview Fewer appeals, reduced
techniques improve incarceration expenses.
witness testimony,
decreasing wrongful
convictions and associated
costs.
Estimated mental‑health‑related costs in the UK: ≈ £15 billion per year (lost
productivity, healthcare, social services). Psychology‑based interventions aim to reduce
this burden.

Case Studies: Strengths & Limitations 📋

Case study: An in‑depth investigation of a single individual or a small group, yielding
rich qualitative data.

Aspect Strengths Limitations

Depth of insight Allows detailed exploration Findings not generalisable
of rare or complex to larger populations.
 phenomena.
Ethical feasibility Enables study of otherwise Retrospective reliance on
unethical or impractical memory can introduce
situations. inaccuracy.
Theory generation Generates hypotheses for Time‑consuming; difficult
future experimental work. to replicate.
Bias risks Can reveal unusual patterns Subject to interviewer
(e.g., Little Hans bias, social desirability,
supporting the Oedipus and interpretative bias.
complex).

Content Analysis 📊
Content analysis: Systematic examination of existing texts/media (e.g., TV ads,
newspaper articles) to infer societal values, beliefs, and prejudices.

How to Conduct a Content Analysis

1. Formulate a hypothesis (what you expect to find).
2. Develop a coding scheme (e.g., 1 = male, 2 = female).
3. Collect material (samples of media, documents).
4. Code the data and record in a table.
5. Analyse using descriptive or thematic methods.
6. Report following the standard scientific‑report format.

Strengths & Limitations

Strengths Limitations
High external validity (real‑world data). Observer bias – mitigated by
inter‑observer reliability checks.
Generates large quantitative & qualitative Researcher may selectively code
datasets. (interpretative bias).
Easy to replicate; no consent issues. Cannot infer causality; only describes
patterns.
Levels of Measurement & Appropriate Statistics 📏
Levels of measurement: Distinguish how data can be quantified and analysed.

Level Description Typical Central Typical Dispersion

Tendency
Nominal Categorical, no Mode –
inherent order (e.g.,
gender).
Ordinal Ranked categories Median Range
(e.g., Likert 1‑5).
Interval Equal‑size numeric Mean Standard
intervals, no true Deviation
zero (e.g.,
temperature °C).
Parametric tests (e.g., t‑test, ANOVA) require interval or ratio data.
Non‑parametric tests (e.g., Mann‑Whitney, Chi‑square) are suitable for ordinal
or nominal data.

Scientific Report Structure 📄

Psychologists publish findings using a standardised scientific‑report format.

Section Core Content

Abstract 150‑200 word summary: aim, hypothesis,
method, results, conclusion.
Introduction Review of relevant literature; narrowing
from broad theory to specific
aims/hypotheses.
Method Design, participants (sampling details),
materials, procedure, ethical
considerations—enough detail for exact
replication.
 Results Descriptive & inferential statistics;
tables/figures; thematic analysis for
qualitative data.
Discussion Interpretation of findings, link to theory,
limitations, future directions, societal
implications.
References Full citations (books, journal articles,
websites) in appropriate style.

Sign Test & Choosing Inferential Tests ✅

Sign Test Procedure
1. State hypotheses (null H vs. alternative H ).
0 ​

1 ​

2. Assign signs to each paired difference (+ if increased, – if decreased; ignore

zeros).
3. Count the less‑frequent sign → this is S .
4. Determine critical S from a binomial table using N (non‑zero pairs) and α =
0.05 .

5. Decision rule:
If S ≤ critical S , reject H (significant difference).
0 ​

Otherwise, fail to reject H . 0 ​

6. Report N , S , critical value, α, and the direction of the effect.

Selecting the Appropriate Test

Data Type Design Test of Difference Test of
Association/Correla
Nominal Unrelated Chi‑square Chi‑square
​ Related Sign test –
Ordinal Unrelated Mann‑Whitney U Spearman’s ρ
​ Related Wilcoxon –
signed‑rank
Interval Unrelated Independent‑samples Pearson r
t
 ​ Related Paired‑samples t –
Memorising this table helps answer exam “choose the correct test” items.

Probability, Significance & Errors 🎯

Significance level (α) commonly set at 0.05 (5 % chance of a Type I error).
p‑value < α → statistically significant → reject H . 0 ​

Confidence level = 1 − α (e.g., 95 % confidence when α = 0.05).

Types of Errors
Error Type Description Consequence
Type I (False Positive) Rejecting a true H Overstating findings; may
lead to wasted resources.
0

(detecting an effect that

​

isn’t there).
Type II (False Negative) Failing to reject a false H 0
Missed discoveries;
(missing a real effect). under‑estimation of true
​

relationships.

Paradigms, Theory Construction & Falsifiability 🔍

Paradigm: Shared set of assumptions, methods, and questions within a scientific
discipline.
Paradigm shift: Fundamental change in these assumptions (Kuhn, 1982).

Psychology faces competing paradigms (biological vs. behavioural),

challenging its status as a unified science.
Theory construction:
1. Gather empirical evidence (observations, experiments).
2. Formulate testable hypotheses derived deductively from the theory.
3. Test; support strengthens the theory, refutation prompts revision.

Falsifiability (Popper, 1934)

A theory is scientific only if it can be potentially proven false.
 Example: Freud’s Oedipus complex lacks clear falsifiable predictions → labelled
pseudoscience by many scholars.
The null hypothesis embodies the falsifiable claim; researchers never claim to
“prove” a hypothesis, only to fail to reject it.

Replicability & Objectivity 🔄

Replicability: Ability of independent researchers to obtain the same results
using the original methods.
Objectivity: Minimising researcher bias (e.g., via blind procedures,
randomisation).
Empirical method: Knowledge must be derived from direct observation or
experience; purely theoretical speculation is insufficient.

Psychology as a Science: Debate ⚖️

Supporting Arguments Counter‑Arguments
Uses empirical methods and statistical Some findings rely on subjective
inference. interpretation (e.g., inferring cognition
from brain scans).
Produces practical applications (e.g., Not all research is generalizable;
CBT). laboratory tasks may lack ecological
validity.
Adopts rigorous experimental designs. Certain areas (e.g., personality) depend
on self‑report, which can be biased.

Ethical Issues & Mitigation Strategies 🛡️

Issue Explanation Typical Solutions
Informed consent Participants must know Provide clear information
study purpose, procedures, sheets; obtain written
risks. consent.
 Deception Deliberate withholding or Use debriefing (verbal or
misleading information. written) immediately after
participation; ensure
deception is justified and
minimal.
Right to withdraw Participants may stop at Reinforce during consent
any time without penalty. and remind throughout.
Privacy & confidentiality Protect personal data from Store data anonymously;
identification. use codes; limit access.
Cost‑benefit analysis Weigh potential harms Submit to ethics
against scientific/social committee; modify design
gains. to reduce risk.
Protection from harm Physical or psychological Provide support resources
injury. (e.g., counselling) if distress
arises.

Self‑Report Techniques & Design 📋

Questionnaires
Open‑ended: Rich qualitative detail; harder to quantify.
Closed‑ended: Easy statistical analysis; limited depth.
Common Scales
Likert (e.g., 1 = Strongly disagree → 5 = Strongly agree).
Rating scales (numeric intensity).
Fixed‑choice (select one option).
Design Checklist
Ensure clarity; avoid jargon, double‑barrels, leading questions.
Sequence: Easy items first → build confidence.
Include filler items to mask the true purpose (reduces demand characteristics).
Conduct a pilot study to detect ambiguities and refine items.
Strengths & Limitations
Strengths Limitations
 Cost‑effective; rapid data collection. Susceptible to social desirability,
acquiescence, and response bias.
Can reach large, geographically dispersed Designing valid items can be
samples. time‑consuming; may lack depth.

Interviews
Type Description Strengths Limitations
Structured Fixed question list; High May restrict depth;
identical order for standardisation; interviewer bias still
all participants. easy to compare possible.
across participants.
Unstructured Open conversation Rich, detailed data; Hard to compare;
guided by flexibility. requires skilled
participant’s interviewers;
responses. time‑intensive.
Semi‑structured Core set of Balances Still demands
questions plus consistency with careful training to
optional probes. depth. avoid bias.
Recording: audio, video, or detailed notes (ensure consent).
Environment: quiet, private room to promote openness.
Neutral language helps build rapport and reduce leading influences.

Reliability Across Methods 🔧

Types of Reliability
Type What It Assesses Common Assessment
Method
Internal reliability Consistency among items Split‑half (correlation
within the same ≥ 0.80).
instrument.
Test‑retest reliability Stability of scores over Same test administered
time. twice; correlation ≥ 0.80.
 Inter‑observer reliability Agreement between Formula:
different observers. Agreements
​
×
Total observations
100 ; aim ≥ 80 %.
External reliability Replicability across Re‑administer test in a new
settings/administrators. context; compare scores.

Improving Reliability
Questionnaires: Replace ambiguous open items with clear closed items; pilot
and revise.
Interviews: Use the same trained interviewer; employ structured protocols;
record and code systematically.
Experiments: Standardise instructions, environment, and equipment; conduct
lab‑based trials where possible.
Observations: Develop detailed coding manuals; train multiple observers and
calculate inter‑observer reliability before data collection.

The sections above build directly on the earlier material (e.g., experimental designs, peer
review) and expand the guide to cover the remaining core concepts from the lecture
transcript.

Validity 🧩
Validity – the degree to which the results of a study accurately represent the
phenomenon they are intended to measure.

Types of Validity
Type Description Typical Assessment
Internal validity Outcomes are caused by Check for confounding
the manipulation of the IV, variables, participant
not by extraneous factors. variables, investigator
bias, and demand
characteristics.
Face validity The measure appears to Expert review or
assess what it claims, lay‑person judgment.
 based on superficial
inspection.
Concurrent validity Correlation between the Pearson r between scores.
new measure and an
established measure taken
at the same time.
Predictive validity Ability of a test to forecast Correlate test scores with
future behaviour or later performance.
outcomes.
External validity Extent to which findings Assess ecological,
generalise beyond the temporal, and population
study’s specific setting, validity.
population, and era.
Ecological validity Degree to which Compare laboratory
experimental conditions behaviours with naturalistic
mimic real‑world observations.
environments.
Temporal validity Stability of findings across Replicate study after a
different historical periods. substantial time lag.
Population validity Generalisability to other Sample diversity analysis;
demographic groups (ages, cross‑cultural replication.
genders, cultures).

Improving Validity in Experimental Research

Control groups – provide a baseline to isolate the IV’s effect.
Standardised procedures – minimise participant reactivity and investigator
bias.
Blind designs
Single‑blind: participants unaware of study aims → reduces demand
characteristics.
Double‑blind: both participants and experimenters blind → reduces
investigator effects as well.

Operationalising Behavioural Categories 📋

 Operationalisation – translating abstract constructs into concrete, observable, and
measurable categories.

Choose unambiguous categories (e.g., pushing vs. vague aggression).

Ensure mutual exclusivity: categories must not overlap (hugging vs. cuddling).
Provide exhaustive coverage: the checklist should capture all possible relevant
behaviours.
Poor operationalisation → inconsistent coding across observers → reduced
reliability and validity.

Questionnaires & Self‑Report Measures ✍️

Many instruments embed a lie scale to detect socially desirable responding.
Anonymity assurances boost honesty, thereby enhancing construct validity.
Design tips:
1. Use clear, single‑barrel items.
2. Randomise item order to curb pattern answering.
3. Pilot test for ambiguity.

Observational Research 👀
Observations that minimise researcher interference tend to have high ecological
validity.

Covert observations (observer undetected) → behaviour remains natural.

Risks to validity:
Overly broad, overlapping, or ambiguous behavioural categories.
Inconsistent coding across observers (see reliability section).

Enhancing Observational Validity

Develop a detailed coding manual with operational definitions.
Train multiple observers and calculate inter‑observer reliability (target ≥ 80 %).
Conduct pilot observations to refine categories before full data collection.

Qualitative Methods & Interpretative Validity 📚

 Qualitative approaches (e.g., case studies, interviews) are prized for ecological and
interpretative validity, capturing participants’ lived realities.

Interpretative validity – the extent to which researchers’ interpretations

reflect participants’ own meanings.
Demonstrated by:
Consistency between researcher’s reporting and participants’ direct
quotes.
Transparent methodological description (e.g., how themes were
derived).

Triangulation 🌐
Triangulation – using multiple data sources or methods to corroborate findings,
thereby bolstering validity.
Example sources: interviews, diary entries, third‑party observations.
Benefits: reduces bias inherent in any single method and strengthens construct
and external validity.

Summary Table: Validity & Strategies Across Methods

Method Primary Validity Threats Key Strategies to Improve
Validity
Experimental Confounds, demand Control groups,
characteristics, investigator standardisation,
bias single/double‑blind
procedures
Questionnaire Social desirability, Lie scales, anonymity, clear
response bias items, pilot testing
Observation Observer influence, Covert observation,
ambiguous categories exhaustive operational
definitions, inter‑observer
reliability
Qualitative (interviews, researcher interpretation Direct participant quotes,
case studies) bias transparent coding,
triangulation
Practical Checklist for Validity (Applicable to Any
Design)
1. Define each variable/behaviour with precise, non‑overlapping criteria.
2. Pilot the measurement tool; revise ambiguous items/categories.
3. Control extraneous variables wherever possible (randomisation,
standardisation).
4. Blind participants and/or experimenters to reduce expectancy effects.
5. Assess reliability (internal, test‑retest, inter‑observer) before full data
collection.
6. Document all procedures thoroughly for reproducibility.
7. Use triangulation when feasible to cross‑validate findings.