Data Science T Distribution Hypothesis Testing 12 min read Updated November 14, 2026
BY: Minsa A (Senior Statistics Editor)
Reviewed By: Statistics Fundamentals Team

How to Read a T Table (Step-by-Step Guide with Examples)

A t table lists critical values of the t distribution, organized by degrees of freedom (rows) and significance level (columns). To read it: determine whether your test is one-tailed or two-tailed, calculate degrees of freedom, locate your α column, then find the intersection. That number is your critical value — compare it to your t-statistic to decide whether to reject H₀.

What Is a T Table? ↓ One vs. Two-Tailed ↓ Degrees of Freedom ↓ 4-Step Method ↓ Worked Examples ↓ Critical Values Table ↓

T Critical Value Lookup

Enter degrees of freedom, significance level, and test type to look up the critical t value with full step-by-step working.

For more options, use the full critical value calculator or the t test calculator.

What Is a T Table?

A t table — also called a Student's t distribution table — lists the critical values of the t distribution for different combinations of degrees of freedom and significance levels. You reach for it in hypothesis testing when you need to know whether a calculated t-statistic clears the threshold for statistical significance, and in confidence intervals when you need the t* multiplier for the margin of error.

Definition — Featured Snippet

A t table lists critical values of the t distribution, organized by degrees of freedom (rows) and significance level (columns). To read it: (1) determine whether your test is one-tailed or two-tailed, (2) calculate your degrees of freedom, (3) locate your significance level column, then (4) find where your df row and α column intersect — that number is your critical value.

Two variables control which cell you read: degrees of freedom (how much information is in your sample) and the significance level (how much evidence you require to reject H₀). The t distribution itself is covered in depth in the t distribution guide. One thing trips people up before they begin: some tables label columns with one-tailed α, others with two-tailed α. Always check the column header before reading a value.

Parts of the T Table

The df Column (Rows)

Each row corresponds to a different degrees of freedom value, starting at df = 1 and running to df = ∞. As df increases, t critical values shrink and converge toward the z critical values. The ∞ row is exactly the standard normal distribution: at two-tailed α = 0.05, it gives 1.960.

The Significance Level (Column Headers)

Each column represents a significance level α. Tables may display this as a decimal (0.05), a percentage (5%), or a confidence level (95%). Read the header carefully: a column labeled "0.05" means something different in a one-tailed table versus a two-tailed one.

The Body — Critical Values

Every cell is the t critical value for that exact combination of df and α. For hypothesis testing: if |t-statistic| > critical value, reject H₀. For confidence intervals: use the critical value as t* in the formula CI = x̄ ± t* × (s / √n).

One-Tailed vs. Two-Tailed T Table

A one-tailed test has a directional hypothesis (H₁: μ > μ₀ or H₁: μ < μ₀), so the rejection region sits entirely in one tail. A two-tailed test is non-directional (H₁: μ ≠ μ₀) and splits the rejection region between both tails. The table in front of you may show either format in its column headers.

ℹ️
Conversion rule — always state this explicitly

If your table uses two-tailed α and you are running a one-tailed test at α = 0.05, look up the column labeled α = 0.10. If your table uses one-tailed α and you are running a two-tailed test at α = 0.05, look up the column labeled α = 0.025.

One-Tailed Test

Directional hypothesis

H₁: μ > μ₀ or H₁: μ < μ₀. All of α is in one tail.

Use two-tailed column = 2 × desired α

α = 0.05 one-tailed → use α = 0.10 column

Two-Tailed Test

Non-directional hypothesis

H₁: μ ≠ μ₀. α is split equally between both tails.

Read the matching α column directly

α = 0.05 two-tailed → use α = 0.05 column

You Are Running Desired α Column to Use in a Two-Tailed T Table
One-tailed test0.10α = 0.20
One-tailed test0.05α = 0.10
One-tailed test0.025α = 0.05
One-tailed test0.01α = 0.02
Two-tailed test0.10α = 0.10 (read directly)
Two-tailed test0.05α = 0.05 (read directly)

Need more detail on when to choose each test type? The one vs. two-tailed tests guide walks through the decision in full.

How to Find Your Degrees of Freedom

Degrees of freedom tell you which row to use. The formula depends on your test design.

One-Sample T Test

df = n − 1 where n = sample size. Example: n = 15 → df = 14.

Independent Samples T Test

df = n₁ + n₂ − 2 where n₁ and n₂ are the sample sizes of each group. Example: n₁ = 12, n₂ = 10 → df = 20.

Paired T Test

df = n − 1 where n = number of pairs, not number of observations. Example: 22 pairs → df = 21.

Test Type Degrees of Freedom Formula Example
One-sample t testdf = n − 1n = 20 → df = 19
Independent samples t testdf = n₁ + n₂ − 2n₁ = 12, n₂ = 10 → df = 20
Paired t testdf = n − 1 (n = pairs)15 pairs → df = 14
Regression / correlationdf = n − 2n = 30 → df = 28

For regression and correlation, df = n − 2; the hypothesis testing hub covers those calculations.

How to Read a T Table — Step by Step

These four steps match the structured data Google uses to generate how-to rich results. Work through them in order every time you open a t table.

  1. 1 Identify your test type (one-tailed or two-tailed). Determine whether your hypothesis is directional (one-tailed) or non-directional (two-tailed). Check whether the t table's column headers show one-tailed or two-tailed α values — this changes which column you use. Apply the conversion rule from the previous section if needed.
  2. 2 Calculate your degrees of freedom. Use the appropriate formula: one-sample df = n − 1; independent samples df = n₁ + n₂ − 2; paired df = n − 1. Locate the matching row in the table.
  3. 3 Choose your significance level (α). Select α — usually 0.05 or 0.01. Find the matching column header. If your table uses two-tailed α and your test is one-tailed, use the column where the header = 2 × your desired α.
  4. 4 Find the critical value. Read the cell where your df row meets your α column. That number is your t critical value. For hypothesis testing: if |t-statistic| > critical value, reject H₀. For confidence intervals, use this value as t*.

Worked Examples

Example 1 — One-Sample Hypothesis Test

Example 1 — One-Sample T Test

Do students study more than the university's claim of 20 hours per week?

n = 15 students, x̄ = 23.1 hours, s = 4.2 hours. One-tailed test (testing μ > 20) at α = 0.05.

1

Test type: one-tailed (H₁: μ > 20)

2

df = n − 1 = 15 − 1 = 14

3

α = 0.05 one-tailed → use two-tailed α = 0.10 column

4

t critical (df = 14, two-tailed α = 0.10) = 1.761

5

t-statistic = (23.1 − 20) / (4.2 / √15) = 3.1 / 1.085 = 2.86

✓ 2.86 > 1.761 → Reject H₀. There is evidence students study more than 20 hours per week.

Example 2 — Confidence Interval

Example 2 — 95% Confidence Interval

Build a 95% CI for mean resting heart rate in adults

n = 20 adults, x̄ = 72.4 bpm, s = 8.1 bpm. Confidence level = 95%.

1

α = 1 − 0.95 = 0.05 (two-tailed — CIs are always two-sided)

2

df = n − 1 = 20 − 1 = 19

3

t* (df = 19, two-tailed α = 0.05) = 2.093

4

Standard error = s / √n = 8.1 / √20 = 8.1 / 4.472 = 1.812

5

Margin of error = 2.093 × 1.812 = 3.792

6

CI = 72.4 ± 3.792 = (68.6 bpm, 76.2 bpm)

✓ 95% CI for mean resting heart rate: (68.6 bpm, 76.2 bpm)

⚠️
Critical formula note — the square root matters

The confidence interval formula is CI = x̄ ± t* × (s / √n). The denominator is the square root of n — the standard error. Dropping the square root and writing s / n dramatically narrows the interval and produces a wrong result. This error appears on several statistics reference sites.

Example 3 — Two Independent Samples

Example 3 — Independent Samples T Test

Do two teaching methods produce different exam scores?

Group A: n₁ = 12 students; Group B: n₂ = 10 students. Two-tailed test at α = 0.05, t-statistic = 2.41.

1

Test type: two-tailed (H₁: μ₁ ≠ μ₂)

2

df = n₁ + n₂ − 2 = 12 + 10 − 2 = 20

3

t critical (df = 20, two-tailed α = 0.05) = 2.086

✓ 2.41 > 2.086 → Reject H₀. The two teaching methods produce significantly different exam scores.

T Table — Key Critical Values (Quick Reference)

The table below uses two-tailed significance levels. All values are verified against R's qt() function. For one-tailed tests, use the column where the header = 2 × your desired α (one-tailed α = 0.05 → use the α = 0.10 column).

Two-tailed critical values of the t distribution. df = 8 at α = 0.05 gives 2.306 — not 2.262 (which belongs to df = 9).
df α = 0.20 α = 0.10 α = 0.05 α = 0.02 α = 0.01
13.0786.31412.70631.82163.657
21.8862.9204.3036.9659.925
31.6382.3533.1824.5415.841
41.5332.1322.7763.7474.604
51.4762.0152.5713.3654.032
61.4401.9432.4473.1433.707
71.4151.8952.3652.9983.499
81.3971.8602.3062.8963.355
91.3831.8332.2622.8213.250
101.3721.8122.2282.7643.169
121.3561.7822.1792.6813.055
151.3411.7532.1312.6022.947
201.3251.7252.0862.5282.845
251.3161.7082.0602.4852.787
301.3101.6972.0422.4572.750
401.3031.6842.0212.4232.704
601.2961.6712.0002.3902.660
1201.2891.6581.9802.3582.617
∞1.2821.6451.9602.3262.576
The ∞ row equals the standard normal (z) distribution. See the full t distribution table for all df values.

When Your Exact df Is Not in the Table

Three situations come up in practice, each with a clear fix.

df falls between two listed values. Use the lower df row. If df = 47, use the df = 40 row (not df = 60). This produces a slightly larger critical value, making rejection marginally harder — the conservative, defensible choice.

df exceeds the last listed row. Use the ∞ row. It equals the z distribution: two-tailed α = 0.05 gives 1.960, two-tailed α = 0.01 gives 2.576.

You need an exact value. Use software. The t distribution calculator works without any coding. For command-line use:

# R — exact critical values qt(0.975, df = 47) # two-tailed α = 0.05 at df = 47 → 2.012 qt(0.995, df = 47) # two-tailed α = 0.01 at df = 47
' Excel formulas =T.INV.2T(0.05, 47) ' two-tailed α = 0.05 at df = 47 =T.INV(0.95, 47) ' one-tailed α = 0.05 at df = 47

T Table vs. Z Table — When to Use Each

Use the t table when the population standard deviation σ is unknown — which covers almost every real-world analysis. Use the z table when σ is known, or as a rule of thumb when n ≥ 30. For a full comparison of the two tests, see the t test vs. z test guide.

SituationUse T TableUse Z Table
Population σ unknown✓
Population σ known✓
Small sample (n < 30)✓
Large sample (n ≥ 30)Either (t is safer)✓ rule of thumb
✓
The ∞ row is the z distribution

At df = ∞ the two-tailed critical values are 1.645 (α = 0.10), 1.960 (α = 0.05), 2.326 (α = 0.02), and 2.576 (α = 0.01) — identical to z-table values. This is why the two tests converge for large samples.

Common Mistakes When Reading the T Table

1
Mixing up one-tailed and two-tailed α columns

Using α = 0.05 from a two-tailed table when you need a one-tailed test at α = 0.05 gives a critical value that is too large, making it harder to detect real effects. Fix: for one-tailed α = 0.05, use the two-tailed α = 0.10 column.

2
Using df = n instead of df = n − 1

For a one-sample t test with n = 20, the correct df is 19. Using df = 20 gives a slightly smaller critical value, making rejection marginally easier but statistically incorrect.

3
Rounding df UP instead of DOWN

When df = 47 is not listed, some people reach for df = 60 (rounding up). The correct move is df = 40 — the more conservative choice that makes rejection slightly harder.

4
Confusing the t-statistic with the t critical value

The t-statistic is what you calculate from your data: t = (x̄ − μ₀) / (s / √n). The t critical value is what you look up in the table. You compare the two — you do not look up the t-statistic.

5
Using s/n instead of s/√n in the confidence interval

The confidence interval formula is CI = x̄ ± t* × (s/√n). The denominator is the square root of n. Dropping the square root produces an interval far too narrow.

Practice Problems

Three problems with fully worked solutions. Click "Show solution" to reveal the working — ungated, no sign-up required.

Problem 1 — One-Tailed Hypothesis Test

A pharmaceutical company tests whether a new drug reduces blood pressure more than the standard treatment. n = 25 patients, t-statistic = 2.15. One-tailed test at α = 0.05. Is the result significant?

df = n − 1 = 25 − 1 = 24
One-tailed α = 0.05 → use two-tailed α = 0.10 column
t critical (df = 24, two-tailed α = 0.10) = 1.711
2.15 > 1.711 → Reject H₀
The new drug produces a significantly greater reduction in blood pressure at α = 0.05.
Problem 2 — 99% Confidence Interval

A researcher needs a 99% confidence interval for average commute time. n = 31, x̄ = 28.4 min, s = 6.7 min. What is the margin of error?

α = 1 − 0.99 = 0.01 (two-tailed)
df = n − 1 = 30
t* (df = 30, two-tailed α = 0.01) = 2.750
SE = 6.7 / √31 = 6.7 / 5.568 = 1.203
Margin of error = 2.750 × 1.203 = 3.308 minutes
99% CI = (25.1 min, 31.7 min). Margin of error: 3.3 minutes.
Problem 3 — Independent Samples, Unlisted df

An independent samples t test: n₁ = 18, n₂ = 16, t-statistic = −1.98, two-tailed α = 0.05. Is the result statistically significant?

df = 18 + 16 − 2 = 32
df = 32 not in table → use next lower: df = 30 (conservative)
t critical (df = 30, two-tailed α = 0.05) = 2.042
|−1.98| = 1.98 < 2.042 → Fail to reject H₀
Not statistically significant at α = 0.05. The two groups do not differ significantly.

Frequently Asked Questions

What is a t table used for? ▼

A t table is used to find critical values of the t distribution. You use it in two main situations: (1) in hypothesis testing, to determine whether your t-statistic is large enough to reject the null hypothesis, and (2) in confidence intervals, to find the multiplier (t*) needed to calculate the margin of error. It is used when the population standard deviation is unknown and you are working with sample data.

What is the difference between a one-tailed and two-tailed t table? ▼

The difference is in how the significance level (α) is split across the tails of the distribution. A one-tailed t table puts all of α in one tail — use it for directional hypotheses (testing whether a value is greater than or less than a reference). A two-tailed t table splits α equally between both tails — use it for non-directional hypotheses (testing whether a value simply differs from a reference). The same critical value table can serve both: for a one-tailed test at α = 0.05, use the two-tailed α = 0.10 column.

What are degrees of freedom in a t table? ▼

Degrees of freedom (df) in a t table represent the number of independent pieces of information used to estimate a parameter. For a one-sample t test, df = n − 1, where n is the sample size. For an independent samples t test, df = n₁ + n₂ − 2. For a paired t test, df = n − 1, where n is the number of pairs. As degrees of freedom increase, the t distribution approaches the standard normal (z) distribution.

What do you do when your exact degrees of freedom is not in the t table? ▼

When your exact degrees of freedom is not listed in the t table, use the next lower value that is listed. For example, if df = 47, use the df = 40 row. This gives a slightly larger (more conservative) critical value, making it harder to reject the null hypothesis — which is the safer approach. For more precision, use software: in R, qt(0.975, df = 47) gives the exact two-tailed critical value at α = 0.05; in Excel, use =T.INV.2T(0.05, 47).

How do I find the critical t value in Excel? ▼

To find a two-tailed critical t value in Excel, use the formula =T.INV.2T(alpha, degrees_freedom). For example, =T.INV.2T(0.05, 20) returns 2.086, the critical value for df = 20 at two-tailed α = 0.05. For a one-tailed critical value, use =T.INV(1-alpha, degrees_freedom) for the upper tail — for example, =T.INV(0.95, 20) returns 1.725.

What is the t critical value at infinite degrees of freedom? ▼

At infinite degrees of freedom (∞), t critical values equal the corresponding z critical values from the standard normal distribution. For two-tailed tests: α = 0.10 gives 1.645, α = 0.05 gives 1.960, α = 0.02 gives 2.326, and α = 0.01 gives 2.576. This is why the rule of thumb says you can use z instead of t when your sample size is very large — the distributions converge.

Key Takeaways

📌

The table is organized by degrees of freedom (rows) and significance level (columns). The cell at their intersection is your critical value.

📌

For a one-tailed test at α = 0.05 using a two-tailed table, read the α = 0.10 column. Always check which format your table uses.

📌

Degrees of freedom: df = n − 1 for one-sample; df = n₁ + n₂ − 2 for independent samples; df = n − 1 (n = pairs) for paired.

📌

When exact df is not listed, round DOWN to the next lower value — this gives a more conservative critical value.

📌

The confidence interval formula is CI = x̄ ± t* × (s/√n). The denominator is √n — the standard error. Never use s/n.

📌

The ∞ row equals the z distribution. At df = ∞: two-tailed α = 0.05 gives 1.960 — identical to the z-table value.

📌

df = 8 at α = 0.05 gives 2.306, not 2.262. The value 2.262 belongs to df = 9 — a common off-by-one-row error on other reference pages.

📌

The t-statistic (calculated from your data) and the t critical value (looked up in the table) are different things. You compare them — you do not look up the t-statistic.