Skip to content
Tutorials Center
My learning

Topics

Lesson 19 of 39 Β· Descriptive Statistics

SPSS Means Procedure, Case Summaries and Custom Tables

Summarize numeric variables by groups in SPSS with Analyze > Compare Means > Means, layers, ANOVA table and eta, Case Summaries, report tables and Custom Tables (CTABLES), with syntax.

Often the first interesting question about numeric data is "How does it differ between groups?" Average income by region, mean score by teaching method, median waiting time by clinic. The Means procedure answers it in a compact table, and a few other tools (Case Summaries, Report Summaries, Custom Tables) produce publication-style layouts. This lesson closes the descriptive-statistics chapter; the formal tests that follow in Chapter 5 build directly on these group summaries.

The Means procedure

Analyze > Compare Means > Means...

  • Dependent List: the numeric variables to summarize.
  • Independent List: the grouping variables (categories).
  • Layer: click Next to add a second layer. Layers nest: the first independent variable defines the main groups and the second splits each of them, producing one table with combined groups.

By default the Report table shows Mean, N and Std. Deviation for each group and a Total row.

Options...

Choose cell statistics from a long list: Mean, Number of Cases, Standard Deviation, Sum, Median, Grouped Median, Std. Error of Mean, Minimum, Maximum, Range, First, Last, Variance, Kurtosis, Skewness, Harmonic Mean, Geometric Mean, Percent of Total Sum, Percent of Total N. Under Statistics for First Layer, tick Anova table and eta to test whether the group means differ, and Test for linearity for ordered groups.

The syntax:

TEXT
MEANS TABLES=income BY region
  /CELLS=MEAN COUNT STDDEV MEDIAN
  /STATISTICS=ANOVA.

MEANS TABLES=income BY region BY gender
  /CELLS=MEAN COUNT.

With BY region BY gender, the second BY creates the second layer (gender within region).

Python Stand-in: a Means table for income by region

Press Run (or Ctrl + Enter) to see the result here.

Program input Β· what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

ANOVA table and eta

When you tick Anova table and eta, SPSS prints a one-way ANOVA table (Between Groups, Within Groups, Total) with an F test and a Measures of Association table with Eta and Eta Squared. Eta squared is the share of the variance in the dependent variable explained by the grouping variable: 0.01 small, 0.06 medium, 0.14 large (a common guideline). It is a quick effect size, and the full procedure with assumptions and post hoc tests is in Chapter 6 on ANOVA.

Python Stand-in: eta squared from sums of squares

Press Run (or Ctrl + Enter) to see the result here.

Program input Β· what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Descriptives per group by other means

You can get group summaries in more than one way. Choose by the layout you need.

Tool Best for
Means A compact table of statistics for one or more variables by one or more groups
Explore with a Factor List Detailed distribution checks per group, with plots
Split File + Descriptives/Frequencies Full output for each group separately
Case Summaries Listing cases with grouped subtotals
Report Summaries in Rows/Columns Hand-formatted reports with titles and summary lines
Custom Tables Publication-quality tables with several variables in rows and columns
Aggregate A new data file holding the group statistics

Case Summaries and Reports

Analyze > Reports > Case Summaries... lists values for each case (or just group summaries) and can show subtotals by a grouping variable. Choose Limit cases to first 100 to avoid huge output, untick Display cases for summaries only, and use Statistics... to pick the subtotals. It is the easy way to look at the cases behind a number: for example, list the ids and incomes of the flagged outliers from cleaning and validating data.

Analyze > Reports > Report Summaries in Rows / in Columns produce text-style reports with page titles, footers and break variables. They are older tools but useful for a formatted listing.

TEXT
SUMMARIZE
  /TABLES=id income region
  /FORMAT=VALIDLIST NOCASENUM TOTAL LIMIT=20
  /TITLE='Largest incomes'
  /CELLS=COUNT MEAN.

Custom Tables

If Custom Tables is licensed (Analyze > Tables > Custom Tables...), you drag variables onto the canvas: categorical variables as rows and columns, scale variables as summaries, and choose statistics like Count, Column N %, Mean or Median in each cell. Features: stacking variables, nesting, totals and subtotals, significance tests of column proportions or means, and excellent formatting. The command is CTABLES:

TEXT
CTABLES
  /TABLE gender [C][COUNT F40.0, COLPCT.COUNT PCT40.1] BY plan
  /CATEGORIES VARIABLES=gender plan ORDER=A KEY=VALUE EMPTY=INCLUDE.

Because Custom Tables can show empty categories and combine many variables in one table, it is the answer when Frequencies and Crosstabs are not flexible enough.

Making group summaries tell the truth

  • Show N in every group. A mean of 3 cases deserves less trust than a mean of 300.
  • Report spread (SD, IQR) and not just the mean, because two groups can share a mean and differ wildly.
  • Use the median for skewed variables such as income or waiting time.
  • Order groups meaningfully (by size, by mean, or by natural order) and label them clearly.
  • Plot the groups too: boxplots or error bars often show more than a table (see the charts chapter).
  • Remember that differences in a sample table are descriptive. To claim a difference in the population you need a test (Chapter 5) and a confidence interval.
Python Stand-in: two groups with the same mean but different spread

Press Run (or Ctrl + Enter) to see the result here.

Program input Β· what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Both teams average 50, yet their experience is entirely different. A means-only table would hide that.

Common mistakes

  • Averaging a variable with undeclared missing codes in every group.
  • Reading differences between group means as significant without a test.
  • Using Means for a categorical dependent variable; use Crosstabs.
  • Forgetting that the layers in Means are nested, not side by side.
  • Ignoring groups with very small N.

Check your understanding

Quick check Β· 4 questions

of 4 answered You scored / 4 Β· perfect!

  1. 1. Where do you find the Means procedure?

    Means lives under Analyze > Compare Means, next to the t tests and One-Way ANOVA.
  2. 2. What does eta squared tell you in the Means output?

    Eta squared is an effect size: between-group variance divided by total variance.
  3. 3. Which tool lets you list the cases behind each group with subtotals?

    Analyze > Reports > Case Summaries lists values for cases and can show grouped statistics.
  4. 4. Two groups have the same mean. What else should a good summary show?

    Group sizes and variability reveal differences that equal means can hide.

Summary

  • Means summarizes numeric variables by the categories of one or more grouping variables, with optional ANOVA table and eta.
  • Layers nest groups; choose cell statistics such as median, SE, minimum and maximum.
  • Case Summaries, Report Summaries and Custom Tables (CTABLES) build listings and publication-style tables.
  • Always show N and spread, and use the median for skewed data.
  • Descriptive group differences need a test before you can generalize them.

Next: Hypothesis Testing and p Values in SPSS