Skip to main content

How to find a p-value on a TI-84

The TI-84 will give you an exact p-value in a few keystrokes once you know which menu to open — and the one piece of syntax that trips almost everyone up is how to type infinity.

6 min read · Last reviewed 7 August 2026

Two menus, two situations

The TI-84 handles p-values in two different places, and which one you want depends on what you already have.

  • DISTR (press 2nd then VARS) — you already have a test statistic and want its tail area.
  • STAT › TESTS — you have raw data or summary statistics and want the calculator to run the whole test.

From a statistic: the DISTR menu

Each function takes a lower bound and an upper bound, and returns the area between them:

StatisticSyntax
znormalcdf(lower, upper)
ttcdf(lower, upper, df)
chi-squareχ²cdf(lower, upper, df)
FFcdf(lower, upper, df1, df2)

normalcdf assumes the standard normal unless you add a mean and standard deviation as two further arguments, so a z-score needs no extra input.

Typing infinity: the 1E99 trick

A tail area runs to infinity, and the TI-84 has no infinity key. The convention is to use 1E99 — a 1 followed by 99 zeros, which is far beyond any distribution's usable range. Type the E with 2nd then the comma key, not the letter E.

So for a right-tailed test with t = 2.31 and df = 27: 

tcdf(2.31, 1E99, 27)
→ .0143842841

For a left-tailed test, the lower bound becomes negative infinity: 

tcdf(-1E99, -2.31, 27)
→ .0143842841

And for a two-tailed test, take the tail beyond the absolute value and double it: 

2 * tcdf(2.31, 1E99, 27)
→ .0287685683

Chi-square and F are already right-tailed, so never double them. For those, the lower bound is your statistic and the upper bound is 1E99.

The mistake almost everyone makes

Getting the bounds the wrong way round returns the complement of what you wanted — a p-value near 1 instead of near 0, or vice versa. If your answer comes out around .98 when you expected something small, you have almost certainly swapped the bounds or dropped a minus sign.

Use the negation key (-) rather than the subtraction key for the negative bound. They look similar and the TI-84 treats them as different operations; subtraction in an argument list is a syntax error.

From data: STAT TESTS

Press STAT, arrow across to TESTS, and the calculator runs the whole test and prints the p-value as p in the results.

  • T-Test — one sample against a target.
  • 2-SampTTest — two independent groups.
  • 2-PropZTest — two proportions, the A/B testing case.
  • χ²-Test — a contingency table you have entered as a matrix.
  • LinRegTTest — the slope and correlation of paired data.

Each prompts for Data or Stats. Choose Stats if you already have the mean and standard deviation; choose Data if your numbers are in lists. You also pick the alternative hypothesis here, so the calculator applies the correct tail without you doubling anything by hand.

On 2-SampTTest there is a Pooled prompt at the bottom. Leave it on No: that gives Welch's test, which does not assume the two groups have equal variances and is the safer default.

Where the TI-84 runs out

The TI-84 displays about ten significant figures and its distribution routines lose accuracy deep in the tail, where the interesting results live. For a statistic large enough to matter it will happily print 0 — the same rounding failure that makes p = .000 appear in published papers.

If you need a value you can report rather than one you can eyeball, put the same statistic and degrees of freedom into the t-score converter here. It computes the tail directly rather than by subtraction, so it returns figures as small as 1.13 × 10⁻¹⁹ instead of collapsing them to zero.

Keep reading

Ready to run the numbers?

Our calculator shows the shaded distribution, the exact p-value, and a plain-English reading of what it supports.

Open the P-Value Calculator

Frequently asked questions

How do I type infinity on a TI-84?

Use 1E99 for positive infinity and -1E99 for negative infinity. Enter the E with 2nd followed by the comma key, which produces the scientific-notation exponent, rather than typing the letter E from the alpha keyboard. 1E99 is far outside the usable range of any of these distributions, so it behaves as infinity for the purposes of a tail area.

What is the difference between normalcdf and normalpdf?

normalcdf gives the area under the curve between two bounds, which is a probability and therefore what you want for a p-value. normalpdf gives the height of the curve at a single point, which is not a probability and is essentially never the answer to a hypothesis-testing question. If you have been using normalpdf for p-values, that is the error.

How do I get a two-tailed p-value on a TI-84?

Compute the upper tail beyond the absolute value of your statistic and multiply by two: 2 * tcdf(2.31, 1E99, 27) returns .0287685683. Alternatively use the STAT TESTS menu and choose the two-sided alternative, which handles it for you. Never double a chi-square or F p-value, since both are already one-tailed by construction.

Should Pooled be Yes or No on 2-SampTTest?

No. Pooled: No runs Welch's t-test, which does not assume the two groups share a variance. It is as accurate as the pooled test when they do and noticeably better when they do not, so there is little to gain from answering Yes. This is the same default the t-test calculator on this site uses.

Why does my TI-84 give p = 0?

Because the true p-value is smaller than the calculator can represent in the tail, so it rounds to zero. It does not mean the result is impossible. Report it as p < .001 and, if you need the actual figure for a paper, recompute it with a tool that evaluates the tail directly rather than by subtracting from one.

Which TI-84 test do I use for a chi-square?

χ²-Test in the STAT TESTS menu, after entering your contingency table as a matrix through 2nd then x-inverse. It uses matrix [A] for the observed counts and writes the expected counts into [B], which is worth checking — the chi-square approximation degrades when any expected count falls below about 5.