Inverse Normal Distribution Calculator

Q: What is the inverse normal distribution?

The inverse normal distribution (also called the probit or quantile function) reverses the normal CDF. Instead of asking 'what is P(X less than x)?', it asks 'for what x does P(X less than x) equal a given probability p?' Formally, it computes x = mu + z_p times sigma, where z_p is the standard normal quantile satisfying P(Z less than z_p) = p. This calculator uses the Beasley-Springer-Moro rational approximation, accurate to within 1e-7.

Q: How do I find the x value for a given probability in a normal distribution?

Enter your distribution mean (mu), standard deviation (sigma), and the desired cumulative probability p (as a percentage). For left-tail: find x such that P(X less than x) = p. For right-tail: find x such that P(X greater than x) = p. The calculator outputs x and the z-score z = (x - mu) / sigma. For example, with mean = 100, SD = 15, and p = 90% left-tail: x = 119.22 and z = 1.2816.

Q: What is the inverse normal formula?

The inverse normal formula is x = mu + sigma times Phi_inverse(p), where Phi_inverse is the standard normal quantile function. For the standard normal (mu = 0, sigma = 1), the formula reduces to x = Phi_inverse(p) = z_p. Common quantiles: Phi_inverse(0.90) = 1.2816, Phi_inverse(0.95) = 1.6449, Phi_inverse(0.975) = 1.9600, Phi_inverse(0.99) = 2.3263, Phi_inverse(0.999) = 3.0902.

Q: How do I find critical z values for hypothesis testing?

Use the Critical Values mode. Enter your significance level alpha (e.g. 5 for 5%) and select the tail type. Two-tailed tests split alpha equally between both tails, giving critical values of plus or minus z_(alpha/2). Left-tailed tests use z_alpha on the left. Right-tailed tests use z_(1-alpha) on the right. For alpha = 5% two-tailed: critical values are -1.9600 and +1.9600. For alpha = 5% right-tailed: critical value is +1.6449.

Q: What is the difference between the normal CDF and inverse normal CDF?

The normal CDF (forward direction) takes an x value and returns a probability: P(X less than x) = Phi((x - mu) / sigma). The inverse normal CDF (backward direction) takes a probability p and returns the x value: x = mu + Phi_inverse(p) times sigma. The two operations are exact mathematical inverses. The Normal Distribution Calculator computes the forward direction; this calculator computes the backward direction.

Q: What is the 95th percentile of the standard normal distribution?

The 95th percentile of the standard normal distribution (mean 0, SD 1) is z = 1.6449. This means P(Z less than 1.6449) = 95%. For a two-tailed 95% test, the critical values are plus or minus 1.9600 (which corresponds to the 97.5th percentile), because each tail holds 2.5% of the area. The value 1.9600 is one of the most commonly used constants in statistics.

Q: How do I find the percentile of a data value using the inverse normal?

The inverse normal finds the value given a percentile, not the percentile given a value. To go from value to percentile, use the normal CDF (forward): percentile = Phi((x - mu) / sigma) times 100. To go from percentile to value (what this calculator does): x = mu + Phi_inverse(percentile / 100) times sigma. For example, the 90th percentile of IQ (mean 100, SD 15) is 100 + 1.2816 times 15 = 119.22.

Q: What is the inverse normal used for in practice?

The inverse normal appears across many fields. In hypothesis testing, it gives critical z-values that define rejection regions. In confidence intervals, it determines the margin of error (z times SE). In quality control, it finds tolerance limits and process capability thresholds. In finance, it computes Value at Risk (VaR) at a given confidence level. In standardized testing, it converts percentile targets into score cutoffs.

Q: What does the probit function mean?

Probit stands for probability unit. The probit of probability p is the inverse standard normal CDF: probit(p) = Phi_inverse(p). For example, probit(0.5) = 0, probit(0.84) = 1, probit(0.975) = 1.96. Probit regression models use this transformation to convert probabilities into a linear scale. The probit function and the logit function are the two most common link functions in binary regression.

Q: Can I use this calculator for the t-distribution?

No. This calculator is for the normal (z) distribution only. The t-distribution has heavier tails and depends on degrees of freedom. When sample size is large (above 30), the t-distribution is very close to the normal, so the critical z-values here are good approximations. For exact t critical values with small samples, use the Critical Value Calculator, which supports z, t, F, and chi-square distributions.

Find the x value or critical z-score for any normal distribution given a cumulative probability, instantly.

Mean (μ)0

-5050

Standard Deviation (σ)1

0.120

Cumulative Probability (%)95

1%99%

Tail Direction

Significance Level α (%)5

1%20%

Test Type

X Value

—

Z-Score

—

P(X < x)

—

P(X > x)

—

Critical Z (lower / left)

—

Critical Z (upper / right)

—

Confidence Level

—

Reject H&sub0; if

—

🔔 What is the Inverse Normal Distribution?

The inverse normal distribution, also called the quantile function or probit function, answers the reverse of the standard normal CDF question. Instead of asking "what probability corresponds to x?", it asks "what x corresponds to this probability?" Formally, given a normal distribution with mean μ and standard deviation σ, and a target cumulative probability p, the inverse normal finds the value x such that P(X ≤ x) = p. The formula is x = μ + z_p × σ, where z_p is the standard normal quantile satisfying P(Z ≤ z_p) = p.

The inverse normal is one of the most practically important functions in statistics. In hypothesis testing, it identifies the critical z-values that define rejection regions: for a 5% two-tailed test, the critical values z = ±1.9600 are the 2.5th and 97.5th percentiles of the standard normal. In confidence interval construction, z_α/2 = 1.9600 for 95% intervals and 2.5758 for 99% intervals determine the margin of error. In quality control, the inverse normal converts defect rate targets into process capability thresholds. In finance, Value at Risk at a given confidence level uses the inverse normal to convert a loss probability into a dollar threshold.

A common misconception is that the inverse normal is only relevant for the standard normal distribution (mean 0, SD 1). In fact, the formula x = μ + z_p × σ applies to any normal distribution. Finding the 90th percentile of IQ scores (mean 100, SD 15) is the same calculation as finding the standard normal 90th percentile and then scaling: z_0.90 = 1.2816, so the 90th percentile IQ = 100 + 1.2816 × 15 = 119.22. This calculator handles both the general case (any μ and σ) and the standard normal (enter μ = 0, σ = 1).

The Critical Values mode is especially useful for researchers and students running z-tests or constructing large-sample confidence intervals. Enter the significance level α and the tail type (two-tailed, left-tailed, or right-tailed), and the calculator immediately returns the critical z-value(s) that determine whether to reject the null hypothesis. Common results: for α = 5% two-tailed, the critical values are ±1.9600; for α = 1% two-tailed, the critical values are ±2.5758; for α = 5% right-tailed, the critical value is +1.6449.

📐 Formula

x = μ + Φ⁻¹(p) × σ

x = the value such that P(X ≤ x) = p

μ = mean of the normal distribution

σ = standard deviation of the normal distribution (must be > 0)

p = target cumulative probability (between 0 and 1)

Φ⁻¹(p) = standard normal quantile (probit) satisfying P(Z ≤ z) = p

Example: For μ = 100, σ = 15, p = 0.90: z = Φ⁻¹(0.90) = 1.2816, so x = 100 + 1.2816 × 15 = 119.22

For right-tail: x = μ + Φ⁻¹(1 − p) × σ

When P(X > x) = p, the left-tail probability is 1 − p. Substitute 1 − p into the formula above.

Critical value (two-tailed test): z_α/2 = Φ⁻¹(1 − α/2), reject H₀ if |z| > z_α/2

Example: α = 0.05 two-tailed: z_0.025 = Φ⁻¹(0.025) = −1.9600, critical values ±1.9600

📖 How to Use This Calculator

Steps

Choose a calculation mode - Select "Find X Value" to find the x-value (or z-score) for a given cumulative probability in any normal distribution. Select "Critical Values" to find the z-values used as rejection thresholds in hypothesis testing.

Enter distribution parameters and probability - In Find X Value mode, enter the mean and standard deviation of your distribution (use 0 and 1 for the standard normal), then enter the cumulative probability as a percentage. Choose "left tail" for P(X < x) = p or "right tail" for P(X > x) = p.

Read the x value and z-score - The X Value result is the point on the distribution axis satisfying your probability. The Z-Score is the standardized version. The verification percentages P(X < x) and P(X > x) confirm the answer is correct.

💡 Example Calculations

Example 1 - IQ Score 90th Percentile

IQ scores follow N(100, 15). Find the score at the 90th percentile.

Set mean = 100, standard deviation = 15, probability = 90%, tail = left. We want the score x such that P(IQ < x) = 90%.

Find the standard normal quantile: z = Φ⁻¹(0.90) = 1.2816.

Convert to the IQ scale: x = 100 + 1.2816 × 15 = 100 + 19.22 = 119.22.

90th percentile IQ score = 119.22 (z = 1.2816)

Try this example →

Example 2 - Bottom 25th Percentile on an Exam

Exam scores follow N(70, 8). Find the score below which 25% of students fall.

Set mean = 70, standard deviation = 8, probability = 25%, tail = left. We want x such that P(score < x) = 25%.

Find the standard normal 25th percentile: z = Φ⁻¹(0.25) = −0.6745.

Convert to the exam scale: x = 70 + (−0.6745) × 8 = 70 − 5.396 = 64.60.

25th percentile score = 64.60 (z = −0.6745)

Try this example →

Example 3 - Manufacturing Defect Threshold (Top 1%)

Widget lengths follow N(500 mm, 10 mm). Find the length exceeded by only 1% of widgets.

Set mean = 500, standard deviation = 10, probability = 1%, tail = right. We want x such that P(length > x) = 1%.

Right-tail probability 1% means left-tail 99%. Find z = Φ⁻¹(0.99) = 2.3263.

Convert to the length scale: x = 500 + 2.3263 × 10 = 500 + 23.26 = 523.26 mm.

Top-1% threshold = 523.26 mm (z = 2.3263)

Try this example →

❓ Frequently Asked Questions

What is the inverse normal distribution calculator used for?+

The inverse normal calculator finds the x value (or z-score) such that the cumulative probability P(X ≤ x) equals a given target. It is used in hypothesis testing to find critical z-values, in confidence interval construction to find z_alpha/2, in percentile calculations for standardized tests (IQ, SAT, GRE), in quality control to set defect thresholds, and in finance to compute Value at Risk. Essentially any time you need to go from a probability to a value on a normal distribution.

What is the inverse normal formula?+

The inverse normal formula is x = μ + Φ⁻¹(p) × σ, where μ is the mean, σ is the standard deviation, p is the target left-tail probability, and Φ⁻¹(p) is the standard normal quantile (probit). For right-tail probability q, substitute p = 1 − q. Common quantiles: Φ⁻¹(0.90) = 1.2816, Φ⁻¹(0.95) = 1.6449, Φ⁻¹(0.975) = 1.9600, Φ⁻¹(0.99) = 2.3263.

How do I find the critical z value for a 95% confidence interval?+

For a 95% confidence interval, use the Critical Values mode with significance level α = 5% and test type = two-tailed. The critical z-value is z = 1.9600, because each tail holds 2.5% of the area. The 95% confidence interval is x-bar ± 1.9600 × SE, where SE is the standard error. For a 99% CI, the critical value is z = 2.5758 (α = 1%, two-tailed).

What is the difference between left-tail and right-tail in the inverse normal?+

Left-tail finds x such that P(X < x) = p, meaning p fraction of the distribution is to the left of x. Right-tail finds x such that P(X > x) = p, meaning p fraction is to the right. For left-tail p = 95%: x is the 95th percentile. For right-tail p = 5%: x is the 95th percentile (same value). Internally, right-tail probability p is converted to left-tail 1 − p before applying the formula.

What is the 99th percentile of the standard normal distribution?+

The 99th percentile of the standard normal (mean 0, SD 1) is z = 2.3263. This means P(Z < 2.3263) = 99%. The 1st percentile is z = −2.3263. Other key percentiles: 90th = 1.2816, 95th = 1.6449, 97.5th = 1.9600, 99.5th = 2.5758, 99.9th = 3.0902. Enter mean = 0 and SD = 1 in this calculator to confirm any of these values.

How do I convert a percentile to a score for a standardized test?+

Enter the test mean and standard deviation in Find X Value mode, set the percentile as the left-tail probability, and read the score. For example, the SAT (mean 1050, SD 217) score at the 75th percentile: z = Φ⁻¹(0.75) = 0.6745, so score = 1050 + 0.6745 × 217 = 1050 + 146.4 = 1196. For GRE Verbal (mean 151, SD 9): 90th percentile = 151 + 1.2816 × 9 = 162.5.

What critical z value corresponds to a 99% confidence level?+

For a 99% confidence interval (two-tailed), the critical z-value is 2.5758, because α = 1% is split equally: 0.5% in each tail. The exact value is Φ⁻¹(0.995) = 2.5758. For a 90% CI: z = 1.6449 (α = 10%, 5% each tail). For a 95% CI: z = 1.9600 (α = 5%, 2.5% each tail). For a 99.9% CI: z = 3.2905 (α = 0.1%, 0.05% each tail).

How is the inverse normal related to the probit function?+

The probit function is exactly the inverse standard normal CDF: probit(p) = Φ⁻¹(p). The term "probit" (probability unit) was coined by Chester Bliss in 1934. Probit regression transforms a probability into a linear predictor using this function. The inverse normal for a general N(μ, σ) distribution is then just x = μ + probit(p) × σ. This calculator computes both the general inverse normal (Find X Value mode) and the probit for hypothesis testing (Critical Values mode).

What is the rejection region for a one-tailed z-test at alpha = 0.05?+

For a right-tailed test at α = 5%, the rejection region is z > 1.6449, which is Φ⁻¹(0.95). For a left-tailed test at α = 5%, the rejection region is z < −1.6449 = Φ⁻¹(0.05). Reject the null hypothesis H₀ if the calculated test statistic falls in the rejection region. Use the Critical Values mode in this calculator and select the appropriate tail type.

Can I find Value at Risk using the inverse normal?+

Yes. If daily portfolio returns follow a normal distribution with mean μ and standard deviation σ, the VaR at confidence level c is VaR = −(μ + Φ⁻¹(1 − c) × σ). For example, with μ = 0.1% and σ = 2%, VaR at 95% confidence: z = Φ⁻¹(0.05) = −1.6449, VaR = −(0.001 + (−1.6449) × 0.02) = −(0.001 − 0.0329) = 3.19%. This means a 5% chance of losing more than 3.19% on any given day.

What numerical method does this calculator use for the inverse normal?+

This calculator uses the Beasley-Springer-Moro (BSM) rational approximation. It partitions the probability range into three regions: the far-left tail (p < 0.02425), the central region (0.02425 ≤ p ≤ 0.97575), and the far-right tail (p > 0.97575). Each region uses a separate rational polynomial fit to Φ⁻¹(p). The maximum absolute error across all three regions is less than 4.5 × 10⁻⁴. The forward verification (CDF check) uses the Abramowitz and Stegun error function approximation, accurate to 1.5 × 10⁻⁷.

How does the inverse normal differ from the inverse t-distribution?+

The inverse t-distribution finds critical values for the t-distribution, which has heavier tails than the normal and depends on degrees of freedom (df). For large df (above 30), the t and z critical values converge: at df = 120, the t critical value for 95% two-tailed is 1.980, very close to z = 1.960. For small samples (df < 30), the t critical values are noticeably larger than z, reflecting greater uncertainty. Use the inverse normal (this calculator) when the population standard deviation is known or the sample is large. Use the inverse t (Critical Value Calculator) for small samples.

🔗 Related Calculators

What is the inverse normal distribution?

The inverse normal distribution (also called the probit or quantile function) reverses the normal CDF. Instead of asking 'what is P(X less than x)?', it asks 'for what x does P(X less than x) equal a given probability p?' Formally, it computes x = mu + z_p times sigma, where z_p is the standard normal quantile satisfying P(Z less than z_p) = p. This calculator uses the Beasley-Springer-Moro rational approximation, accurate to within 1e-7.

How do I find the x value for a given probability in a normal distribution?

Enter your distribution mean (mu), standard deviation (sigma), and the desired cumulative probability p (as a percentage). For left-tail: find x such that P(X less than x) = p. For right-tail: find x such that P(X greater than x) = p. The calculator outputs x and the z-score z = (x - mu) / sigma. For example, with mean = 100, SD = 15, and p = 90% left-tail: x = 119.22 and z = 1.2816.

What is the inverse normal formula?

The inverse normal formula is x = mu + sigma times Phi_inverse(p), where Phi_inverse is the standard normal quantile function. For the standard normal (mu = 0, sigma = 1), the formula reduces to x = Phi_inverse(p) = z_p. Common quantiles: Phi_inverse(0.90) = 1.2816, Phi_inverse(0.95) = 1.6449, Phi_inverse(0.975) = 1.9600, Phi_inverse(0.99) = 2.3263, Phi_inverse(0.999) = 3.0902.

How do I find critical z values for hypothesis testing?

Use the Critical Values mode. Enter your significance level alpha (e.g. 5 for 5%) and select the tail type. Two-tailed tests split alpha equally between both tails, giving critical values of plus or minus z_(alpha/2). Left-tailed tests use z_alpha on the left. Right-tailed tests use z_(1-alpha) on the right. For alpha = 5% two-tailed: critical values are -1.9600 and +1.9600. For alpha = 5% right-tailed: critical value is +1.6449.

What is the difference between the normal CDF and inverse normal CDF?

The normal CDF (forward direction) takes an x value and returns a probability: P(X less than x) = Phi((x - mu) / sigma). The inverse normal CDF (backward direction) takes a probability p and returns the x value: x = mu + Phi_inverse(p) times sigma. The two operations are exact mathematical inverses. The Normal Distribution Calculator computes the forward direction; this calculator computes the backward direction.

What is the 95th percentile of the standard normal distribution?

The 95th percentile of the standard normal distribution (mean 0, SD 1) is z = 1.6449. This means P(Z less than 1.6449) = 95%. For a two-tailed 95% test, the critical values are plus or minus 1.9600 (which corresponds to the 97.5th percentile), because each tail holds 2.5% of the area. The value 1.9600 is one of the most commonly used constants in statistics.

How do I find the percentile of a data value using the inverse normal?

The inverse normal finds the value given a percentile, not the percentile given a value. To go from value to percentile, use the normal CDF (forward): percentile = Phi((x - mu) / sigma) times 100. To go from percentile to value (what this calculator does): x = mu + Phi_inverse(percentile / 100) times sigma. For example, the 90th percentile of IQ (mean 100, SD 15) is 100 + 1.2816 times 15 = 119.22.

What is the inverse normal used for in practice?

The inverse normal appears across many fields. In hypothesis testing, it gives critical z-values that define rejection regions. In confidence intervals, it determines the margin of error (z times SE). In quality control, it finds tolerance limits and process capability thresholds. In finance, it computes Value at Risk (VaR) at a given confidence level. In standardized testing, it converts percentile targets into score cutoffs.

What does the probit function mean?

Probit stands for probability unit. The probit of probability p is the inverse standard normal CDF: probit(p) = Phi_inverse(p). For example, probit(0.5) = 0, probit(0.84) = 1, probit(0.975) = 1.96. Probit regression models use this transformation to convert probabilities into a linear scale. The probit function and the logit function are the two most common link functions in binary regression.

Can I use this calculator for the t-distribution?

No. This calculator is for the normal (z) distribution only. The t-distribution has heavier tails and depends on degrees of freedom. When sample size is large (above 30), the t-distribution is very close to the normal, so the critical z-values here are good approximations. For exact t critical values with small samples, use the Critical Value Calculator, which supports z, t, F, and chi-square distributions.

How accurate is the inverse normal calculation?

This calculator uses the Beasley-Springer-Moro rational approximation, which has a maximum absolute error of less than 4.5e-4 across the full range (0, 1). For the central region (probability 0.025 to 0.975), the error is less than 1e-7. The Abramowitz and Stegun error function approximation used for verification (the forward CDF) has a maximum error of 1.5e-7. Together, verification output confirming P(X less than x) matches the input probability to 4 decimal places.

📌 Quick Tips

💡The standard normal (mean 0, SD 1) 95th percentile is z = 1.6449. For the 97.5th percentile (used in two-tailed 95% tests), z = 1.9600. These two values appear constantly in statistics.

💡For a right-tail probability, enter the tail area on the right. For example, P(X greater than x) = 5% means enter probability = 5% with right tail, giving the 95th percentile x value.

💡The inverse normal is also called the probit function, percent-point function (PPF), or quantile function. All of these refer to finding the x value such that P(X less than or equal to x) equals your given probability.

💡Common critical z-values to remember: 90% CI uses z = 1.6449; 95% CI uses z = 1.9600; 99% CI uses z = 2.5758. Use the Critical Values mode to verify these instantly.

💡To find percentile cutoffs for standardized tests, enter the test mean and SD, then the desired percentile as the left-tail probability. For example, to find the SAT score at the 75th percentile with mean 1050 and SD 217, enter probability = 75%.