Sampling Design: Types, Methods & Applications in Business Research

What is Sampling Design?

Sampling design refers to the process of selecting a subset of individuals, objects, or events from a larger population to represent the entire group for research purposes. It ensures that the sample is representative, feasible, and cost-effective while minimizing errors.

Why is Sampling Important?

• Cost-Effective: Studying an entire population (census) is often impractical.
• Time-Saving: Reduces the time required for data collection.
• Accuracy: A well-designed sample can provide results as accurate as a full census.

---

Population vs. Sample

Example:

• Population: All registered voters in Nepal (for an election study).
• Sample: 5,000 randomly selected voters from Kathmandu, Pokhara, and Biratnagar.

---

Types of Sampling Methods

Sampling methods are broadly classified into two categories:

1. Probability Sampling (Every member has a known chance of selection).
2. Non-Probability Sampling (Selection is based on convenience or judgment).

Comparison Table

---

Probability Sampling Methods

1. Simple Random Sampling

• Definition: Every member of the population has an equal chance of being selected.
• How it works:
  • Assign a unique number to each population member.
  • Use a random number generator to select samples.
• Example:
  • eSewa User Study: If eSewa wants to survey 1,000 users out of 10 million, it assigns random numbers to users and selects those matching the generated list.
• Advantages:
  • Unbiased, easy to implement.
• Disadvantages:
  • May not represent subgroups well.

---

2. Stratified Sampling

• Definition: Population is divided into homogeneous subgroups (strata), and samples are taken from each stratum.
• How it works:
  • Divide population into strata (e.g., age groups, income levels).
  • Randomly select samples from each stratum.
• Example:
  • Nepalese Bank Loan Study: A bank divides loan applicants into:
    • Low-income (< Rs. 50,000/month)
    • Middle-income (Rs. 50,000–200,000/month)
    • High-income (> Rs. 200,000/month)
  • Then randomly selects 100 applicants from each group.
• Advantages:
  • Ensures representation of all subgroups.
• Disadvantages:
  • More complex and costly than simple random sampling.

---

3. Cluster Sampling

• Definition: Population is divided into heterogeneous clusters, and entire clusters are randomly selected.
• How it works:
  • Divide population into clusters (e.g., schools, districts).
  • Randomly select clusters and study all members within them.
• Example:
  • NTC Customer Satisfaction Survey: NTC divides Nepal into 7 provinces and randomly selects 2 provinces. It then surveys all customers in those provinces.
• Advantages:
  • Cost-effective for large or dispersed populations.
• Disadvantages:
  • Less precise than stratified sampling.

---

4. Systematic Sampling

• Definition: Select every k-th member from a list after a random start.
• How it works:
  • Determine sampling interval $k=\frac{N}{n}$ (where $N$ = population size, $n$ = sample size).
  • Randomly select a starting point, then pick every k-th member.
• Example:
  • Daraz Order Processing Study: Daraz has 1 million orders. To study 1,000 orders, it selects every 1,000th order starting from a random number (e.g., order #456).
• Advantages:
  • Simple and efficient.
• Disadvantages:
  • Risk of periodicity bias (if the list has hidden patterns).

---

Non-Probability Sampling Methods

1. Convenience Sampling

• Definition: Samples are selected based on ease of access.
• How it works:
  • Use readily available subjects (e.g., students, friends).
• Example:
  • Pathao Driver Survey: A researcher surveys Pathao drivers waiting at a single pickup point in Thapathali.
• Advantages:
  • Quick and inexpensive.
• Disadvantages:
  • Highly biased, not generalizable.

---

2. Snowball Sampling

• Definition: Initial samples refer the researcher to other potential subjects.
• How it works:
  • Start with a few respondents, who then recommend others.
• Example:
  • Underground Economy Study: A researcher interviews a few black-market traders, who introduce them to more traders.
• Advantages:
  • Useful for hard-to-reach populations.
• Disadvantages:
  • Risk of bias from referral networks.

---

3. Quota Sampling

• Definition: Samples are selected to meet predefined quotas (e.g., age, gender).
• How it works:
  • Set quotas for subgroups (e.g., 50% male, 50% female).
  • Select samples until quotas are filled.
• Example:
  • Nepalese Election Poll: A pollster ensures 30% samples are from Terai, 40% from Hills, and 30% from Mountains.
• Advantages:
  • Ensures representation of key groups.
• Disadvantages:
  • Non-random selection can introduce bias.

---

4. Purposive Sampling

• Definition: Samples are selected based on specific characteristics relevant to the study.
• How it works:
  • Researcher handpicks subjects who meet criteria.
• Example:
  • Toyota Hybrid Car Study: Researchers select only Toyota Prius owners for a study on hybrid vehicle satisfaction.
• Advantages:
  • Targets specific groups of interest.
• Disadvantages:
  • Not generalizable to broader populations.

---

Sampling Errors and Biases

Types of Errors

Common Biases

• Selection Bias: Certain groups are over/under-represented.
  • Example: A survey on smartphone use conducted only in urban areas.
• Response Bias: Respondents answer dishonestly or incorrectly.
  • Example: People lying about income in a government survey.
• Non-Response Bias: Those who respond differ from non-respondents.
  • Example: Only wealthy individuals reply to a luxury brand survey.

---

Determining Sample Size

Sample size depends on:

1. Population Size (N): Larger populations require larger samples.
2. Confidence Level: Higher confidence (e.g., 95%) requires larger samples.
3. Margin of Error (E): Smaller margins require larger samples.
4. Population Variability: More variability (e.g., income levels) requires larger samples.

Formula for Sample Size (Simple Random Sampling)

$$n=\frac{N\times X^2\times (1-\frac{X}{N})}{E^2\times (N-1)+X^2}$$ Where:

• $n$ = sample size
• $N$ = population size
• $X$ = number of successes in the population (if known)
• $E$ = margin of error (e.g., 5% = 0.05)

Example:

• Nepalese Youth Unemployment Study:
  • Population ($N$) = 5 million youth.
  • Confidence level = 95% ($X=1.96$ for Z-score).
  • Margin of error ($E$) = 3% (0.03).
  • Sample size ($n$) ≈ 1,068.

---

In the Real World

1. eSewa (Nepal):

   • Stratified Sampling: eSewa divides its user base by age (18–30, 31–50, 50+) and region (Kathmandu, Pokhara, Terai) before selecting samples for usability testing. This ensures feedback from diverse user groups.

2. Daraz (Nepal/Global):

   • Systematic Sampling: Daraz uses systematic sampling to audit 1% of all orders for quality control. Every 100th order is checked, ensuring a representative sample of customer experiences.

3. Nabil Bank (Nepal):

   • Cluster Sampling: Nabil Bank selects entire branches (clusters) from different regions (e.g., Kathmandu, Biratnagar, Butwal) to study customer satisfaction. This reduces costs while covering diverse geographic areas.

4. Google (Global):

   • Snowball Sampling: Google uses snowball sampling to identify influential bloggers for product reviews. Initial reviewers recommend others, expanding the sample organically.

5. NTC (Nepal):

   • Stratified Sampling: NTC divides its customer base by service type (landline, mobile, internet) and region before conducting satisfaction surveys to ensure balanced representation.

---

Worked Example: Kathmandu Traffic Congestion Study

Scenario: A researcher wants to study traffic congestion in Kathmandu. The population is all vehicles in Kathmandu Valley (~1.5 million). The budget allows for 500 samples.

Steps:

1. Define Population: All registered vehicles in Kathmandu.
2. Choose Sampling Method: Stratified Sampling (to represent different vehicle types and regions).
   • Strata:
     • Private cars
     • Public buses
     • Motorcycles
     • Commercial vehicles
   • Regions: Kathmandu, Lalitpur, Bhaktapur.
3. Allocate Samples:
   • Private cars: 200
   • Public buses: 100
   • Motorcycles: 150
   • Commercial vehicles: 50
4. Random Selection: Use random number tables to select vehicles from each stratum.
5. Data Collection: Survey drivers at key checkpoints (e.g., Ring Road, Tribhuvan International Airport).

Why Stratified?

• Ensures motorcycles (a major traffic contributor) and public buses (affecting commuters) are represented.
• Avoids bias toward only private car users.

---

Case Study: Himalayan Java’s Market Research

Company: Himalayan Java (Nepal’s leading coffee brand). Challenge: Determine which new coffee flavor to launch in Pokhara. Sampling Approach:

1. Population: Coffee drinkers in Pokhara (~200,000).
2. Method: Quota Sampling (to ensure representation by age and income).
   • Quotas:
     • Age 18–30: 40%
     • Age 31–50: 40%
     • Age 50+: 20%
     • Income: Low (< Rs. 30,000), Medium (Rs. 30,000–100,000), High (> Rs. 100,000).
3. Sample Size: 500 respondents.
4. Execution:
   • Survey conducted at cafes, supermarkets, and universities.
   • Interviewers ensured quotas were met before stopping data collection.
5. Result:
   • Caramel Mocha was the top choice among all groups, leading to its nationwide launch.

Why Quota Sampling?

• Cost-effective: No need for complex random selection.
• Representative: Ensured feedback from all age and income groups.

---

Exam Tip

What Examiners Look For

1. Understanding Definitions:

   • Clearly distinguish between probability and non-probability sampling.
   • Know the difference between stratified and cluster sampling (homogeneous vs. heterogeneous groups).

2. Application to Real Scenarios:

   • Always relate sampling methods to Nepali businesses (e.g., eSewa, Daraz, NTC).
   • Example question: "How would Nabil Bank use stratified sampling to study loan defaults?" Answer: Divide borrowers by income, loan amount, and region, then randomly sample from each group.

3. Calculations:

   • Be ready to calculate sample size using the formula.
   • Example: "A population of 5,000 with a 95% confidence level and 5% margin of error requires a sample size of __?" Answer: Use the formula to derive n ≈ 370.

4. Advantages/Disadvantages:

   • For each method, list at least two pros and two cons.
   • Example for convenience sampling:
     • ✅ Quick and cheap.
     • ❌ Highly biased, not generalizable.

5. Diagrams and Tables:

   • Draw mindmaps for sampling types (probability vs. non-probability).
   • Use tables to compare methods (as shown above).
   • Label real-world examples clearly (e.g., "eSewa uses stratified sampling").

6. Common Pitfalls:

   • Avoid: Saying "random sampling" without specifying if it’s simple, stratified, or cluster.
   • Avoid: Ignoring biases in non-probability methods.
   • Avoid: Forgetting to justify why a method is chosen (e.g., "stratified sampling ensures representation of all income groups").

High-Scoring Answer Structure

Use this template for exam questions:

1. Define the method (1 mark).
2. Explain how it works (2 marks).
3. Give a Nepali business example (2 marks).
4. List advantages and disadvantages (2 marks).
5. Justify why this method is best for the scenario (2 marks).

Example Question: "Explain systematic sampling with an example from a Nepalese company. What are its limitations?"

Model Answer:

Systematic sampling is a probability method where every k-th member is selected from a list after a random start. For example, Daraz might select every 1,000th order to audit customer service quality. The sampling interval (k) is calculated as $k=\frac{N}{n}$, where $N$ is the total population (e.g., 1 million orders) and $n$ is the sample size (e.g., 1,000 orders).

Advantages:

• Simple and cost-effective.
• Ensures even coverage of the population.

Disadvantages:

• Periodicity bias: If the list has hidden patterns (e.g., orders on Mondays are similar), the sample may not be representative.
• Less control over sample distribution compared to stratified sampling.

Justification: Daraz uses this method because it’s efficient for large datasets and provides a random yet structured way to monitor quality without excessive cost. However, for diverse subgroups (e.g., urban vs. rural customers), stratified sampling would be better.

---

Final Tip: Always link theory to practice. Examiners love answers that show you understand how businesses like eSewa or Nabil Bank apply sampling in real life.