SPSS tutorials website header logo SPSS TUTORIALS BASICS ANOVA REGRESSION FACTOR CORRELATION

SPSS Data Analysis – Simple Roadmap

  1. Set Up Project Folder and Open Data;
  2. SPSS Data File Inspection;
  3. SPSS Categorical Variable Inspection;
  4. SPSS Metric Variable Inspection;
  5. Optionally: Edit Data;
  6. Choose and Run Tables/Charts/Tests.

1. Set Up Project Folder and Open Data

The biggest waste of time and effort in SPSS is probably not keeping projects organized. A related pitfall is not regularly making backup copies of the entire project. Avoiding this starts with setting up a project folder that'll contain all of your data -original and edited-, syntax and output files.
We recommend you never edit your original data and keep it in a safe place. For me, that's usually a subfolder called “ori”, short for “original data”. Make sure that the project contains all files you'd like to backup -and nothing else.
Done setting up a decent project folder? Then let's go and open the data.

Screenshot of well organized SPSS project folder Keeping this project nicely organized saved me way more time than it cost me.

2. SPSS Data File Inspection

At this point we know which variables in our data -possibly all- we're actually going to use. A sound way to proceed from here is inspecting our data visually. Some things we need to know are

If you encounter any such issues, fix them right away. The sooner you troubleshoot such issues, the less time and effort they'll cost you.

Excessively Long Variable Names Shortening these variable names and applying variable labels saves more effort than it costs

At this point our data should be technically in order. So what about the contents of our variables? I suggest you carefully check these for categorical variables and metric variables separately.

3. SPSS Categorical Variable Inspection

We inspect categorical variables by

A single line FREQUENCIES command suffices for many variables in one go. Issues we typically look for are:

Reverse Coded Variable in SPSS Reverse coded variable - not really wrong but inconvenient nevertheless.

If any such issues are present, try and fix them. If they can't be fixed, perhaps take some notes so you won't have any nasty surprises later on.

4. SPSS Metric Variable Inspection

We inspect metric variables by

Note that you can run many histograms with a single line FREQUENCIES command as shown in Creating Histograms in SPSS. Histograms basically tell you all you need to know. Issues to look out for are

Next, a basic DESCRIPTIVES table comes in handy for checking the completeness of a set of variables. It'll also allow for a quick comparison of means and standard deviations.
After completing these steps, we can be confident that our data are sound. Nothing incorrect or unusual can mess up any newly created variables or test results anymore. Now -and only now- should we proceed with editing or analyzing our data. As a bonus, we also know what our data basically look like.

5. Optionally: Edit Data

Perhaps your research questions relate to variables that still need to be created or adjusted. Well, this is the moment to do so. Our most read tutorials on common data adjustments are

Computing Means in SPSS While Excluding Missing Values A nice -syntax only- trick for excluding cases with many missings when computing means

Hope those will get you started. Really, do adjust your data if needed. This often results in much nicer output with much less effort.

6. Choose and Run Tables, Charts & Tests

First off, which tables, charts and tests are appropriate is a complicated question that doesn't have a simple answer. Oftentimes, different approaches are equally defensible.
In any case, the simplest analysis techniques examine each variable separately. These are called univariate analyses (“univariate” means “for one variable”). As shown below, we should at least distinguish categorical from metric variables.

Minimal Overview Univariate Analyses

LevelTableChartTest
CategoricalFREQUENCIESBar chart frequenciesBinomial test (2 categories)
Chi-square goodness-of-fit test (3+ categories)
MetricDESCRIPTIVESHistogramOne-sample t-test (mean)
Kolmogorov-Smirnov test (distribution)

A next step could be to examine if 2 variables are associated in any way. This involves bivariate analyses (“bivariate” means “for 2 variables”). Distinguishing categorical from metric variables once again, we arrive at the simple overview below.

Minimal Overview Bivariate Association Analyses

Variable AVariable BTableChartTest
CategoricalCategoricalCROSSTABSStacked bar chart percentagesChi-square independence test
MetricCategoricalMEANSBar chart means by categoryIndependent samples t-test (2 categories)
One-way ANOVA (3+ categories)
MetricMetricCORRELATIONSScatterplotCorrelation test (non directional)
Simple linear regression (directional)

If you properly understand these tests, you'll start to see that most statistical tests are variations on these big 5 tests. For example,

Is that all? No, not quite. First off, we only mentioned categorical and metric variables. Ideally, we'd distinguish

We don't always need to treat these all separately but doing so results in a much more complete overview. We're working on it but it'll take another while.
For now, perhaps consult Which Statistical Test Should I Use?, part of which is shown below. Unfortunately, this overview is limited to statistical significance tests and does not suggest which tables and charts to use.

Screenshot from overview tutorial statistical tests Simple overview statistical comparison tests

Thanks for reading!

Tell us what you think!

*Required field. Your comment will show up after approval from a moderator.

THIS TUTORIAL HAS 11 COMMENTS: