“Write short note : Research Design” (1985)

A research design is the overall plan connecting a research question to the evidence that will answer it, and not every plan that gets the researcher from question to data collection is equally sound.
Cook and Campbell defined validity as “the best available approximation to the truth or falsity of a given inference,” a standard that doubles as the central criterion any design must meet.
Beyond validity, a good design must also satisfy reliability, feasibility, and fit with the nature of the research question itself.
This note sets out these four evaluative criteria — what actually makes a research design good, as distinct from merely describing what a design is or listing its sequential stages.

Internal validity asks whether the design permits the causal or descriptive claim it makes — whether the relationship observed can genuinely be attributed to what the researcher believes is producing it, rather than to some uncontrolled alternative explanation.
External validity asks whether findings generalise beyond the specific sample and setting studied to the broader population the researcher wishes to speak about.
Construct validity concerns whether the operational indicators chosen actually capture the theoretical concept intended — a poorly designed questionnaire item can produce reliable but entirely invalid data.
A design weak on validity produces confident-looking findings that answer the wrong question, a more serious failure than producing no findings at all.

Reliability concerns whether the design, if implemented again under the same conditions, would yield the same results — the consistency of the instrument and procedure rather than the truth of what it measures.
Test-retest reliability checks whether the same instrument applied to the same subjects at two points yields comparable results; internal consistency checks whether items measuring the same concept correlate with one another.
Reliability is necessary but not sufficient for validity — an instrument can be highly consistent yet consistently measuring the wrong thing.
Structured, standardised designs tend to maximise reliability, often at some cost to the validity gains that more flexible, in-depth designs can offer.

No design is good in the abstract; it must be executable within the time, funding, and access actually available to the researcher.
Scientific inquiry should not proceed where adequate data are doubtful to obtain, time constraints make thorough execution impossible, or the cost of the inquiry outweighs the value of what it can plausibly yield.
Ethical feasibility is equally binding: a design requiring deception or exposure of vulnerable respondents to harm may be analytically elegant yet practically unusable, which is why institutional ethics review exists.
Access is a frequently underrated constraint — a design premised on cooperation from a closed community or politically sensitive group can collapse entirely if that access is denied midway through fieldwork.

A design’s soundness cannot be judged independent of the question it serves — the same design can be excellent for one research problem and entirely unsuited to another.
A question that is exploratory in nature, seeking to understand a poorly mapped phenomenon, is best served by a flexible, open-ended design capable of following unexpected leads as they surface in the field.
A question that is explanatory or causal in nature, seeking to establish why a relationship holds, generally demands a more structured design with clearer controls over alternative explanations and more standardised measurement.
Judging a flexible ethnographic design as weak simply because it lacks the statistical rigor of a structured survey design mistakes a criterion appropriate to one kind of question for a universal standard — appropriateness to the question, not adherence to any single design template, is itself a mark of design quality.

Contemporary mixed-methods research increasingly builds validity and appropriateness together into a single design — for instance a sequential design in which an exploratory qualitative phase generates categories that a subsequent quantitative phase tests for generalisability — treating the criteria discussed here as complementary requirements to be satisfied jointly rather than as separate boxes to tick.

A research design earns the label “good” not by following a fixed template but by satisfying validity, reliability, feasibility, and appropriateness to the specific question simultaneously.
These criteria can pull in different directions — the standardisation that strengthens reliability can weaken the contextual validity a flexible design offers — so design choice is ultimately an exercise in judgement, not mechanical compliance.
Evaluating a design against these four criteria, rather than treating any single one as sufficient, remains the most reliable way of distinguishing a genuinely sound research design from a merely plausible one.

Write short note : Research Design

Download PDF