Key Takeaways
- A confidence interval gives a range likely to contain the true population parameter
- 95% confidence means if you repeated the sampling 100 times, ~95 intervals would contain the true mean
- Larger sample sizes produce narrower (more precise) confidence intervals
- Higher confidence levels produce wider intervals (trading precision for certainty)
- The margin of error equals z-score times standard error
What Is a Confidence Interval?
| Confidence Level | Alpha (α) | Z-score (z*) | Notes |
|---|---|---|---|
| 80% | 0.20 | 1.282 | Less common; wider intervals sacrificed for smaller sample |
| 85% | 0.15 | 1.440 | Rarely used in formal research |
| 90% | 0.10 | 1.645 | Common in business and industry |
| 95% | 0.05 | 1.960 | Most common in scientific literature |
| 99% | 0.01 | 2.576 | High-stakes research; medical, pharmaceutical |
| 99.5% | 0.005 | 2.807 | Rarely used; very conservative |
| 99.9% | 0.001 | 3.291 | Extreme caution; particle physics standard |
| Note: Z-score is the number of standard deviations from the mean in a standard normal distribution. For a 95% CI, α = 0.05, so α/2 = 0.025 — we use z* = 1.96 (the value where 97.5% of the normal curve falls below, leaving 2.5% in each tail). | |||
| Sample Size (n) | Margin of Error | Notes |
|---|---|---|
| 30 | ±0.356 | Small sample; wide margin |
| 50 | ±0.277 | |
| 100 | ±0.196 | Common minimum for many studies |
| 200 | ±0.138 | Meaningful improvement; diminishing returns begin |
| 500 | ±0.088 | Solid precision for many applications |
| 1,000 | ±0.062 | Polling standard; ≈ ±3% for proportions |
| 2,000 | ±0.044 | Large survey; national polling level |
| 5,000 | ±0.028 | Very high precision; scientific study |
| 10,000 | ±0.020 | Diminishing returns; rarely worth cost |
| Note: ME = z* × (σ/√n) = 1.96 × (1/√n). Halving the margin of error requires quadrupling the sample size. For proportion estimates (p̂), σ is replaced with √(p̂(1−p̂)), which is maximized at p̂ = 0.5 (σ_max ≈ 0.5). | ||
| Scenario | Formula | When to Use | Notes |
|---|---|---|---|
| Mean (σ known) | x̄ ± z* × (σ/√n) | Large sample; population σ known | Use Z-distribution; σ = population standard deviation |
| Mean (σ unknown) | x̄ ± t* × (s/√n) | Any sample; σ unknown (typical) | Use t-distribution; df = n−1; s = sample standard deviation |
| Proportion | p̂ ± z* × √(p̂(1−p̂)/n) | Binary outcomes (yes/no, success/fail) | Requires np̂ ≥ 10 and n(1−p̂) ≥ 10 |
| Difference of means | (x̄₁−x̄₂) ± t* × SE | Comparing two group means | SE = pooled or Welch standard error |
| Note: The t-distribution approaches the Z-distribution as sample size increases (t* ≈ z* for n > 100). For proportions near 0 or 1 (rare events), use Wilson interval or exact binomial CI instead of the standard normal approximation. | |||
A confidence interval (CI) is a range of values that is likely to contain an unknown population parameter. Instead of estimating the parameter by a single value, an interval is given along with a confidence level that specifies how certain we are that the parameter lies within the interval.
For example, if you calculate a 95% confidence interval for the mean height of adult males and get [5'8", 5'11"], you can say: "We are 95% confident that the true average height of adult males in the population falls between 5'8" and 5'11"."
The Confidence Interval Formula
CI = x +/- z * (s / sqrt(n))
Z-Score Reference Table
| Confidence Level | Z-Score | Alpha (a) |
|---|---|---|
| 80% | 1.282 | 0.20 |
| 90% | 1.645 | 0.10 |
| 95% | 1.960 | 0.05 |
| 99% | 2.576 | 0.01 |
| 99.9% | 3.291 | 0.001 |
How to Interpret Confidence Intervals
A common misconception is that a 95% confidence interval means there's a 95% probability that the true parameter is within that specific interval. The correct interpretation is:
- If we repeated our sampling procedure many times and calculated a 95% CI each time, approximately 95% of those intervals would contain the true population parameter.
- Any single interval either contains the true value or it doesn't - we just don't know which case applies.
- The 95% refers to the reliability of the method, not the probability for any single interval.
Factors That Affect Confidence Interval Width
1. Sample Size (n)
Larger sample = Narrower interval. Doubling your sample size reduces the margin of error by about 30% (dividing by sqrt(2)). This is why larger studies provide more precise estimates.
2. Confidence Level
Higher confidence = Wider interval. A 99% CI is wider than a 95% CI because we need a larger range to be more certain we've captured the true value.
3. Standard Deviation
Higher variability = Wider interval. When data is more spread out, there's more uncertainty about where the true mean lies.