Research in Education I: Scientific Method, Types and Approaches to Research, Variables, Hypotheses, Sampling and Tools

Unit 6 opens research methodology proper: what makes a method scientific, the ways research is classified by purpose and design, the vocabulary of variables and hypotheses that every study is built from, how a sample is drawn from a universe, and the tools and techniques that actually collect data. The paper asks for the defining characteristic of the scientific method, the correct classification of a described study, and which type of variable, hypothesis, sampling technique or tool a described scenario names; every term here carries its precise definition and its neighbour it is most often confused with.

1. Meaning and scope of educational research, and the scientific method

Educational research is the systematic, controlled, empirical investigation of educational phenomena, aimed at describing, explaining, predicting or improving practice, and it covers every level of the system this course has surveyed — the learner, the teacher, the curriculum, the institution and policy itself. Its scope runs from basic questions about how learning happens to applied questions about which intervention raises achievement, and to action research on a single teacher's own classroom. The scientific method is characterised by replicability (another researcher following the same steps should reach the same finding), precision (careful, unambiguous definition of what is being measured and how), falsifiability (a genuine scientific claim must be capable, in principle, of being shown false by evidence, following Karl Popper's criterion) and parsimony (the simplest explanation consistent with the evidence is preferred over a needlessly complex one).

Classifying research: by scientific method, by aim, and by approach and design
Classification	Types
Types of scientific method	Exploratory (investigates a little-understood phenomenon to generate hypotheses, not test them), explanatory (tests why a relationship holds, establishing cause), descriptive (systematically describes a phenomenon's characteristics as it currently exists, without manipulating variables)
Aims of research as a scientific activity	Problem-solving (addressing a specific practical difficulty), theory building (developing or refining explanatory frameworks), prediction (forecasting future outcomes from established relationships)
Types of research by purpose	Fundamental/basic (advances knowledge for its own sake, no immediate application intended), applied (addresses a specific practical problem directly), action (a practitioner systematically studies and improves their own immediate practice, often over a short, iterative cycle)
Approaches to educational research	Quantitative (numerical data, statistical analysis, seeks generalisable, objective patterns), qualitative (non-numerical data — words, images, observations — seeks depth, meaning and context in a specific setting)
Designs in educational research	Descriptive (surveys, case studies, correlational studies describing what exists), experimental (manipulates an independent variable under controlled conditions to test cause), historical (reconstructs and interprets past events from records and documents to understand present practice)

⚠️ Five overlapping classifications, five different questions

A single real study can be, say, applied (by purpose), quantitative (by approach) and experimental (by design) all at once — these are five independent axes, not five mutually exclusive labels for one thing. A question naming 'systematic description of current classroom practice with no manipulation of variables' is testing descriptive design specifically, distinct from the descriptive TYPE of scientific method, which is a narrower, related but separately named idea in the syllabus's own first sub-item. Read carefully which axis a question is asking about before choosing an option from a different axis that also sounds plausible.

2. Concepts, constructs and variables, and hypotheses

A concept is a general, abstract idea formed from observing a class of objects or events (e.g., 'motivation'); a construct is a concept deliberately, formally defined for scientific use, often built from more than one concept and given a specific operational meaning so it can be measured ('academic motivation' operationalised as a score on a specific validated scale); a variable is any characteristic or condition that can take different values across cases and that is actually measured or manipulated in a study.

Types of variables and types of hypotheses
Term	Meaning
Independent variable	The presumed cause, manipulated or selected by the researcher (a new teaching method)
Dependent variable	The presumed effect, measured to see if it changed (test score)
Extraneous variable	A variable outside the researcher's interest that could also affect the dependent variable and must be controlled (students' prior ability)
Intervening variable	An unobserved, theoretical variable that mediates between the independent and dependent variable, explaining the mechanism of the effect (e.g., 'increased engagement' between a new method and higher scores)
Moderator variable	A variable that changes the strength or direction of the relationship between independent and dependent variables (the new method may work better for younger students — age moderates the effect)
Null hypothesis	States there is no relationship or no difference between variables — the statement statistical tests are actually set up to reject or fail to reject
Research hypothesis (alternative hypothesis)	States the relationship or difference the researcher actually expects to find, based on theory or prior evidence
Directional hypothesis	Specifies not just that a difference exists but its direction (Method A will produce HIGHER scores than Method B)
Non-directional hypothesis	States only that a difference exists, without predicting which direction (Method A and Method B will produce different scores)

🧠 Intervening versus moderator, and directional versus null

An intervening variable explains HOW or WHY an effect happens (the mechanism, unobserved and internal to the causal chain); a moderator variable explains WHEN or FOR WHOM an effect is stronger or weaker (a pre-existing condition that changes the relationship's strength, often observed and external). 'Better scores because students were more engaged' → intervening (engagement). 'Better scores, but only for younger students' → moderator (age). A hypothesis predicting 'A will outperform B' is directional; one predicting merely 'A and B will differ' is non-directional; the null hypothesis, tested statistically, always says no difference exists at all.

3. Universe, sample, and techniques of sampling

The universe (population) is the complete set of individuals, events or objects a study is interested in generalising to; a sample is the smaller subset actually studied, chosen because studying the entire universe is usually impossible or impractical. A good sample is representative (mirrors the universe's relevant characteristics), adequate in size (large enough for stable estimates and the planned statistical tests), and drawn by a procedure free of systematic bias.

Probability and non-probability sampling techniques
Technique	How it selects
Simple random sampling (probability)	Every member of the universe has an equal, known chance of selection (lottery, random-number table)
Stratified random sampling (probability)	The universe is divided into homogeneous strata (by gender, grade, region) and random samples drawn from each, ensuring proportional representation
Cluster sampling (probability)	The universe is divided into naturally occurring clusters (schools, districts) and whole clusters are randomly selected, useful when a full list of individuals is unavailable
Systematic sampling (probability)	Every kth member is selected from a list after a random start (e.g., every 10th student on a roll)
Convenience sampling (non-probability)	Whoever is easiest to reach is included (students in the researcher's own class); fast and cheap but weak generalisability
Purposive/judgemental sampling (non-probability)	The researcher deliberately selects cases believed to be especially information-rich for the research question (expert teachers for a study of best practice)
Snowball sampling (non-probability)	Existing participants refer additional participants, useful for reaching a hard-to-identify or small population (teachers using a rare pedagogical technique)
Quota sampling (non-probability)	A fixed quota is set for each subgroup (50 male, 50 female teachers) and filled non-randomly, resembling stratified sampling in structure but without random selection within strata

🎯 Why probability sampling is preferred when it is feasible

Only probability sampling lets a researcher calculate a known margin of error and legitimately generalise findings from the sample back to the whole universe with a stated confidence level; non-probability techniques are chosen when a full list of the universe does not exist, when time and resources are limited, or when the research question specifically needs information-rich or hard-to-reach cases rather than statistical generalisability — which is exactly why qualitative studies (Unit 6's second half) lean toward purposive and snowball sampling while quantitative surveys lean toward simple random or stratified sampling.

4. Tools and techniques of research: validity, reliability, standardisation

Validity — the degree to which a tool actually measures what it claims to measure (content validity: does it cover the intended content domain; construct validity: does it truly capture the theoretical construct; criterion validity: do its scores correlate with an established outside measure).
Reliability — the consistency of a tool's measurements: test-retest reliability (same tool, same people, two occasions, consistent scores), inter-rater reliability (different raters agree), and internal consistency (items within the tool correlate with each other, e.g., Cronbach's alpha).
Standardisation — the process of establishing uniform administration and scoring procedures and norms (typical scores for a defined population) so results from different administrations of the same tool remain comparable.

Types of tools and techniques of research
Tool / technique	What it captures
Rating scale	A respondent or observer rates a trait or performance on a graded scale (poor to excellent)
Attitude scale	Measures direction and intensity of feeling toward an object (Likert-type agree-disagree items, Thurstone or semantic differential formats)
Questionnaire	A set of written questions, self-administered, for gathering standardised data from many respondents efficiently
Aptitude test	Measures potential capacity to learn or perform in a specific domain, predicting future performance rather than assessing past learning
Achievement test	Measures learning already accomplished against a defined curriculum or content domain
Inventory	A structured self-report list (often of interests, personality traits or habits) that a respondent checks against their own characteristics
Observation	Systematic watching and recording of behaviour as it naturally occurs, structured (with predefined categories) or unstructured
Interview	Direct verbal interaction to gather in-depth information, structured (fixed questions), semi-structured (a guide with flexibility) or unstructured (open conversation)
Projective technique	An ambiguous stimulus (an inkblot, an incomplete sentence, a picture) to which a respondent's response is thought to reveal unconscious attitudes or personality traits not accessible through direct questioning

⚠️ Aptitude versus achievement, and reliability versus validity

Aptitude looks forward (can this student succeed in engineering?); achievement looks backward (how much of the class-10 syllabus has this student mastered?). Reliability and validity are related but not the same: a tool can be highly reliable (consistently gives the same score) yet invalid (that consistent score does not measure what it claims to) — a bathroom scale stuck five kilograms too high is perfectly reliable but not valid. Validity requires reliability as a precondition, but reliability alone never guarantees validity.

Key takeaways

The scientific method requires replicability, precision, falsifiability (Popper) and parsimony; research is classified by scientific-method type (exploratory, explanatory, descriptive), aim (problem-solving, theory-building, prediction), purpose (fundamental, applied, action), approach (quantitative, qualitative) and design (descriptive, experimental, historical) — five independent axes.
A variable is independent (cause), dependent (effect), extraneous (must be controlled) or intervening/moderator (explains the mechanism or changes the strength); a hypothesis is null, research/alternative, directional or non-directional.
A sample is drawn from the universe by probability techniques (simple random, stratified, cluster, systematic — allowing generalisation) or non-probability techniques (convenience, purposive, snowball, quota — chosen for reach or depth over generalisability).
A good tool is valid (measures what it claims), reliable (consistent) and standardised (uniform administration and norms); tools include rating and attitude scales, questionnaires, aptitude and achievement tests, inventories, observation, interviews and projective techniques.

Practice questions (8)

Attempt each one before opening the answer. Every explanation names the tempting wrong option as well as the right one, because that is where marks are lost.

Karl Popper's criterion that a genuine scientific claim must be capable, in principle, of being shown false by evidence is called:
1. Parsimony
2. Falsifiability
3. Replicability
4. Precision
Show answer
Answer: B — Falsifiability
Falsifiability is Popper's criterion. Parsimony prefers the simplest adequate explanation; replicability means another researcher should reach the same finding; precision requires unambiguous definition of what is measured.
A teacher systematically studies and improves her own immediate classroom practice over a short, iterative cycle. This is best classified, by purpose, as:
1. Fundamental research
2. Applied research
3. Action research
4. Historical research
Show answer
Answer: C — Action research
Action research is a practitioner's own immediate, iterative practice-improvement study. Fundamental research advances knowledge without immediate application; applied research addresses a practical problem more broadly than one's own classroom; historical research reconstructs past events, an unrelated design.
In an experiment testing whether a new teaching method raises test scores, the new teaching method is the:
1. Dependent variable
2. Independent variable
3. Extraneous variable
4. Moderator variable
Show answer
Answer: B — Independent variable
The teaching method is the presumed cause, manipulated by the researcher — the independent variable. Test scores are the dependent variable (the measured effect); an extraneous variable would be something like prior ability that must be controlled; a moderator variable would change the strength of the relationship for a subgroup.
A study finds a new teaching method raises scores, and further finds that this happens because it increases student engagement, which in turn raises scores. 'Increased engagement' here is functioning as a:
1. Extraneous variable
2. Moderator variable
3. Intervening variable
4. Independent variable
Show answer
Answer: C — Intervening variable
This is the classic intervening-variable pattern: engagement mediates the causal chain, explaining the mechanism between method (independent) and scores (dependent). A moderator would instead show the effect is stronger for one subgroup than another; an extraneous variable is an unwanted outside influence to control for.
A researcher who divides a state's schools into naturally occurring groups (rather than individual students) and randomly selects whole schools to study, because no complete list of individual students exists, is using:
1. Stratified random sampling
2. Cluster sampling
3. Purposive sampling
4. Quota sampling
Show answer
Answer: B — Cluster sampling
Cluster sampling selects whole naturally occurring groups, useful exactly when an individual-level list is unavailable. Stratified sampling divides the universe into strata and samples randomly WITHIN each; purposive and quota sampling are non-probability techniques driven by researcher judgement or fixed quotas, not random cluster selection.
A test designed to predict how well a student is likely to succeed in a future engineering programme, rather than measure content already learned, is a:
1. Achievement test
2. Aptitude test
3. Attitude scale
4. Projective technique
Show answer
Answer: B — Aptitude test
Aptitude tests predict future capacity to learn or perform. Achievement tests measure already-accomplished learning; attitude scales measure feeling toward an object; projective techniques use ambiguous stimuli to reveal unconscious traits, unrelated to predicting academic success directly.
A tool that consistently gives the same score every time it is administered to the same group, but that score does not actually measure the trait it claims to measure, is described as:
1. Valid but unreliable
2. Reliable but invalid
3. Both valid and reliable
4. Neither valid nor standardised
Show answer
Answer: B — Reliable but invalid
Consistency of scores is reliability; whether the score measures the intended trait is validity. A tool can be highly reliable yet invalid, as the bathroom-scale example in this chapter illustrates — reliability does not guarantee validity.
A hypothesis stating 'Method A will produce higher scores than Method B', specifying both that a difference exists and its direction, is a:
1. Null hypothesis
2. Non-directional hypothesis
3. Directional hypothesis
4. Intervening hypothesis
Show answer
Answer: C — Directional hypothesis
A directional hypothesis specifies the direction of the expected difference. A non-directional hypothesis would predict only that A and B differ, without saying which is higher; a null hypothesis predicts no difference at all; 'intervening hypothesis' is not a standard category — intervening describes a type of variable, not a hypothesis.