How to Find SD | Standard Deviation Made Clear

Standard deviation looks scary until you tie it to one plain question: “How far do values tend to sit from the mean?” If two data sets share the same mean, the one with the wider spread has the larger SD.

Below you’ll learn how to compute SD by hand, with a calculator or spreadsheet, and from frequency tables. You’ll also learn how to read the result so you can spot wrong answers fast.

What SD Measures And What It Does Not

SD measures spread. Start with the mean, find each deviation (value minus mean), square those deviations, add them, divide, then take a square root so the answer returns to the original unit.

SD is not a score of “good” or “bad.” A higher SD can be normal if the data cover a wider range. A lower SD can be normal if the data cluster tightly.

SD Vs Variance

Variance is spread in squared units. SD is the square root of variance, so it’s easier to interpret because it matches your unit again.

Sample SD Vs Population SD

You’ll see s for sample SD and σ for population SD. A sample is a subset you measured. A population is the full set you care about. Sample SD divides by n − 1. Population SD divides by n.

Before You Calculate, Pick The SD Type

Many wrong answers come from choosing the wrong SD type. Use these cues:

• Sample SD is common in homework when data come from a survey, experiment, or small set taken from a larger group.
• Population SD fits when you truly have every value in the population you’re describing.

In software, look for “sample,” “population,” STDEV.S, STDEV.P, s, or σ.

How to Find SD By Hand

Hand calculation is slower, but it builds intuition and helps you double-check tool output.

Step 1: Compute The Mean

Add the values and divide by the count. With 2, 4, 4, 4, 5, 5, 7, 9 the mean is 5.

Step 2: Compute Deviations

Subtract the mean from each value. You’ll get negatives below the mean and positives above it.

Step 3: Square Deviations

Square each deviation so negatives don’t cancel positives. Bigger gaps weigh more once squared.

Step 4: Sum The Squared Deviations

Add all squared deviations. This is the sum of squares.

Step 5: Divide To Get Variance

Divide by n for population variance or by n − 1 for sample variance.

Step 6: Square Root For SD

Take the square root of variance. That result is SD.

Mini Check With The Same Data

For 2, 4, 4, 4, 5, 5, 7, 9 the squared deviations sum to 32. Population variance is 32 ÷ 8 = 4, so population SD is 2. Sample variance is 32 ÷ 7 ≈ 4.571, so sample SD is √4.571 ≈ 2.138.

Quick sanity check: values farthest from the mean are 2 and 9, which sit 3 and 4 units away, so an SD a bit above 2 is believable.

Finding Standard Deviation (SD) From Different Data Set Types

Real assignments present data in different forms. The core idea stays the same: distance from the mean.

Raw Lists

Use the hand steps or a stats tool that takes a list. If your list is long, a spreadsheet is the cleanest option.

Frequency Tables

If a value repeats, treat it as a weight. Compute the weighted mean first. Then multiply each squared deviation by its frequency before you add.

Grouped Data (Intervals)

When values come in bins like 10–19, 20–29, use each bin’s midpoint as a stand-in value. That gives an estimate because the exact values inside each bin are unknown.

Data You Have	How To Get SD	Common Trap
Short raw list	Hand steps or calculator stats mode	Wrong SD type setting
Long raw list	Spreadsheet SD function	Range misses some rows
Value + frequency	Weighted mean, then weighted squared deviations	Using an unweighted mean
Grouped intervals	Midpoints + frequencies	Treating the result as exact
Variance given	Square root of variance	Forgetting the square root
Two groups to compare	Compute both with the same method	Mixing sample and population SD
Outliers present	SD plus a resistant spread check	Outliers inflate SD
Only summary stats	Use the given SD if provided	Confusing SD with standard error

How To Get SD Fast In Excel And Google Sheets

Put your numbers in one column, then use a built-in function.

• Sample SD:=STDEV.S(A2:A99)
• Population SD:=STDEV.P(A2:A99)

Audit the range before you trust the result. A skipped row can throw SD off more than the mean. Also watch for cells stored as text.

If you want the formal definition and formula in one place, the NIST e-Handbook states SD as the square root of variance and shows the standard sample formula in its spread section: NIST’s “Measures of Scale”.

How To Find SD On A Calculator Or TI-84

Most calculators have a one-variable statistics mode. You enter a list, then read the SD output.

What To Look For On The Screen

• Sample SD often appears as sx or Sx.
• Population SD often appears as σx.

If your calculator shows only one SD value, check whether it is set to sample SD by default.

TI-84 Steps (One-Var Stats)

1. STAT → EDIT, enter data in L1.
2. STAT → CALC → 1-Var Stats, choose L1, press ENTER.
3. Read Sx and σx in the results list.

SD Vs Standard Error And Z-Scores

SD describes how spread out individual data values are. Standard error describes how spread out a statistic is, usually the sample mean.

If you see “SE” or “standard error,” do not swap it with SD. A common classroom link is SE = s / √n. As the sample size grows, SE shrinks, even if SD stays the same, because the mean becomes more stable.

Z-scores also connect to SD. A z-score counts how many SDs a value sits above or below the mean. If a value is one SD above the mean, its z-score is 1. This is handy when you compare values from different scales.

Shortcut Method When You Have Sums, Not A Full List

Some problems give you Σx and Σx² (sum of values and sum of squared values). You can still get SD without writing every deviation.

First compute the mean: ̄x = Σx / n. Then compute variance using the shortcut: s² = [Σx² − (Σx)² / n] / (n − 1) for a sample, or divide by n for a population. Then take the square root for SD.

This method saves time, but keep parentheses straight. A missed set of brackets can flip the answer.

How To Interpret SD In Plain Language

SD is easiest to read next to the mean. “Mean 50, SD 2” means most values sit close to 50. “Mean 50, SD 20” means values are scattered across a wider span.

Use SD As A Typical Distance

Many class problems treat SD as a typical distance from the mean. That doesn’t mean every value lands that far away. It’s a compact way to describe spread with one number.

Use SD For Comparisons

SD shines when you compare two groups. Compute both SDs with the same SD type and the same unit, then compare the sizes.

Know When SD Can Mislead

SD reacts strongly to extreme values. One large outlier can make SD jump. If your data have sharp outliers, also check a resistant measure like the interquartile range or median absolute deviation so you can see the spread near the center.

Slipup	What It Does	Fix
Wrong SD type (sample vs population)	SD comes out too big or too small	Match s/σ to the task
Square root taken at the wrong time	Variance and SD get mixed up	Sum, divide, then take the root
Mean computed wrong	All deviations shift	Recheck the sum and count
Frequency table not weighted	Spread is distorted	Use weighted mean and weights in the sum
Spreadsheet range misses data	SD ignores values	Select the whole column range
Rounding mid-calculation	Final SD drifts	Round at the end
Outlier treated as a normal point	SD inflates	Check a resistant spread measure too

Two Worked Examples You Can Reuse

Use these as templates for your own numbers.

Example 1: Sample SD From A Short List

Data: 3, 5, 8. Mean is 16 ÷ 3 = 5.333. Squared deviations sum to 12.667. Sample variance is 12.667 ÷ 2 = 6.333. Sample SD is √6.333 ≈ 2.517.

Example 2: Frequency Table SD

Value 1 occurs twice, value 4 occurs once. The mean is (1·2 + 4·1) ÷ 3 = 2. Deviations are −1 and 2. Squared deviations are 1 and 4. Weighted sum is 1·2 + 4·1 = 6. Population variance is 6 ÷ 3 = 2. Population SD is √2 ≈ 1.414.

A Fast SD Check Before You Submit

• If all values match, SD is 0.
• SD cannot be negative.
• SD uses the same unit as the data.
• Switching from sample SD to population SD (same data) makes the population SD smaller.

If your answer breaks one of these rules, revisit the divisor (n vs n − 1), the square-then-root order, and the exact data range used in your tool.

Need more practice with spread and SD in textbook form? OpenStax walks through measures of spread and SD with worked tables and interpretation notes: OpenStax “Measures of the Spread of the Data”.