To calculate RSD on a scientific calculator, switch to single-variable statistics mode, enter your values, read the mean (x̄) and the sample standard deviation (sx or Sx), then calculate sx ÷ x̄ × 100. For six titration volumes of 12.45, 12.52, 12.38, 12.60, 12.47 and 12.41 mL, the mean is 12.4717, sx is 0.07935 and the RSD is 0.64%.
Almost no scientific calculator shows %RSD directly, so the last step is always a short division. The steps below work on any calculator with a statistics mode. There is also a manual method for calculators without one.
The Formula Your Calculator Needs
Here s is the standard deviation and x̄ is the arithmetic mean. A scientific calculator’s statistics mode gives you both numbers. It does not usually combine them, so you finish the calculation yourself.
The only decision you have to make is which standard deviation to use. Most calculators show two, and picking the wrong one is the most common mistake. Step 4 below covers how to tell them apart.
Step 1: Switch to Single-Variable Statistics Mode
Every calculator with statistics functions has a mode for a single list of numbers. The name varies:
| Brand family | What the mode is called | Where to find it |
|---|---|---|
| Casio (MS models) | SD mode | The MODE key |
| Casio (ClassWiz models) | Statistics, 1-Variable | The main menu or HOME screen |
| Texas Instruments TI-30XS MultiView | Data editor and 1-Var Stats | The data and stat keys |
| Sharp WriteView models | SD (single-variable statistics) | The MODE key |
| TI-84 Plus family | List editor and 1-Var Stats | The STAT key |
Choose the single-variable option. Paired or two-variable modes (often labelled REG or 2-Var) are for regression and expect x and y values.
Exact key sequences differ between models, even within one brand. For model-specific steps, see the guides for Casio fx-991MS, fx-991EX and fx-991CW calculators and the TI-84 Plus. Your calculator’s manual is the final reference for anything not covered there.
Step 2: Clear Any Old Data
Calculators keep statistics data from earlier problems, and new values are simply added to the old ones. Clear the data before you start. Depending on the model this is a “clear statistics” option, a “delete all” command in a data editor, or clearing a list.
Some calculators clear the data for you in certain situations. Casio’s fx-991EX manual, for example, says the Statistics Editor is emptied when you leave Statistics Mode. Sharp’s manual says statistical data is erased when the submode is changed. Neither is a reliable habit to depend on, so clear the data yourself every time.
Step 3: Enter the Data
Enter each value once, confirming each with the entry key your calculator uses. On some models that is = or ENTER. On older Casio models it is a DT key, which shares the M+ key.
Use the six titration volumes as a practice set:
| Replicate | Volume (mL) |
|---|---|
| 1 | 12.45 |
| 2 | 12.52 |
| 3 | 12.38 |
| 4 | 12.60 |
| 5 | 12.47 |
| 6 | 12.41 |
When you finish, check the count. Every statistics mode reports n, the number of values entered. If n is not 6, a value is missing, doubled or left over from a previous problem.
If your calculator shows a frequency column, leave every frequency at 1 unless you are deliberately entering repeated values as a frequency table.
Step 4: Read the Mean and the Correct Standard Deviation
Open the statistics results. Look for the mean, labelled x̄, and two standard deviations. Their symbols depend on the brand and the age of the model:
| Calculator family | Sample SD (n − 1) | Population SD (n) |
|---|---|---|
| TI-84 Plus family and TI-30XS MultiView | Sx | σx |
| Casio ClassWiz (fx-991EX, fx-991CW) and newer MS manuals | sx | σx |
| Older Casio models (for example the original fx-991ES and fx-991MS manuals) | xσn-1 or σn-1 | xσn or σn |
| Sharp WriteView (for example EL-W516XG) | sx | σx |
| HP 35s | sx | σx |
The symbols are taken from each manufacturer’s manual. The rule behind them is consistent: an s or an n−1 means the sample standard deviation, and a plain σ or σn means the population standard deviation. The HP 35s user’s guide is one of the few that states the divisor outright, describing sx as calculated with n − 1 and σx with n.
For replicate measurements and most experimental data, use the sample standard deviation. Your values are a sample of what the method or process could produce, not the whole population. Use the population standard deviation only when the data set is the entire group you want to describe. The sample vs population standard deviation guide explains the difference in detail.
For the titration data, the calculator reports:
| Statistic | Symbol | Value |
|---|---|---|
| Number of values | n | 6 |
| Mean | x̄ | 12.471666667 |
| Sample standard deviation | sx | 0.079351539 |
| Population standard deviation | σx | 0.072437713 |
Step 5: Convert the Standard Deviation to %RSD
Divide the sample standard deviation by the mean and multiply by 100:
Report it as 0.64% to two decimal places. If you had used σx, the answer would be 0.58%, which is why the choice in Step 4 matters.
Three checks before you divide
A few seconds of checking catches most calculator errors before they reach your report:
- n matches your data. Six values entered should show n = 6.
- The mean makes sense. Most statistics modes also show Σx, the sum. Σx ÷ n should equal x̄, and x̄ should sit between your smallest and largest values. For the titration data, 74.83 ÷ 6 = 12.4717.
- sx is slightly larger than σx. With the same data, the sample standard deviation is always larger than the population one, because it divides by n − 1 instead of n. If the value you picked is the smaller of the two, you have the population SD.
If any check fails, fix the data before calculating the RSD.
Use the stored values instead of retyping
Most statistics modes store x̄ and sx as variables that you can recall inside an expression. Recalling them avoids copying rounded numbers off the screen. Typing 0.0794 ÷ 12.47 × 100, for example, gives 0.637% rather than 0.636%, and larger rounding produces larger errors.
Where exactly the stored variables live depends on the model: a statistics variable menu, an OPTN or CATALOG menu, or a VARS key. The Casio and TI-84 guides linked above show the exact keys for those models.
Order of operations
Scientific calculators evaluate division and multiplication from left to right, so sx ÷ x̄ × 100 needs no brackets. Typing sx ÷ (x̄ × 100) by mistake gives an answer 10,000 times too small.
Complete Worked Example: What the Calculator Is Doing
Knowing what happens behind the statistics keys helps you spot errors. Here is the same calculation done by hand for the titration data.
Mean: the sum is 74.83 mL, so x̄ = 74.83 ÷ 6 = 12.471667 mL.
| Value (mL) | Deviation from mean | Squared deviation |
|---|---|---|
| 12.45 | −0.021667 | 0.000469 |
| 12.52 | 0.048333 | 0.002336 |
| 12.38 | −0.091667 | 0.008403 |
| 12.60 | 0.128333 | 0.016469 |
| 12.47 | −0.001667 | 0.000003 |
| 12.41 | −0.061667 | 0.003803 |
| Total | 0 | 0.031483 |
Sample variance: 0.031483 ÷ (6 − 1) = 0.0062967.
Sample standard deviation: √0.0062967 = 0.079352 mL.
RSD: 0.079352 ÷ 12.471667 × 100 = 0.636%.
The deviations add up to zero, which confirms the mean is right. The result matches the statistics mode exactly, because the calculator runs the same formula.
If Your Calculator Has No Statistics Mode
Basic scientific calculators without statistics functions can still produce an exact RSD. It just takes more key presses.
- Add the values and divide by n to get the mean. Store the mean in a memory if your calculator has one, so you do not retype it.
- Subtract the mean from each value and square the result. Write each squared deviation down, or add each one to the memory with the M+ key.
- Add the squared deviations. For the titration data the total is 0.0314833.
- Divide by n − 1 for the sample variance: 0.0314833 ÷ 5 = 0.0062967.
- Take the square root for the sample standard deviation: 0.079352.
- Divide by the mean and multiply by 100: 0.079352 ÷ 12.471667 × 100 = 0.636%.
Keep at least five or six significant figures in each intermediate result. The deviations for tightly clustered data are small, and rounding them early is the fastest way to a wrong answer.
A shortcut formula, variance = (Σx² − (Σx)²/n) ÷ (n − 1), needs only the sum and the sum of squares. It is quicker on paper but can lose precision on a calculator when the values are large and close together, so the deviation method above is safer for lab data.
Rounding: When and How Much
- Keep full precision for the mean and standard deviation.
- Round only the final RSD, usually to two decimal places or to the precision your method or course requires.
- If your calculator is set to a fixed number of decimal places (Fix mode), results are rounded on the display but kept in full internally. Recalled variables keep full precision too.
Rounding matters most when the RSD is small. For the titration data, dividing a rounded SD of 0.08 by a rounded mean of 12.5 gives 0.64%, which happens to agree. With 0.08 ÷ 12.4 it becomes 0.65%. Unrounded values avoid the question entirely.
Troubleshooting
| Problem | Likely cause | Fix |
|---|---|---|
| n is larger than the number of values you entered | Old data left in the calculator | Clear the statistics data and re-enter |
| RSD is slightly lower than the expected answer | Population SD (σx) used instead of sample (sx) | Use the SD marked s, Sx or σn-1 |
| RSD is 100 times too large or too small | Multiplied by 100 twice, or left out the × 100 | Check the expression: SD ÷ mean × 100 |
| Answer appears as a fraction or root | Calculator is in an exact or natural display format | Use the model’s decimal conversion key or setting |
| Mean looks right but SD is too large | A value typed wrongly, or a frequency other than 1 | Review the data list and the frequency column |
| RSD is hundreds of percent | Mean close to zero | Check whether RSD is meaningful for the data (see below) |
A huge RSD is usually a warning, not a result. RSD divides by the mean, so data centred near zero gives very large or unstable percentages, and negative means give negative ones. For blanks, baseline readings or data that can be negative, report the standard deviation instead. The home page explains the limitations of RSD.
Is There a Calculator With an RSD Key?
We checked the statistics chapters of the TI-84 Plus family guidebooks, the TI-30XS MultiView guide, the Casio fx-991EX, fx-991ES and fx-991CW manuals, Sharp’s WriteView guide and the HP 35s user’s guide. None lists an RSD or coefficient of variation function. They all report the mean and the standard deviations, and leave the division to you.
That is a small inconvenience, but it has an upside: you always see the standard deviation you are dividing, so the sample or population choice stays visible.
Verify Your Result Online
Enter the same values in the RSD Calculator with the Sample (n − 1) option selected. It shows the mean, standard deviation and %RSD together with each step of the working. If your calculator and the RSD Calculator disagree, compare the three numbers one at a time: the mean first, then the standard deviation, then the final division.
Frequently Asked Questions
Can I calculate RSD on a calculator without a statistics mode?
Yes. Work out the mean, subtract it from each value, square the differences, add them, divide by n − 1 and take the square root to get the sample standard deviation. Then divide by the mean and multiply by 100. It takes longer, but gives the same result.
Which button gives the sample standard deviation?
Look for sx, Sx or, on older Casio models, σn-1 or xσn-1. Those divide by n − 1. The symbols σx, σn or xσn mean the population standard deviation, which divides by n. For replicate measurements, the sample version is usually the right one.
Is there a scientific calculator with an RSD button?
None of the mainstream models we checked in the manufacturers' manuals, including the TI-84 Plus family, TI-30XS MultiView, Casio ClassWiz and ES models, Sharp WriteView models and the HP 35s, has a dedicated RSD or coefficient of variation function. You divide the standard deviation by the mean yourself.
Why is my calculator's RSD different from my textbook?
The most common reasons are using the population standard deviation instead of the sample one, rounding the mean or SD before dividing, and leftover data from a previous problem. Check n first, then the SD symbol, then use unrounded values.