Guides / descriptive statistics spss
Descriptive statistics in SPSS
SPSS offers three different dialogs that all produce descriptive statistics, and they are not interchangeable. Picking the wrong one is why people end up computing a mean for a variable that needed a median, or hunting for a frequency table the dialog cannot produce.
Go to Analyze > Descriptive Statistics > Descriptives, move your variable into the Variable(s) box and click OK. The default output includes the mean, standard deviation, minimum and maximum. For the median, use Frequencies instead and tick Median under Statistics, since the Descriptives dialog does not offer it.
Orientation
Which dialog to use
Three dialogs, three purposes. This is the source of most of the confusion.
| Dialog | Gives you | Use it when |
|---|---|---|
| Descriptives | Mean, SD, min, max, range | You want summary numbers for continuous variables. No median |
| Frequencies | Everything, plus frequency tables | You want the median, quartiles, mode, or a count per category |
| Explore | Everything, plus plots and normality tests | You want descriptives split by group, or you are checking assumptions |
This catches almost everyone once. Analyze → Descriptive Statistics → Descriptives has no median option at all. Use Frequencies and tick Median under Statistics.
Descriptives
Mean and standard deviation
The simplest dialog in SPSS.
Several at once is fine — each gets a row in the output.
Mean, standard deviation, minimum and maximum are on by default. Add Skewness and Kurtosis if you want distribution shape.
The output is one compact table.
SPSS will compute a mean for anything numeric, including a variable where 1 = female and 2 = male. The result is arithmetically valid and completely meaningless. Check the measure level in Variable View before running descriptives.
Frequencies
Median, quartiles and mode
This is the dialog that has the median.
For a continuous variable with many values, untick Display frequency tables or you will get hundreds of rows.
Quartiles gives you the 25th, 50th and 75th percentiles — the 50th is the median, and the gap between the 25th and 75th is the interquartile range.
SPSS reports quartiles as percentiles rather than labelling an IQR. Compute it yourself: IQR = 75th percentile − 25th percentile.
Categorical data
Frequency tables
For categorical variables the frequency table is the descriptive statistic. Analyze → Descriptive Statistics → Frequencies with Display frequency tables ticked.
| Column | What it is |
|---|---|
| Frequency | Raw count in that category |
| Percent | Percentage of all cases, including missing |
| Valid Percent | Percentage of non-missing cases — usually the one to report |
| Cumulative Percent | Running total, meaningful only for ordered categories |
Percent and Valid Percent differ whenever data are missing. Valid Percent is almost always the figure you want, but the difference between them tells the reader how much was missing — so report the n alongside it.
Split output
Descriptives by group
Two ways, and the second is usually better.
Outcome into Dependent List, grouping variable into Factor List. You get full descriptives per group plus boxplots.
Applies to every subsequent analysis until you switch it off — which is easy to forget and produces confusing output later.
It stays active across analyses and there is no obvious indicator once you have moved on. If your output starts arriving in unexpected sections, check whether Split File is still on.
Which to report
Mean or median?
| Report the mean when | Report the median when |
|---|---|
| The variable is continuous and roughly symmetric | The distribution is skewed |
| You will run parametric tests on it | The variable is ordinal — a Likert item, a ranking |
| Outliers are absent or genuinely representative | Outliers are present and would distort the mean |
| Reporting alongside a t-test or ANOVA | Reporting alongside Mann-Whitney or Kruskal-Wallis |
Choosing a rank-based test because the data are skewed and then reporting group means contradicts your own reasoning, and reviewers notice. If you ran Mann-Whitney, report medians and interquartile ranges.
Send your variables and design. A named statistician confirms the right summaries and tests, and returns annotated output with APA tables.
Get a fixed quoteAPA style
How to report it
| Situation | How to write it |
|---|---|
| Continuous, symmetric | M = 24.8, SD = 9.6 |
| Skewed or ordinal | Mdn = 24, IQR 15–31 |
| Categorical | 62.4% (n = 143) agreed |
| With sample size | M = 24.8 (SD = 9.6), n = 167 |
Answers
Frequently asked questions
How do I calculate the mean in SPSS?
Analyze > Descriptive Statistics > Descriptives, move the variable into Variable(s) and click OK. The mean, standard deviation, minimum and maximum appear by default.
How do I find the median in SPSS?
Use Frequencies, not Descriptives — the Descriptives dialog has no median option. Analyze > Descriptive Statistics > Frequencies, then Statistics > tick Median. For a continuous variable, untick 'Display frequency tables' first.
What is the difference between Percent and Valid Percent in SPSS?
Percent is calculated over all cases including missing values; Valid Percent excludes missing. Valid Percent is usually the figure to report, and the difference between the two tells the reader how much data was missing.
How do I get descriptive statistics by group in SPSS?
Analyze > Descriptive Statistics > Explore, with the outcome in Dependent List and the grouping variable in Factor List. This is generally better than Data > Split File, which stays active across later analyses and is easy to forget about.
Should I report the mean or the median?
The mean for continuous, roughly symmetric variables analysed with parametric tests. The median with an interquartile range for skewed or ordinal variables, and whenever you have used a rank-based test such as Mann-Whitney or Kruskal-Wallis.
How do I get the interquartile range in SPSS?
Frequencies > Statistics > tick Quartiles. SPSS reports the 25th, 50th and 75th percentiles rather than labelling an IQR, so subtract the 25th from the 75th yourself.