Converting a Z Cutoff Back to Original Units

Why Convert a Z Cutoff?

A z-score reports a value’s location relative to a distribution’s mean, in standard deviation units. That is useful for comparing positions, but many decisions need a cutoff in the variable’s original units: a test score in points, a measurement in centimeters, or a time in minutes. If the normal model and a z cutoff are known, the original-scale cutoff can be recovered directly.

This is the reverse of standardizing a value. Instead of starting with an observed value and asking how many standard deviations it lies from the mean, we start with that standardized distance and find the corresponding value in the original scale. The sign of the z-score tells us whether the value is below or above the mean; the standard deviation determines how far away it is in the original units.

The Conversion Formula

Definition: If $X$ follows a normal model with mean $\mu$ and standard deviation $\sigma>0$, then a z cutoff $z$ corresponds to the original-scale cutoff $$ x=\mu+z\sigma. $$ The value $x$ is measured in the same units as $X$.

The formula comes from the z-score transformation $z=(x-\mu)/\sigma$. It applies when the mean and standard deviation describe the normal population or model under consideration, and the given z-score is on the same standard normal scale. In an application, check that the variable is being modeled as normal and that $\mu$ and $\sigma$ are the appropriate parameters. The standard deviation must be positive. A positive z cutoff produces a value above the mean; a negative z cutoff produces a value below the mean.

Why the Formula Works

Standardizing a value subtracts the mean and divides by the standard deviation. To undo those operations, first multiply the z-score by the standard deviation, then add the mean. The algebra makes clear both the order of operations and why the units return to their original scale.

z=\frac{x-\mu}{\sigma} \quad\Longrightarrow\quad z\sigma=x-\mu \quad\Longrightarrow\quad x=\mu+z\sigma.

Here $z$ has no units, while $\sigma$ has the units of $X$. Thus $z\sigma$ is a distance in the original units, and adding that distance to $\mu$ gives an original-scale value. For $z=0$, the formula gives $x=\mu$, as it should. For $z<0$, $z\sigma$ is negative and the cutoff lies below the mean.

Worked Example: A Cutoff Above the Mean

Suppose, hypothetically, that scores in a specified population are modeled by $X\sim N(72,10^2)$, where scores are measured in points. Find the score corresponding to $z=1.35$.

Check the model and units. The problem specifies a normal model with mean $\mu=72$ points and standard deviation $\sigma=10$ points. The cutoff is 1.35 standard deviations above the mean.

Substitute into the conversion formula.

x=\mu+z\sigma =72+(1.35)(10) =72+13.5 =85.5\text{ points}.

Under this model, a score of 85.5 points is 1.35 standard deviations above the mean. It is the cutoff corresponding to $z=1.35$; scores above this cutoff have z-scores greater than 1.35. The converted value is in points, not standard deviation units.

Negative Cutoffs and Values Below the Mean

A negative z-score does not mean that the original measurement is negative. It means that the value is below the mean. The conversion formula handles this through the sign of $z\sigma$: for a negative z-score, subtract the magnitude of that distance from the mean. The result may still be positive, depending on the variable and its mean.

Worked Example: A Lower Cutoff in a Measurement Distribution

Imagine a hypothetical population whose resting pulse rates are modeled by $X\sim N(68,9^2)$, with pulse rate measured in beats per minute. Find the cutoff at $z=-1.20$.

Check the model and units. The specified model is normal, with $\mu=68$ beats per minute and $\sigma=9$ beats per minute. Since the z-score is negative, the cutoff should be below 68 beats per minute.

Calculate the distance from the mean. The standardized distance in original units is $z\sigma=(-1.20)(9)=-10.8$ beats per minute.

Recover the cutoff.

x=\mu+z\sigma =68+(-1.20)(9) =68-10.8 =57.2\text{ beats per minute}.

In this hypothetical normal model, 57.2 beats per minute is 1.20 standard deviations below the mean. The negative z-score describes the cutoff’s position, not a negative pulse rate. The calculation also confirms the expected direction: 57.2 is less than the mean of 68.

When Both Ends of a Range Are Given

The conversion works separately for each z cutoff. If a range is defined by a lower and an upper z-score, convert both endpoints using the same $\mu$ and $\sigma$. Because $\sigma>0$, larger z-scores correspond to larger original-scale values, so the lower endpoint remains the lower endpoint.

Worked Example: Converting Two Z Cutoffs

Suppose, hypothetically, that a process duration is modeled by $X\sim N(24,3.5^2)$, with times measured in minutes. Convert the z-score interval from $-1.645$ to $1.645$ into minutes.

Check the model and units. The given normal model has $\mu=24$ minutes and $\sigma=3.5$ minutes. Both z cutoffs use these same parameters.

Convert the lower endpoint.

x_{\mathrm{low}} =24+(-1.645)(3.5) =24-5.7575 =18.2425\text{ minutes}.

Convert the upper endpoint.

x_{\mathrm{high}} =24+(1.645)(3.5) =24+5.7575 =29.7575\text{ minutes}.

The original-scale interval is approximately 18.24 to 29.76 minutes. These are the values corresponding to the stated z boundaries in this model. If those boundaries were chosen to enclose a particular standard normal area, the converted interval would enclose the same model area; the conversion changes the scale, not the area.

Interpreting and Checking a Converted Cutoff

A useful interpretation connects the converted value to the variable, its units, and its location relative to the mean. For example, say “85.5 points is the cutoff 1.35 standard deviations above the modeled mean of 72 points,” rather than merely saying “the answer is 85.5.” The second statement includes a number but does not identify what it measures or why it is the relevant cutoff.

Before accepting a result, check its direction and size. A positive z-score should yield a value above $\mu$; a negative z-score should yield a value below $\mu$. The distance from the mean should be $|z|\sigma$. For instance, if $|z|=2$, the cutoff must be two standard deviations from the mean, not two original measurement units from it. Retain units for $\mu$, $\sigma$, and the final value, while remembering that z itself is unitless.

Common Mistakes

Using the inverse formula incorrectly. A common wrong step is to divide the z-score by the standard deviation or to use $x=\mu-z\sigma$ for every cutoff. The correct rearrangement is $x=\mu+z\sigma$; the sign of $z$ already determines whether to move above or below the mean.
Treating a z-score as an original-scale value. A cutoff of $z=1.35$ is not 1.35 points, minutes, or centimeters above zero. It is 1.35 standard deviations from the mean.
Dropping the negative sign. Replacing $z=-1.20$ with $1.20$ changes a lower cutoff into an upper cutoff. Keep the sign when substituting.
Mixing units or parameters. Use the mean and standard deviation for the specified variable and population, and keep their units consistent. A standard deviation from a different population does not convert the given z-score correctly.
Rounding too early. Keep full precision during the multiplication and addition, then round the final cutoff to a sensible number of significant figures for the context.

Key takeaway: To return from a z cutoff to the original scale, multiply by the model standard deviation and add the model mean: $x=\mu+z\sigma$. The sign locates the cutoff relative to the mean, and the final value carries the units of the original variable.

Check Your Understanding

For each question, show the conversion and state the result in context.

A hypothetical normal model has mean 50 and standard deviation 6 units. What original-scale value corresponds to $z=0.75$?
A variable is modeled as normal with mean 120 milligrams and standard deviation 15 milligrams. Find the cutoff corresponding to $z=-1.4$, and say whether it is above or below the mean.
A normal model has mean 8.0 seconds and standard deviation 0.5 seconds. Convert the z-score interval from $-2$ to $1$ into seconds.
Explain why a negative z-score does not necessarily produce a negative original-scale value.
A learner calculates a cutoff by dividing $z$ by $\sigma$, then adding $\mu$. Identify the error and write the correct formula.