Skip to main content

How to find the p-value in SPSS

SPSS never uses the words "p-value" in its output. It calls the column Sig., prints it to three decimals, and shows the most important results as ".000" — a number that cannot be true.

7 min read · Last reviewed 7 August 2026

The p-value is the "Sig." column

SPSS labels p-values Sig., short for significance. Wherever you see that heading, that is your p-value. Depending on the procedure and your SPSS version it may be written out more fully as Asymptotic Significance (2-sided), Sig. (2-tailed), or split into One-Sided p and Two-Sided p in version 27 and later.

TestMenu pathRow to read
t-testAnalyze › Compare Means › Independent-Samples T TestSig. (2-tailed)
Paired tAnalyze › Compare Means › Paired-Samples T TestSig. (2-tailed)
Chi-squareAnalyze › Descriptive Statistics › Crosstabs › StatisticsPearson Chi-Square
CorrelationAnalyze › Correlate › BivariateSig. (2-tailed)
ANOVAAnalyze › Compare Means › One-Way ANOVASig.

".000" is not zero

SPSS prints the Sig. column to three decimal places. When a p-value is smaller than 0.0005 it rounds to .000 — which does not mean the p-value is zero. A p-value can never be exactly zero, because there is always some chance of extreme data under the null.

Never report p = .000. Reviewers flag it, and it is mathematically false. Report p < .001 instead, which is what the number actually tells you.

If you want the real value, double-click the output table to enter edit mode, then double-click the cell itself — SPSS shows more decimal places than it prints. You can also right-click the column, choose Cell Properties, and raise the decimal count. Alternatively, take the test statistic and degrees of freedom straight from the same table and put them into the calculator on this site, which reports exact values far below .001 instead of rounding them away. That rounding is precisely the failure this calculator was built to avoid — the methodology page explains the numerics.

Independent-samples t-test: which row?

The Independent-Samples T Test table gives you two rows and they have different p-values. Which one you read depends on Levene's Test for Equality of Variances, printed at the left of the same table.

  • Equal variances assumed — the pooled, Student's t-test.
  • Equal variances not assumed — Welch's t-test, with fractional degrees of freedom.

The traditional advice is to read Levene's own Sig. value and pick the first row if it is above .05. Current statistical practice is simpler and safer: read the second row every time. Welch's test performs as well as the pooled test when variances really are equal, and much better when they are not, so choosing on the basis of a preliminary test adds a decision without adding accuracy. This calculator defaults to Welch for the same reason — see the t-test calculator.

Getting a one-tailed p-value

SPSS reports two-tailed p-values by default for t-tests and correlations. For a one-tailed test, halve the reported value — but only if the difference runs in the direction you predicted. If it runs the other way, your one-tailed p-value is greater than .5 and the result is not significant no matter how large the statistic looks.

Decide the direction before you see the data. Halving a two-tailed p-value after noticing which way the effect went is a way of turning p = .08 into p = .04 without any new evidence. The one-tailed vs two-tailed guide covers when a one-tailed test is legitimate.

Checking an SPSS result

Every SPSS table prints the test statistic and its degrees of freedom next to the Sig. column. That is everything you need to reproduce the p-value independently: take t and df to the t-score converter, or χ² and df to the chi-square converter. If the two disagree, the usual cause is a one-tailed versus two-tailed mismatch, or reading the pooled row when you meant the Welch row.

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

Where is the p-value in SPSS output?

It is the column headed "Sig." — SPSS never calls it a p-value. In t-test output it appears as "Sig. (2-tailed)", in Crosstabs as "Asymptotic Significance (2-sided)", and in SPSS 27 and later some procedures split it into "One-Sided p" and "Two-Sided p" columns. Whichever label appears, that number is the p-value.

What does Sig. .000 mean in SPSS?

It means the p-value is smaller than 0.0005 and has rounded to three decimals, not that it is zero. A p-value can never be exactly zero. Report it as p < .001. To see the actual value, double-click the table and then the cell, or take the test statistic and degrees of freedom to a calculator that does not round at three decimals.

Should I use equal variances assumed or not assumed?

Read the "equal variances not assumed" row, which is Welch's t-test. It is as accurate as the pooled test when the variances happen to be equal and considerably more reliable when they are not, so there is little reason to choose the pooled row on the basis of Levene's test. Welch is why SPSS sometimes shows fractional degrees of freedom such as 26.751.

How do I get a one-tailed p-value in SPSS?

Halve the reported two-tailed value, but only when the observed difference goes in the direction you predicted in advance. If it goes the other way, the correct one-tailed p-value is 1 minus half the two-tailed value, which will exceed .5. Deciding the direction after seeing the data invalidates the test.

Why does SPSS give a different p-value from R?

Most often because a different test was run. SPSS shows pooled and Welch rows together and it is easy to read the wrong one, while R's t.test uses Welch by default and SciPy uses pooled by default. Chi-square is the other common cause: SPSS reports the uncorrected Pearson chi-square and a separate continuity-corrected row for 2 x 2 tables, and R applies the correction by default.

Does SPSS report effect sizes?

Version 27 and later report Cohen's d and Hedges' g alongside t-tests automatically; earlier versions do not, and you have to compute them yourself. This matters because a p-value alone cannot tell you whether an effect is large enough to care about. The effect size calculator here computes d and g with confidence intervals from the same summary statistics SPSS prints.