9  Methods of Research: Experimental, Descriptive, Historical, Qualitative and Quantitative methods

9.1 The Five Examined Methods

The syllabus names five methods of research — Experimental, Descriptive, Historical, Qualitative, and Quantitative. Each has a distinct question form, data type, and inferential strength. PYQs typically test: (a) recognising which method fits a described study, (b) naming the seminal designs (Solomon Four-Group, Cohort, Panel, Ethnography), and (c) distinguishing internal vs external validity.

TipFive Methods at a Glance
Method Asks Data Strength
Experimental “Does X cause Y?” Manipulated + measured numeric Strong causal inference
Descriptive “What is happening?” Numeric or categorical Accurate portrayal
Historical “What happened, and why?” Documents, artefacts, oral records Reconstruction of past
Qualitative “What does it mean to them?” Words, images, observation Depth of meaning
Quantitative “How much, how many, how related?” Numbers Generalisation, testability 

flowchart TB
  M{Methods of<br/>Research} --> E[Experimental<br/>Cause-Effect]
  M --> D[Descriptive<br/>What is]
  M --> H[Historical<br/>What was]
  M --> QL[Qualitative<br/>Meaning]
  M --> QN[Quantitative<br/>Measure]
    classDef default fill:#003366,color:#ffffff,stroke:#ffcc00,stroke-width:3px,rx:10px,ry:10px;

9.2 The Experimental Method

9.2.1 Defining Features

TipThree Defining Features of an Experiment
  1. Manipulation of one or more independent variables (IV).
  2. Control of extraneous variables.
  3. Random assignment to conditions (in true experiments).

9.2.2 Variables

TipVariable Types in an Experiment
Variable Role
Independent (IV) Manipulated by researcher
Dependent (DV) Measured; presumed effect
Extraneous / Confounding Unwanted; threatens validity
Control / Constant Held steady
Moderator Changes the IV→DV relationship
Mediator Carries the IV’s effect through to DV

9.2.3 Three Classical Experimental Designs (Campbell & Stanley, 1963)

Donald Campbell and Julian Stanley in Experimental and Quasi-Experimental Designs for Research (1963) classified designs and named the threats to validity.

TipThree Families — Pre-, True-, and Quasi-Experimental
  • Pre-experimental — weakest. No control or randomisation. Examples: One-shot case study (X-O), one-group pre-test–post-test (O-X-O), static-group comparison.
  • True experimental — strong. Random assignment. Examples: Post-test-only control group (R-X-O / R-O), Pre-test–post-test control group, Solomon Four-Group.
  • Quasi-experimental — no random assignment. Examples: Non-equivalent control group, time-series, regression-discontinuity.

9.2.4 The Solomon Four-Group Design

Richard L. Solomon (1949) — the most rigorous experimental design, controls for both selection and testing effects.

TipSolomon Four-Group Layout
Group Pre-test Treatment Post-test
1 O₁ X O₂
2 O₃ O₄
3 X O₅
4 O₆

The four groups isolate treatment effect, pre-test effect, and their interaction.

9.2.5 Threats to Internal Validity — Campbell & Stanley’s Eight

TipEight Threats to Internal Validity
  1. History — outside events during the study.
  2. Maturation — natural change in subjects (growth, fatigue).
  3. Testing — pre-test affects post-test.
  4. Instrumentation — measurement tool changes.
  5. Statistical regression — extreme scores drift toward the mean.
  6. Selection — non-equivalent groups at start.
  7. Mortality / Attrition — differential dropout.
  8. Interaction of selection × maturation, etc.

9.2.6 Threats to External Validity (Generalisation)

TipExternal Validity Threats
  • Hawthorne effect — subjects change because they know they’re being observed.
  • John Henry effect — control-group competitiveness.
  • Pre-test sensitisation — pre-test alters response to treatment.
  • Experimenter bias / Pygmalion (Rosenthal & Jacobson, 1968) — expectations affect outcome.
  • Demand characteristics — subjects guess hypothesis and oblige.
  • Setting-specific findings — lab vs real world.

9.3 The Descriptive Method

Descriptive research asks “what is” without manipulating variables. It is the most common method in education and social science.

9.3.1 Sub-types

TipDescriptive Method Sub-types
  • Survey research — sample-based estimation of population traits.
  • Case study — bounded, in-depth.
  • Correlational study — degree of relationship without cause.
  • Comparative study — across two or more groups.
  • Documentary analysis — content analysis of texts.
  • Normative survey — establishes a norm or standard.
  • Developmental study — Cross-sectional / Longitudinal / Trend / Cohort / Panel.

9.3.2 The Survey

TipAnatomy of a Good Survey
  • Sampling frame — list from which sample is drawn.
  • Sampling method — probability vs non-probability.
  • Sample size — power calculation; rule-of-thumb ≥10 per item.
  • Instrument — questionnaire, interview schedule, scale.
  • Pilot — test on small subset.
  • Mode — postal, phone, online, in-person.
  • Response rate — minimum 60–70 % for credibility.
  • Analysis — descriptive stats + inferential tests.

9.3.3 Correlational Study

Reports the strength and direction of an association, not causation. Pearson’s r (parametric), Spearman’s ρ (non-parametric ranks), Kendall’s τ. Effect sizes: small = 0.10, medium = 0.30, large = 0.50 (Cohen, 1988).

9.4 The Historical Method

Historical research reconstructs past events using documentary evidence — to understand how the present came to be.

9.4.1 Five Steps of Historical Research

TipFive Steps of Historical Method
  1. Identifying the research problem and period.
  2. Sourcing — locate documents, artefacts, witnesses.
  3. Criticising — external (authenticity) + internal (credibility) criticism.
  4. Synthesising — narrative construction.
  5. Reporting — argued historical account.

9.4.2 Sources

TipTwo Source Categories
Source What it is Examples
Primary First-hand, contemporaneous Manuscripts, letters, photos, eyewitness accounts, original government documents
Secondary Interpretive accounts of primary sources Textbooks, biographies, review articles

9.4.3 External vs Internal Criticism

TipTwo Layers of Criticism
Type Asks Tools
External Is the document genuine? Carbon-dating, handwriting analysis, provenance
Internal Is its content credible and accurate? Cross-check with other sources, author motive, context

9.4.4 Strengths and Limitations

Strengths: only way to study what has already happened; reveals long-term trends. Limitations: surviving record is selective and biased; researcher must interpret without direct verification.

9.5 The Qualitative Method

Qualitative research seeks meaning — how participants understand their world.

9.5.1 Five Major Qualitative Designs (Creswell)

TipFive Major Qualitative Designs
Design Asks Example
Phenomenology What is the lived experience of X? Experience of first-year doctors
Ethnography What is the culture of this group? Tribal community study
Grounded theory What theory emerges from the data? Theory of teacher resilience
Case study What can we learn from this bounded case? A school’s reform process
Narrative inquiry What story does the participant tell? Refugee life stories

9.5.2 Qualitative Data Collection

TipCommon Qualitative Data Collection Modes
  • Interview — structured, semi-structured, unstructured, depth, key-informant.
  • Focus group — 6–12 participants, group dynamic, moderator.
  • Observation — participant vs non-participant; overt vs covert.
  • Document analysis — text, image, audio, video.
  • Field notes & memos — researcher’s reflexive record.
  • Photovoice / arts-based — participatory visual.

9.5.3 Qualitative Data Analysis

TipSteps of Qualitative Analysis

Familiarisation → Open coding → Axial coding → Selective coding → Theme generation → Theory or rich description

Tools: NVivo, ATLAS.ti, MAXQDA, Dedoose. Approaches: thematic analysis (Braun & Clarke, 2006), content analysis, narrative analysis, discourse analysis, IPA (Interpretative Phenomenological Analysis).

9.5.4 Trustworthiness — Lincoln & Guba (1985)

Already covered in Topic 7: Credibility · Transferability · Dependability · Confirmability. Strategies: triangulation, member checking, peer debriefing, thick description, audit trail, prolonged engagement, negative case analysis, reflexivity.

9.5.5 Sample Sizes — Saturation

Qualitative sampling is purposive. The stopping rule is theoretical saturation — when new data add no new themes. Typical: 6–15 for phenomenology, 20–30 for grounded theory, single-case for case study, full community for ethnography.

9.6 The Quantitative Method

Quantitative research measures variables and uses statistics to estimate population parameters and test hypotheses.

9.6.1 Sub-Methods

TipQuantitative Sub-Methods
Sub-method Question
Descriptive (quantitative) What is the distribution?
Correlational How strongly related?
Causal-comparative / Ex-post-facto What caused this effect? (no manipulation)
Experimental Does X cause Y? (with manipulation)
Survey research How does the population look on these variables?
Meta-analysis What does the body of studies say?
Econometrics / Modelling How do variables behave systemically?

9.6.2 Scales of Measurement (S.S. Stevens, 1946)

Stanley Smith Stevens (1946, “On the Theory of Scales of Measurement”) defined the four scales:

TipStevens’s Four Scales
Scale Property Example Permitted statistics
Nominal Categories only Gender, religion Mode, χ²
Ordinal Order, no equal intervals Likert rank, race position Median, Spearman’s ρ
Interval Equal intervals, no true zero Temperature °C, IQ Mean, SD, Pearson’s r, t-test
Ratio Equal intervals + true zero Length, weight, income All — including geometric mean, ratios

Memory cue: N · O · I · R (the French word for “black”).

9.6.3 Statistical Toolkit

TipStatistical Tests by Question
  • Describe one variable — frequency, mean, SD, percentiles.
  • Compare two means — t-test (paired / independent).
  • Compare 3+ means — ANOVA; post-hoc Tukey HSD.
  • Two categorical variables — Chi-square.
  • Strength of relation — Pearson’s r, Spearman’s ρ.
  • Predict outcome — Regression (linear / logistic / multiple).
  • Reduce dimensions — Factor analysis, PCA.
  • Test agreement — Cohen’s κ.
  • Non-parametric — Mann-Whitney U, Wilcoxon, Kruskal-Wallis.

9.6.4 Sampling — Quantitative

TipQuantitative Sampling — Probability vs Non-Probability
Probability Non-probability
Simple random Convenience
Stratified random Purposive
Systematic Quota
Cluster Snowball
Multi-stage Judgement

Rule: only probability sampling allows statistical generalisation.

9.7 Mixed-Methods Research

Combines quantitative and qualitative within a single study. John W. Creswell and Vicki L. Plano Clark (2007) named four core designs:

TipFour Core Mixed-Methods Designs
Design Sequence What it does
Convergent Parallel Quant ⇄ Qual simultaneous Compare results side by side
Explanatory Sequential Quant → Qual Use Qual to explain Quant findings
Exploratory Sequential Qual → Quant Use Qual to develop a Quant instrument
Embedded One within the other Supplemental dataset within main design

The paradigmatic root of mixed methods is pragmatism (Peirce, James, Dewey) — method serves the question.

9.8 Choosing the Right Method — A Decision Framework

TipSeven Filters for Method Choice
  1. Question form — “Is there an effect?” → Experimental; “What does it mean?” → Qualitative.
  2. State of literature — empty → Exploratory/Qualitative; mature → Confirmatory/Quantitative.
  3. Variable control feasibility — Yes → Experiment; No → Quasi or Survey.
  4. Time scale — Past → Historical; Now → Cross-sectional; Change → Longitudinal.
  5. Sample access — Few participants → Case/Phenomenology; Many → Survey.
  6. Resources — funding, time, instruments.
  7. Ethical / regulatory feasibility — IRB approval, vulnerable groups.

9.9 How the Methods Interrelate

flowchart LR
  H[Historical] -.-> D[Descriptive]
  D --> C[Correlational]
  C --> EX[Experimental]
  D -.-> QL[Qualitative]
  EX -.-> MM[Mixed-methods]
  QL -.-> MM
  D --> QN[Quantitative]
  MM --> T((Theory))
  QN --> T
  QL --> T
    classDef default fill:#003366,color:#ffffff,stroke:#ffcc00,stroke-width:3px,rx:10px,ry:10px;

Most mature research programmes cycle: historical → descriptive → correlational → experimental → mixed-methods → theory revision.

9.10 Theory Anchors at a Glance

TipPersons, Years and Key Ideas
Person Year Contribution PYQ hook
R.A. Fisher 1925, 1935 Statistical methods; design of experiments Father of modern experimental design
Campbell & Stanley 1963 Pre-/True-/Quasi-experimental designs; 8 threats Internal validity
Richard L. Solomon 1949 Solomon Four-Group Design Controls testing effect
S.S. Stevens 1946 NOIR scales of measurement Nominal-Ordinal-Interval-Ratio
Rosenthal & Jacobson 1968 Pygmalion / experimenter expectancy Threat to validity
Glaser & Strauss 1967 Grounded Theory Qualitative theory generation
John Creswell 2007 4 core mixed-methods designs (with Plano Clark) Mixed-methods designs
Braun & Clarke 2006 Thematic analysis in psychology 6-step thematic analysis
Lincoln & Guba 1985 4 trustworthiness criteria Qualitative quality
Jacob Cohen 1988 Effect size conventions Small / medium / large
Hawthorne (Mayo studies) 1924–32 Observation alters behaviour External validity threat

9.11 Practice Questions

Q 01 Method Easy

Which method provides the STRONGEST evidence for cause-effect relationships?

  • ADescriptive survey
  • BHistorical research
  • CExperimental research
  • DEthnographic case study
View solution
Correct Option: C
Manipulation + control + randomisation = strongest causal inference. Experimental.
Q 02 Variable Medium

In an experiment, the variable MANIPULATED by the researcher is called the:

  • ADependent variable
  • BIndependent variable
  • CExtraneous variable
  • DModerator variable
View solution
Correct Option: B
Independent variable is manipulated. Dependent variable is measured.
Q 03 Solomon Medium

How many groups are there in the Solomon Four-Group Design?

  • ATwo
  • BThree
  • CFour
  • DSix
View solution
Correct Option: C
Richard L. Solomon (1949) — four groups isolate treatment effect, pre-test effect, and their interaction.
Q 04 Validity Medium

Outside events that occur during a study and might affect the dependent variable threaten the experiment through:

  • AMaturation
  • BHistory
  • CSelection
  • DMortality
View solution
Correct Option: B
History threat — uncontrolled outside events during the study. From Campbell & Stanley's 8 threats.
Q 05 Hawthorne Hard

"Subjects change their behaviour because they know they are being observed" is called the:

  • APygmalion effect
  • BHawthorne effect
  • CJohn Henry effect
  • DHalo effect
View solution
Correct Option: B
Hawthorne effect — named after Elton Mayo's Hawthorne Works studies (1924–32). Pygmalion = experimenter expectancy; John Henry = control-group competitiveness.
Q 06 Quasi-experiment Medium

A quasi-experimental design DIFFERS from a true experiment primarily in that it:

  • AHas no dependent variable
  • BHas no random assignment to groups
  • CHas no independent variable
  • DCannot use statistical analysis
View solution
Correct Option: B
Quasi-experimental designs manipulate the IV but use intact (non-randomised) groups — e.g., comparing two existing classes.
Q 07 Descriptive Easy

A study reports the literacy rate of every district in Tamil Nadu in 2024. This is BEST described as:

  • AExperimental
  • BDescriptive (Survey)
  • CHistorical
  • DPhenomenological
View solution
Correct Option: B
"What is" — no manipulation, sampled at one time = Descriptive / Survey.
Q 08 Historical Medium

In historical research, EXTERNAL criticism establishes:

  • AWhether the document's content is credible
  • BWhether the document is genuine
  • CWhether the author's interpretation is correct
  • DWhether the conclusions can be generalised
View solution
Correct Option: B
External criticism = authenticity (is the document genuine?). Internal criticism = credibility of its content.
Q 09 Primary Source Easy

Which of the following is a PRIMARY source for historical research?

  • AA textbook chapter on the freedom struggle
  • BA biography written 50 years after the event
  • CAn eyewitness diary entry from 1947
  • DA review article in a history journal
View solution
Correct Option: C
Eyewitness, contemporaneous = primary source. Others are secondary.
Q 10 Qualitative Medium

A researcher lives with a hill community for 14 months to describe their cultural practices. This is:

  • APhenomenology
  • BEthnography
  • CGrounded theory
  • DNarrative inquiry
View solution
Correct Option: B
Long immersion to describe culture = Ethnography.
Q 11 Saturation Hard

In qualitative research, "theoretical saturation" means:

  • AThe sample size hits a statistical power requirement
  • BThe data show normal distribution
  • CNew data no longer add new themes or categories
  • DThe researcher has interviewed at least 30 people
View solution
Correct Option: C
Saturation — when new interviews/observations stop yielding new themes. The stopping rule for purposive sampling.
Q 12 Stevens NOIR Medium

Temperature measured in degrees Celsius is on which scale of measurement?

  • ANominal
  • BOrdinal
  • CInterval
  • DRatio
View solution
Correct Option: C
°C has equal intervals but no true zero (0 °C ≠ absence of temperature) = Interval. Temperature in Kelvin would be Ratio.
Q 13 Stevens Hard

The four scales of measurement — Nominal, Ordinal, Interval, Ratio — were formalised in 1946 by:

  • AKarl Pearson
  • BS.S. Stevens
  • CR.A. Fisher
  • DJacob Cohen
View solution
Correct Option: B
Stanley Smith Stevens, Harvard psychologist, "On the Theory of Scales of Measurement", Science (1946). Mnemonic: NOIR.
Q 14 Statistic Medium

To compare the means of THREE or more independent groups on a continuous outcome, the appropriate test is:

  • AIndependent-samples t-test
  • BChi-square
  • CANOVA
  • DPearson's r
View solution
Correct Option: C
ANOVA (Analysis of Variance) compares 3+ means; post-hoc Tukey HSD finds which pair differs. (Two means → t-test; categorical → χ².)
Q 15 Mixed Methods Medium

A researcher first runs a quantitative survey, then uses follow-up interviews to explain the unexpected findings. This is BEST classified as:

  • AConvergent Parallel design
  • BExplanatory Sequential design
  • CExploratory Sequential design
  • DEmbedded design
View solution
Correct Option: B
Quant → Qual to explain = Explanatory Sequential (Creswell & Plano Clark, 2007).
Q 16 Sampling Medium

Which of the following is NON-probability sampling?

  • ASimple random
  • BStratified random
  • CSnowball
  • DCluster
View solution
Correct Option: C
Snowball = non-probability (one participant refers the next; used for hidden / hard-to-reach populations). Random / stratified / cluster are probability methods.
Q 17 Correlation Hard

Cohen's (1988) conventions for the size of a Pearson's r correlation classify "0.30" as:

  • ATrivial
  • BSmall
  • CMedium
  • DLarge
View solution
Correct Option: C
Cohen: small ≈ 0.10, medium ≈ 0.30, large ≈ 0.50.
Q 18 Grounded Theory Medium

In grounded theory, the analytic technique of repeatedly comparing data segments against each other to refine emergent categories is called:

  • ATriangulation
  • BConstant comparison
  • CMember checking
  • DReflexivity
View solution
Correct Option: B
Constant comparison — Glaser & Strauss (1967). The signature method of grounded theory.
Q 19 Categorisation Medium

Match each study to its method:

(i) Tests two teaching methods on randomly assigned Class 8 sections (a) Qualitative
(ii) Interviews 15 first-year doctors on burnout (b) Historical
(iii) Reconstructs the Kothari Commission (1964–66) from archival documents (c) Experimental
(iv) Surveys 2000 households for mobile use (d) Descriptive
  • A(i)-c, (ii)-a, (iii)-b, (iv)-d
  • B(i)-a, (ii)-b, (iii)-c, (iv)-d
  • C(i)-d, (ii)-c, (iii)-a, (iv)-b
  • D(i)-b, (ii)-d, (iii)-a, (iv)-c
View solution
Correct Option: A
Random assignment → experimental; depth interviews → qualitative; archival → historical; large survey → descriptive.
Q 20 Validity Hard

Campbell and Stanley (1963) identified threats to internal and external validity. The "Pygmalion / experimenter expectancy effect" is best understood as a threat to:

  • AInternal validity only
  • BExternal validity only
  • CBoth internal and external validity
  • DNeither — it is a sampling issue
View solution
Correct Option: C
Experimenter expectancy (Rosenthal & Jacobson, 1968) confounds the IV-DV relationship (internal) AND limits generalisation to non-expectant settings (external).

9.12 Quick Recall

ImportantQuick recall
  • Five methods: Experimental · Descriptive · Historical · Qualitative · Quantitative.
  • Experiment essentials: Manipulation · Control · Random assignment.
  • Variable types: IV · DV · Extraneous · Control · Moderator · Mediator.
  • Campbell & Stanley (1963): Pre- / True- / Quasi-experimental. 8 internal-validity threats: History · Maturation · Testing · Instrumentation · Statistical regression · Selection · Mortality · Interaction.
  • External validity threats: Hawthorne · Pygmalion (Rosenthal & Jacobson 1968) · John Henry · Demand characteristics · Pre-test sensitisation.
  • Solomon Four-Group (1949): 4 groups, controls testing × treatment interaction.
  • Descriptive types: Survey · Case · Correlational · Comparative · Documentary · Normative · Developmental (Cross-sectional / Longitudinal / Trend / Cohort / Panel).
  • Historical method 5 steps: Identify → Source → Criticise → Synthesise → Report. Two criticisms: External (authenticity) · Internal (credibility). Sources: Primary vs Secondary.
  • Qualitative — 5 designs (Creswell): Phenomenology · Ethnography · Grounded theory · Case study · Narrative inquiry.
  • Qualitative analysis: Familiarisation → Open → Axial → Selective coding → Themes. Tools: NVivo, ATLAS.ti, MAXQDA, Dedoose. Method: Braun & Clarke (2006) thematic analysis.
  • Lincoln & Guba (1985): Credibility · Transferability · Dependability · Confirmability. Strategies: triangulation, member checking, peer debriefing, thick description, audit trail, reflexivity.
  • Quantitative sub-methods: Descriptive · Correlational · Causal-comparative · Experimental · Survey · Meta-analysis · Modelling.
  • Stevens NOIR (1946): Nominal · Ordinal · Interval · Ratio.
  • Stats: 2 means → t-test; 3+ means → ANOVA; categorical → χ²; relation → Pearson’s r / Spearman’s ρ; predict → regression.
  • Cohen (1988) effect sizes: small 0.10 · medium 0.30 · large 0.50.
  • Sampling: Probability (random, stratified, systematic, cluster, multi-stage) vs Non-probability (convenience, purposive, quota, snowball, judgement).
  • Mixed-methods (Creswell & Plano Clark, 2007): 4 designs — Convergent Parallel · Explanatory Sequential · Exploratory Sequential · Embedded.