Confidence intervals & margin of error • 2026 edition
Basic Sample Size: \( n = \frac{Z^2 \cdot p(1-p)}{E^2} \)
With Population Correction: \( n = \frac{n_0}{1 + \frac{n_0-1}{N}} \)
Margin of Error: \( E = Z \cdot \sqrt{\frac{p(1-p)}{n}} \)
Confidence Level Z-scores: 90% (1.645), 95% (1.96), 99% (2.576)
Finite Population Correction: \( n_{adj} = \frac{n}{1 + \frac{n-1}{N}} \)
Sample size calculation determines the minimum number of observations needed to achieve reliable statistical results. It depends on confidence level, margin of error, population proportion, and population size. Larger samples provide more precise estimates but require more resources.
| Confidence Level | Margin of Error | Sample Size |
|---|
Sample size is the number of observations or participants in a study required to achieve statistically significant results. It determines the precision and reliability of statistical estimates. A properly calculated sample size ensures that a study has sufficient power to detect meaningful effects while controlling for Type I and Type II errors.
\( n = \frac{Z^2 \cdot p(1-p)}{E^2} \)
Where Z is the Z-score for the confidence level, p is the expected proportion, and E is the margin of error.
\( n_{adj} = \frac{n}{1 + \frac{n-1}{N}} \)
Used when population size N is known and relatively small compared to the calculated sample size.
Which of the following would result in the SMALLEST required sample size?
The answer is C) 90% confidence level, 5% margin of error. Sample size is directly proportional to confidence level (higher confidence = larger sample) and inversely proportional to the square of the margin of error (smaller margin = larger sample). Therefore, the combination of lowest confidence level (90%) and largest margin of error (5%) will yield the smallest sample size.
The sample size formula shows that sample size increases with higher confidence levels (because the Z-score is larger) and decreases with larger margins of error (because the denominator E² is larger). When choosing parameters, researchers must balance the need for precision and confidence against resource constraints. The relationship between parameters is crucial for making informed decisions about study design.
Confidence Level: Probability that the true parameter falls within the confidence interval
Margin of Error: Range of values above and below the sample statistic
Sample Size: Number of observations in a study
• Sample size ∝ Confidence level
• Sample size ∝ 1/(Margin of Error)²
• Sample size ∝ p(1-p)
• Higher confidence = larger sample
• Smaller error = larger sample
• Proportion of 50% maximizes sample size
• Confusing direct and inverse relationships
• Forgetting that margin of error is squared
• Not considering population size effects
Calculate the sample size needed for a 95% confidence level with a 4% margin of error and an expected proportion of 30%. Use the basic sample size formula. Show your work.
Step 1: Identify the parameters
Confidence level = 95%, so Z = 1.96
Margin of error (E) = 4% = 0.04
Expected proportion (p) = 30% = 0.30
Step 2: Apply the formula
n = (Z² × p × (1-p)) / E²
n = (1.96² × 0.30 × (1-0.30)) / 0.04²
n = (3.8416 × 0.30 × 0.70) / 0.0016
n = (3.8416 × 0.21) / 0.0016
n = 0.806736 / 0.0016
n = 504.21
Step 3: Round up to whole number
n = 505 participants needed
The sample size formula balances precision and confidence. The Z-score represents the desired confidence level, the proportion p(1-p) reflects the expected variability in the data, and the margin of error E defines the desired precision. The formula shows that sample size increases quadratically as we demand more precision (smaller E).
Z-Score: Number of standard deviations from the mean
Proportion: Fraction of population with specific characteristic
Sample Size Formula: n = (Z² × p × (1-p)) / E²
• Always round sample size up
• Use Z-score for confidence level
• Square the margin of error in denominator
• Round up to ensure minimum sample
• Use p=0.5 for conservative estimate
• Remember margin of error is squared
• Forgetting to square the Z-score
• Not squaring the margin of error
• Forgetting to round up the result
A market researcher wants to estimate the percentage of customers who prefer online shopping with 95% confidence and a margin of error of 3%. Based on previous studies, about 60% of customers prefer online shopping. What sample size is needed? If the company has 10,000 customers, how does this affect the sample size?
Step 1: Calculate basic sample size
Z = 1.96 (for 95% confidence)
p = 0.60, 1-p = 0.40
E = 0.03
n = (1.96² × 0.60 × 0.40) / 0.03²
n = (3.8416 × 0.24) / 0.0009
n = 0.921984 / 0.0009 = 1,024.43
Basic sample size = 1,025
Step 2: Apply finite population correction
N = 10,000
n_adj = n / (1 + (n-1)/N)
n_adj = 1,025 / (1 + (1,024)/10,000)
n_adj = 1,025 / (1 + 0.1024) = 1,025 / 1.1024 = 929.8
Adjusted sample size = 930
The finite population correction reduces the required sample size from 1,025 to 930.
This problem demonstrates how finite population size affects sample size requirements. When the population is relatively small compared to the calculated sample size, we can reduce the sample size using the finite population correction formula. This adjustment recognizes that sampling a significant portion of the population provides more information than sampling from an infinite population.
Finite Population Correction: Adjustment for known population size
Survey Design: Planning methodology for data collection
Market Research: Collecting and analyzing consumer data
• Use finite correction when n/N > 0.05
• Always round sample size up
• Correction reduces required sample size
• Apply finite correction for small populations
• Use p=0.5 for maximum sample size estimate
• Consider practical constraints when designing surveys
• Forgetting to apply finite population correction
• Using wrong formula for finite populations
• Not considering practical sampling constraints
A quality control manager needs to determine the sample size for inspecting products. The manager wants 99% confidence that the sample proportion of defective items is within 2% of the true proportion. Historical data suggests about 8% of items are defective. How many items should be inspected? What would happen if the confidence level was reduced to 95%?
Step 1: Calculate sample size for 99% confidence
Z = 2.576 (for 99% confidence)
p = 0.08, 1-p = 0.92
E = 0.02
n = (2.576² × 0.08 × 0.92) / 0.02²
n = (6.635776 × 0.0736) / 0.0004
n = 0.488393 / 0.0004 = 1,220.98
For 99% confidence: 1,221 items needed
Step 2: Calculate sample size for 95% confidence
Z = 1.96 (for 95% confidence)
n = (1.96² × 0.08 × 0.92) / 0.02²
n = (3.8416 × 0.0736) / 0.0004
n = 0.282741 / 0.0004 = 706.85
For 95% confidence: 707 items needed
Reducing confidence level from 99% to 95% reduces sample size from 1,221 to 707 items (about 42% reduction).
This example shows how confidence level dramatically affects sample size requirements. The Z-score increases from 1.96 (95%) to 2.576 (99%), which squares to approximately 3.84 to 6.64, more than doubling the required sample size. This demonstrates the cost-benefit trade-off in statistical design: higher confidence comes at the expense of significantly larger samples.
Quality Control: Process of maintaining product standards
Defective Items: Products that fail to meet quality standards
Statistical Trade-offs: Balancing precision, confidence, and cost
• Higher confidence = much larger sample
• Z-score increases exponentially with confidence
• Consider practical constraints in quality control
• Evaluate cost-benefit of higher confidence levels
• Use historical data for proportion estimates
• Consider stratified sampling for quality control
• Not considering practical feasibility of large samples
• Using unrealistic proportion estimates
• Forgetting to square the Z-score in calculations
Which expected proportion would require the LARGEST sample size, assuming all other parameters remain constant?
The answer is C) 50%. The sample size formula includes the term p(1-p), which represents the variance of a proportion. This value reaches its maximum when p = 0.5 (50%). At 50%, p(1-p) = 0.5 × 0.5 = 0.25. At 10%, p(1-p) = 0.1 × 0.9 = 0.09. At 25%, p(1-p) = 0.25 × 0.75 = 0.1875. At 75%, p(1-p) = 0.75 × 0.25 = 0.1875. Therefore, 50% requires the largest sample size.
The p(1-p) term in the sample size formula represents the variance of a binomial proportion. This value peaks at p = 0.5 and decreases symmetrically as p moves away from 0.5. This is because the maximum uncertainty occurs when the probability of success equals the probability of failure (50-50 chance). For proportions near 0% or 100%, there's less uncertainty, requiring smaller samples for the same level of precision.
Proportion Variance: p(1-p) term in sample size calculation
Binomial Proportion: Outcome with two possible values
Maximum Uncertainty: Occurs at 50% probability
• p(1-p) is maximized at p = 0.5
• Maximum occurs at 50% proportion
• Symmetric around 50%
• Use p = 0.5 for conservative sample size estimate
• Proportions near 0% or 100% require smaller samples
• Maximum uncertainty at 50%
• Not understanding why 50% requires largest sample
• Confusing the p(1-p) relationship
• Forgetting that variance peaks at 50%
Q: What is the difference between confidence level and confidence interval?
A: Confidence level and confidence interval are related but different concepts:
For example, a 95% confidence level means that if we repeated the study 100 times, about 95 of the confidence intervals would contain the true value.
Q: How do researchers determine appropriate sample sizes in practice?
A: Researchers consider multiple factors:
Often, researchers start with calculated minimum sample size and adjust based on practical feasibility.