Simple Random Sample: Every individual in the population has an equal chance of being selected, and every possible combination of individuals has an equal chance of forming the sample.
- Assign a unique number to each of the 1200 students (1 to 1200)
- Use a random number generator to select 100 unique numbers
- Select the students corresponding to those numbers
Each student receives a unique ID from 1 to 1200
Use random number table, calculator, or computer to select 100 numbers
Contact students whose IDs match the selected random numbers
Assign each student a number (1-1200), use random selection to choose 100 students. This ensures equal probability for all students.
• Equal Probability: Every individual has the same chance of selection
• Independence: Selection of one individual doesn't affect others
• Representative: Large samples reflect population characteristics
Stratified Sample: Population divided into distinct subgroups (strata) based on shared characteristics, then simple random samples taken from each stratum proportional to its size.
Four strata: Freshmen (40%), Sophomores (30%), Juniors (20%), Seniors (10%)
Freshmen: 40% of 200 = 80, Sophomores: 30% of 200 = 60, etc.
Randomly select 80 freshmen, 60 sophomores, 40 juniors, and 20 seniors
Select 80 freshmen, 60 sophomores, 40 juniors, and 20 seniors using simple random sampling within each group.
• Proportional Allocation: Sample size from each stratum reflects population proportion
• Within-Stratum Homogeneity: Individuals within strata are similar to each other
• Between-Stratum Heterogeneity: Strata are different from each other
Cluster Sample: Population divided into clusters (usually geographic), entire clusters randomly selected, and all members of selected clusters included in sample.
Each of the 50 elementary schools represents a cluster
Randomly select 10 schools from the 50 available
Survey all students in the 10 selected schools
Randomly select 10 schools from 50, then survey all students in those schools. This is efficient and cost-effective.
• Cluster Homogeneity: Each cluster should represent the entire population
• Within-Cluster Heterogeneity: Members of each cluster should vary widely
• Cost Efficiency: Reduces travel and administrative costs
Proportional Allocation: Sample sizes from each stratum are proportional to the stratum's size in the population.
18-30: 25% of 1200 = 0.25 × 1200 = 300
31-50: 35% of 1200 = 0.35 × 1200 = 420
51-65: 25% of 1200 = 0.25 × 1200 = 300
65+: 15% of 1200 = 0.15 × 1200 = 180
300 + 420 + 300 + 180 = 1200 ✓
Select 300 from 18-30, 420 from 31-50, 300 from 51-65, and 180 from 65+ age groups.
• Proportional Formula: Sample size = (Stratum percentage) × (Total sample size)
• Verification: Sum of all stratum samples equals total sample size
• Representativeness: Proportional allocation maintains population structure
Method Selection Criteria: Choose sampling method based on population characteristics, resource constraints, and precision requirements.
200 schools across 20 districts - geographically dispersed
Traveling to 200 schools would be expensive and time-consuming
Simple random would require visiting scattered schools; cluster sampling allows efficient grouping
Cluster sampling by districts would be most practical and cost-effective
Cluster sampling by districts is recommended because it's cost-effective while maintaining reasonable accuracy.
• Cost-Benefit Analysis: Balance precision needs with resource constraints
• Geographic Considerations: Cluster sampling efficient for spatially distributed populations
• Accuracy Trade-offs: Understand sampling method impact on precision
Population: The entire group of individuals or items we want to study and draw conclusions about
Sample: A subset of the population selected for study
Parameter: A numerical characteristic of the population (e.g., population mean μ)
Statistic: A numerical characteristic of the sample (e.g., sample mean x̄)
Bias: Systematic error that causes sample results to consistently deviate from true population values
- Representativeness: A good sample should mirror the population's characteristics
- Randomness: Chance selection reduces bias and allows statistical inference
- Sample Size: Larger samples generally provide more accurate estimates
- Sampling Frame: Complete list of all population members eligible for selection
• Randomization: Essential for reducing bias and enabling statistical inference
• Proportionality: Stratified sampling maintains population proportions in the sample
• Efficiency: Cluster sampling maximizes data collection efficiency for geographically dispersed populations
• Accuracy vs Cost: Balance desired precision with available resources
• Generalizability: Results can only be generalized to the population sampled
Questions & Answers
Question: How do I know which sampling method to use for a given situation? Is there a decision tree I can follow?
Answer: Yes, here's a practical decision framework:
- Homogeneous Population? → Simple Random Sampling
- Distinct Subgroups? → Stratified Sampling (if you want representation from all groups)
- Geographically Dispersed? → Cluster Sampling (for efficiency)
- Large Population? → Consider systematic sampling as an alternative
Additional Factors: Consider budget, time constraints, and required precision. Simple random is ideal but often impractical for large populations.
Example: Surveying all employees in a company → Simple random. Surveying customers by region → Cluster. Surveying students by grade level → Stratified.
Question: What's the difference between stratified sampling and quota sampling? Aren't they the same thing?
Answer: These are fundamentally different methods with important distinctions:
- Stratified Sampling: Uses random selection within each stratum - rigorous and unbiased
- Quota Sampling: Uses convenience or judgment sampling to fill quotas - non-probability method
Key Difference: In stratified sampling, you randomly select from each group. In quota sampling, you select the first available individuals until you reach your quota.
Example: For 100 women and 100 men - Stratified: Randomly select from all women/men. Quota: Ask until you find 100 women and 100 men.
Only stratified sampling allows for probability-based statistical inference!
Question: How does sample size affect the accuracy of results, and how do I calculate the optimal sample size?
Answer: Sample size has a crucial relationship with accuracy:
- Direct Relationship: Larger samples = smaller sampling error
- Diminishing Returns: Accuracy improves with √n, so doubling sample size only improves accuracy by ~40%
- Formula: Standard error = σ/√n, where σ is population standard deviation
Optimal Sample Size Calculation: Depends on desired margin of error, confidence level, and population variability. For proportions: n = (Z²pq)/E² where Z is confidence multiplier, p is estimated proportion, q=1-p, E is margin of error.
In practice, researchers balance statistical requirements with budget and time constraints!