One Sample Z-Test Calculator

Test whether a sample mean differs from a hypothesized population mean when the population standard deviation is known, using z = (sampleMean - mu0) / (populationSD / sqrt(n)).

📌Presets

Load a realistic z-test case, then adjust the sample mean, hypothesized mean, known population SD, sample size, tail, confidence level, and alpha.

z Statistic -- standard normal score

p-Value -- selected tail area

Mean CI -- confidence interval

Decision -- compare p to alpha

Enter a sample mean, hypothesized mean, population standard deviation, and sample size to calculate the z-test.

⚙Z-Test Inputs

Test label Short label used in the result statement.

Measurement units Shown beside mean differences and intervals.

Sample mean (sampleMean) The observed mean from the sample.

Hypothesized mean (mu0) The null-hypothesis population mean.

Population standard deviation Use known sigma, not sample SD.

Sample size (n) Standard error is sigma / sqrt(n).

Alternative hypothesis Tail choice controls the p-value.

Significance level alpha Decision rule: reject H0 when p <= alpha.

Confidence level CI is for the population mean.

Decimal places Controls z, p-value, CI, and SE rounding.

Interval display Difference interval centers on sampleMean - mu0.

Reference table focus Updates the quick reference table.

📋Live Tail Breakdown Grid

Measure

Value

Formula

Tail Area

Alpha Check

Use

Mean difference

sampleMean - mu0

center

direction

Standard error

sigma / sqrt(n)

scale

denominator

Left p

normal CDF(z)

P(Z <= z)

left tests

Right p

1 - CDF(z)

P(Z >= z)

right tests

Two p

2 x min tails

both sides

two-tailed

Critical z

cutoff

📊Output Summary Cards

--standard error

--mean difference

--CI margin

--CI z critical

--left p-value

--right p-value

--two-tail p

--alpha rule

🗄Z-Test Reference Tables

Setup	Use When	Statistic	Standard Error	P-Value Rule
One sample z-test	Population sigma is known	z = (sampleMean - mu0) / SE	sigma / sqrt(n)	Normal tail area
One sample t-test	Population sigma is unknown	t = (sampleMean - mu0) / SE	s / sqrt(n)	Student t with df = n - 1
Right-tailed z-test	Question asks whether mean is greater	Positive z supports H1	Known sigma scale	p = 1 - CDF(z)
Left-tailed z-test	Question asks whether mean is less	Negative z supports H1	Known sigma scale	p = CDF(z)
Two-tailed z-test	Question asks whether mean differs	Large abs(z) supports H1	Known sigma scale	p = 2 x (1 - CDF(abs(z)))
Mean confidence interval	Estimate the population mean	sampleMean +/- z* x SE	sigma / sqrt(n)	Use central z*

Alpha	One-Tail Cutoff	Two-Tail Cutoff	Central Confidence	Use For	Known Sigma Check
0.10	1.282	1.645	90%	Exploratory screens	sigma set before sampling
0.05	1.645	1.960	95%	Common mean tests	process SD documented
0.025	1.960	2.241	97.5%	Stricter one-tail rules	not sample SD
0.01	2.326	2.576	99%	Strict claims	stable population spread
0.001	3.090	3.291	99.9%	Very rare tail evidence	normal model suitable

Reference	Tail / Confidence	Alpha	Lower z	Upper z	Decision Cue
Common	Two-tail	0.05	-1.960	1.960	abs(z) >= 1.960

🧮Formula Notes

Test statistic: z = (sampleMean - mu0) / (populationSD / sqrt(n)). The p-value is read from the standard normal distribution using the selected left, right, or two-tailed alternative.

Confidence interval: mean CI = sampleMean +/- z* x populationSD / sqrt(n), where z* is the central normal critical value for the selected confidence level.

Use known population SD. If sigma is estimated from the same sample, a one sample t-test is usually the better match.

Set the alternative first. A one-tailed z-test should be chosen before looking at whether the sample mean moved up or down.

It is a factory floor. There is a clipboard. Bottles roll along on a conveyor belt. Is it what the bottle say? It is fifty milliliters. Your sample consists of 64 bottles that average 49.8. Maybe that’s just a rounding error. Maybe it’s just some sensor glitch. Maybe it’s a real process failure.

How do you know? The one sample z-test replace your intuition with evidence and transforms a vague feeling into a statistical verdict. But only if you have the correct inputs.

How to Use the One Sample Z-Test

What’s needed? The test require knowledge of the standard deviation for the population. This means you don’t estimate the spread based off your current sample. You know the naturaly variation in the system, because you have historic data or some other sort of certified process specification.

When you guess the spread, you’re using your current sample. In that case, you should of run a t-test. Otherwise you’re assuming you know the terrain. That make a difference in the math.

That’s why the calculator above runs the numbers for you. You give it the known standard deviation. It saves you having to look up tables manualy.

How much does it matter? Let’s find out.

Plug in the mean of your sample. Plug in your target hypothesis. Plug in the standard deviation you know about. The calculator spits out the standard error, or the noise floor of your measurement. It take the population spread and divides it by the square root of the number in your sample. The bigger the sample, the more confident we feels. The error term shrinks. We can see smaller changes because they stands out against this backdrop. With ten bottle, zero point two milliliters gets lost. With a thousand bottles, it stand out.

Next, select which way the tail face. Decide before seeing results. Decide depending on what you’re asking. Worry about low values? Left-tailed test. Care about both high and low values? Two-tailed test. Violates honesty (or efficiency). Don’t get this one confused after the fact. Statistical sin. Like moving the goalposts. P-value depend on your selection. Pick the side before starting.

So what does the confidence interval provide? It is a range that includes plausible values of the true mean. Your target value falls within this range. There’s no reason to reject the null hypothesis. In other words, the data match your target. But it’s not perfectly on-target.

Then there are the critical cutoffs from the reference table. These list common alpha levels such as zero point zero five. Here you can verify your z-score and determine whether or not it cross over into rejection territory.

A stable process is where this test shines. Look, you’re monitoring call center response time. You are monitoring your manufacturing line. You are looking at standardized test scores. All of these have long histories. We know what their standard deviations looks like.

The process is new or chaotic. The known sigma is no longer real. Don’t force the z-test here. Let the history and data speaks for themselves.

That is what the z-test do. It separates noise from signal. It works off of a known baseline. It turns measurements into a decision rule. You evaluate a survey result. You audit a production run. The tool does the work. You bring the historical context. You bring the right questions.

The bottles on the belt didn’t care about your hunch. They cared about the numbers.