Beginner

26 min read

Collection of Data — Primary & Secondary Sources, Sampling Methods

Master Collection of Data for CBSE Class 11 Statistics for Economics. Covers primary data vs secondary data definition differences with Indian examples NSO NSSO RBI Census, precautions for secondary data, five methods of primary data collection direct personal indirect oral enumerators mail telephone merits limitations, qualities of good questionnaire pilot survey, census vs sample investigation comparison, population sample sampling unit frame, random sampling simple random systematic k=N/n formula stratified cluster, non-random purposive quota convenience, sampling error vs non-sampling error with 7 board-level questions.

New

Every statistical investigation begins with a single question: where does the data come from? Data is the raw material of all economic analysis — without reliable, well-collected data, no meaningful conclusion is possible. This chapter answers the foundational question of how data is collected. It covers the two fundamental sources of data (primary and secondary), the methods of primary data collection, the qualities of a good questionnaire, and the critical distinction between studying an entire population (census) versus studying a representative part of it (sampling). For CBSE Class 11, this topic produces 1-mark definitions, 3-mark comparisons, and 4-mark explanations in board exams — the distinctions between primary/secondary and census/sample are particularly reliable sources of questions.

1. What is Data?

Data are facts and figures — raw, unprocessed numerical information collected for a specific purpose. In economics, data could be prices of commodities, output of industries, wages of workers, literacy rates, or population figures.

Types of Data — Quantitative and Qualitative

Type	Definition	Example
Quantitative	Data that can be expressed numerically — measured or counted	Income ₹35,000; Temperature 38°C; 500 students enrolled
Qualitative	Data describing attributes or characteristics that cannot be directly measured in numbers	Gender (male/female); Religion; Satisfaction level (good/bad)

2. Sources of Data — Primary and Secondary

Data can be obtained from two broad sources:

Basis	Primary Data	Secondary Data
Definition	Data collected for the first time, directly from original sources, by the investigator for a specific purpose	Data that have already been collected and published by someone else for their own purpose; used in a secondary capacity
Original?	Yes — first-hand, fresh, original	No — second-hand; already processed
Collection method	Surveys, interviews, questionnaires, observation — directly by investigator	Published reports, government databases, newspapers, journals
Cost and time	Expensive and time-consuming	Cheaper and quicker — already available
Accuracy	More accurate — tailored to exact purpose	May not fit the exact purpose; must be evaluated for reliability
Precautions	Proper questionnaire design, trained investigators, adequate sample	Must check source reliability, definition used, time period, and purpose of original collection
Indian examples	NSO/NSSO directly conducting surveys on household expenditure, employment	A researcher using RBI Annual Report data; a student using Census 2011 population figures

⚠️ Important: The same data can be primary for one user and secondary for another. Example: When NSSO collects household data by direct survey → Primary for NSSO. When a researcher uses the published NSSO report → Secondary for that researcher.

Sources of Secondary Data — Where to Find It

Source Type	Examples
Government publications	Census of India; National Sample Survey (NSS); RBI Handbook of Statistics; Economic Survey of India; NITI Aayog reports; Annual Survey of Industries
International organisations	World Bank Development Report; IMF World Economic Outlook; UN Human Development Report; WHO Health Statistics
Semi-government / autonomous bodies	NABARD reports; SEBI Annual Reports; Stock Exchange data
Private publications	Newspapers (The Hindu, Economic Times); research journals; industry associations; NGO reports

Precautions When Using Secondary Data

Before using secondary data, a researcher must verify:

Reliability of source: Is it a reputable government or international agency, or an anonymous website?
Suitability for the current purpose: Were the definitions used the same as what you need? (e.g., poverty line definition may vary)
Adequacy: Does the data cover the required time period, region, and population?
Accuracy: Were the methods of collection described? Were there known biases?
Time period: Is the data sufficiently recent, or outdated?

3. Methods of Collecting Primary Data

When primary data is needed, the investigator must choose the most appropriate method of collection. Five main methods exist:

Method	How it works	Merits	Limitations
1. Direct Personal Investigation	Investigator personally visits each respondent and collects data face-to-face	Most reliable and accurate; can clarify doubts; observe reactions	Very expensive and time-consuming; personal bias possible; limited geographic coverage
2. Indirect Oral Investigation	Investigator interviews third parties (witnesses, experts) who have knowledge of the subject rather than the concerned persons themselves	Useful when direct contact with concerned persons is not possible; wider coverage	Data may be biased or inaccurate (hearsay); witnesses may not have full information
3. Questionnaire through Enumerators	Trained enumerators visit respondents and fill the questionnaire by asking questions	Good response rate; enumerators can explain questions; covers illiterate respondents	Training enumerators is costly; enumerator bias possible
4. Questionnaire by Mail/Post	Questionnaire is sent to respondents by post/email; they fill it and return	Cheap; covers wide geographic area; respondent has time to think	Low response rate; only literate respondents; no one to clarify doubts; late/no returns
5. Telephone / Online Interview	Data collected via telephone calls or online forms (increasingly common)	Quick, cheap, wide reach; easy to compile digital data	Limited to those with phones/internet; no in-depth probing; possible refusals

4. Qualities of a Good Questionnaire

A questionnaire is a set of questions used to collect data from respondents. Its quality determines the quality of the entire survey. Characteristics of a good questionnaire:

Clear and simple language: Questions should be easily understood — no ambiguous, complex or technical language.
Logical sequence: Questions should flow in a logical order — general to specific.
Limited number of questions: Short questionnaires get better response rates — avoid unnecessary questions.
No leading questions: Questions must not suggest the desired answer (e.g., "Don't you agree that prices are too high?" is a leading question).
Specific and objective: Questions must have clear, definite answers — avoid vague questions like "How often do you buy vegetables?"
Confidentiality: Personal or sensitive questions should assure respondents that their answers are confidential.
Pre-tested (Pilot Survey): The questionnaire should be tested on a small sample before the main survey — called a pilot survey — to identify problems and refine questions.

What is a Pilot Survey?

A Pilot Survey (or Pre-test) is a small-scale trial run of the questionnaire on a limited number of respondents before the main survey. It helps:

Identify ambiguous or confusing questions
Test the time required to complete the survey
Reveal any missing important questions
Check if the method of collection is feasible

5. Census vs Sample Investigation

Once the source and method are decided, the investigator must choose how many units to study. Two approaches exist:

Basis	Census (Complete Enumeration)	Sample Investigation
Definition	Every unit of the population is studied	Only a representative subset (sample) of the population is studied
Coverage	100% — no unit is excluded	Partial — only selected units
Cost and time	Very expensive and time-consuming	Cheaper and faster
Accuracy	High — no sampling error; but non-sampling errors can occur	Subject to sampling error; but intensive study possible
When used	Small population; when complete coverage is essential (Census of India)	Large populations; when cost/time is a constraint; destructive testing
Indian example	Census of India (every 10 years — studies every household)	NSS surveys; agricultural surveys; election exit polls

Key Terms — Population, Sample, Sampling Unit

Term	Definition	Example
Population (Universe)	The complete set of ALL units under study	All students in India; all factories in a state
Sample	A representative subset of the population selected for study	500 students selected from all students in India
Sampling Unit	The individual unit selected for the sample	A household, a farm, a student
Sampling Frame	The complete list of all units in the population from which the sample is drawn	Electoral rolls; list of all schools in a district

6. Sampling Methods

Sampling methods fall into two broad categories:

A. Random (Probability) Sampling

Every unit in the population has a known, non-zero probability of being selected. Selection is free from personal bias. Results can be used to make statistical inferences about the whole population.

Type	How it works	Example
Simple Random Sampling	Every unit has an equal chance; selection by lottery method (chits drawn) or random number table	Names of 1000 employees written on chits; 50 drawn at random for a survey
Systematic Sampling	Every kth unit selected from a numbered list; k = N/n (N = population size, n = sample size); first unit randomly chosen from 1 to k	N=500, n=50 → k=10. If first unit = 7, select: 7, 17, 27, 37... (every 10th unit)
Stratified Sampling	Population divided into non-overlapping subgroups (strata) based on a characteristic; random sample drawn from each stratum	Strata: urban / rural. Random sample of 200 from urban and 300 from rural; ensures both groups are represented
Cluster Sampling	Population divided into clusters (natural groupings); entire clusters randomly selected and all units within studied	India's 700 districts = clusters; 50 districts randomly chosen; all households in those districts surveyed

B. Non-Random (Non-Probability) Sampling

Not every unit has a known probability of selection. Based on the researcher's judgement, convenience, or quotas. Results cannot be used to make statistical inferences about the full population. Cheaper and faster, but potentially biased.

Type	How it works	When used
Purposive / Judgement	Researcher deliberately selects units believed to be most representative	Selecting expert economists for a panel discussion
Quota Sampling	Field workers fill specified quotas for each group (e.g., 50 men, 50 women)	Market research surveys; opinion polls
Convenience Sampling	Whoever is easily accessible is selected	Student surveys interviewing friends or classmates

7. Sampling and Non-Sampling Errors

Error Type	Definition	Present in Census?	Present in Sample?
Sampling Error	Difference between the sample estimate and the true value of the population parameter — arises because a sample (not the full population) is studied	❌ No (no sampling involved)	✅ Yes — inherent in sampling
Non-Sampling Error	Errors arising from defects in data collection regardless of whether census or sample — due to faulty questionnaire, enumerator bias, respondent lying, processing mistakes	✅ Yes	✅ Yes

How to reduce sampling error: Increase the sample size — larger samples are more representative. However, increasing sample size also increases cost.

Key insight: A large sample with non-sampling errors can be less accurate than a smaller, well-designed sample. Quality of data collection matters as much as quantity.

Collection of Data — Quick Classification Reference

Situation	Classification	Reason
NSO conducts a field survey on wages	Primary (for NSO)	Collected directly for the first time
A researcher uses NSO's published wage report	Secondary (for researcher)	Already published by someone else
N=400, n=40 → systematic sampling, k = ?	k = 10	k = N/n = 400/40
Sampling error present in census?	No	Census studies entire population — no sampling
Non-sampling error present in census?	Yes	Enumerator bias, wrong answers exist in any study
Population divided into strata; random sample from each	Stratified Sampling	Each sub-group (stratum) represented in sample
Trial run of questionnaire on small group	Pilot Survey	Pre-test before main survey to identify problems

Board Exam Power Tips — Avoid These Common Mistakes

Primary vs Secondary depends on WHO collected it, not WHAT it is. The same data (e.g., NSS employment figures) is primary for the NSS organisation that collected it and secondary for any researcher, planner or student who uses the published report. Board MCQs frequently test this with "which of the following is secondary data for a researcher?" — always ask for whom?
Sampling Error ≠ Mistake. Sampling error is not a mistake — it is the natural, unavoidable difference between the sample estimate and the true population value. It exists even with a perfectly conducted sample survey. It can be reduced by increasing sample size but cannot be eliminated entirely (unlike non-sampling errors, which can be reduced by better methods).
Non-sampling errors can occur in BOTH census AND sample surveys. A common misconception is that census has no errors. Census has no sampling error but is very prone to non-sampling errors — enumerator bias, respondent errors, processing mistakes — because of the enormous scale of operation.
Systematic Sampling formula: k = N/n — always calculate this. Board numericals frequently give N (population) and n (sample size) and ask for the sampling interval k. Example: N = 600, n = 30 → k = 20. Then if first unit randomly drawn = 5, the sample includes 5, 25, 45, 65... Show this working in the answer.
A leading question is a flaw in a questionnaire — always a wrong answer in MCQs. "Don't you think inflation is too high?" is a leading question — it pushes the respondent toward a "yes" answer. Any question that suggests the answer or uses emotionally loaded language is leading and must be avoided.
Stratified ≠ Quota sampling. Both divide the population into groups — but stratified sampling uses random selection within each stratum (probability method), while quota sampling uses non-random filling of quotas by field workers (non-probability method). This distinction is tested in MCQs.

Practice Questions (CBSE Board Level)

Q1 (1 mark): Distinguish between Primary and Secondary data in one sentence each.

Explanation:

Primary data: Data collected for the very first time, directly from the original source by the investigator for a specific, predetermined purpose—meaning it is entirely original and first-hand.

Secondary data: Data that has already been collected, compiled, and published by someone else for their own original purpose—meaning it is second-hand information being reused by a different investigator.

Q2 (3 marks): Distinguish between Census and Sample Investigation.

Explanation:

Basis	Census Investigation	Sample Investigation
Coverage	Every single unit of the population is studied (100% coverage).	Only a representative subset (sample) of the population is studied.
Cost and Time	Highly expensive and extremely time-consuming.	Significantly cheaper and much faster to execute.
Sampling Error	Completely absent — because no sampling is involved.	Present — the sample estimate will naturally differ slightly from the true population value.
Example	The Decennial Census of India (every household is formally counted).	NSS (National Sample Survey) household expenditure surveys.

Q3 (2 marks): In a systematic sampling of a population of 600 workers, a sample of 30 is to be drawn. Find the sampling interval ( $k$ ). If the first unit selected is 8, list the first five units in the sample.

Explanation:

The formula for the Sampling Interval ( $k$ ) is:
$k = \frac{N}{n}$
Where $N$ is total population and $n$ is sample size.

$k = \frac{600}{30} = 20$ .

The first unit selected randomly is 8. Subsequent units are selected by continuously adding the interval $k$ (20):
Unit 1 = 8
Unit 2 = 8 + 20 = 28
Unit 3 = 28 + 20 = 48
Unit 4 = 48 + 20 = 68
Unit 5 = 68 + 20 = 88

First five units: 8, 28, 48, 68, 88.

Q4 (MCQ — 1 mark): Which of the following is a clear example of Primary data?

A) Data on India's GDP directly from the RBI Annual Report.
B) Population data from the Census of India 2011 utilized by a university researcher.
C) A student directly interviewing local farmers about their annual income.
D) Employment statistics published online by the NSSO.

Answer: C) A student directly interviewing local farmers about their annual income.

Explanation: This scenario involves fresh, first-hand data collection directly from the original source (the farmers) exclusively for the student's specific research purpose—which is the exact definition of primary data. Options A, B, and D all involve leveraging published data that was originally collected by someone else (RBI, Government, NSSO); therefore, they represent secondary data for the end user.

Q5 (3 marks): What is a Pilot Survey? Why must it be conducted before the main survey?

Explanation:

A Pilot Survey is a small-scale trial run or "pre-test" of the questionnaire and overall data collection methodology, conducted on a limited, representative number of respondents before the actual main survey begins.

Reasons for conducting a pilot survey:

Identify ambiguous questions: It acts as a safety net to reveal questions that respondents misunderstand, interpret differently than intended, or simply find difficult/offensive to answer. These can be reworded or deleted.
Estimate time and resources: It provides a realistic, real-world estimate of exactly how long the survey takes to complete per person, helping finalize budget and staffing needs.
Detect missing areas: Crucial topics may have been overlooked during the questionnaire design phase. Pilot testing often reveals these logical gaps.
Test feasibility: It confirms that the chosen collection method (e.g., mail vs. in-person) actually works in practice, not just in theory.

Q6 (MCQ — 1 mark): Sampling error is present strictly in:

A) Census only
B) Sample survey only
C) Both census and sample survey
D) Neither census nor sample survey

Answer: B) Sample survey only.

Explanation: Sampling error arises fundamentally because only a fraction or part of the population is being studied. By definition, the sample estimate will naturally differ slightly from the true population value. Since a Census studies all units in the population, there is absolutely no sampling involved, and therefore no sampling error. (Note: Both census and sample surveys suffer from non-sampling errors like enumerator bias, refusal to answer, or data entry mistakes).

Q7 (4 marks): Explain any four methods of collecting primary data. State one merit and one limitation for each.

Explanation:

Direct Personal Investigation: The investigator personally visits and interviews each respondent face-to-face.
Merit: Most accurate and reliable, as doubts can be clarified immediately by reading body language.
Limitation: Extremely expensive and time-consuming; physically limited to small geographic areas.
Indirect Oral Investigation: Third parties or witnesses who have knowledge of the subject (e.g., asking a union leader about workers' conditions) are interviewed instead of the directly concerned persons.
Merit: Highly useful when direct access to the concerned persons is impossible or dangerous.
Limitation: Information may be heavily biased or incomplete since it relies entirely on second-hand accounts.
Questionnaire through Enumerators: Trained enumerators (field workers) visit respondents and physically fill out the questionnaire by asking questions out loud.
Merit: High response rate and successfully covers illiterate respondents.
Limitation: Hiring and properly training enumerators adds massive costs; enumerator bias/prompting may taint the data.
Questionnaire by Mail/Post: The physical or digital questionnaire is sent directly to respondents, who fill it out and return it independently.
Merit: Very cheap and can easily cover a massive, nationwide geographic area.
Limitation: Notoriously low response rates; strictly limited to literate respondents; no investigator is present to clarify complex questions.

Related Study Material

More Economics Notes More Statistics For Economics Notes All Collection Of Data Topics

Practice Questions

Economics Practice Collection Of Data Questions

Topic Information

DifficultyBeginner

Est. Read Time26 minutes

CategoryCollection Of Data

Quick Actions

Track Your Learning

Study Tips

Take notes of key points while reading

Practice related questions after studying

Review topics marked as "Learning" regularly

Test your knowledge with practice questions

Edvaya Target

Detailed Notes & Solved Examples

Master complex concepts with our rich, topic-wise study notes. Access hundreds of step-by-step solved examples and interactive flashcards right from your phone.

Edvaya Target

Edvaya Aspire

Take a Free Mock Test

Target Subjects

Aspire Subjects

Subject Mastery

Collection of Data — Primary & Secondary Sources, Sampling Methods

1. What is Data?

Types of Data — Quantitative and Qualitative

2. Sources of Data — Primary and Secondary

Sources of Secondary Data — Where to Find It

Precautions When Using Secondary Data

3. Methods of Collecting Primary Data

4. Qualities of a Good Questionnaire

What is a Pilot Survey?

5. Census vs Sample Investigation

Key Terms — Population, Sample, Sampling Unit

6. Sampling Methods

A. Random (Probability) Sampling

B. Non-Random (Non-Probability) Sampling

7. Sampling and Non-Sampling Errors

Collection of Data — Quick Classification Reference

Practice Questions (CBSE Board Level)

Related Study Material

Practice Questions

Topic Information

Quick Actions

Track Your Learning

Study Tips

Edvaya Target

We Value Your Privacy