Conducting Univariate Analysis with SPSS

If you’ve ever collected survey data and stared at rows of numbers wondering where to even begin, you’re not alone. Before running any complex statistical tests, researchers need a clear picture of what each variable in their dataset looks like on its own. That’s exactly what univariate analysis delivers. In SPSS, it’s the essential first move – examining one variable at a time to understand its distribution, central tendency, and spread. This post walks you through the full process: generating frequency tables, calculating descriptive statistics, and – most importantly – knowing how to read and interpret what SPSS gives you.

What is univariate analysis and why does it matter? 🔗

Univariate analysis means analyzing a single variable in isolation. It doesn’t explore relationships between variables – that comes later with bivariate or multivariate analysis. Instead, it focuses on describing a variable’s basic characteristics: what values appear, how often, and how spread out those values are.

According to LibreTexts Social Data Analysis, univariate analysis is the first step in any quantitative analysis project – not just useful for understanding the data itself, but also critical preparation for all subsequent bivariate and multivariate work. In sociology, you might use it to examine the age distribution of survey respondents, the frequency of different education levels, or the average household income in a community. Without this foundation, the more complex analyses that follow lack context.

Another often-overlooked benefit: univariate analysis helps you spot data errors early. As noted in the Open Education Network’s SPSS Data Management guide, examining frequency distributions can reveal unexpected categories, miscoded values, or implausible outliers that need to be addressed before the analysis moves forward.

Matching your statistics to your variable type 🔗

Before generating any output in SPSS, you need to identify what level of measurement each variable operates at. This is not a technicality – it directly determines which statistics are appropriate and which are meaningless. The four levels are: nominal (categories with no order, e.g. religion, gender), ordinal (ordered categories, e.g. education level), and continuous (interval or ratio, e.g. age, income).

As SUNY Empire State’s SPSS eTutor warns, SPSS will calculate statistics even when they are not appropriate for the variable type. For instance, computing a mean for a nominal variable like religious affiliation produces a number, but that number is meaningless. It’s the researcher’s responsibility – not SPSS’s – to apply the right statistics for the right variable. A useful rule of thumb: nominal variables only support a mode; ordinal variables support mode and median; continuous variables support all three measures of central tendency including the mean.

Similarly, Simple Stats Tools reinforces this clearly: if you request a mean for race/ethnicity, SPSS will give you one – but interpreting it as an average race is a logical impossibility. Knowing your data is step one.

Generating frequency tables in SPSS 🔗

Frequency tables are the go-to tool for categorical variables (nominal and ordinal). They show exactly how often each value or category appears in your dataset, expressed both as a raw count and as a percentage. For categorical data on variables like gender, political affiliation, or education level, a frequency table gives you the clearest snapshot of your sample’s composition.

The navigation path in SPSS is straightforward: Analyze → Descriptive Statistics → Frequencies. According to Kent State University’s SPSS Tutorials, the Frequencies procedure produces summary measures including tallies and proportions, as well as two types of graphs suited to categorical data – bar charts and pie charts. Pie charts work well when your variable has only a few categories; for variables with many categories, a bar chart is the better choice to avoid a cluttered display.

When reading SPSS frequency table output, you will typically see four columns:

• Frequency – the raw count of how many cases fall into each category.
• Percent – the percentage of all cases (including missing values) in that category.
• Valid Percent – the percentage calculated only from valid (non-missing) responses. This is usually the more meaningful figure to report.
• Cumulative Percent – a running total of valid percentages from row to row, useful for ordinal variables.

Missing data also appears in the output. SPSS separates valid responses from missing ones, allowing you to see both how many people answered a question and how many did not. This matters: a large number of missing cases on a variable is itself an important finding that may affect how you proceed.

It’s also worth noting that the Frequencies tool is preferable to the separate Descriptives tool for most purposes, as it offers far more control and a wider range of options. Reserve the Descriptives path primarily if you specifically need Z-scores.

Calculating univariate statistics: measures of central tendency 🔗

Once you have a frequency table for categorical variables, the next layer of univariate analysis involves computing numerical summaries – particularly for continuous variables. Within the Frequencies dialog, click the Statistics button to access these options.

The three key measures of central tendency, as outlined by the University of Utah Department of Sociology, are:

• Mean – the arithmetic average of all values. Best used with continuous variables where the distribution is not heavily skewed. Sensitive to extreme outliers; a few very high or very low values can pull the mean significantly in one direction.
• Median – the middle value when all cases are ranked from lowest to highest. Appropriate for ordinal and continuous variables. Less affected by outliers than the mean, making it the better central tendency measure for skewed distributions.
• Mode – the most frequently occurring value. The only measure of central tendency appropriate for nominal variables, though it can be used across all variable types.

When mean and median are close to each other, the distribution is likely roughly symmetrical. When they diverge significantly, the distribution is probably skewed, and the median becomes the more honest summary of the “typical” case. According to SPSS Services, if your distribution is roughly normal, report the mean and standard deviation; if it’s skewed or has notable outliers, report the median and interquartile range instead.

Measures of dispersion: understanding data spread 🔗

Central tendency tells you where the center of the data lies. Measures of dispersion tell you how spread out the data is around that center. Both are essential – a mean without a measure of spread can be highly misleading. Two datasets can share the same mean while having very different distributions.

According to the Social Data Analysis textbook, the key dispersion measures available in SPSS’s Frequencies → Statistics menu include:

• Standard deviation – the most commonly used measure of spread for continuous variables. It reflects the average distance of each data point from the mean. A small standard deviation means the data points cluster tightly around the mean; a large one indicates wide spread.
• Variance – the square of the standard deviation. Less intuitively interpretable on its own, but important for inferential statistics.
• Range – the difference between the highest and lowest values. Simple and easy to report, but unstable because it depends entirely on the two most extreme values in the dataset.
• Interquartile range (IQR) – the difference between the 75th percentile (Q3) and the 25th percentile (Q1). More stable than the range and especially useful when the distribution is skewed or contains outliers.
• Minimum and maximum – the lowest and highest recorded values, useful for spotting data entry errors or extreme outliers.

For nominal and ordinal variables, dispersion statistics like the standard deviation are not appropriate. As the Open Education Network SPSS guide explains, for categorical variables you work only with frequencies and percentages; for scale (interval/ratio) variables, you calculate mean, standard deviation, minimum, and maximum.

Skewness and kurtosis: reading the shape of your distribution 🔗

Two additional statistics in the SPSS output deserve attention when working with continuous variables: skewness and kurtosis. They describe the shape of the distribution, which matters because many inferential tests assume a roughly normal (bell-shaped) distribution.

Skewness measures asymmetry. A positive skewness value means the tail of the distribution extends to the right – a few unusually high values are pulling the mean upward. Negative skewness means the tail extends to the left. Kurtosis measures the peakedness of the distribution: values above zero indicate a sharper peak than a normal curve; values below zero indicate a flatter distribution. As PubAdmin Institute notes, these statistics are important signposts – significant skewness or kurtosis may indicate that a data transformation is needed before running parametric statistical tests.

SPSS can also generate histograms with a normal curve overlay, which makes it visually easy to assess whether your continuous variable is approximately normally distributed. Access this through: Analyze → Descriptive Statistics → Explore → Plots → Histogram.

Interpreting SPSS output: turning numbers into meaning 🔗

Generating statistics in SPSS is the easy part. Interpreting them meaningfully is where the real analytical skill lies. A few principles to keep in mind:

Always contextualize your numbers within the research domain. An average age of 35 means something very different in a study of first-year university students versus one on retirement planning. A standard deviation of 15 points on an IQ test is expected; the same standard deviation on a 5-point Likert scale question would be impossible and would signal a data problem.

Compare mean and median to gauge skewness before formal tests. If these two values are close, your distribution is probably symmetrical enough to use the mean as your headline figure. If they diverge, lean on the median.

Read Valid Percent, not Percent, in frequency tables. Valid Percent excludes missing data from the denominator, giving you a more accurate picture of the actual responses in your sample.

Don’t ignore missing data. If a substantial proportion of your cases are missing on a variable, that needs to be addressed – either through imputation, exclusion, or at minimum, transparent reporting – before any conclusions are drawn.

As Doing Social Research cautions: always use your analytical judgment and never rely blindly on whatever SPSS produces. SPSS will calculate a mean for gender if you ask it to; it will not tell you that the result is nonsensical. That critical awareness is yours to maintain throughout.

Choosing the right visualizations 🔗

Visual output complements the numerical statistics and often communicates patterns faster than tables alone. Within SPSS’s Frequencies dialog, clicking Charts gives you the option to generate bar charts, pie charts, or histograms alongside your statistical output.

• Use bar charts or pie charts for nominal and ordinal variables – they show the relative frequency of each category clearly.
• Use histograms for continuous variables – they display the distribution shape, allow you to spot skewness, and can include a normal curve overlay to assess normality visually.
• Use box plots to examine spread and identify outliers in continuous variables. Access these via: Graphs → Legacy Dialogs → Box Plot.

For researchers preparing results for academic publication or thesis work, the Explore function (Analyze → Descriptive Statistics → Explore) offers the richest output in a single run – producing detailed descriptives, normality test results (including the Shapiro-Wilk test), stem-and-leaf plots, and boxplots together. This is particularly valuable for confirming whether your continuous variables meet the assumptions required for further parametric testing.

A practical step-by-step summary 🔗

To pull the full process together, here is a concise workflow for conducting univariate analysis in SPSS:

1. Identify the level of measurement for each variable before generating any statistics.
2. Navigate to Analyze → Descriptive Statistics → Frequencies.
3. Move your variable(s) into the variables box. Ensure Display frequency tables is checked for categorical variables.
4. Click Statistics and select the appropriate measures: Mode for nominal; Mode + Median for ordinal; Mean, Median, Mode, Standard Deviation, Range, Skewness, and Kurtosis for continuous variables.
5. Click Charts and select Bar/Pie chart for categorical variables, or Histogram for continuous variables.
6. Click OK to run the analysis. Review the output in the Output Viewer.
7. Interpret each statistic in the context of your research question and variable type – and write up findings in plain language, not just as restated numbers.

Thorough univariate analysis at the start of a project saves significant time later. It catches data problems early, informs decisions about which inferential tests are appropriate, and gives researchers a nuanced command of their dataset before more complex analyses begin.

What do you think? When you examine a variable’s distribution and find it is heavily skewed, how does that change the way you would report your findings – and which subsequent statistical tests would you reconsider using? If you were designing a survey study, which variables in your dataset do you think are most likely to require special attention during univariate analysis, and why?

References

1. https://stats.libretexts.org/Bookshelves/Applied_Statistics/Social_Data_Analysis:_Qualitative_and_Quantitative_Approaches_(Arthur_and_Clark)/03:_Quantitative_Data_Analysis_with_SPSS/3.02:_Quantitative_Analysis_with_SPSS-_Univariate_Analysis
2. https://manifold.open.umn.edu/read/chapter-5-data-management-and-cleaning-in-spss
3. https://subjectguides.sunyempire.edu/c.php?g=659059&p=4626896
4. https://pressbooks.bccampus.ca/simplestats/chapter/3-7/
5. https://libguides.library.kent.edu/SPSS/FrequenciesCategorical
6. https://soc.utah.edu/sociology3112/central-tendency-variability.php
7. https://spssservices.com/univariate-analysis-spss-guide/
8. https://pressbooks.ric.edu/socialdataanalysis/chapter/quantitative-analysis-with-spss-univariate-analysis/
9. https://pubadmin.institute/research-methodologies/univariate-analysis-spss
10. https://doingsocialresearch.com/calculating-measures-of-central-tendency-in-spss/