Sampling Method – Types, Advantages & Drawbacks
"Tasting one spoon of soup tells you if the whole pot needs salt!" - That's sampling! 🥄
What is Sampling?
[!NOTE] Definition: A method where only a selected portion (sample) of the population is studied, and conclusions are drawn about the entire population.
Population: The complete group we want to study Sample: A representative subset of the population
Real-Life Example:
- Blood Test: Doctor takes 5ml blood, not all blood from your body!
- Food Tasting: Chef tastes one spoonful, not the entire dish
- TRP Ratings: Survey 1000 households to estimate what 30 crore households watch
Key Terms 📖
| Term | Meaning | Example |
|---|---|---|
| Population/Universe | Total group to study | All voters in India (90 crore) |
| Sample | Selected part | 10,000 voters surveyed |
| Sampling Unit | Individual element | One voter |
| Sample Size | Number of units in sample | n = 10,000 |
| Sampling Frame | List of all units | Electoral roll |
| Parameter | Population characteristic | Average income of all Indians |
| Statistic | Sample characteristic | Average income of sampled 10,000 |
Why Use Sampling? 🤔
graph TD
A[Why Sampling?] --> B[Saves Money]
A --> C[Saves Time]
A --> D[Practical Feasibility]
A --> E[Detailed Study Possible]
B --> B1[Survey 1000 vs 1 crore people]
C --> C1[Results in days vs years]
D --> D1[Destructive testing - bulbs, crackers]
E --> E1[More detailed questions possible]
Types of Sampling Methods
A. Probability Sampling (Random)
Every unit has a known, non-zero chance of being selected.
1️⃣ Simple Random Sampling
- Method: Lottery, Random number table
- Example: Pick 50 students from 500 by drawing chits
[!TIP] How to do:
- Number all units (1 to N)
- Use random number table or software
- Select required sample size
Advantages:
- ✅ Unbiased
- ✅ Easy to understand
- ✅ Statistical theory applicable
Disadvantages:
- ❌ Need complete list (Sampling Frame)
- ❌ May not represent subgroups
2️⃣ Systematic Sampling
- Method: Select every kth unit
- Formula: k = N/n (k = sampling interval)
- Example: From 1000 students, select every 10th = 100 students
Steps:
- Calculate k = 1000/100 = 10
- Randomly pick starting point (say, 7)
- Select: 7, 17, 27, 37... (every 10th)
Advantages:
- ✅ Easy to execute
- ✅ Spreads across population
Disadvantages:
- ❌ If list has pattern, bias may occur
- ❌ Example: If every 10th house is a corner house!
3️⃣ Stratified Sampling
- Method: Divide population into strata (homogeneous groups), then sample from each
Example:
graph TD
A[College Students: 1000] --> B[Arts: 400]
A --> C[Science: 300]
A --> D[Commerce: 300]
B --> B1[Sample 40]
C --> C1[Sample 30]
D --> D1[Sample 30]
style B1 fill:#90EE90
style C1 fill:#90EE90
style D1 fill:#90EE90
Types:
- Proportionate: Sample size proportional to stratum size
- Disproportionate: Equal from each stratum
Advantages:
- ✅ Ensures representation of all groups
- ✅ More accurate than simple random
Disadvantages:
- ❌ Need prior knowledge of strata
- ❌ Complex
4️⃣ Cluster Sampling
- Method: Divide area into clusters, select few clusters, study ALL units in selected clusters
Example:
graph LR
A[India] --> B[Divide into Districts]
B --> C[Randomly select 50 districts]
C--> D[Survey ALL villages in these 50]
Difference from Stratified:
- Stratified: Sample from ALL strata
- Cluster: Study FEW clusters completely
Advantages:
- ✅ Economical (travel to fewer areas)
- ✅ No need for complete list
Disadvantages:
- ❌ Higher sampling error
- ❌ Clusters may not represent population
B. Non-Probability Sampling (Non-Random)
Not every unit has a chance. Selection is subjective/convenient.
1️⃣ Convenience Sampling
- Method: Select whoever is easily available
- Example: Interview people in a mall (whoever agrees)
Use: Quick, preliminary studies
❌ Drawback: Highly biased
2️⃣ Judgment/Purposive Sampling
- Method: Expert selects "typical" units
- Example: Political analyst selects 10 "bellwether" constituencies
✅ Advantage: Expert knowledge used ❌ Drawback: Subjective, biased
3️⃣ Quota Sampling
- Method: Fixed quota for each group (like stratified but non-random)
- Example: Interview 30 males, 30 females (first 30 of each you find)
Used in: Market research, opinion polls
4️⃣ Snowball Sampling
- Method: Respondents refer other respondents
- Example: Study drug users - one user refers other users
Use: Hard-to-reach populations
Comparison Table: Probability vs Non-Probability
| Aspect | Probability | Non-Probability |
|---|---|---|
| Selection | Random | Subjective |
| Bias | Low | High |
| Statistical Inference | Possible | Not possible |
| Cost | Higher | Lower |
| Examples | Random, Stratified | Convenience, Quota |
| Use | Scientific research | Market research |
Advantages of Sampling ✅
[!NOTE] Key Benefits:
- Economical: Costs 1/100th of census
- Quick: Results in weeks vs years
- Feasible: Can study large populations
- Detailed: Can ask more questions
- Destructive Testing: Only option (can't test all bulbs!)
- Flexibility: Can repeat frequently
Example: Exit polls during elections = sample 10,000 voters in 2 hours!
Disadvantages of Sampling ❌
[!CAUTION] Limitations:
- Sampling Error: Sample may not perfectly represent population
- Need Expertise: Wrong sampling = wrong conclusions
- Limited Detailed Analysis: Can't study very small subgroups
- Sampling Frame: Need a good list (often unavailable)
Example: If you survey only mobile phone owners about poverty, you'll get wrong results (poor people may not have phones!)
Factors Affecting Sample Size
graph TD
A[What Determines Sample Size?] --> B[Variability in Population]
A --> C[Desired Precision]
A --> D[Confidence Level]
A --> E[Budget]
B --> B1[More variety = Larger sample]
C --> C1[More accuracy = Larger sample]
D --> D1[Higher confidence = Larger sample]
E --> E1[More money = Larger sample]
Rule of Thumb:
- Minimum: 30 (for statistical tests)
- Small survey: 100-500
- National survey: 1000-5000
- Election polls: 10,000+
Exam Quick Points 📝
[!IMPORTANT] For Exam Remember:
Probability Sampling:
- Simple Random = Lottery
- Systematic = Every kth
- Stratified = Divide into groups, sample from all
- Cluster = Divide into groups, study few groups completely
When to Use:
- Large population → Sampling
- Small population → Census
- Destructive testing → MUST use sampling
- High precision needed → Census (if possible)
Summary
- Sampling = Study part, conclude about whole
- Types: Probability (Random, Systematic, Stratified, Cluster) & Non-Probability (Convenience, Judgment, Quota, Snowball)
- Advantages: Quick, cheap, practical
- Disadvantages: Sampling error, needs expertise
- Key: Sample must be representative!
Quote: "A good sample is a miniature photograph of the population!" 📸
Test Your Knowledge
Question 1 of 5
1. In which sampling method is every kth unit selected?