T Table: Student’s t Distribution

T table of critical t values by degrees of freedom, with a bar chart showing critical t falling towards z = 1.960

A t table lists the critical values of Student’s t distribution, the cut-off scores your t statistic must reach for a result to count as statistically significant. With 18 degrees of freedom and a two-tailed test at 0.05, that number is 2.101.

Key Takeaways

  • A t table gives the critical value of t for your degrees of freedom, significance level and number of tails. If your calculated t (ignoring its sign in a two-tailed test) is equal to or larger than that value, the result is significant.
  • To read it, pick the column for your alpha and tails, then go down to the row for your degrees of freedom. If your exact df is missing, use the next smaller df.
  • The t distribution looks like the normal curve but with heavier tails, so its critical values are larger. As degrees of freedom rise, the t values shrink towards the z values of 1.645 and 1.960.

Most people arrive at a t table with a number already in hand. They have run a t-test, worked out a t statistic, and now need to know whether it is big enough to matter. The t table answers that one question. It does not tell you how to calculate t, and it cannot tell you whether you chose the right test. It tells you the threshold.

This page puts the full table first, because that is what most visitors need. Every value in it was computed directly from the t distribution rather than copied from another table, and it matches the reference table published by the US National Institute of Standards and Technology. Below the table is a free t table calculator that gives the critical value for any degrees of freedom, including the ones a printed table skips, and turns your t statistic into an exact p value.

After that, the article explains how to read the table step by step, which degrees of freedom to use for each kind of t-test, three worked examples, what the t distribution actually is, how it differs from the normal distribution used in a z table, and why it carries the odd name “Student’s t”. If you are studying psychology statistics at A level, AP or degree level, this covers everything the t table asks of you.

T table of critical t values by degrees of freedom, with a bar chart showing critical t falling towards z = 1.960
Small samples need a larger t to reach significance. By 120 degrees of freedom the critical value is within 1% of the normal curve’s 1.960.

T Table: Critical Values of the t Distribution

The t table below gives critical values of t for degrees of freedom from 1 to 30, then 40, 50, 60, 80, 100, 120 and infinity. The top header row is the significance level for a one-tailed test and the second header row is the matching level for a two-tailed test. The highlighted column most people need is 0.05 two-tailed, which is the same column as 0.025 one-tailed.

One-tailed α0.10.050.0250.010.0050.0005
Two-tailed α0.20.10.050.020.010.001
dfCritical value of t
13.0786.31412.70631.82163.657636.619
21.8862.9204.3036.9659.92531.599
31.6382.3533.1824.5415.84112.924
41.5332.1322.7763.7474.6048.610
51.4762.0152.5713.3654.0326.869
61.4401.9432.4473.1433.7075.959
71.4151.8952.3652.9983.4995.408
81.3971.8602.3062.8963.3555.041
91.3831.8332.2622.8213.2504.781
101.3721.8122.2282.7643.1694.587
111.3631.7962.2012.7183.1064.437
121.3561.7822.1792.6813.0554.318
131.3501.7712.1602.6503.0124.221
141.3451.7612.1452.6242.9774.140
151.3411.7532.1312.6022.9474.073
161.3371.7462.1202.5832.9214.015
171.3331.7402.1102.5672.8983.965
181.3301.7342.1012.5522.8783.922
191.3281.7292.0932.5392.8613.883
201.3251.7252.0862.5282.8453.850
211.3231.7212.0802.5182.8313.819
221.3211.7172.0742.5082.8193.792
231.3191.7142.0692.5002.8073.768
241.3181.7112.0642.4922.7973.745
251.3161.7082.0602.4852.7873.725
261.3151.7062.0562.4792.7793.707
271.3141.7032.0522.4732.7713.690
281.3131.7012.0482.4672.7633.674
291.3111.6992.0452.4622.7563.659
301.3101.6972.0422.4572.7503.646
401.3031.6842.0212.4232.7043.551
501.2991.6762.0092.4032.6783.496
601.2961.6712.0002.3902.6603.460
801.2921.6641.9902.3742.6393.416
1001.2901.6601.9842.3642.6263.390
1201.2891.6581.9802.3582.6173.373
∞1.2821.6451.9602.3262.5763.291
Critical values of Student’s t distribution, computed to three decimal places. Your t must be equal to or greater than the value shown to be significant.

Each value is the point on the t distribution that cuts off the stated area in the upper tail (for one-tailed) or splits it equally between both tails (for two-tailed). Because the t distribution is symmetrical, the table only shows positive numbers. A negative critical value is the same number with a minus sign in front. The final row, marked infinity, is identical to the standard normal distribution, which is why it reads 1.645, 1.960 and 2.576, the familiar critical values of z.

T Table Calculator: Critical Values and P Values

This t table calculator does the two jobs people most often want from a t table. The first tab finds the critical value of t for any degrees of freedom, so you do not have to round down when your df falls between rows. The second tab converts your t statistic and degrees of freedom into an exact p value, one-tailed or two-tailed, which a printed table can only bracket.

T Table Calculator

Find the critical value of t for any degrees of freedom, or turn your t statistic into an exact p value. Results match the printed t table above to three decimal places.

The calculator uses the same mathematics as statistical software: it works out the area under the t curve directly, so its answers agree with SPSS, R and Excel's T.INV.2T and T.DIST.2T functions. If your answer differs from a textbook table in the third decimal place, the textbook has usually rounded.

How to Read a T Table

To read a t table, you need three pieces of information: whether your hypothesis is one-tailed or two-tailed, the significance level you are testing at, and your degrees of freedom. The columns are set by the first two and the row by the third. The critical value sits where they cross.

  1. Decide on one or two tails. A directional hypothesis ("scores will be higher after training") is one-tailed. A non-directional hypothesis ("scores will differ") is two-tailed. This must be decided before you look at the data.
  2. Choose the significance level. In psychology this is almost always 0.05, but 0.01 is used when a mistake would be costly or many tests are being run.
  3. Find the column. Use the one-tailed header row for a one-tailed test, or the two-tailed header row for a two-tailed test, and go to the column for your alpha.
  4. Work out the degrees of freedom. For a related t-test this is the number of pairs minus 1. For an independent t-test it is the total number of participants minus 2. More detail is in the next section.
  5. Read across to the critical value. Go down the df column to your row and read the number in your alpha column.
  6. Compare. If your calculated t is equal to or larger than the critical value, the result is significant and you can reject the null hypothesis. In a two-tailed test, ignore the minus sign on your t before comparing.

Reading the T Table for a Two-Tailed Test

For a two-tailed test, read the column labelled with your alpha in the two-tailed header row. At 0.05 two-tailed, the 5% rejection region is split into 2.5% at each end of the distribution, which is why this column is also labelled 0.025 one-tailed. With 10 degrees of freedom the critical value is 2.228, so a t of 2.40 or a t of minus 2.40 would both be significant, while a t of 1.90 would not.

Reading the T Table for a One-Tailed Test

For a one-tailed test, read the column labelled with your alpha in the one-tailed header row. All 5% of the rejection region sits at one end, so the critical value is smaller and easier to reach: with 10 degrees of freedom it is 1.812 rather than 2.228. There is one extra condition. The result must also go in the direction you predicted. A large t in the opposite direction is not significant in a one-tailed test, however big it is.

Reading the T Table for a P Value

A printed t table cannot give an exact p value, but it can tell you which two values your p falls between. Find your degrees of freedom row, then move along it until you find the two critical values your t sits between. Your p value lies between the significance levels at the top of those two columns.

For example, a two-tailed t of 2.50 with 12 degrees of freedom sits between 2.179 (the 0.05 column) and 2.681 (the 0.02 column). So p is less than 0.05 but greater than 0.02, which would be written 0.02 < p < 0.05. The exact value from the calculator is p = .028. For an exact figure in a report, use the calculator above or your software output rather than the table.

What If Your Degrees of Freedom Are Not in the Table?

If your exact degrees of freedom are missing from the t table, use the next smaller df that is listed. With 35 degrees of freedom, for example, use the row for 30. The critical value for 30 df (2.042) is slightly larger than the true value for 35 df (2.030), so using it makes the test a little stricter, never more lenient. That is called being conservative, and it is the convention examiners expect. The calculator above removes the problem entirely by giving the exact value for any df.

Degrees of Freedom for the T Table

Degrees of freedom for the t table depend on which t-test you ran, and getting them wrong is the most common reason students read the wrong row. In plain terms, degrees of freedom are the number of values in your data that are free to vary once you have estimated what you need from them. Each mean you estimate from the sample uses up one degree of freedom.

TestDegrees of freedomExample
One-sample t-testn − 125 participants: df = 24
Related (paired) t-testNumber of pairs − 115 participants tested twice: df = 14
Independent t-testn1 + n2 − 2Two groups of 10: df = 18
Welch's t-test (unequal variances)Calculated by formula, often not a whole numberSoftware reports it, e.g. df = 15.7
Significance of Pearson's rn − 220 participants: df = 18

The pattern is that you lose one degree of freedom for every mean you estimate. A one-sample or repeated measures design estimates one mean (of the scores, or of the differences), so df is n minus 1. An independent groups design estimates two means, one per group, so df is the total minus 2.

The last row surprises some students. The significance of a Pearson correlation coefficient is tested with a t statistic, using the formula t = r × √(n − 2) ÷ √(1 − r²), and the result is checked against the t table with n − 2 degrees of freedom. A correlation of 0.45 from 20 people gives t = 2.14, which beats the two-tailed critical value of 2.101 at df = 18, so it is significant at 0.05.

Welch's t-test is the version software uses when the two groups have unequal variances. Its degrees of freedom come from a formula that usually produces a decimal, such as 15.7. A printed table cannot handle that, so round down to the whole number below (15) or type the decimal into the calculator, which accepts it.

T Table Worked Examples

The three worked examples below show the t table in use for an independent t-test, a related t-test and a confidence interval. They use invented but realistic psychology data, and every figure was checked by calculation.

Example 1: Independent T-Test, Two-Tailed

A researcher compares memory scores for 10 participants who learned a word list in silence and 10 who learned it with music playing. The hypothesis is non-directional: the two conditions will differ. The calculated t is 2.31.

  • Tails: two-tailed, because no direction was predicted.
  • Alpha: 0.05.
  • Degrees of freedom: 10 + 10 − 2 = 18.
  • Critical value: row 18, 0.05 two-tailed column = 2.101.
  • Decision: 2.31 is greater than 2.101, so the result is significant. The null hypothesis is rejected.

The exact p value is .033, so it would be reported as t(18) = 2.31, p = .033. Notice that 2.31 would not have been significant at 0.01, where the critical value for df 18 is 2.878.

Example 2: Related T-Test, One-Tailed

Fifteen children complete a reading task before and after a six-week phonics programme. The hypothesis is directional: scores will be higher after the programme. The calculated t is 1.90, and scores did go up.

  • Tails: one-tailed, because a direction was predicted.
  • Alpha: 0.05.
  • Degrees of freedom: 15 pairs − 1 = 14.
  • Critical value: row 14, 0.05 one-tailed column = 1.761.
  • Decision: 1.90 is greater than 1.761 and in the predicted direction, so the result is significant.

This example shows why tails matter. Had the hypothesis been two-tailed, the critical value would have been 2.145, and the same t of 1.90 would not have been significant. That is exactly why the choice must be made before the data are collected. Switching to one-tailed after seeing the result, to scrape under the line, is a form of p-hacking. The exact one-tailed p here is .039.

Example 3: Using the T Table for a 95% Confidence Interval

A t table is also used to build confidence intervals around a mean when the population standard deviation is unknown, which is nearly always. The formula is the sample mean plus or minus t* multiplied by the standard error, where t* is the two-tailed critical value for your confidence level.

Suppose 25 nursery practitioners rate their job satisfaction, with a mean of 52 and a standard deviation of 10. The standard error is 10 ÷ √25 = 2. For a 95% confidence interval you need the 0.05 two-tailed column, and with 25 − 1 = 24 degrees of freedom that gives t* = 2.064. The margin of error is 2.064 × 2 = 4.13, so the 95% confidence interval runs from 47.87 to 56.13.

Confidence levelT table column to uset* at df = 24
90%0.10 two-tailed (0.05 one-tailed)1.711
95%0.05 two-tailed (0.025 one-tailed)2.064
99%0.01 two-tailed (0.005 one-tailed)2.797
99.9%0.001 two-tailed (0.0005 one-tailed)3.745

The rule is simple: for a C% confidence interval, use the two-tailed column for alpha = 1 − C/100. A 95% interval uses 0.05, a 99% interval uses 0.01. Many t tables print a "confidence level" row at the bottom for exactly this reason.

What Is the Student's t Distribution?

The Student's t distribution is a bell-shaped probability distribution used when you estimate a population mean from a small sample and do not know the population's standard deviation. It looks like the normal distribution, symmetrical and centred on zero, but its tails are heavier. More of its area lies far from the centre.

The heavier tails exist because of extra uncertainty. When you calculate a t statistic, you divide by the standard error, and the standard error itself depends on the sample's standard deviation. In a small sample that standard deviation is a shaky estimate. Sometimes it comes out too small, which makes t look bigger than it should. The t distribution builds in that wobble, so extreme values are more likely than the normal curve would suggest, and the critical values have to be further out to keep the error rate at 5%.

How Degrees of Freedom Change the Shape

The t distribution is not one curve but a family of curves, one for each number of degrees of freedom. With very few degrees of freedom the tails are fat and the critical values are large. As degrees of freedom increase, the sample standard deviation becomes a more reliable estimate, the tails thin out, and the curve closes in on the normal distribution.

Degrees of freedomCritical t (0.05, two-tailed)How far above z = 1.960
24.303120% higher
52.57131% higher
102.22814% higher
302.0424% higher
1201.9801% higher
Infinity1.960Identical to z

The practical lesson is that sample size matters enormously at the small end and hardly at all at the large end. Going from 2 to 10 degrees of freedom almost halves the critical value. Going from 30 to 120 trims it by about 3%. This is one reason psychology studies with fewer than ten participants per group struggle to find significant effects: the bar is simply set higher.

The heavier tails can also be put in probability terms. With 5 degrees of freedom, the chance of a t value more than 3 units from zero in either direction is about 3.0%. Under the normal distribution, the chance of a z more than 3 from zero is about 0.27%. The t distribution with 5 df makes such extreme values roughly eleven times more likely.

T Distribution vs Normal (Z) Distribution

The t distribution is used instead of the normal (z) distribution when the population standard deviation is unknown and has to be estimated from the sample, which describes almost every psychology study. The z distribution is used when the population standard deviation is genuinely known, as with standardised tests such as IQ scales, which are designed to have a standard deviation of 15.

Featuret distributionNormal (z) distribution
ShapeBell-shaped, heavier tailsBell-shaped, thinner tails
ParametersDegrees of freedomNone (mean 0, SD 1)
Population SDUnknown, estimated from sampleKnown
Critical value, 0.05 two-tailedDepends on df (2.228 at df 10)Always 1.960
Typical useT-tests, confidence intervals for small samplesZ-tests, standardised scores, large samples

A common rule of thumb says you can switch from t to z once the sample reaches 30. The table shows why that rule exists and also why it is only approximate: at 30 degrees of freedom the critical t of 2.042 is still about 4% above 1.960. Software always uses the t distribution when the standard deviation is estimated, whatever the sample size, and that is the safer habit. For more on z scores and the z table, see the z-score guide.

Why Is It Called Student's t? Gosset and the Guinness Brewery

The t distribution is called Student's t because its discoverer, William Sealy Gosset, published under the pen name "Student". Gosset was a chemist who joined Arthur Guinness, Son and Co. in Dublin in 1899, after taking a first-class degree in chemistry at Oxford (Zabell, 2008). Brewing research produced exactly the kind of data that the statistics of the day handled badly: small numbers of samples of barley, hops and malt, where large-sample methods gave misleading answers.

His solution appeared in 1908 in the journal Biometrika, under the title "The probable error of a mean" (Student, 1908). To check his mathematics, Gosset ran one of the earliest statistical simulations, done entirely by hand. He wrote the height and finger measurements of 3,000 criminals on cards, shuffled them, drew them in random order, and divided them into 750 samples of 4, then compared the spread of the results with his formula (Zabell, 2008).

The pseudonym is often explained as Guinness hiding a trade secret from rival brewers. The earliest published account, by the statistician Harold Hotelling in 1930, gives a different reason: the brewery had a rule forbidding its chemists to publish, and Gosset was allowed to publish only under a pen name, to avoid difficulties with the rest of the staff (Zabell, 2008). In all, Gosset published 14 papers in Biometrika, all but one as "Student".

Gosset's own statistic was not quite today's t. He worked with a quantity he called z. It was the statistician Ronald Fisher who reworked it into the t we use now and placed it at the centre of significance testing in his 1925 textbook Statistical Methods for Research Workers (Fisher, 1925; Zabell, 2008). Joan Fisher Box, Fisher's daughter and biographer, describes how quickly Fisher recognised the importance of Gosset's small-sample work and began extending it (Box, 1987). Her paper, Guinness, Gosset, Fisher, and Small Samples, is free to read and is the best short history of the t-test.

One footnote shows that even famous examples can go wrong. Gosset illustrated his paper with data from a 1905 study by Cushny and Peebles on the sleep of 10 patients given two drugs, which he described as two forms of the same compound. Fisher reused the example in his textbook. In 1934 a reader wrote to point out that the two drugs were in fact different compounds altogether, and Fisher quietly relabelled them "A" and "B" in later editions (Zabell, 2008).

How to Report a t Value in APA Style

In APA style, a t-test result is reported as the letter t in italics, the degrees of freedom in brackets, the t value to two decimal places, and the exact p value. The American Psychological Association's manual asks for exact p values to two or three decimal places, with values below .001 written as p < .001 (American Psychological Association, 2020).

  • t(18) = 2.31, p = .033
  • t(14) = 1.90, p = .039, one-tailed
  • t(28) = −4.12, p < .001

Three details catch students out. P values have no leading zero, because a probability cannot exceed 1. The critical value from the t table is not normally reported; it is a decision tool, not a result. And a one-tailed test should be labelled as such, since readers assume two-tailed unless told otherwise. A full guide to writing up results is in the article on psychology report sections, and the meaning of p itself is covered in statistical significance and p values.

Significance is also not the whole story. A t value says whether a difference is unlikely to be chance; it does not say whether the difference is large enough to matter. APA style expects an effect size, such as Cohen's d, alongside every t-test.

Common Mistakes When Using a T Table

The most common t table mistakes are using the wrong row, the wrong column, or the wrong comparison. All of them are easy to avoid once you know to look for them.

  • Using n instead of degrees of freedom. With 20 participants in an independent design the row is 18, not 20.
  • Reading the one-tailed header for a two-tailed test. The 0.05 one-tailed column gives a smaller critical value and makes results look significant when they are not.
  • Comparing the sign as well as the size in a two-tailed test. A t of −2.50 is just as significant as +2.50. Use the absolute value.
  • Ignoring direction in a one-tailed test. A large t in the unpredicted direction is not significant under a one-tailed hypothesis.
  • Rounding df up. With 35 df, use the row for 30, not 40. Rounding up makes the test too lenient.
  • Getting the comparison backwards. For t, bigger is more significant: the calculated value must be equal to or greater than the critical value. Some tests, such as the Mann-Whitney U test and the Wilcoxon signed-rank test, work the other way round, which is a frequent source of exam errors.
  • Using a t table when a t-test was the wrong choice. T-tests assume roughly normal data measured at interval or ratio level. If those assumptions fail, a non-parametric test may be better. The guide to choosing a statistical test walks through the decision.

Conclusion

A t table answers one question: is your t statistic large enough to count as significant? To use it, decide on one or two tails before collecting data, pick your significance level, work out your degrees of freedom, and read the critical value where the column and row meet. If your t equals or beats it, the result is significant.

Behind the table is the t distribution, a heavy-tailed cousin of the normal curve that Gosset worked out for the small samples of a Dublin brewery and Fisher turned into the backbone of modern significance testing. Its critical values fall as samples grow, closing in on the z values of 1.645 and 1.960. For degrees of freedom the table skips, or for an exact p value, the calculator on this page does the work in one click.

Frequently Asked Questions

What is a t table in statistics?

A t table in statistics is a reference chart of critical values from Student's t distribution. Each entry is the smallest t statistic that counts as significant for a given number of degrees of freedom, significance level and number of tails. Researchers compare their calculated t against it to decide whether to reject the null hypothesis.

What is the t critical value for a 95% confidence interval?

The t critical value for a 95% confidence interval comes from the 0.05 two-tailed column, and it depends on degrees of freedom. It is 2.228 at 10 df, 2.086 at 20 df, 2.042 at 30 df and 2.000 at 60 df, falling towards 1.960 for very large samples.

What is the difference between the t critical value and the t statistic?

The t statistic is calculated from your data and measures how far apart the means are relative to the variability. The t critical value is read from the t table and is the threshold that statistic must reach. If the statistic equals or exceeds the critical value, the result is significant.

Why is the t distribution wider than the normal distribution?

The t distribution has heavier tails because it allows for error in estimating the standard deviation from a small sample. That extra uncertainty makes extreme t values more likely, so the cut-offs for significance sit further from zero. With more data the estimate improves and the curve narrows towards the normal shape.

When do you use a t table instead of a z table?

Use a t table when the population standard deviation is unknown and you have estimated it from your sample, as in nearly every t-test. Use a z table when the population standard deviation is known, such as for scores on a standardised test, or when working with z scores directly.

How do you find degrees of freedom for a t-test?

  1. One-sample test: subtract 1 from the number of scores.
  2. Related or paired test: subtract 1 from the number of pairs.
  3. Independent test: add both group sizes and subtract 2.

Is a bigger t value better?

A bigger t value, ignoring its sign, means stronger evidence against the null hypothesis, because the difference between means is large compared with the noise in the data. It does not mean the effect is important. A tiny difference can produce a large t in a very big sample, which is why effect sizes are reported too.

What does a negative t value mean?

A negative t value means the first mean in the calculation was smaller than the second, so the sign depends only on which group was entered first. In a two-tailed test, compare the number without its minus sign to the t table. In a one-tailed test, check the difference runs the way you predicted.

References

  • American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.). American Psychological Association.
  • Box, J. F. (1987). Guinness, Gosset, Fisher, and small samples. Statistical Science, 2(1), 45-52.
  • Fisher, R. A. (1925). Statistical methods for research workers. Oliver and Boyd.
  • NIST/SEMATECH. (2012). Critical values of the Student's t distribution. In e-Handbook of statistical methods (Section 1.3.6.7.2). National Institute of Standards and Technology.
  • Student. (1908). The probable error of a mean. Biometrika, 6(1), 1-25.
  • Zabell, S. L. (2008). On Student's 1908 article "The probable error of a mean". Journal of the American Statistical Association, 103(481), 1-7.

Further Reading and Research

Recommended Articles

Suggested Books

  • Field, A. (2024). Discovering Statistics Using IBM SPSS Statistics (6th ed.). Sage.
    • The standard undergraduate psychology statistics text, with a full chapter on comparing two means and a friendly explanation of the t distribution.
  • Coolican, H. (2019). Research Methods and Statistics in Psychology (7th ed.). Routledge.
    • Covers significance testing, critical value tables and the t-test in the depth needed for A level and first-year degree work.
  • Salsburg, D. (2001). The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century. W. H. Freeman.
    • A readable history of modern statistics, including Gosset's work at Guinness and his relationship with Fisher.

Recommended Websites

  • NIST/SEMATECH e-Handbook: Critical Values of the Student's t Distribution
    • The US government reference table of t critical values, with instructions for one-sided and two-sided tests.
  • Project Euclid: Statistical Science archive
    • Free access to Joan Fisher Box's history of Gosset, Guinness and the origins of the t-test.
  • Khan Academy: Statistics and probability
    • Free video lessons on the t distribution, t-tests and confidence intervals.

Kathy Brodie

Kathy Brodie is an Early Years Professional, Trainer and Author of multiple books on Early Years Education and Child Development. She is the founder of Early Years TV and the Early Years Summit.

Kathy's Author Profile
Kathy Brodie

To cite this article please use:

Early Years TV T Table: Student’s t Distribution. Available at: https://www.earlyyears.tv/t-table/ (Accessed: 16 November 2026).