Skip to content

Sources, Acquisition, Classification and Types of Data

UGC NET Paper 1: primary and secondary sources, census and sample, methods of collection, classification, tabulation, and quantitative vs qualitative data.

  • 19 min read
  • Updated
On this page · 14 sections
  1. Prerequisites
  2. The Hook: Where Did That Number Come From?
  3. 1. What “Data” Means
  4. 2. Sources of Data
  5. 3. Acquisition of Data: How Data Are Collected
  6. 4. Classification of Data
  7. 5. Quantitative and Qualitative Data
  8. Memory Tricks
  9. Don’t Confuse
  10. Common Mistakes
  11. Practice
  12. Quick Revision Table
  13. Key Takeaways
  14. What’s Next

Prerequisites

The Hook: Where Did That Number Come From?


1. What “Data” Means

Data are recorded facts or observations: numbers, words, images, sounds or measurements, collected for a purpose. Data become information when they are organised and given context, and information becomes knowledge when it is understood and applied.

Working definitions for the exam
  • Datum / data: an individual observation / a set of observations. In formal writing “data” is plural (“the data are”), though “the data is” is now widely accepted.
  • Variable: a characteristic that can take different values (age, income, marks).
  • Attribute: a qualitative characteristic that can only be present or absent, or fall into categories (literate or not, religion, blood group).
  • Population (universe): the whole group about which information is wanted.
  • Sample: a part of the population actually studied.
  • Parameter vs statistic: a measure of the population (parameter) vs the same measure computed from a sample (statistic).

The common ladder of understanding is Data → Information → Knowledge → Wisdom, often called the DIKW hierarchy. Exam questions use it to test whether you know that raw data come at the bottom and need processing before they inform anything.


2. Sources of Data

Primary and secondary data

BasisPrimary dataSecondary data
MeaningCollected first-hand by the investigator for the present purposeCollected earlier by someone else, reused for a new purpose
OriginalityOriginalAlready “processed” by another agency
Cost and timeHighLow
SuitabilityExactly fits the study’s purposeMay not fit; definitions and units may differ
Accuracy checkInvestigator controls itMust be checked before use
ExamplesYour own survey, interview, experiment, field observationCensus reports, NSO survey reports, RBI publications, journal articles, company annual reports

Sources of secondary data:

  • Published: government reports (Census, NSO survey reports, Economic Survey), publications of the RBI and international bodies (World Bank, IMF, UN agencies), journals, books, newspapers, reports of commissions and committees.
  • Unpublished: records and files of offices, theses, records of trade bodies and firms, research notes.

Before using secondary data, test them for: suitability (do the definitions, units and coverage fit?), adequacy (is the coverage, period and detail enough?) and reliability (who collected them, by what method, with what bias?).

Internal and external data

For an organisation, internal data come from its own records (sales registers, admission records, attendance), while external data come from outside (government statistics, market surveys). This classification is common in business statistics and appears in Paper 1 as a distractor.

Important Indian data sources

SourceAgencyWhat it gives
Census of IndiaOffice of the Registrar General and Census Commissioner, India (Ministry of Home Affairs)Population, housing, literacy, workers, migration; decennial; the last completed census was 2011 (the next has reference date 1 March 2027)
Sample Registration System (SRS)Registrar General of IndiaBirth rate, death rate, infant mortality between censuses
National Sample Survey rounds, Periodic Labour Force Survey (PLFS), Annual Survey of IndustriesNational Statistics Office (NSO), Ministry of Statistics and Programme Implementation (MoSPI)Employment, consumption, industry
National accounts (GDP), Consumer Price Index (CPI), Index of Industrial ProductionNSO, MoSPIMacro-economic statistics
Wholesale Price Index (WPI)Office of the Economic Adviser, DPIIT (Ministry of Commerce and Industry)Wholesale prices
National Family Health Survey (NFHS)Ministry of Health and Family Welfare, with IIPS Mumbai as nodal agencyHealth, nutrition, fertility
AISHE (All India Survey on Higher Education)Ministry of EducationInstitutions, enrolment, teachers in higher education
UDISE+Ministry of EducationSchool education
Banking and monetary dataReserve Bank of IndiaCredit, deposits, money supply, interest rates

The NSO was formed in 2019 when the Central Statistics Office (CSO) and the National Sample Survey Office (NSSO) were merged under MoSPI. The statistical system’s legal backbone is the Collection of Statistics Act, 2008. India marks Statistics Day on 29 June, the birth anniversary of P.C. Mahalanobis, founder of the Indian Statistical Institute.


3. Acquisition of Data: How Data Are Collected

Census vs sample enquiry

Census (complete enumeration)Sample enquiry
Every unit of the population is studiedOnly a representative part is studied
More accurate in principle; no sampling errorHas sampling error, which can be estimated
Costly, slow, needs a large trained staffCheaper, quicker, needs fewer but better-trained staff
Suits small or very important populations (national census)Suits large populations and destructive testing (testing bulbs, tasting food)

Both suffer from non-sampling errors: wrong answers, faulty questions, recording and processing mistakes, non-response. A census can actually have more non-sampling error than a well-run sample because its staff is larger and less trained. Sampling methods (random, stratified, cluster, quota, snowball) are treated in detail in Methods and Steps of Research.

Methods of collecting primary data

Indian statistics textbooks list five classic methods:

MethodHow it worksSuited to
Direct personal investigationThe investigator personally contacts every informantSmall, local, intensive enquiries
Indirect oral investigationInformation from third parties (witnesses) who know the factsWhen informants are unwilling or unable to answer (enquiries into addiction, accidents)
Information from local correspondentsLocal agents send periodic estimatesRegular rough estimates (crop conditions, local prices)
Mailed questionnaireQuestionnaire sent to informants who fill it themselvesLarge, scattered, literate populations
Schedules through enumeratorsTrained enumerators ask questions and fill the scheduleLarge enquiries including non-literate respondents; the Census uses this approach

Other modern modes of acquisition

  • Observation and experiment: recording behaviour or outcomes directly (classroom observation, field trials).
  • Administrative data: records generated by running a programme: school admissions, hospital registers, tax returns, scheme beneficiary lists.
  • Transaction and digital trace data: payments, mobile records, website logs.
  • Sensors and remote sensing: weather stations, air-quality monitors, satellite imagery.
  • Web scraping and APIs: extracting data from websites or receiving it through a programming interface from an open-data portal.
  • Crowdsourcing: many volunteers contribute observations (citizen science, mapping by volunteers).

From raw data to usable data

After collection, data are edited (checked for completeness, consistency and accuracy), coded (answers turned into numerical or category codes), classified, tabulated and then analysed. Cleaning (removing duplicates, fixing errors, handling missing values) is the modern name for much of the editing step.

graph LR
    A["Define purpose
and units"] --> B["Choose source:
primary or secondary"] B --> C["Collect:
census or sample"] C --> D["Edit and code
(clean)"] D --> E["Classify"] E --> F["Tabulate"] F --> G["Present:
graphs, maps"] G --> H["Analyse and
interpret"]

4. Classification of Data

Classification is the process of arranging data into groups or classes according to common characteristics. It condenses the mass of raw data, brings out similarities and differences, makes comparison possible and prepares the ground for tabulation.

Four bases of classification

BasisGroups data byExample
Geographical (spatial)PlaceState-wise literacy rates
Chronological (temporal)TimeYear-wise enrolment, 2015–2025 (a time series)
QualitativeAn attribute that cannot be measured numericallyStudents by gender, religion, or stream
QuantitativeA measurable variable, in class intervalsStudents by marks: 0–10, 10–20, …

Qualitative classification is simple (dichotomous) when there are only two classes (literate / illiterate) and manifold when an attribute is divided into many classes, or when several attributes are used one after another (sex, then literacy within each sex, and so on).

A good classification is: unambiguous, exhaustive (every item has a class), mutually exclusive (no item fits two classes), stable over time, and suitable for the purpose.

Frequency distribution: the vocabulary

Suppose the marks of 20 students are:

12, 25, 37, 41, 18, 29, 33, 45, 22, 38, 27, 31, 14, 49, 35, 26, 39, 43, 28, 34

Grouped in classes of width 10:

Marks (class)Tally count (frequency)Cumulative frequency (less than upper limit)Class mark
10–203315
20–306925
30–4071635
40–5042045
Total20
TermMeaningIn the table
Class limitsThe two end values of a class20 and 30
Class width (size, interval)Upper limit − lower limit10
Class mark (mid-value)\( \frac{\text{lower limit} + \text{upper limit}}{2} \)\( \frac{20+30}{2} = 25 \)
FrequencyNumber of items in a class6 in 20–30
Cumulative frequencyRunning total of frequencies16 up to 40
Relative frequencyFrequency ÷ total\( \frac{7}{20} = 35\% \) for 30–40

Exclusive and inclusive classes. In the exclusive method (10–20, 20–30) the upper limit of one class is the lower limit of the next, and a value equal to the upper limit (20) goes into the next class. In the inclusive method (10–19, 20–29) both limits belong to the class. To convert inclusive classes to continuous ones for a histogram, subtract half the gap (0.5) from each lower limit and add it to each upper limit: 9.5–19.5, 19.5–29.5. An open-end class has no lower or upper limit (“below 10”, “60 and above”).

How many classes? Textbooks quote Sturges’ rule: \( k = 1 + 3.322 \log_{10} N \). For \( N = 100 \), \( k \approx 1 + 3.322 \times 2 = 7.6 \), so about 8 classes. In practice 5 to 15 classes are usual.

Tabulation

Tabulation is the orderly arrangement of classified data in rows and columns. It comes after classification. The parts of a statistical table are:

PartWhereWhat it holds
Table numberTopFor reference
TitleTopWhat, where, when: brief and complete
Head noteBelow the titleUnits (“figures in ₹ crore”)
CaptionsColumn headingsWhat each column shows
StubsRow headings (left column)What each row shows
BodyMain partThe figures
FootnoteBelow the bodyClarification of any item
Source noteBottomWhere the data came from

A simple (one-way) table classifies by one characteristic; a two-way table by two (students by stream and gender); a manifold (complex) table by three or more. The most important comparisons should sit side by side in columns, and totals usually come at the end of rows and columns.


5. Quantitative and Qualitative Data

This is the second official line. The distinction drives everything else: which average you may use, which graph you may draw and which test you may apply.

BasisQuantitative (numerical) dataQualitative (categorical) data
NatureNumbers that measure or countCategories, labels, words, descriptions
Answers“How much?”, “How many?”“What kind?”, “Which group?”
ExamplesHeight, income, marks, number of booksGender, religion, language, grade A/B/C, interview transcripts
Levels of measurementInterval, ratioNominal, ordinal
Summary measuresMean, median, standard deviationFrequency, percentage, mode (median for ordinal)
Usual graphsHistogram, frequency polygon, line chart, scatter plotBar chart, pie chart
Research traditionSurveys, experiments, statisticsInterviews, observation, coding into themes

Discrete and continuous quantitative data

  • Discrete data take separate, countable values, usually whole numbers: number of students, number of books, number of accidents.
  • Continuous data can take any value in a range, limited only by the precision of measurement: height, weight, time, temperature.

Levels of measurement (NOIR)

S.S. Stevens (1946) gave the four levels; the same scheme is used in the research methods topic.

LevelType of dataWhat you can sayExamplePermitted average
NominalQualitativeEqual / not equal (labels only)Blood group, roll number, PIN codeMode
OrdinalQualitative (ordered)Greater / less (order, unequal gaps)Ranks, grades, Likert responsesMedian, mode
IntervalQuantitativeDifferences are meaningful; no true zeroTemperature in °C, calendar yearsMean, median, mode
RatioQuantitativeRatios are meaningful; true zeroIncome, weight, age, marks out of 100All, including geometric mean

40 °C is not “twice as hot” as 20 °C, because 0 °C is not “no heat” (interval scale). A person earning ₹40,000 does earn twice as much as one earning ₹20,000 (ratio scale).

Other ways of typing data

PairFirstSecond
Time-series vs cross-sectionalOne unit over many periods (India’s GDP, 2000–2025)Many units at one point in time (GDP of all states in 2024–25)
Panel (longitudinal)The same units observed over several periods (the same 1,000 households surveyed every year)—
Structured vs unstructuredFits rows and columns (spreadsheets, databases)No fixed format (text, images, video, social media posts); semi-structured sits between (JSON, XML, emails with headers)
Raw vs processedAs collectedEdited, classified, summarised
Data vs metadataThe observationsData about data: who collected them, when, definitions, units

Qualitative data in research

Qualitative data are not just “categories”. In research they are rich, non-numerical material: interview transcripts, field notes, documents, photographs. They are analysed by coding (labelling segments of text), grouping codes into themes, and interpreting patterns, often with software such as NVivo or ATLAS.ti. They can be quantified (counting how often a theme appears), but the analysis aims at meaning rather than measurement.

Big data

Very large, fast-moving and varied data sets are called big data. Their defining features are usually summarised as the three Vs: volume, velocity and variety, with veracity (trustworthiness) and value often added to make five. Big data matter to governance and are taken up again in Data and Governance.


Memory Tricks

Don’t Confuse

ThisvsThat
Primary: collected by the user for this purposevsSecondary: collected by someone else, reused
Questionnaire: respondent fillsvsSchedule: enumerator fills
Direct personal investigation: investigator meets informantsvsIndirect oral investigation: investigator questions third parties
Census: every unit, no sampling errorvsSample: part of the units, sampling error
Classification: grouping into classesvsTabulation: arranging classified data in rows and columns
Captions: column headingsvsStubs: row headings
Exclusive classes (10–20, 20–30)vsInclusive classes (10–19, 20–29)
Discrete: countedvsContinuous: measured
Interval: no true zero (°C)vsRatio: true zero (income)
Time-series: one unit, many periodsvsCross-sectional: many units, one period
Variable: measured numericallyvsAttribute: qualitative characteristic

Common Mistakes


Practice

All questions below are original practice questions written in the UGC NET format. They are not previous-year questions.

Level 1: Foundation

  1. A researcher downloads district-level tables from the Census website. For her, these are ______ data.
  2. Find the class mark of the class 45–55.
  3. Name the level of measurement of the ranks given to students in a debate.
Answers
  1. Secondary data: collected by the Census office, reused by the researcher.
  2. \( \frac{45 + 55}{2} = 50 \).
  3. Ordinal: ranks show order but not equal gaps.

Level 2: Exam-style MCQs

Q1. (Practice) A study of drug abuse among college students collects information by interviewing wardens, teachers and parents rather than the students themselves. The method is:

  • (a) Direct personal investigation
  • (b) Indirect oral investigation
  • (c) Mailed questionnaire
  • (d) Information from local correspondents
Answer and explanation

(b) Indirect oral investigation. Information is gathered from third parties who know the facts, which is the usual method when the persons concerned are unwilling to give information directly.

Q2. (Practice) Which of the following is an example of continuous data?

  • (a) Number of colleges in a district
  • (b) Number of books in a library
  • (c) Time taken by students to finish a test
  • (d) Number of questions attempted
Answer and explanation

(c). Time is measured and can take any value in a range. The other three are counts, hence discrete.

Q3. (Practice) Consider the following statements:

  1. In a statistical table, stubs are the headings of rows.
  2. The source note is placed immediately below the title.
  3. Classification precedes tabulation.

Which of the statements given above is/are correct?

  • (a) 1 only
  • (b) 1 and 3 only
  • (c) 2 and 3 only
  • (d) 1, 2 and 3
Answer and explanation

(b) 1 and 3 only. The head note (units) sits below the title; the source note is at the bottom of the table.

Q4. (Practice) On which scale of measurement is “temperature in degrees Celsius” measured?

  • (a) Nominal
  • (b) Ordinal
  • (c) Interval
  • (d) Ratio
Answer and explanation

(c) Interval. Differences are meaningful, but 0 °C is not an absence of temperature, so ratios such as “twice as hot” are not.

Q5. (Practice) In a frequency distribution with classes 0–10, 10–20, 20–30, …, an observation of exactly 20 is placed in:

  • (a) 10–20
  • (b) 20–30
  • (c) Either class, at the investigator’s choice
  • (d) Both classes, with half weight each
Answer and explanation

(b) 20–30. In exclusive classes the upper limit is excluded from a class and belongs to the next.

Level 3: Assertion–reason, match and sequence

For assertion–reason questions, use these options:

  • (a) Both A and R are true, and R is the correct explanation of A
  • (b) Both A and R are true, but R is not the correct explanation of A
  • (c) A is true, but R is false
  • (d) A is false, but R is true

Q1. (Practice) Assertion (A): Sample surveys can sometimes give more accurate results than a complete census. Reason (R): A sample needs a smaller field staff, which can be better trained and supervised, so non-sampling errors can be reduced.

Answer and explanation

(a). A census avoids sampling error but needs a huge staff, which raises non-sampling error. A smaller, better-trained team can more than make up for the sampling error a sample introduces. R explains A.

Q2. (Practice) Assertion (A): The arithmetic mean is an appropriate average for data on blood groups. Reason (R): Blood groups are measured on a nominal scale.

Answer and explanation

(d). A is false: nominal data permit only the mode. R is true.

Q3. (Practice) Match List I with List II.

List I (Data)List II (Level of measurement)
A. Religion of respondentsI. Ratio
B. Grades A, B, C, DII. Interval
C. Year of birth (calendar year)III. Nominal
D. Monthly incomeIV. Ordinal
  • (1) A-III, B-IV, C-II, D-I
  • (2) A-IV, B-III, C-II, D-I
  • (3) A-III, B-IV, C-I, D-II
  • (4) A-III, B-II, C-IV, D-I
Answer and explanation

(1) A-III, B-IV, C-II, D-I. Calendar years are interval data: the starting point of the calendar is a convention, not “no time”, so 2000 is not “twice” 1000 in any meaningful sense.

Q4. (Practice) Arrange the following stages of handling data in the correct order:

  • A. Tabulation

  • B. Collection

  • C. Editing and coding

  • D. Classification

  • E. Interpretation

  • (1) B, C, D, A, E

  • (2) B, D, C, A, E

  • (3) C, B, D, A, E

  • (4) B, C, A, D, E

Answer and explanation

(1) B, C, D, A, E. Collect, then edit and code, then classify, then tabulate, and only then interpret.


Quick Revision Table

FactDetail
Primary vs secondaryFirst-hand for this purpose vs collected by others and reused
Test secondary dataSuitability, adequacy, reliability
Five classic primary methodsDirect personal, indirect oral, local correspondents, mailed questionnaire, schedules via enumerators
Census vs sampleAll units, no sampling error vs part of units, sampling error; both have non-sampling error
NSOFormed 2019 by merging CSO and NSSO, under MoSPI
Statistics Day29 June, birth anniversary of P.C. Mahalanobis
Bases of classificationGeographical, chronological, qualitative, quantitative
Class mark(lower limit + upper limit) ÷ 2
Sturges’ rule\( k = 1 + 3.322 \log_{10} N \)
Parts of a tableNumber, title, head note, captions, stubs, body, footnote, source note
Discrete vs continuousCounted vs measured
NOIRNominal, ordinal, interval (no true zero), ratio (true zero)
Big dataVolume, velocity, variety (+ veracity, value)

Key Takeaways

What’s Next