28 Sources, Acquisition and Classification of Data

Data are facts, observations or measurements recorded for analysis. The path from raw data to useful information runs through three stages: identifying sources, choosing methods of acquisition, and classifying the data so it can be analysed.

28.1 Sources of Data

Primary vs Secondary Sources

Source	Definition	Examples	Strength	Limitation
Primary	Collected first-hand by the researcher	Survey, experiment, observation, interview	Tailored to the question; full control over quality	Time-consuming; expensive
Secondary	Collected by someone else, repurposed	Census, NSSO, AISHE, journal articles, RBI reports	Quick, low-cost, often large-scale	Original purpose may not match the new question; quality outside researcher’s control

Major Indian Secondary-Data Sources

Census of India — decennial demographic data (Office of the Registrar General and Census Commissioner).
National Sample Survey Office (NSSO) — household consumption, employment, health, education.
National Statistical Office (NSO) — successor body that consolidates several survey programmes.
All India Survey on Higher Education (AISHE) — annual data on higher education institutions.
Reserve Bank of India (RBI) — financial and monetary statistics.
Open Government Data Platform (data.gov.in) — central government datasets.
State of Forest Report (Forest Survey of India) — forest cover and biodiversity.
Sample Registration System (SRS) — births, deaths, fertility.
Periodic Labour Force Survey (PLFS) — labour-market indicators.
National Family Health Survey (NFHS) — health, nutrition, fertility.

28.2 Methods of Data Acquisition

Six Working Methods of Data Acquisition

Method	What it captures	Best when
Survey / Questionnaire	Self-reported opinions, behaviour, demographics	Large samples, literate populations
Interview	Spoken responses; structured / semi-structured / unstructured	Sensitive topics, low-literacy contexts
Observation	Behaviour as it happens	Classroom interaction, child development
Experiment	Manipulated cause and observed effect	Causal claims
Document analysis	Existing records	Historical, policy work
Sensor / device data	Real-time machine readings	Traffic, weather, IoT, fitness trackers

28.2.1 Census vs Sample Survey

Census vs Sample

Approach	Coverage	Cost	Accuracy	Use
Census	Every unit in the population	High	Highest (no sampling error)	Decennial population count, foundational planning data
Sample survey	A representative subset	Lower	Subject to sampling error	Routine household, business and labour surveys

28.3 Classification of Data

Data may be classified along several independent dimensions.

28.3.1 By Source

Internal — generated by an organisation itself (sales records, payroll).
External — drawn from sources outside the organisation (government reports, market research).

28.3.2 By Nature

Quantitative vs Qualitative

Type	Description	Examples
Quantitative	Numerical measurements	Height, marks, income
Qualitative / Categorical	Categories or attributes	Gender, religion, occupation

28.3.3 By Measurement Level (Stevens 1946)

Four Levels of Measurement

Level	What it permits	Example
Nominal	Naming and counting	Gender, religion
Ordinal	Rank ordering	Educational attainment, satisfaction (Likert)
Interval	Equal intervals; no absolute zero	Temperature in Celsius, year
Ratio	Equal intervals; absolute zero	Income, height, weight, age

28.3.4 By Time

Time-Based Classification

Type	Description	Example
Time-series	Same indicator measured at successive time points	Monthly inflation, annual GDP
Cross-section	Different units at the same time	Household survey on a single date
Panel / Longitudinal	Same units at successive time points	NSSO panel, NFHS rounds

28.3.5 By Frequency

Discrete — countable values (number of children).
Continuous — values on a continuum (height, time).

flowchart TB
  D[Data Classification] --> S[By Source]
  D --> N[By Nature]
  D --> M[By Measurement Level]
  D --> T[By Time]
  S --> S1[Internal]
  S --> S2[External]
  N --> N1[Quantitative]
  N --> N2[Qualitative]
  M --> M1[Nominal · Ordinal · Interval · Ratio]
  T --> T1[Time-series]
  T --> T2[Cross-section]
  T --> T3[Panel]
    classDef default fill:#003366,color:#ffffff,stroke:#ffcc00,stroke-width:3px,rx:10px,ry:10px;

28.4 Working Stages: Editing, Coding, Tabulation

After collection, raw data goes through three working stages before analysis.

Three Stages of Data Preparation

Stage	Purpose
Editing	Identify and correct errors, missing values, inconsistencies
Coding	Assign numeric labels to categorical responses (M = 1, F = 2)
Tabulation	Arrange in tables, frequency distributions, cross-tabs

28.5 Frequency Distribution

A frequency distribution counts how often each value (or class of values) appears in the data.

Frequency Distribution Vocabulary

Class interval — a range of values (e.g., 50–60).
Class boundary — exact upper and lower limits (49.5–60.5).
Class width — upper − lower (60 − 50 = 10).
Class mark / mid-point — average of upper and lower (55).
Frequency (f) — number of observations in the class.
Cumulative frequency — sum of frequencies up to and including the class.
Relative frequency — frequency as a proportion of total.

28.6 Practice Questions

Q 01 Primary vs Secondary Easy

A researcher conducts her own household survey on consumption patterns. The data she collects is classified as:

APrimary data
BSecondary data
CTertiary data
DArchival data

View solution

Correct Option: A

Data collected first-hand by the researcher is primary data.

Q 02 Census Easy

The Census of India is conducted at intervals of:

AFive years
BTen years (decennial)
CTwenty years
DAnnually

View solution

Correct Option: B

The Census of India is decennial — conducted every 10 years.

Q 03 Levels of Measurement Medium

A data set lists "religion" as Hindu / Muslim / Christian / Other. This is at which level of measurement?

ANominal
BOrdinal
CInterval
DRatio

View solution

Correct Option: A

Categories with no inherent order = nominal.

Q 04 AISHE Medium

AISHE — the All India Survey on Higher Education — is conducted by:

AReserve Bank of India
BMinistry of Education, Government of India
CUNICEF
DWorld Bank

View solution

Correct Option: B

AISHE is conducted by the Ministry of Education, Government of India.

Q 05 Time Classification Medium

Annual GDP figures of India from 1991 to 2024 are an example of:

ACross-section data
BTime-series data
CPanel data
DCategorical data

View solution

Correct Option: B

Same indicator measured at successive time points = time-series data.

Q 06 Discrete vs Continuous Easy

Which of the following is an example of *continuous* quantitative data?

ANumber of children in a household
BHeight of a person
CType of religion
DMarital status

View solution

Correct Option: B

Height can take any value on a continuum. Number of children is discrete; religion and marital status are categorical.

Q 07 Class Width Medium

In the class interval 50–60, the class width is:

View solution

Correct Option: B

Class width = upper limit − lower limit = 60 − 50 = 10.

Q 08 Stages of Preparation Easy

Assigning the labels "M = 1, F = 2" to gender responses is which stage of data preparation?

AEditing
BCoding
CTabulation
DSampling

View solution

Correct Option: B

Coding = assigning numeric labels to categorical responses.

Quick recall

Two sources: Primary (first-hand) vs Secondary (re-used).
Indian secondary sources: Census, NSSO/NSO, AISHE, RBI, NFHS, PLFS, data.gov.in.
Six acquisition methods: Survey, Interview, Observation, Experiment, Document analysis, Sensor data.
Classification dimensions: Source, Nature, Measurement level, Time, Frequency.
Stevens’s levels: Nominal · Ordinal · Interval · Ratio (NOIR).
Time forms: Time-series · Cross-section · Panel.
Three preparation stages: Editing · Coding · Tabulation.