Standard deviation vs standard error is a distinction nursing students need when reading a research table, describing participants or writing quantitative findings. Both statistics can sit beside a mean. However, they answer different questions. Standard deviation describes variation between observations. Standard error describes the estimated precision of a statistic, such as a sample mean.
Confusing them can make a diverse group of patients appear much more alike than it is. It can also turn a statement about an average into an unsupported claim about individual people. This guide explains the difference through hypothetical nursing examples, transparent calculations and practical reporting decisions.
The focus is choosing and interpreting SD and SE. For a broader analysis workflow, use our SPSS nursing dissertation data analysis guide. All numerical examples below are invented for teaching. They are not results from patients or published studies.
Standard deviation vs standard error: the quick comparison
Ask what the number is supposed to describe before deciding which statistic belongs in a table. If your question concerns differences among people, measurements or observations, SD is usually relevant. If your question concerns uncertainty in an estimated average, SE may be relevant.
| Question | Standard deviation | Standard error of the mean |
|---|---|---|
| What does it describe? | Spread of observed values around their mean | Estimated sampling variability of the mean |
| Typical reporting purpose | Describe the sample | Describe precision of the estimated mean |
| Units | Same units as the measured variable | Same units as the mean |
| Effect of increasing sample size | Does not automatically become smaller | Usually becomes smaller if other conditions remain comparable |
| Common mistake | Treating it as uncertainty in the mean | Treating it as variation between individuals |
For example, a mean waiting time of 30 minutes with SD 12 minutes describes observed variation in waiting times. A mean of 30 minutes with SE 2 minutes describes precision of the estimated average. Those two statements are compatible with a sample of 36 independent observations.
Neither statistic, by itself, establishes good care, poor care or an effective intervention. Interpretation still depends on the outcome, measurement process and study design.
What standard deviation tells a nursing student
Standard deviation summarises how spread out numerical observations are around their mean. A small SD indicates relatively concentrated values. A larger SD indicates greater spread, although the size must be judged against the measurement scale.
Consider two hypothetical groups with the same mean knowledge score of 70 out of 100. Group A has SD 4. Group B has SD 18. The averages match, but the second group has much more varied scores. Teaching needs may therefore differ even before you compare average performance.
OpenStax explains SD as a measure of data spread and distinguishes sample and population calculations (OpenStax, n.d.). In a dissertation, most datasets represent a sample, so the sample SD is commonly appropriate. Check what the software reports rather than switching formulas without a reason.
SD is expressed in the original units. If the variable is measured in minutes, SD is also in minutes. Variance, by comparison, uses squared units. This is one reason SD is easier to explain to a reader.
However, SD does not tell you the minimum or maximum. It does not identify which patient waited longest. A summary table therefore needs enough context to show what was measured, who contributed data and whether unusual values require further examination.
What standard error tells a nursing student
Standard error concerns a statistic rather than the individual observations themselves. Imagine repeating the same sampling process many times. Each sample would produce an average. Those averages would vary. The standard error of the mean estimates that sampling variability.
You do not have to collect all those repeated samples to calculate an estimated SE. For a simple sample of independent observations, the sample SD and sample size provide the usual estimate. The important point is what that estimate means.
Suppose one nursing student recruits 25 eligible participants and another recruits 100 under comparable conditions. Their estimated means might have different precision even if their participant scores show a similar spread. A larger sample can improve precision without making the participants more similar.
Penn State’s statistical teaching materials distinguish a sampling distribution from the distribution of individual values and identify SE as the spread of the sampling distribution (Penn State, n.d.). Keep that distinction in mind when reading an output table.
SE does not measure questionnaire validity, representativeness or freedom from bias. A precise average from a poorly selected sample can still be a poor estimate for the intended population. Therefore, do not use a small SE as a substitute for discussing recruitment and measurement limitations.
Calculate SD with a small worked example
Use five hypothetical waiting times: 10, 12, 14, 16 and 18 minutes. Their mean is 14 minutes. Subtract that mean from each value to obtain deviations of −4, −2, 0, 2 and 4.
Next, square the deviations: 16, 4, 0, 4 and 16. Their sum is 40. For the sample variance, divide by n − 1. Here, n is 5, so the divisor is 4. The sample variance is therefore 10 minutes squared.
Finally, take the square root of 10. The sample SD is approximately 3.16 minutes. The calculation shows why SD reflects differences between the observations and their mean.
Sample SD = square root of [sum of squared deviations ÷ (n − 1)].
Do not divide by n simply because there are five values. Dividing by n is the corresponding population-variance calculation when the observations constitute the entire population of interest. Specify the intended population before making that choice.
For real dissertation data, software reduces arithmetic mistakes. Nevertheless, a short hand calculation helps you recognise impossible output. For example, a negative SD is not a valid result. It should trigger a check of the copied value, column heading or calculation.
Calculate SE without confusing the two statistics
For the same five independent waiting-time observations, calculate the estimated standard error of the mean by dividing SD by the square root of n.
SE of the mean = SD ÷ √n.
Using the previous example, SE = 3.16 ÷ √5, which is approximately 1.41 minutes. SD and SE have the same units here, but their purposes remain different.
A second hypothetical example makes the sample-size effect clearer. Suppose a sample has SD 12 minutes. With n = 36 independent observations, SE is 12 ÷ 6 = 2 minutes. With n = 144 and the same SD, SE is 12 ÷ 12 = 1 minute.
Quadrupling the sample size halves SE in this simplified comparison. It does not halve the variation between people. Both samples still have SD 12 minutes.
The formula is not a universal instruction for every dataset. Repeated observations from the same patient, clustering within wards, survey weights and other design features can require different SE estimates. Match the calculation to the analysis plan. Do not treat every spreadsheet row as an independent participant.
For planning decisions, consult our separate nursing dissertation sample-size guide rather than increasing recruitment until a preferred number appears.
Choose SD or SE according to the reporting purpose
Start by completing this sentence: “I want this table to show…” If the answer is “the characteristics and variability of the participants”, use a descriptive summary that suits the variable. For a reasonably interpretable mean, that commonly includes SD and the valid sample size.
If the answer is “how precisely we estimated the population mean”, report the estimate with an appropriate uncertainty measure. A confidence interval often communicates that purpose more directly than SE alone. Our confidence intervals guide covers the interpretation in more detail.
A methods chapter should explain why each summary is appropriate. Avoid writing that SD was selected because SPSS displayed it first. Your choice should follow the nature of the data and the question.
For categorical variables, frequencies and percentages are generally more meaningful than a mean of arbitrary category codes. For strongly skewed continuous data, a median and interquartile range may describe the distribution more clearly.
This does not mean an SD cannot be calculated for skewed data. It means a mathematically available summary may still leave the reader with an incomplete picture. Inspect the distribution and choose a summary that helps answer the nursing question.
Standard deviation vs standard error in results tables
Column headings should remove ambiguity. Write “Mean”, “SD” and “n”, or “Mean” and “SE”, rather than using an unexplained plus-or-minus symbol. Define abbreviations in a table note if the assessment style requires it.
| Illustrative variable | Valid n | Mean | SD | SE |
|---|---|---|---|---|
| Waiting time, minutes | 36 | 30.0 | 12.0 | 2.0 |
| Knowledge score, points | 100 | 70.0 | 15.0 | 1.5 |
This teaching table includes both statistics to show the distinction. Your dissertation does not automatically need both columns. Include only the measures that support the reporting purpose and your brief.
Always use the valid n for the relevant variable. A project may recruit 100 people but have only 81 complete observations for one measurement. Using the recruitment total in the SE formula would give the wrong result.
Apply consistent decimal places within comparable columns. Keep unrounded values during calculations, then round for presentation. Precision should reflect the measurement, rather than suggesting that a questionnaire score is known to an unrealistic number of decimal places.
For table selection and chapter organisation, see our SPSS results chapter guide.
Write clear sentences about variability and precision
A descriptive sentence could read: “Among 36 participants, mean waiting time was 30.0 minutes (SD 12.0).” This states the sample size, average, units and spread. It does not claim that everyone waited around 30 minutes.
A precision-focused sentence could read: “The estimated mean waiting time was 30.0 minutes, with a standard error of 2.0 minutes.” If you report a confidence interval instead, ensure it comes from the correct method and valid dataset.
An inaccurate sentence would be: “Waiting times varied by only two minutes because the standard error was two.” That uses uncertainty in the mean to describe variation among people. In the example, the relevant spread statistic is SD 12 minutes.
Another inaccurate statement is: “The small standard error proves the sample represents all patients.” Representativeness depends on the sampling process, eligibility criteria, response and setting. SE does not prove those conditions.
Try explaining the sentence without abbreviations before placing it in your dissertation. If the explanation refers to individual scores, check whether you have used SD. If it refers to an estimated average, check whether you have used an appropriate precision measure.
Read error bars before interpreting a figure
Error bars can represent SD, SE or confidence intervals. Their appearance alone does not identify the statistic. Read the caption and methods before deciding what a narrow or wide bar means.
Suppose two graphs show the same sample mean. One uses SD bars and the other uses SE bars. The SE bars will usually be narrower for n greater than one. That change in appearance does not mean the underlying patients are less diverse.
Likewise, do not declare a statistically significant group difference solely because two error bars appear separate. The meaning of overlap depends on what the bars represent, the design and the statistical comparison.
When preparing your own figure, label the error bars explicitly. Include units on the outcome axis, identify the groups and explain whether n refers to patients, measurements or another unit. A good caption should allow the reader to understand the graphic without guessing.
If a published paper leaves the error-bar definition unclear, record that limitation rather than supplying an assumption. In a critical appraisal, ambiguity in reporting is a finding about the clarity of the evidence.
Check assumptions and avoid common calculation errors
First, check independence. Measurements collected repeatedly from the same patient are related. Entering them as unrelated observations can overstate precision. The appropriate analysis may estimate uncertainty at the participant level or explicitly model repeated measurements.
Second, distinguish SD of an outcome from SD of a change score. In a pre–post study, the spread of individual changes is a separate quantity. It cannot generally be obtained by subtracting the two outcome SDs.
Third, check missing values and filters. If software analyses only a subgroup, its valid n may differ from the number of visible rows. Document any exclusions and confirm the sample-size column before calculating SE.
Fourth, avoid treating a smaller SD as inherently better. A narrower range of pain scores could reflect a more homogeneous group, a limited scale or restricted recruitment. The clinical interpretation comes from the context.
Finally, do not change statistics because one creates a more persuasive-looking graph. Select the reporting approach before seeing which presentation appears favourable. Keep an analysis record so that a supervisor can trace the final numbers to the approved dataset.
A practical editing checklist for your dissertation
Review each table and sentence using the following sequence. It takes less effort than correcting several chapters after the same confusion has spread through the work.
- Identify the variable and its units.
- Confirm whether the summary describes observations or an estimate.
- Check the valid sample size for that variable.
- Confirm the correct SD or SE column in the software output.
- Review independence, clustering and repeated measurements.
- Label table columns and error bars accurately.
- Remove claims that a small SE proves validity or representativeness.
- Keep rounding consistent and preserve the original output.
Imagine a supervisor asks, “What does 2.0 mean in this table?” A strong answer names the statistic, states its purpose and explains how the number was obtained. An answer such as “SPSS produced it” leaves the central methodological decision unexplained.
Also compare the abstract, results and discussion. They should describe the same statistic consistently. A table labelled SD should not become SE when the finding is paraphrased in another chapter.
Frequently asked questions
Is standard error always smaller than standard deviation?
For the simple formula SE = SD ÷ √n, SE is smaller when n is greater than one and SD is positive. However, that relationship does not justify substituting SE for SD. More complex designs also need the appropriate design-based calculation.
Should I report mean plus or minus SD?
You can, if your style guide allows it and you clearly define the notation. Separate Mean and SD columns are often easier to read. The main requirement is that the statistic and units are unambiguous.
Does a large sample make standard deviation smaller?
Not automatically. A larger sample can estimate the population spread more reliably, but the population itself does not become less variable because more people are measured.
Can I calculate standard error from standard deviation?
Yes, for a simple independent sample mean, divide SD by √n. Use the valid n for the outcome. For clustered, weighted or repeated-measure data, obtain the SE through the appropriate analysis.
Does standard deviation prove normality?
No. Many distributions have an SD. Assess distributional shape separately where relevant. Our SPSS normality guide explains that assessment.
Conclusion
The central distinction in standard deviation vs standard error is the reporting purpose. SD describes spread among observations. SE describes estimated uncertainty in a statistic. Clear labels, a valid sample size and a design-appropriate calculation help you communicate both accurately.
Get feedback on your statistical reporting
Unsure whether your table needs SD, SE or a confidence interval? Request focused statistical-reporting feedback with your assessment brief and permitted output. Support should help you understand and defend your own analysis, without changing findings to fit an expected result.
References
- OpenStax (n.d.) ‘2.7 Measures of the spread of the data’, Statistics. Available at: OpenStax statistics textbook (Accessed: 2 October 2026).
- Penn State (n.d.) ‘4 Sampling distributions’, STAT 500: Applied Statistics. Available at: Penn State STAT 500 (Accessed: 2 October 2026).
- Altman, D.G. and Bland, J.M. (2005) ‘Standard deviations and standard errors’, BMJ, 331(7521), p. 903. doi:10.1136/bmj.331.7521.903.
