CI for a Mean — Confidence Interval for a Mean (t-interval) - College Statistics

SECTION 1

Historical Context & Motivation

Statistical inference rests on a deceptively simple question: given a finite sample from a larger population, how confidently can we estimate the true population parameter? By the late nineteenth century, mathematicians had established the normal distribution as a cornerstone of probability theory, and the Central Limit Theorem guaranteed that sample means would follow an approximately normal shape for sufficiently large samples. However, a critical gap remained: these results presumed that the population standard deviation σ was known, a condition rarely satisfied in practice. In most real-world experiments—especially in agriculture, medicine, and quality control—researchers work with small samples and must estimate σ from the data itself, introducing additional uncertainty that the standard normal curve fails to capture.

1809 • Gauss & the Normal Curve

Carl Friedrich Gauss formalizes the method of least squares and the normal (Gaussian) distribution, establishing the theoretical foundation for modeling measurement errors in astronomical observations.

1908 • Student's t-Distribution

William Sealy Gosset, publishing under the pseudonym "Student" while working at Guinness Brewery in Dublin, derives the t-distribution to handle small-sample inference when σ is unknown.

1925 • Fisher's Formalization

Ronald A. Fisher provides rigorous mathematical proofs for the t-distribution and introduces degrees of freedom as a key parameter, cementing the framework in mainstream statistical practice.

1937 • Neyman's Confidence Interval Theory

Jerzy Neyman publishes a landmark paper formally defining the confidence interval as a repeated-sampling concept, unifying estimation theory and giving the t-interval its modern interpretation.

1970s–present • Computational Era

Software packages (SAS, SPSS, R, Python) automate t-interval calculations, enabling routine use in clinical trials, social science research, and industrial quality control across every quantitative discipline.

Gosset's original problem was strikingly practical: with only a handful of barley yield measurements, how could he estimate the true mean yield and attach a meaningful measure of precision? The z-interval, which relies on knowing σ, systematically understated the true uncertainty in small samples, producing intervals that were too narrow and overconfident. The t-interval resolved this by substituting the sample standard deviation s for σ and using a distribution with heavier tails—the t-distribution—to account for the extra variability introduced by estimating σ. This section of the course equips you to construct and interpret such intervals, one of the most frequently used tools in applied statistics.

SECTION 2

Core Principles & Definitions

Before constructing a t-interval, it is essential to understand the conceptual machinery that makes the procedure valid. A confidence interval is not a probability statement about a single fixed parameter; rather, it describes a procedure that, when repeated across many independent samples, captures the true parameter a specified percentage of the time. The confidence level (commonly 90%, 95%, or 99%) quantifies the long-run success rate of this procedure. Five foundational ideas underpin the t-interval for a mean.

Point Estimate: x̄

The sample mean x̄ serves as the single best guess for the unknown population mean μ. It is the center of every confidence interval.

Standard Error: s / √n

Because σ is unknown, we estimate the standard deviation of x̄ using the standard error SE = s / √n, where s is the sample standard deviation and n is the sample size.

Degrees of Freedom: df = n − 1

The shape of the t-distribution depends on degrees of freedom (df = n − 1). Smaller df yields heavier tails; as df → ∞, the t-distribution converges to the standard normal.

Critical Value: t*

The critical value t* is the number of standard errors we extend in each direction from x̄. It is determined by the confidence level and df, and is always larger than the corresponding z* for the same confidence level.

Margin of Error: E = t* · (s / √n)

The margin of error E determines the half-width of the interval. Increasing n or decreasing the confidence level shrinks E; increasing s or raising the confidence level expands it.

✦

KEY TAKEAWAY

Think of constructing a confidence interval like casting a fishing net. The sample mean x̄ is where you aim the net, and the margin of error determines the net's width. A wider net (higher confidence level or more variability) is more likely to capture the fish (the true mean μ), but it gives you a less precise catch. The t-interval uses a slightly wider net than the z-interval because estimating σ with s introduces additional uncertainty—analogous to casting your net in murkier water where the fish's exact location is harder to judge.

SECTION 3

Visual Explanation — The t-Distribution and Interval Construction

The visual below contrasts the standard normal distribution (z) with t-distributions at various degrees of freedom. Notice how the t-distribution has heavier tails than the normal curve, especially when df is small. This additional probability mass in the tails reflects the extra uncertainty from estimating σ with s. As degrees of freedom increase, the t-distribution gradually approaches the standard normal, which is why the z-interval and t-interval produce nearly identical results for large samples.

As degrees of freedom decrease, the t-distribution's tails become heavier (more probability in the extremes), and the peak flattens. The blue solid curve is the standard normal z; dashed curves show t-distributions with df = 10 (violet), df = 3 (pink), and df = 1 (amber, also known as the Cauchy distribution). For df > 30, the t and z curves become nearly indistinguishable.

The practical implication is immediate: for a given confidence level, the critical value t* will always be larger than z*, producing a wider interval. With df = 10 and a 95% confidence level, t* ≈ 2.228 compared to z* = 1.960—a difference of about 14%. With df = 3, t* balloons to 3.182. This widening is not a deficiency; it is the honest acknowledgment that estimating σ from a small sample introduces genuine uncertainty that must be reflected in the interval.

SECTION 4

Mathematical Framework

The derivation of the t-interval begins with the sampling distribution of the mean. When a random sample of size n is drawn from a population with mean μ and standard deviation σ, the quantity Z = (x̄ − μ) / (σ / √n) follows a standard normal distribution. In practice, σ is unknown and is replaced by s, the sample standard deviation. The resulting quantity follows the t-distribution with n − 1 degrees of freedom.

T-STATISTIC

t = (x̄ − μ) / (s / √n)

where x̄ = sample mean, μ = population mean, s = sample standard deviation, n = sample size. This statistic follows a t-distribution with df = n − 1.

Inverting this relationship—solving for μ—yields the confidence interval formula. We seek the range of μ values for which the observed t-statistic would fall between −t* and +t*, where t* is the critical value corresponding to the desired confidence level C.

CONFIDENCE INTERVAL FOR μ

x̄ ± t* · (s / √n)

Equivalently: (x̄ − t* · s/√n, x̄ + t* · s/√n). The value t* is the critical value from the t-distribution with df = n − 1 that captures the central C% of the area.

MARGIN OF ERROR

E = t* · (s / √n)

The margin of error E is the half-width of the interval. Doubling the sample size reduces E by a factor of √2 ≈ 1.414, not by half—a consequence of the square root relationship.

SAMPLE STANDARD DEVIATION

s = √[ Σ(xᵢ − x̄)² / (n − 1) ]

The division by n − 1 (Bessel's correction) ensures that s² is an unbiased estimator of σ². This is the source of the n − 1 degrees of freedom in the t-distribution.

⚠

Conditions for Validity

The t-interval requires three conditions: (1) the data come from a simple random sample (or a randomized experiment); (2) the observations are independent—when sampling without replacement, n should be less than 10% of the population; (3) the population distribution is approximately normal, or the sample size is large enough (n ≥ 30 is a common guideline) for the CLT to compensate for moderate non-normality. For small samples, check for severe skewness or outliers using a histogram or normal probability plot.

SECTION 5

Conditions, Mechanics, and the t-Table

In practice, constructing a t-interval involves a systematic sequence: verify conditions, compute the standard error, look up the critical value t* from a t-table (or software), calculate the margin of error, and form the interval. The diagram below illustrates this workflow and highlights the decision points where you choose between a z-interval and a t-interval.

The key decision node is whether σ is known. In virtually all real research scenarios, σ is unknown, directing you to the t-interval branch. The bottom three boxes show the computational steps: compute SE, find t*, then form the interval.

Excerpt from the t-Table

The table below shows selected critical values t* for common confidence levels and degrees of freedom. In practice, you may also use technology—R's qt() function, Python's scipy.stats.t.ppf(), or a TI calculator's invT() command—to obtain exact values for any df.

df	90% (t*)	95% (t*)	99% (t*)
3	2.353	3.182	5.841
5	2.015	2.571	4.032
10	1.812	2.228	3.169
20	1.725	2.086	2.845
30	1.697	2.042	2.750
∞ (z*)	1.645	1.960	2.576

Selected critical values from the t-distribution. The bottom row (df = ∞) gives the corresponding z* values for reference.

SECTION 6

Worked Example — Estimating Mean Study Hours

A university researcher randomly selects 16 students and records the number of hours each student studies per week. The sample yields a mean of x̄ = 14.5 hours and a sample standard deviation of s = 4.2 hours. Construct a 95% confidence interval for the true mean weekly study hours μ of the student population.

95% t-Interval for Mean Study Hours

1
Step 1 — Verify Conditions

The sample is a simple random sample of n = 16 students. Assuming the university has well over 160 students, the 10% condition is satisfied (16 < 10% of population), ensuring independence. With n = 16 (a small sample), we need the population distribution to be approximately normal. Suppose a histogram of the data shows no severe skewness or outliers—the normality condition is reasonably met.

All three conditions (SRS, independence, normality) are satisfied. ✓
2
Step 2 — Identify Given Values

From the problem statement: x̄ = 14.5, s = 4.2, n = 16, and the confidence level C = 95%.

x̄ = 14.5, s = 4.2, n = 16, C = 0.95
3
Step 3 — Compute Degrees of Freedom

The degrees of freedom for a one-sample t-interval are df = n − 1 = 16 − 1 = 15.

df = 15
4
Step 4 — Find the Critical Value t*

Using a t-table or technology, the critical value for a 95% confidence level with df = 15 is t* = 2.131. This means we need 2.131 standard errors in each direction to capture 95% of the t-distribution's central area.

t* = 2.131
5
Step 5 — Compute the Standard Error

SE = s / √n = 4.2 / √16 = 4.2 / 4 = 1.05.

SE = 1.05 hours
6
Step 6 — Calculate the Margin of Error

E = t* × SE = 2.131 × 1.05 ≈ 2.238.

E ≈ 2.238 hours
7
Step 7 — Form the Confidence Interval

The 95% confidence interval is x̄ ± E = 14.5 ± 2.238, giving us the interval (14.5 − 2.238, 14.5 + 2.238) = (12.262, 16.738).

95% CI: (12.262, 16.738) hours
8
Step 8 — Interpret in Context

We are 95% confident that the true mean weekly study hours for all students at this university lies between 12.26 and 16.74 hours. This means that if we were to repeat this sampling procedure many times, approximately 95% of the resulting intervals would contain the true population mean μ.

Interpretation: 95% confident that μ ∈ (12.26, 16.74).

💡

Comparison to z-Interval

Had we mistakenly used z* = 1.960 instead of t* = 2.131, the margin of error would have been 1.960 × 1.05 = 2.058, yielding a narrower interval of (12.442, 16.558). The t-interval is about 8.7% wider, reflecting the additional uncertainty from estimating σ. This difference matters most for small samples and diminishes as n grows.

SECTION 7

z-Interval vs. t-Interval — Strengths and Limitations

Understanding when to use the z-interval versus the t-interval is a common source of confusion in introductory statistics courses. The following table provides a side-by-side comparison of the two procedures across several important dimensions. In virtually all practical research scenarios, the t-interval is the appropriate choice because σ is almost never known.

Feature	z-Interval	t-Interval
σ known?	Yes — σ is known or given	No — σ is estimated by s
Distribution used	Standard normal (z)	t-distribution with df = n − 1
Critical value	z* (e.g., 1.960 for 95%)	t* (depends on df; always ≥ z*)
Interval width	Narrower for a given n	Wider — accounts for estimating σ
Sensitivity to n	Width shrinks as 1/√n	Width shrinks as 1/√n, plus t* → z* as df → ∞
Normality requirement	Population normal or n large (CLT)	Same, but more sensitive for small n
Practical use	Rare — mostly textbook exercises	Standard in applied research

Comparison of z-interval and t-interval for estimating a population mean.

✦

KEY TAKEAWAY

The z-interval is like measuring with a ruler whose markings you trust completely (σ known), while the t-interval is like measuring with a ruler you manufactured yourself from the sample data (s estimates σ). Your handmade ruler might be slightly off, so you wisely add extra buffer to your measurement range. As you make more rulers (larger n), your confidence in each ruler's accuracy grows, and the extra buffer shrinks until the two approaches become indistinguishable.

Limitations of the t-Interval

Sensitivity to outliers: Both x̄ and s are heavily influenced by extreme values. A single outlier can dramatically inflate s, widening the interval, or shift x̄, mislocating the interval entirely.
Normality assumption: For very small samples (n < 15), the t-interval's validity depends strongly on the population being approximately normal. Strongly skewed or heavy-tailed distributions can cause the actual coverage rate to differ substantially from the nominal confidence level.
Single population only: The one-sample t-interval addresses a single mean. Comparing two means requires the two-sample t-interval or paired t-interval, with their own assumptions.

SECTION 8

Connections to Advanced Theory

The one-sample t-interval is the gateway to a family of inferential procedures that permeate advanced statistics. Understanding its structure—point estimate plus or minus a critical value times a standard error—reveals a template that generalizes broadly. The table below maps the one-sample t-interval to several advanced extensions you will encounter later in your studies.

Procedure	Parameter	Key Difference from One-Sample t
Paired t-interval	μ_d (mean of paired differences)	Applied to difference scores; same formula with d̄ and $s_d$
Two-sample t-interval	μ₁ − μ₂	Uses pooled or Welch SE; df calculated via Satterthwaite approximation
Regression CI for β₁	Slope coefficient β₁	SE derived from residuals; df = n − 2
Bootstrap CI	Any parameter	Nonparametric; uses resampling instead of t-distribution; no normality assumption
Bayesian credible interval	μ (posterior)	Incorporates prior beliefs; interpretation is a direct probability statement about μ

Extensions and alternatives to the one-sample t-interval.

The relationship between confidence intervals and hypothesis tests is particularly worth noting. A 95% confidence interval for μ contains all values of μ₀ for which a two-sided t-test at α = 0.05 would fail to reject the null hypothesis H₀: μ = μ₀. This duality means that constructing a confidence interval simultaneously answers every possible two-sided hypothesis test at the corresponding significance level—a powerful conceptual unification that deepens as you study ANOVA, regression, and general linear models.

ℹ

Looking Ahead

When sample sizes are very small and the normality assumption is questionable, modern alternatives such as the bootstrap confidence interval offer robust estimation without distributional assumptions. Additionally, Bayesian methods provide credible intervals whose interpretation—"there is a 95% probability that μ lies in this range given the observed data"—addresses a question many students intuitively want the frequentist CI to answer but that it technically does not.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

A student claims: "There is a 95% probability that the true population mean falls within my 95% confidence interval." Explain precisely what is wrong with this interpretation and provide the correct one.

Show answer

The statement treats μ as a random variable and the interval as fixed, but in frequentist statistics, μ is a fixed (albeit unknown) constant. The confidence interval is the random entity because it depends on the particular sample drawn. The correct interpretation is: "We are 95% confident that the interval (L, U) contains μ," meaning that if we repeated the sampling procedure and constructed a confidence interval each time, approximately 95% of those intervals would contain the true μ. Any single interval either contains μ or it does not—we simply cannot assign a probability to that event within the frequentist framework. This distinction is subtle but foundational: the 95% refers to the reliability of the method, not to any single interval's contents.

Question 2: BASIC CALCULATION

A random sample of n = 25 light bulbs has a mean lifetime of x̄ = 1,200 hours with s = 100 hours. Construct a 95% confidence interval for the true mean lifetime μ. Use t* = 2.064 for df = 24.

Show answer

First, compute the standard error: SE = s / √n = 100 / √25 = 100 / 5 = 20 hours. Next, calculate the margin of error: E = t* × SE = 2.064 × 20 = 41.28 hours. The confidence interval is x̄ ± E = 1,200 ± 41.28 = (1,158.72, 1,241.28). We are 95% confident that the true mean lifetime of light bulbs in this population lies between approximately 1,158.7 and 1,241.3 hours. Note that using z* = 1.960 would have given E = 39.20, yielding a slightly narrower interval of (1,160.80, 1,239.20), which would understate the true uncertainty since σ is unknown.

Question 3: INTERMEDIATE

A nutritionist measures the daily caloric intake of 12 randomly selected adults and obtains x̄ = 2,150 calories and s = 320 calories. (a) Construct a 90% confidence interval for μ. (b) How would the interval change if the sample size were increased to 48 while x̄ and s remained the same? Use t* = 1.796 for df = 11 and t* = 1.678 for df = 47.

Show answer

(a) With n = 12, df = 11, and t* = 1.796: SE = 320 / √12 = 320 / 3.464 ≈ 92.38. E = 1.796 × 92.38 ≈ 165.91. The 90% CI is 2,150 ± 165.91 = (1,984.09, 2,315.91). (b) With n = 48, df = 47, and t* = 1.678: SE = 320 / √48 = 320 / 6.928 ≈ 46.19. E = 1.678 × 46.19 ≈ 77.51. The 90% CI is 2,150 ± 77.51 = (2,072.49, 2,227.51). Quadrupling the sample size from 12 to 48 cuts the standard error in half (since √48 = 2√12) and also reduces t* (from 1.796 to 1.678), producing a margin of error that is less than half of the original. The interval narrows from a width of about 332 calories to about 155 calories, illustrating the dual benefit of increased sample size.

Question 4: APPLIED

An environmental engineer collects dissolved oxygen readings (in mg/L) from 20 randomly selected sites along a river. The data yield x̄ = 7.8 mg/L and s = 1.4 mg/L. Regulatory standards require that the mean dissolved oxygen exceed 6.5 mg/L. Construct a 99% confidence interval for the true mean (use t* = 2.861 for df = 19) and discuss whether the data provide evidence that the river meets the regulatory standard.

Show answer

SE = 1.4 / √20 = 1.4 / 4.472 ≈ 0.3130. E = 2.861 × 0.3130 ≈ 0.8953. The 99% CI is 7.8 ± 0.895 = (6.905, 8.695). Because the entire 99% confidence interval lies above the regulatory threshold of 6.5 mg/L, we have strong evidence that the true mean dissolved oxygen concentration exceeds the standard. Even the lower bound of the interval (6.905) is above 6.5. Equivalently, a one-sided hypothesis test H₀: μ ≤ 6.5 versus Hₐ: μ > 6.5 at α = 0.005 (corresponding to the lower tail of the 99% two-sided CI) would reject H₀. The engineer can report with high confidence that the river meets the dissolved oxygen requirement, though continued monitoring is prudent because seasonal or weather-related fluctuations could affect future readings.

Question 5: CRITICAL THINKING

A researcher reports a 95% confidence interval for mean reaction time as (245 ms, 305 ms) based on n = 9 subjects. A reviewer notes that the data contained one extreme outlier (a reaction time of 580 ms). (a) Explain how the outlier likely affected both x̄ and s, and consequently the confidence interval. (b) Propose two alternative approaches the researcher could use to obtain a more reliable interval. (c) If the outlier is removed (n = 8), the remaining data have x̄ = 252 ms and s = 22 ms. Construct the new 95% CI (t* = 2.365 for df = 7) and compare it to the original.

Show answer

(a) The outlier of 580 ms would pull x̄ upward (inflating the center of the interval) and dramatically inflate s (since it is far from the mean), which in turn increases the standard error and margin of error. The combined effect shifts the interval to the right and widens it, potentially misrepresenting the typical reaction time. (b) Two alternatives: First, the researcher could use a bootstrap confidence interval, which is more robust to non-normality and can accommodate outliers without distributional assumptions. Second, the researcher could use a trimmed mean (e.g., 10% trimmed) with a Winsorized standard deviation, which limits the influence of extreme values while retaining all data. (c) With the outlier removed: SE = 22 / √8 = 22 / 2.828 ≈ 7.78. E = 2.365 × 7.78 ≈ 18.40. The new 95% CI is 252 ± 18.40 = (233.6, 270.4). This interval is dramatically narrower (width ≈ 37 ms vs. 60 ms originally) and centered about 23 ms lower. The contrast underscores how a single outlier in a sample of 9 can substantially distort t-interval results, highlighting the importance of checking conditions—especially normality—before applying the procedure.

SUMMARY

Summary — Confidence Interval for a Mean (t-interval)

Summary

The t-interval for a population mean is the standard inferential tool when the population standard deviation σ is unknown—the typical situation in practice. The formula x̄ ± t* · (s / √n) combines the point estimate x̄ with a margin of error that accounts for both sampling variability (through s / √n) and the additional uncertainty of estimating σ (through the t-distribution's heavier tails relative to the normal curve). The degrees of freedom (df = n − 1) govern the shape of the t-distribution: smaller df means wider intervals, and as df → ∞, the t-interval converges to the z-interval.

Three conditions must be verified before construction: the data must come from a random sample, observations must be independent (10% condition for sampling without replacement), and the population should be approximately normal or the sample size large enough for the CLT to apply. Interpretation is crucial: a 95% confidence level means that 95% of all intervals constructed by this method would capture the true μ—it is a statement about the procedure's long-run reliability, not a probability about any single interval. This foundational tool generalizes to paired t-intervals, two-sample t-intervals, and regression inference, making it an essential building block for the rest of the statistics curriculum.