Our Z score Calculator is a free statistical tool that helps you standardize data by converting raw scores into z-scores. Z-scores measure how far a data point is from the mean in terms of standard deviations.
This calculator doesn't change your data distribution—instead, it transforms values to a common scale (mean = 0, standard deviation = 1), making it easy to compare measurements from different datasets, identify outliers, and calculate probabilities.
What is a Z score?
A z-score (also called a standard score) is a statistical measurement that tells you how many standard deviations a data point is from the mean of your dataset.
It allows you to:
- Standardize data - Convert raw scores to a common scale (mean=0, std dev=1)
- Compare values - Compare observations from different datasets directly.
- Identify outliers - Spot unusual or extreme values easily.
- Calculate probabilities - Determine what percentage of data falls below a value.
Key Concept:
- Z = 0 → Value equals the mean.
- Z > 0 → Value is above the mean.
- Z < 0 → Value is below the mean.
- |Z| > 3 → Extreme value (potential outlier - less than 0.3% of data)
Z score Formula, Explained
There isn't a single z-score formula — there are three, and which one applies depends entirely on which of the three input modes above matches your situation.
Z-score Example
Example 1: A Single Test Score
Scenario: A student scores 92 on a national exam. The exam's population mean is 80, with a standard deviation of 6. How does this score compare to everyone else who took it?
Step 1 — Subtract the mean from the score: 92 − 80 = 12
Step 2 — Divide by the standard deviation: 12 / 6 = 2.0
Interpretation: A z-score of 2.0 puts this student roughly 2 standard deviations above average — about the 97.7th percentile, meaning only around 2.3% of test-takers scored higher. It's a strong result and well within the expected range; only values beyond ±3 would typically warrant closer scrutiny.
Example 2: Comparing a Sample to a Known Standard
Scenario: A manufacturer's bolts are specified to have a population mean diameter of 10mm with a standard deviation of 0.4mm. A sample of 16 bolts pulled from the line averages 10.2mm. Has the process actually drifted, or is this normal variation?
Step 1 — Apply the standard error formula: standard error = 0.4 / √16 = 0.1
Step 2 — Calculate z: z = (10.2 − 10) / 0.1 = 2.0
Interpretation: Even though the raw difference (0.2mm) looks tiny, a z-score of 2.0 for a sample mean is statistically meaningful — there's roughly a 5% chance of seeing a deviation this large purely by random variation, which is usually enough to flag the process for a closer look in a quality-control setting.
Example 3: A Full Dataset (Matching the Calculator's Default)
Scenario: Using the calculator's own default dataset — 2, 4, 3, 34, 23.
Step 1 — Find the mean: (2 + 4 + 3 + 34 + 23) / 5 = 13.2
Step 2 — Find the population standard deviation: ≈ 12.98
Step 3 — Calculate a z-score for each value:
| Value | Deviation from Mean | Z-Score |
|---|
| 2 | −11.2 | −0.86 |
| 4 | −9.2 | −0.71 |
| 3 | −10.2 | −0.79 |
| 34 | +20.8 | 1.60 |
| 23 | +9.8 | 0.75 |
Interpretation: None of these cross the |z| > 3 outlier threshold, but 34 is clearly the standout value of the set at z = 1.60; it's the farthest from the group's center of gravity by a meaningful margin, even though it's not extreme enough to call a true statistical outlier on its own.
How to Interpret a Z-Score
| Z-Score Range | Position | Approx. Share of Data | Read |
|---|
| Below −3 | Far below mean | < 0.15% | Extreme — worth investigating |
| −3 to −2 | Well below mean | ~2.1% | Notable, possible outlier |
| −2 to −1 | Below mean | ~13.6% | Lower than typical, not unusual |
| −1 to +1 | Near mean | ~68.3% | Typical range |
| +1 to +2 | Above mean | ~13.6% | Higher than typical, not unusual |
| +2 to +3 | Well above mean | ~2.1% | Notable, possible outlier |
| Above +3 | Far above mean | < 0.15% | Extreme — worth investigating |
This follows the empirical rule (sometimes called the 68-95-99.7 rule): in a normal distribution, about 68.27% of values fall within one standard deviation of the mean, 95.45% within two, and 99.73% within three. Anything beyond ±3 standard deviations covers less than 0.3% of the data combined, which is the statistical basis for treating |z| > 3 as a reasonable, if somewhat arbitrary, outlier cutoff.
Z-Score to Percentile Reference
A z-score and a percentile describe the same position in a distribution from two different angles. This table covers the values people look up most often:
| Z-Score | Percentile (% below) | Common Context |
|---|
| −3.00 | 0.13% | Far below average |
| −2.00 | 2.28% | Well below average |
| −1.00 | 15.87% | Below average |
| 0.00 | 50.00% | Exactly average |
| +1.00 | 84.13% | Above average |
| +1.28 | ~90.0% | 90th-percentile cutoff |
| +1.645 | ~95.0% | One-tailed 95% confidence cutoff |
| +1.96 | ~97.5% | Two-tailed 95% confidence cutoff |
| +2.00 | 97.72% | Well above average |
| +3.00 | 99.87% | Far above average |
Need a value not listed here? Run the figure through the
percentile calculator for an exact result.
When Z-Scores Don't Work Well
Z-scores are simple to calculate, which is exactly why they're easy to misapply. A few situations where the number you get back may be technically correct but practically misleading:
- Skewed data. The percentile interpretations above assume a roughly normal distribution. Apply them to strongly skewed data — income or reaction times are classic examples — and the percentile read can be noticeably off, even though the arithmetic is fine.
- Very small samples. With fewer than about 30 observations, a sample standard deviation is a rough estimate of the true population value. A t-distribution accounts for that extra uncertainty; a plain z-score doesn't.
- The wrong standard deviation. Mixing up a sample standard deviation with a population standard deviation — or using one from the wrong reference group entirely — produces a z-score that's confidently wrong rather than obviously wrong.
- Correlated or time-series data. The standard z-score assumes independent observations. Daily stock prices or repeated measurements on the same subject violate that assumption and usually call for a different method.
- Heavy outliers already in the data. A single extreme value inflates the standard deviation, which in turn shrinks everyone else's z-score — sometimes masking the very outliers you're trying to detect. The median absolute deviation is a more robust alternative when this is a concern.
Frequently Asked Questions
What's the difference between a z-score and a t-score?
A z-score uses the population standard deviation; a t-score uses the sample standard deviation and adjusts for the extra uncertainty that comes with estimating it, particularly in smaller samples. As a rule of thumb, use a t-score whenever the true population standard deviation isn't known and the sample is under about 30.
Can a z-score be negative?
Yes. A negative z-score simply means the value falls below the mean — it isn't a sign of an error, and depending on what's being measured, "below average" isn't necessarily a bad outcome.
What does a z-score of 0 mean?
It means the value is exactly equal to the mean of the dataset — the precise center of the distribution.
Why is |z| > 3 the usual outlier cutoff?
In a normal distribution, values beyond three standard deviations from the mean make up less than 0.3% of all observations combined. That rarity is what makes ±3 a practical, widely-used threshold — though it's a convention, not a hard statistical law, and some fields use ±2 instead.
Does converting to z-scores change the shape of my data?
No. Standardizing rescales the values to a mean of 0 and a standard deviation of 1, but it doesn't change the underlying distribution's shape. Skewed data stays skewed after converting to z-scores only the scale changes, not the pattern.
How do I convert a z-score into a percentile?
Look the z-score up in a standard normal (z) table, or use the percentile conversion table on this page for the most commonly needed values. A z-score of 1.96, for instance, corresponds to roughly the 97.5th percentile.
Very Poor
Poor
Average
Good
Excellent
Feedback Submitted Successfully.
Feedback Submitted Successfully.
