Skip to content
Tutorials Center
My learning

Topics

Lesson 15 of 39 ยท Descriptive Statistics

SPSS Frequencies: Tables, Percentiles, Mode, Median and Charts

Master Analyze > Descriptive Statistics > Frequencies: tables, percent versus valid percent, quartiles and percentiles, mode and median, histograms and bar charts, format options and syntax.

Frequencies is the first procedure you run on almost every variable. It counts values, shows percentages, and can compute a whole menu of summary statistics and charts. For categorical variables it is your main descriptive tool. For numeric variables it shows the shape through percentiles and a histogram. This lesson goes beyond the quick tour in the first session.

Open the dialog

Analyze > Descriptive Statistics > Frequencies.... Move one or more variables into Variable(s). The checkbox Display frequency tables is on by default; turn it off for continuous variables with many values, where you only want the statistics and chart. Three buttons hold the options: Statistics..., Charts... and Format....

Statistics

Group Choices
Percentile Values Quartiles; Cut points for n equal groups; Percentile(s) you type and Add
Central Tendency Mean, Median, Mode, Sum
Dispersion Std. deviation, Variance, Range, Minimum, Maximum, S.E. mean
Distribution Skewness, Kurtosis
Values are group midpoints Treat the data as grouped when the values are interval midpoints

Charts

Choose Bar charts, Pie charts or Histograms (optionally With normal curve). For bar and pie charts, Chart Values can be Frequencies or Percentages. Use bar charts for categories and ordinal scales, histograms for scale variables, and pie charts rarely: they are hard to compare and cannot show more than a few slices well.

Format

  • Order by: ascending or descending values (the default shows categories in code order) or ascending/descending counts (to rank categories by size).
  • Multiple Variables: Compare variables (one table per variable) or Organize output by variables (one summary table).
  • Suppress tables with more than n categories (default 10).

The frequency table

The table has the columns Frequency, Percent, Valid Percent and Cumulative Percent. Valid Percent excludes missing values; Cumulative Percent is the running total of Valid Percent. Cumulative Percent is meaningful only for ordinal and scale variables, where the order has a meaning.

For a numeric variable that is stored with value labels, the table lists the labels in order of the codes. If a category has zero cases it will not appear at all; to show empty categories you need to build the table another way (for example with Custom Tables).

Percentiles and quartiles

A percentile is a value below which a given percent of the cases fall. The 50th percentile is the median, the 25th and 75th are the quartiles (Q1, Q3), and Q3 - Q1 is the interquartile range (IQR), a robust measure of spread.

SPSS computes percentiles with a definition called HAVERAGE (weighted average). For N valid cases sorted from smallest to largest and a percentile p (a fraction between 0 and 1):

  1. compute the position (N + 1) * p;
  2. if the position is a whole number, take that case's value;
  3. otherwise interpolate linearly between the two neighbouring cases;
  4. positions below 1 give the minimum, and above N give the maximum.

Python's statistics.quantiles(..., method='exclusive') uses the same rule, and so does numpy.percentile(..., method='weibull'). Other programs (Excel's PERCENTILE.INC) use (N - 1) * p + 1, so results can differ slightly. That is not an error; the definitions are different.

Python Stand-in: percentiles the way Frequencies computes them

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

Program input ยท 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.

The "exclusive" quartiles match SPSS; the "inclusive" ones are the Excel-style values and differ. Remember this when you compare SPSS to a spreadsheet.

Mode, median and mean

  • The mode is the most frequent value. If several values tie, SPSS shows the smallest and adds a footnote that multiple modes exist.
  • The median is the middle value (the 50th percentile). It resists outliers.
  • The mean is the average and is pulled toward extreme values.

For a right-skewed variable such as income, the mean is greater than the median. Comparing them is a quick check of symmetry. SPSS also reports skewness and kurtosis (see descriptives and z-scores).

Python Stand-in: mode with ties, median and mean

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

Program input ยท 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.

Syntax

TEXT
FREQUENCIES VARIABLES=age income
  /FORMAT=NOTABLE
  /NTILES=4
  /PERCENTILES=10 90
  /STATISTICS=MEAN MEDIAN MODE STDDEV MINIMUM MAXIMUM SKEWNESS KURTOSIS
  /HISTOGRAM NORMAL
  /ORDER=ANALYSIS.

FREQUENCIES VARIABLES=region
  /BARCHART PERCENT
  /FORMAT=DFREQ.

/FORMAT=NOTABLE suppresses the table; /NTILES=4 requests quartiles; /FORMAT=DFREQ orders categories by descending frequency. The Statistics table in the output lists the requested statistics, one column per variable, along with N Valid and N Missing.

Multiple responses and other cases

For a question where respondents could tick several options, ordinary Frequencies counts each tick-box variable separately. Use Analyze > Multiple Response > Define Variable Sets and Frequencies there to get percentages of respondents and of responses. For long lists of numeric values, bin the variable first (see recode and binning). For a weighted table, use Weight Cases first (see select, split and weight).

Reading the output wisely

  • Check N Valid against your expectations.
  • Look at the percent missing; large amounts of missing data bias results.
  • For categorical variables, check that every code has a label.
  • Read the histogram: Is it symmetric? Are there gaps, spikes or clear outliers? Do not trust the mean of a multimodal variable.
  • Report Valid Percent when missing values exist, and say how many were missing.
Python Stand-in: Valid Percent and the missing row

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

Program input ยท 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.

Common mistakes

  • Reading Percent when Valid Percent is the relevant column.
  • Asking for a frequency table of a continuous variable with 500 distinct values.
  • Using the mean and standard deviation for a clearly skewed variable.
  • Comparing SPSS quartiles with Excel's and thinking one is wrong.
  • Using a pie chart for 12 categories.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. If SPSS finds several modes, which value does the Frequencies Statistics table show?

    SPSS reports the smallest of the tied modes and adds a footnote.
  2. 2. Which position does the default SPSS percentile definition use for N = 9 and p = 50%?

    The HAVERAGE definition uses the position (N+1)p, here 10 * 0.5 = 5, the middle value.
  3. 3. Which chart is the most appropriate for a continuous variable such as age?

    A histogram groups values into intervals and shows the shape of the distribution.
  4. 4. For the table Format option, which setting ranks categories from most to least common?

    Order by Descending counts sorts by frequency, largest first.

Summary

  • Frequencies gives tables, percentiles, central tendency, spread, shape statistics and charts.
  • Percent includes missing cases; Valid Percent and Cumulative Percent do not.
  • SPSS percentiles use position (N+1)p; spreadsheets often use a different rule.
  • Mode, median and mean answer different questions; compare them to judge skewness.
  • Turn off the table for continuous variables and read the histogram.

Next: Descriptives and Z-Scores