Reporting Mean and Standard Deviation With Correct Precision

{
“title”: “Reporting Mean and Standard Deviation With Correct Precision”,
“slug”: “reporting-mean-and-standard-deviation-precision”,
“excerpt”: “Learn the correct rules for reporting mean and standard deviation, including significant figures, rounding conventions, and standards from NIST, GUM, and ISO.”,
“seo_title”: “Mean & SD Precision: How to Report Correctly | SigFig Reference”,
“meta_description”: “Master the rules for reporting mean and standard deviation with correct significant figures. Includes standards, examples, and common pitfalls.”,
“content”: “

Rule Statement

When reporting a mean and standard deviation (SD), the precision of both values must be consistent. The standard rule, endorsed by major metrology and statistics guides, is to round the standard deviation to one or two significant figures (depending on the context and sample size), and then round the mean to the same decimal place as the final rounded SD. This ensures that the uncertainty is expressed with appropriate precision and that the mean does not imply false precision beyond the uncertainty.

For most scientific and engineering applications, one significant figure for the SD is sufficient, especially when the sample size is small (n < 10). For larger samples (n ≥ 10), two significant figures are often recommended to avoid loss of information. The mean is then reported to the same decimal place as the SD. For example, if the SD is rounded to 0.3 (one decimal place), the mean must also be reported to one decimal place.

This rule is a direct consequence of the concept of significant figures: the uncertainty (SD) defines the last reliable digit of the measurement. Reporting more digits in the mean would be misleading.

Worked Examples

Example 1: Small Sample (n = 5)

Measured values: 10.2, 10.5, 10.3, 10.1, 10.4

1. Calculate mean: (10.2+10.5+10.3+10.1+10.4)/5 = 51.5/5 = 10.30
2. Calculate standard deviation (sample): s = 0.1581 (using formula)
3. Round SD to one significant figure: 0.1581 → 0.2 (since first non-zero digit is 1, we round to 0.2 because 0.1581 is closer to 0.2 than 0.1? Actually 0.1581 rounds to 0.2 if we keep one sig fig? 0.1581 has first sig fig 1, second is 5, so round up to 0.2. Yes.)
4. Round mean to same decimal place as SD: SD is 0.2 (one decimal place), so mean = 10.3
5. Report: Mean = 10.3 ± 0.2

Example 2: Larger Sample (n = 20)

Measured values (hypothetical): mean = 25.678, SD = 1.2345

1. Since n = 20 (≥10), use two significant figures for SD: 1.2345 → 1.2 (two sig figs: first two digits are 1 and 2, third is 3, so round down to 1.2)
2. Round mean to same decimal place as SD: SD is 1.2 (one decimal place), so mean = 25.7
3. Report: Mean = 25.7 ± 1.2

Example 3: Very Small SD

Mean = 0.0004567, SD = 0.00001234

1. Round SD to one significant figure: 0.00001234 → 0.00001 (since first sig fig is 1, second is 2, so round to 0.00001)
2. Round mean to same decimal place: SD has 5 decimal places (0.00001), so mean = 0.00046 (round to 5 decimal places: 0.0004567 → 0.00046)
3. Report: Mean = 0.00046 ± 0.00001

Counter-Examples

Common errors arise from reporting too many digits or inconsistent precision.

• Over-precise mean: Reporting mean = 10.3000 ± 0.2 is wrong because the mean implies precision to 0.0001, while the SD only justifies 0.1. The extra zeros are not significant.
• Under-precise SD: Reporting SD = 0.1581 (with four decimal places) when the mean is reported to one decimal place is inconsistent. The SD should be rounded to match the mean’s decimal place, or the mean should be rounded to match the SD’s precision.
• Inconsistent decimal places: Reporting mean = 10.3 ± 0.25 is wrong because the SD has two decimal places while the mean has one. Both must have the same decimal place.
• Using too many significant figures for SD: For n=5, reporting SD = 0.1581 (four sig figs) is excessive. The uncertainty itself is uncertain; one or two sig figs are sufficient.

Convention Comparison Table

Standards Citation

Several international standards and guides provide explicit guidance on reporting uncertainty and significant figures.

• NIST Technical Note 1297 (1994), Section 7.5: “The numerical value of the uncertainty should be given to two significant figures, except when the leading digit is 1, in which case three significant figures may be used.” This applies to expanded uncertainty, but the same principle is often applied to standard deviation.
• JCGM 100:2008 (GUM), Section 7.2.6: “The numerical values of the estimate y and its standard uncertainty u(y) should be given to an appropriate number of significant figures. In general, it is sufficient to quote u(y) to two significant figures.”
• ASTM E29-13, Section 6.4: “The standard deviation should be reported to one or two significant figures, and the mean should be reported to the same decimal place as the standard deviation.”
• ISO 80000-1:2009, Section 7.3.4: “The uncertainty of a measurement result should be expressed to two significant figures, and the result should be rounded to the same decimal place.”

Common Mistakes

1. Reporting too many decimal places in the mean – e.g., 10.3000 ± 0.2. The mean should be rounded to the same decimal place as the SD.
2. Reporting SD with too many significant figures – e.g., 0.1581 instead of 0.2. The SD itself is an estimate, so 1-2 sig figs is enough.
3. Mixing decimal places between mean and SD – e.g., 10.3 ± 0.25. Both must have the same decimal place.
4. Ignoring the leading digit rule – When the SD starts with 1 or 2, it may be better to use 2 sig figs (e.g., 0.12 rather than 0.1) to avoid rounding away too much information.
5. Rounding the mean before calculating SD – Always compute statistics from unrounded data, then round the final results.
6. Using the same number of significant figures for SD as for the mean – The mean may have many digits, but the SD should be reported with fewer.

Practice Problems

1. Data: 12.1, 12.3, 12.2, 12.4, 12.0. Calculate mean and SD, then report with correct precision (use 1 sig fig for SD).
2. Data: 0.0045, 0.0047, 0.0046, 0.0044, 0.0045. Report mean ± SD with 2 sig figs for SD.
3. A sample of 30 measurements yields mean = 45.6789 and SD = 0.04567. Report the result according to GUM (2 sig figs for SD).

Answers: (1) Mean = 12.2, SD = 0.158 → 0.2, report 12.2 ± 0.2. (2) Mean = 0.00454, SD = 0.000114 → 0.00011 (2 sig figs), mean = 0.00454, report 0.00454 ± 0.00011. (3) SD = 0.04567 → 0.046 (2 sig figs), mean = 45.679, report 45.679 ± 0.046.

Quick Reference Table

FAQ

Why can’t I report the mean with more digits than the SD?

The SD represents the uncertainty in the measurement. Reporting more digits in the mean implies a precision that is not supported by the data. The last digit of the mean should be in the same decimal place as the SD to reflect that uncertainty.

Should I always use two significant figures for the SD?

Not always. For small samples (n < 10), one significant figure is often sufficient because the SD itself is imprecise. For larger samples, two significant figures are recommended to preserve information. The GUM suggests using two significant figures for uncertainty, but allows one when the leading digit is 3 or more.

What if the SD is zero?

If the SD is zero (all measurements identical), it usually indicates a lack of variation, but this is rare in real data. Report the mean with the same precision as the raw data (e.g., if data are to 0.1, report mean to 0.1).

How do I handle rounding when the SD has a leading digit of 1?

When the SD begins with 1, rounding to one significant figure can lose too much information (e.g., 0.15 → 0.2, which is a 33% change). In such cases, use two significant figures (e.g., 0.15) to reduce rounding error. This is recommended in the GUM and NIST guidelines.

Sources & Further Reading

• NIST Technical Note 1297, “Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results,” 1994.
• JCGM 100:2008, “Evaluation of Measurement Data – Guide to the Expression of Uncertainty in Measurement (GUM).”
• ASTM E29-13, “Standard Practice for Using Significant Digits in Test Data to Determine Conformance with Specifications.”
• ISO 80000-1:2009, “Quantities and units – Part 1: General.”
• Altman, D.G., & Bland, J.M. (2005). “Statistics notes: How to obtain the P value from a confidence interval.” BMJ, 330(7495), 373. (For related rounding practices.)

For more detailed guidance on significant figures and rounding, explore our significant figures rules and rounding methods articles.

“,
“categories”: [
“Standard Deviation Reporting”,
“Significant Figures”,
“Rounding Rules”,
“Measurement Uncertainty”,
“Physics”,
“Chemistry”,
“Engineering”
],
“tags”: [
“mean”,
“standard deviation”,
“significant figures”,
“rounding”,
“uncertainty”,
“GUM”,
“NIST”,
“ASTM E29”,
“ISO 80000”,
“precision”
],
“image_prompt”: “An educational infographic showing a set of measurements, a calculator, and a table with a mean and standard deviation correctly rounded. The image includes a ruler, a bell curve, and a magnifying glass highlighting the decimal places. Clean, modern design with blue and gray tones, text labels like ‘Mean ± SD’ and ‘Correct Precision’.”,
“quick_facts”: [
{
“label”: “SD Significant Figures”,
“value”: “1 or 2, depending on sample size and leading digit”
},
{
“label”: “Mean Rounding”,
“value”: “Same decimal place as the rounded SD”
},
{
“label”: “Small Sample (n<10)",
"value": "Use 1 significant figure for SD"
},
{
"label": "Large Sample (n≥10)",
"value": "Use 2 significant figures for SD"
},
{
"label": "Leading Digit 1 or 2",
"value": "Prefer 2 significant figures to avoid rounding loss"
},
{
"label": "GUM Recommendation",
"value": "Uncertainty to 2 significant figures, result to same decimal place"
},
{
"label": "Common Error",
"value": "Reporting mean with more decimal places than SD"
},
{
"label": "ASTM E29",
"value": "SD to 1-2 sig figs, mean to same decimal place"
}
],
"related_terms": [
{
"term": "Significant Figures",
"definition": "The digits in a number that carry meaning contributing to its measurement precision, including all certain digits plus the first uncertain digit."
},
{
"term": "Standard Deviation (SD)",
"definition": "A measure of the dispersion or variability in a set of data, indicating how much individual measurements differ from the mean."
},
{
"term": "Measurement Uncertainty",
"definition": "A parameter, associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be attributed to the measurand."
},
{
"term": "GUM",
"definition": "Guide to the Expression of Uncertainty in Measurement, an international standard (JCGM 100:2008) for evaluating and expressing measurement uncertainty."
}
],
"references": [
"NIST Technical Note 1297 (1994), Section 7.5",
"JCGM 100:2008 (GUM), Section 7.2.6",
"ASTM E29-13, Section 6.4",
"ISO 80000-1:2009, Section 7.3.4",
"Altman, D.G., & Bland, J.M. (2005). Statistics notes. BMJ."
],
"faq": [
{
"question": "Why can't I report the mean with more digits than the SD?",
"answer": "The SD represents the uncertainty in the measurement. Reporting more digits in the mean implies a precision that is not supported by the data. The last digit of the mean should be in the same decimal place as the SD to reflect that uncertainty."
},
{
"question": "Should I always use two significant figures for the SD?",
"answer": "Not always. For small samples (n < 10), one significant figure is often sufficient because the SD itself is imprecise. For larger samples, two significant figures are recommended to preserve information. The GUM suggests using two significant figures for uncertainty, but allows one when the leading digit is 3 or more."
},
{
"question": "What if the SD is zero?",
"answer": "If the SD is zero (all measurements identical), it usually indicates a lack of variation, but this is rare in real data. Report the mean with the same precision as the raw data (e.g., if data are to 0.1, report mean to 0.1)."
},
{
"question": "How do I handle rounding when the SD has a leading digit of 1?",
"answer": "When the SD begins with 1, rounding to one significant figure can lose too much information (e.g., 0.15 → 0.2, which is a 33% change). In such cases, use two significant figures (e.g., 0.15) to reduce rounding error. This is recommended in the GUM and NIST guidelines."
}
],
"related_articles": [
"Significant Figures: Rules and Examples",
"Rounding Methods: Half-Up, Half-Down, and Banker's Rounding",
"Measurement Uncertainty: A Practical Guide",
"How to Use a Significant Figures Calculator Effectively"
]
}