Reporting Descriptive Statistics in SPSS
A Participant Characteristics Table
Reporting descriptive statistics in SPSS ends in two products, a participant characteristics table and a paragraph that describes the sample. Counts, summaries, missing cases and decimals have to agree between the table, the text and the output file.
IN SHORT
Who Was Studied, Counted in Ways a Reader Can Verify
A sample description states who or what was analyzed, in counts and summaries that a reader can check against the data. Categorical variables appear as n (%), continuous variables as a mean with its standard deviation or a median with its quartiles, according to the distribution, and every variable shows how many cases are missing. For reporting descriptive statistics in SPSS, the values come from Frequencies, Descriptives and Explore under Analyze, Descriptive Statistics, and menu labels differ slightly between versions. The finished table is assembled outside SPSS, so the task is transfer and consistency. At OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, in-house PhD-level statisticians build sample tables of this kind from the output, or check a draft table against it.
In This Guide
RELATED SUPPORT
Where the Sample Table Meets the Analysis
A variable with extensive missing values calls for a documented decision first, and the analysis that follows the sample description is its own piece of work.
Missing Data Analysis
The pattern and likely mechanism of missing values examined in your data, so the per-variable counts in the table rest on a documented decision.
Quantitative Data Analysis
The analysis that follows the sample description, planned and carried out for your design and reported with estimates and intervals.
BEFORE YOU BUILD IT
Four Choices Behind Every Row
These choices are made once for the whole table, because a reader compares rows with one another and expects a single convention throughout.
Mean or Median Choose by the shape of each variable and the way it will be used, not by a test p-value alone, and give the same summary to the same kind of variable throughout.
Denominator of the Percentages Percentages are taken from valid cases or from all cases. SPSS prints both as Percent and Valid Percent, so the table says which of the two it reports, in a note if not in a heading.
Display of Missing Data Show the number missing for each variable, in a column or in a note, so that a changing n from one row to the next is visible and not left to the reader to find.
Decimal Places Report to the precision the measurement supports. Counts of persons have no decimals, and a mean from a whole-number scale rarely needs more than two, whatever SPSS prints.
IN SPSS
Taking Each Value From Its Source Table
IBM SPSS prints the parts of the table in separate procedures, and the choice of summary by variable type is explained in Descriptive Statistics in SPSS and APA Tables.
Take Counts and Valid Percent
From the Frequencies table of each categorical variable, take the Frequency and the Valid Percent, or the Percent when missing cases are shown as a category.
Check: Add the categories and the missing row. The sum should equal the number of rows in the data file, unless a filter or weight is active, and a different total points to a coding problem.
Take N, Mean and Std. Deviation
From the Descriptive Statistics table, take N, Mean and Std. Deviation for each continuous variable that is reasonably symmetric.
Tip: Valid N (listwise) at the foot of that table counts the cases complete on every variable listed. It is not the n of any one row, so it belongs in a note and not in the table body.
Take the Median and the Quartiles
For skewed variables, request Quartiles under Statistics in Frequencies, or take the percentiles from Explore, and read the median beside them.
Watch: Programs compute percentiles by slightly different rules, so quartiles from SPSS can differ noticeably in small samples from R or Stata on the same data. Name the program in the methods.
Take the Missing Count Per Variable
Take the Valid and Missing figures from the Statistics table of Frequencies, for every variable that appears in the table.
Check: Compare each missing count with what the questionnaire or record form would lead you to expect. A much higher count points to a skip pattern or a coding error to settle before reporting.
Repeat the Run for Each Group
For a table by group, rerun the same procedures with Split File, organizing output by groups, or place the grouping variable in the Factor list of Explore.
Watch: Explore drops a case from every variable listed when listwise exclusion is left on, so the n in a table built from one run can fall below the counts that Frequencies gives.
Reconcile Every Cell With the Output
Fill the table shell below from your notes, then check each finished cell against the output file for the same n and the same rounding.
Tip: Keep a short list that ties every table cell to a named output table and run. A reviewer’s query about a figure is then answered from the source and not from memory.
READING THE OUTPUT
Where Each Descriptives Value Goes in the Table
Frequencies, Descriptives and Explore supply different cells, and the table built from them keeps one convention for each type of variable.
| Output value | Where SPSS prints it | Where it goes and what to check |
|---|---|---|
| Frequency | Frequencies table, one per categorical variable | The n of each category; the counts add up to the valid cases plus the missing |
| Percent and Valid Percent | Frequencies table | The % in n (%); Valid Percent is quoted when missing cases are shown apart, and the table says so |
| Valid and Missing | Statistics table of Frequencies | The n or missing column; a changing n down the table is shown and not hidden |
| Mean and Std. Deviation | Descriptive Statistics table of Descriptives | M (SD) of a continuous variable, with the same decimals for every row on the same scale |
| Median and Percentiles 25 and 75 | Statistics table of Frequencies, or Percentiles table of Explore | Mdn with its quartiles for a skewed variable; the percentile rule differs between programs |
| Minimum and Maximum | Descriptive Statistics table | The range where asked for, and a check for impossible values before anything is reported |
| Valid N (listwise) | Foot of the Descriptive Statistics table | Cases complete on all listed variables, given in a note and not used as the n of a row |
| Skewness and Kurtosis, with Std. Error | Descriptives, Options | Support for choosing mean or median; the values seldom appear in the sample table |
REPORTING
A Sample Paragraph and a Participant Characteristics Table
The paragraph reports the size and main features of the sample, the table carries every variable, and the two agree to the last digit.
In a descriptive statistics APA table, continuous variables appear as M (SD) or Mdn (IQR), categorical variables as n (%), and the sample size heads each column. Percentages are rounded the same way throughout, usually to one decimal or to whole numbers, and means keep the precision of the measurement. The frame below describes a sample divided into two groups, and a single-group study drops the group columns. The number missing for each variable sits in the table or in its note, so a reader can see which rows rest on fewer cases than the total.
The sample comprised ___ [participants/cases], of whom ___ (___%) were [category] and ___ (___%) were [category]. The mean [variable] was ___ (SD = ___), and the median [skewed variable] was ___ (Q1 to Q3 = ___ to ___). Data were missing for ___ (___%) cases on [variable] and ___ (___%) on [variable]; percentages are of valid cases unless stated otherwise, and the characteristics of the two groups are given in Table 1.
| Table 1. Participant Characteristics, Overall and by [Group] | |||
|---|---|---|---|
| Characteristic | Overall (N = ___) | [Group A] (n = ___) | [Group B] (n = ___) |
| [Continuous variable], M (SD) | ___ (___) | ___ (___) | ___ (___) |
| [Skewed variable], Mdn (IQR) | ___ (___ to ___) | ___ (___ to ___) | ___ (___ to ___) |
| [Category variable], n (%) | |||
| [Category 1] | ___ (___%) | ___ (___%) | ___ (___%) |
| [Category 2] | ___ (___%) | ___ (___%) | ___ (___%) |
| Missing, n (%) on [variable] | ___ (___%) | ___ (___%) | ___ (___%) |
| Note. M = mean; SD = standard deviation; Mdn = median; IQR = interquartile range, shown as the first and third quartiles. Percentages are of valid cases. | |||
No effect size belongs in a sample description, because nothing is being inferred, so the summary itself is named in the row label or the note: M and SD, Mdn and IQR, n and %. Where the sample was weighted, the note says whether the percentages are weighted and the counts unweighted, because the two will not match.
COMMON MISTAKES
Faults a Reader Finds in a Sample Table
Most of these show up when the table is read from the top with the output file open beside it, row against row.
A Shifting n Left Unflagged A table whose n changes from row to row without comment suggests a complete sample. Each row states its own n, or a missing column accompanies the table.
Group Columns Without Their Own n A by-group table whose column headings omit each group's n leaves the reader unable to turn percentages back into counts. Each column heading carries its n, and the overall column its N.
Categories Merged Without a Note Small categories pooled into Other, for confidentiality or because counts are sparse, leave totals that cannot be matched with the instrument. The note names what was merged and why, and the merged row keeps its own n.
Precision the Scale Cannot Support A mean quoted to four decimals from a five-point item implies accuracy the scale does not hold. Decimals follow the measurement and stay fixed within a column.
Baseline Differences Tested in a Trial Significance tests of baseline characteristics in a randomized trial add little, since chance produced any difference. The table describes each arm and the text says so.
Rounding That Does Not Add Up Percentages rounded one by one may total 99.9 or 100.1. A table note saying that percentages may not sum to 100 because of rounding spares the reader a query.
Percentages that will not match the output file? Independent Statistical Review traces every cell to the output table it came from.
FREQUENTLY ASKED QUESTIONS
Open Points in Reporting Descriptive Statistics in SPSS
A sample table has to show missing data for each variable, handle select-all items and keep one rounding rule from the first row to the last.
Where do I show missing data in a characteristics table?
Show it for each variable, either as a Missing column or as a note giving n (%) of the cases lacking that value. The percentages in the body are then calculated from valid cases, and the note says so. A missing category inside the percentage column is also acceptable when its denominator is stated, but mixing the two conventions across rows is not.
How are "select all that apply" items shown in the table?
One row per option, giving the n (%) of respondents who chose it, with a note that the percentages can sum to more than 100. The denominator is the number who answered the item, not the number of ticks. Multiple Response Frequencies prints both a percent of responses and a percent of cases, and the table uses the percent of cases.
How many decimals does a characteristics table need?
Enough to match the measurement and no more. Counts have none, percentages take one decimal or whole numbers, and means and standard deviations usually take one or two decimals. One rule is applied down each column, and the rounding in the paragraph matches the table, because a reader who finds two versions of one figure doubts both.
How is a sample table prepared from my data file?
Given the cleaned file and a list of the characteristics to describe, an in-house PhD-level statistician at OnlineSPSS runs the summaries, builds the table and matches each cell to the output. A table you have already drafted is checked the same way. Results Reporting & Interpretation is the service for this stage. Which characteristics to present is your choice.
How do you protect my research data and confidentiality?
Your data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the dataset or the output you have, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Does Every Cell Trace
to the Output?
A characteristics table is accepted more readily when each cell has a named source in the output. Where a draft already exists, sending it with the output file shows which rows are still in question.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014