Skip to content

Research Aptitude Formula Sheet and Short Notes

One-page CSIR NET Part A research aptitude revision: study types, hypotheses, variables, controls, sampling, statistics, ethics, authorship and metrics.

  • 6 min read
  • Updated
On this page · 10 sections
  1. Part A in one table
  2. Types of research and study designs
  3. Research process and hypotheses
  4. Variables
  5. Reasoning
  6. Controls and design
  7. Sampling and bias
  8. Measurement and statistics
  9. Ethics, publication and metrics
  10. Where to go next

Every definition, formula and trap from Research Process, Hypotheses and Scientific Reasoning, Experimental Design, Sampling and Data Analysis and Research Ethics, Publication and Scientific Communication, on one page.

Part A in one table

Per the NTA information bulletin for December 2026, the same for all five subjects:

GivenAttemptCorrectWrongMaximum
2015+2−0.530

Only the first 15 answered questions are evaluated. Research aptitude is named on CSIR-HRDG’s subject pages, not in the bulletin, and has no official topic list. Details on the General Aptitude overview.

Types of research and study designs

AxisTypes
PurposeBasic (understanding) vs applied (practical problem); action research (practitioners improve their own practice)
DepthExploratory → descriptive → explanatory
DataQuantitative vs qualitative vs mixed
ControlExperimental (researcher assigns) vs observational
TimeCross-sectional (one point) vs longitudinal (followed over time)
LogicDeductive (theory → test) vs inductive (observations → generalisation)
  • Cohort: exposure → forward to outcome. Case–control: outcome → back to exposure. Cross-sectional: both at once.
  • Only random assignment breaks the link between treatment and everything else.
  • Evidence ladder (interventions): opinion < case reports < cross-sectional < case–control < cohort < RCT < systematic review / meta-analysis.

Research process and hypotheses

Problem → literature review → hypothesis and objectives → design → data collection → analysis → interpretation → report → new questions.

TermKey point
HypothesisTestable, falsifiable, specific, parsimonious; stated before the test
\( H_0 \)No effect; you reject or fail to reject it, never prove it
\( H_1 \)An effect exists; one-tailed (direction stated) or two-tailed
TheoryBroad, well-tested explanation (not a guess)
LawDescription of a regularity
Operational definitionExactly how a variable is measured

Variables

VariableRole
IndependentChanged by the researcher, or used to group subjects
DependentMeasured outcome
ControlledHeld constant
ConfoundingVaries with the independent variable and affects the outcome
MediatingOn the causal path between the two
ModeratingChanges the size or direction of the effect

Reasoning

FormValid?
If P then Q; P; so Q (modus ponens)Valid
If P then Q; not Q; so not P (modus tollens)Valid
If P then Q; Q; so P (affirming the consequent)Invalid
If P then Q; not P; so not Q (denying the antecedent)Invalid
  • A confirmed prediction supports; a failed prediction can refute.
  • Association may be causation, reverse causation, confounding, chance or selection.
  • Bradford Hill considerations: strength, consistency, specificity, temporality (essential), gradient, plausibility, coherence, experiment, analogy.

Controls and design

ControlRules out
NegativeFalse positive from background or contamination
PositiveFalse negative from a failed assay
VehicleEffect of the solvent
ShamEffect of the procedure
PlaceboEffect of expectation
Loading / internalDifferences in amount of material
  • Fisher’s principles: replication, randomisation, local control (blocking).
  • Experimental unit = smallest unit independently treated. Sub-samples are technical replicates; counting them as independent is pseudoreplication.
  • Designs: CRD (uniform units), RBD (one gradient), Latin square (two gradients), factorial (interactions; combinations = product of levels), paired / crossover.
  • Blinding: single (subjects), double (subjects + assessors).
  • Look-alike effects: placebo, Hawthorne, regression to the mean, observer bias.

Sampling and bias

MethodRule
Simple randomEvery sample of size n equally likely
SystematicRandom start, every k-th, \( k = N/n \)
StratifiedRandom units from every stratum; proportional allocation \( n_i = n \frac{N_i}{N} \)
ClusterAll units from randomly chosen clusters
Convenience, purposive, quota, snowballNon-probability; weak for generalisation

Bias: selection, non-response, survivorship, recall, measurement, confirmation, publication. A bigger sample cuts random error, not bias.

Measurement and statistics

ScaleZeroExample
Nominal—Blood group
Ordinal—Disease grade
IntervalArbitrary°C
RatioTrueMass, K, concentration
  • Accuracy = near the true value (systematic error spoils it; fix by calibration). Precision = near each other (random error spoils it; repeating helps).
  • Reliability ≈ consistency; validity ≈ measuring the right thing.
$$\boxed{s^2 = \frac{\sum (x_i - \bar{x})^2}{n-1}} \qquad \boxed{\text{SE} = \frac{s}{\sqrt{n}}} \qquad \boxed{\text{CV} = \frac{s}{\bar{x}}\times 100\%}$$
  • Normal data: about 68 / 95 / 99.7% within 1 / 2 / 3 SD. 95% CI \( \approx \bar{x} \pm 1.96\,\text{SE} \).
  • Right skew: mean > median > mode.
  • \( -1 \leq r \leq 1 \); \( r \) measures linear association only; \( r^2 \) = share of variation explained.
  • Simpson’s paradox: a trend in every subgroup reverses on pooling when a confounder is unevenly spread.
\( H_0 \) true\( H_0 \) false
RejectType I (\( \alpha \)), false positivePower \( 1 - \beta \)
Fail to rejectCorrectType II (\( \beta \)), false negative
$$\boxed{\text{Expected false positives} = m\alpha} \qquad \boxed{P(\geq 1) = 1 - (1-\alpha)^m} \qquad \boxed{\text{Bonferroni: } \alpha/m}$$
MeasureFormula
Sensitivity\( \frac{TP}{TP + FN} \)
Specificity\( \frac{TN}{TN + FP} \)
PPV\( \frac{TP}{TP + FP} \) (falls with prevalence)
NPV\( \frac{TN}{TN + FN} \)

Ethics, publication and metrics

TopicKey point
MisconductFFP: fabrication (invent), falsification (alter or omit), plagiarism (take credit)
QRPsp-hacking, HARKing, selective reporting, salami slicing, duplicate publication, gift and ghost authorship
Similarity indexScreening figure only
UGC 2018 levels0: ≤ 10%; 1: > 10–40%; 2: > 40–60%; 3: > 60%
Human researchRespect for persons, beneficence, justice; informed voluntary consent, right to withdraw; IEC review; child assent
Animal research3Rs: replacement, reduction, refinement; IAEC review
BiosafetyIBSC for recombinant DNA work; BSL-1 to BSL-4
DataBound, dated notebook; raw data kept unaltered; FAIR: findable, accessible, interoperable, reusable
Authorship (ICMJE)All four: contribution, drafting/revising, final approval, accountability
Paper structureIMRaD: Introduction, Methods, Results and Discussion
SourcesPrimary (original data), secondary (reviews), tertiary (textbooks, encyclopaedias)
Peer reviewSingle-blind, double-blind, open; preprints are not peer reviewed
After publicationCorrection (honest error), expression of concern, retraction
Open accessGold, green, hybrid, diamond; beware predatory journals
PatentNovel, inventive step, industrial application; about 20 years; file before publishing
CopyrightProtects expression, not ideas
$$\boxed{\text{IF}_Y = \frac{\text{citations in } Y \text{ to items of } Y-1,\ Y-2}{\text{citable items of } Y-1,\ Y-2}}$$
  • h-index: largest h with h papers cited at least h times. i10: papers with at least 10 citations.
  • Impact factor describes a journal; h-index describes an author.

Where to go next

Advertisement
Advertisement