Variable Labels and Value Labels in SPSS (Why They Matter)

The Difference Between a Code and Its Meaning

Open a typical SPSS dataset and you will see numbers: 1s and 2s in one column, 1s through 5s in another. To SPSS, those numbers are perfectly clear. To a human reading the output, they are meaningless — unless someone took the time to label them. That is what variable labels and value labels are for, and they are the single easiest way to turn cryptic output into results you can actually read.

New SPSS users often skip labels because the analysis runs fine without them. The analysis does run fine. The problem comes later: when you stare at a table headed “q14” with rows labelled 1, 2, 3, 4, and 5, and you can no longer remember what the question asked or which end of the scale meant agreement. Labels fix that, and they cost only a few minutes per dataset. This tutorial explains the difference between the two kinds of label, shows you how to set both click by click and with syntax, and explains how they change your output, charts, and reporting.

Variable Labels vs Value Labels: What’s the Difference?

The distinction is simple once you see it, and confusing only because the two names sound alike.

A variable label describes the variable itself. It is the full, human-readable version of a short variable name. The variable name might be sat_job, but its variable label is “Overall satisfaction with current job”. The variable name might be q7, but its variable label is “How often do you exercise in a typical week?” Variable names must be short and follow SPSS naming rules — no spaces, no special characters — so the label carries the real meaning.

A value label describes what each code in a variable means. If gender is coded 1 for female and 2 for male, the value labels are “Female” and “Male”. If a Likert item is coded 1 to 5, the value labels are “Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, and “Strongly agree”. Value labels turn numbers back into words wherever SPSS displays results.

Think of it this way: the variable label answers “what was asked?”, and the value labels answer “what does each answer mean?” You need both. A labelled variable with unlabelled values still produces tables full of bare numbers, and labelled values on an unlabelled variable still leave you guessing which question the table belongs to.

Why Labels Matter More Than You Think

Your output becomes readable at a glance. Frequency tables, crosstabs, and charts display labels instead of codes. Instead of a table for “q14” with rows 1–5, you get a table titled with the question wording and rows reading “Strongly disagree” through “Strongly agree”. Anyone — including you in six months — can read it without a codebook open beside it.

You make fewer mistakes. It is easy to misread a table when 1 means “yes” in one variable and “no” in another, which happens more often than anyone admits. Labels make reversed coding visible: if a table shows “Strongly agree” where you expected disagreement, you catch the problem immediately instead of reporting it backwards.

Your work is reproducible. A dataset with labels is partly self-documenting. A colleague, a supervisor, or a future you can open the file and understand it without hunting for a separate codebook document. Labels travel with the .sav file when you share it.

Reporting gets faster. When output already shows the question wording and answer labels, writing up results is mostly a matter of copying values into sentences rather than decoding tables first. That is also where labelling pays for itself during a thesis deadline.

How to Set Variable Labels (Click by Click)

Open your dataset and switch to Variable View using the tab at the bottom of the Data Editor. Each row here is one variable, and each column is one property of that variable.

Step 1: Find the Label column

In Variable View, locate the Label column. It sits between the Type and Values columns. If your screen is narrow, scroll right — the column is there, just sometimes off screen.

Step 2: Type the full description

Click the Label cell for a variable and type the full wording: the question as participants saw it, or a clear description of what the variable measures. For example, for a variable named age, the label might be “Age in years at time of survey”. For a variable named income_cat, it might be “Annual household income category”. Keep labels informative but not rambling — one clear sentence is ideal.

Step 3: Repeat for every variable

Work down the list methodically. It is tempting to label only the variables you are analysing today, but the ones you skip are exactly the ones you will puzzle over later. When you are done, click back to Data View — nothing looks different there, because labels show up in output, not in the raw data grid.

One habit worth building: label variables the moment you create them. A computed variable named wellbeing_total deserves its label — “Total wellbeing score (sum of 5 items, range 5–25)” — at the moment of creation, while you still remember what went into it.

How to Set Value Labels (Click by Click)

Step 1: Open the Value Labels dialog

In Variable View, click the Values cell for the variable you want to label. A small grey button with three dots appears in the cell — click it to open the Value Labels dialog.

Step 2: Add each code and its label

In the dialog, type the numeric code in the Value box (for example, 1), type its meaning in the Label box (for example, Female), and click Add. The pair appears in the list at the bottom. Repeat for every code: 2 = Male, and so on. For a five-point Likert item, add all five pairs, including the midpoint — an unlabelled middle category is a common source of misread tables.

Step 3: Confirm and move on

Click OK. The Values cell now shows a summary like {1, Female}… to confirm labels are attached. Repeat for each categorical variable. Continuous variables such as age in years or test scores do not need value labels — every value is itself meaningful — so skip those and save your time for the coded variables.

Watch out for one trap: value labels do not change the underlying numbers. Labelling 1 as “Female” does not turn the data into text, and your statistics still run on the numbers. That is exactly what you want — labels change display, never data.

Setting Labels with Syntax

If you have many variables to label, or you want a record of your labels, syntax is faster and more reliable than clicking through dialogs. Paste-ready examples:

VARIABLE LABELS
  gender "Participant gender"
  sat_job "Overall satisfaction with current job"
  q14 "I feel confident using statistical software".
VALUE LABELS
  gender 1 "Female" 2 "Male"
  /q14 1 "Strongly disagree" 2 "Disagree" 3 "Neither agree nor disagree"
         4 "Agree" 5 "Strongly agree".
EXECUTE.

Note the structure: VARIABLE LABELS lists each variable name followed by its label in quotes. VALUE LABELS lists each variable followed by its code-label pairs, with a slash before each new variable after the first. Run the syntax, then check Variable View to confirm the labels landed. Save the syntax file with your project — it doubles as a codebook.

Syntax also makes it easy to reuse labels across datasets. If you run the same questionnaire each semester, the same VALUE LABELS block works every time, and your output stays consistent from one wave to the next.

How Labels Change Your Output

The payoff appears the first time you run an analysis. Run Analyze > Descriptive Statistics > Frequencies on a labelled variable and compare it with an unlabelled one. The labelled table carries the question wording as its title and the answer meanings as row labels. Run a crosstab on two labelled variables and both the rows and columns read as words, not codes.

Charts benefit too. A bar chart of a labelled variable shows category names on the axis rather than bare numbers, which means the chart is presentable without manual relabelling in another program. If a chart still shows numbers, check whether that variable actually has value labels — the usual cause is a variable you meant to label but missed.

There is one display setting worth knowing: in the output viewer, SPSS can show values, labels, or both. If your tables ever show codes when you expected labels, go to Edit > Options > Output and check the pivot-table labelling settings. Labels you set are still there; the display option just decides whether output uses them.

Common Labelling Mistakes to Avoid

Labelling only some categories. A variable coded 0, 1, and 2 where only 1 and 2 are labelled will show a bare 0 in every table. Label every code that occurs, including 0 and any “prefer not to say” or “not applicable” codes.

Inconsistent labels across variables. If 1 means “Yes” in one variable and “No” in another, say so clearly in each variable’s labels — and think twice before designing a questionnaire that way. Consistency in coding prevents the classic reversed-result error.

Labels that are too long. A 200-character label wraps awkwardly in tables and charts. Keep labels to a sentence. Put extra detail — the exact questionnaire wording, the source scale — in your codebook or a separate note, not in the label itself.

Forgetting missing-value codes. If 99 means “missing” in your data, label it “Missing” and ideally declare it as a user-missing value under Analyze settings so it is excluded from statistics. An unlabelled 99 otherwise looks like a real score and wrecks means and correlations.

Assuming labels survive every file format. Labels are stored in SPSS .sav files. If you export to Excel or CSV and re-import, labels are usually lost. Always keep the .sav file as your master copy, and re-apply labels from syntax if you ever rebuild the dataset.

Reporting in APA Style (7th ed.)

Labels do not usually appear verbatim in a results section, but they shape how you describe your variables, and that description belongs in the Method section. A reader should be able to match every variable in your tables to a clear definition in the text. For example:

“Job satisfaction was measured with a single item (‘Overall, how satisfied are you with your current job?’) rated on a five-point scale from 1 (strongly disagree) to 5 (strongly agree). Gender was recorded as a categorical variable (1 = female, 2 = male).”

That sentence is only possible because the variable and value labels were set properly in the first place. When you report results, use the label wording — “satisfaction with current job” rather than “sat_job” — and give the scale anchors when you first mention a measure, so readers know which direction is which. Tables should carry the same wording as the text; a table headed with a cryptic variable name while the text uses plain language forces readers to translate, and translation is where errors creep in.

Key Takeaways

  • Variable labels describe the variable (the question asked); value labels describe what each code means (the answers).
  • Set variable labels in the Label column of Variable View, and value labels through the Values cell dialog — or faster, with VARIABLE LABELS and VALUE LABELS syntax.
  • Labels change display, never data: your statistics still run on the numbers underneath.
  • Label every code that occurs, including 0 and missing-value codes, and keep labels to one clear sentence.
  • Labelled output reads at a glance, prevents reversed-coding mistakes, and makes your dataset self-documenting for anyone who opens it later.

Frequently Asked Questions

Do I need labels for continuous variables like age?

Variable labels, yes — “Age in years” is clearer than “age”. Value labels, no: each value of a continuous variable is already meaningful, so there is nothing to decode.

Can I change a label after analysing?

Yes, freely. Changing a label never changes your data or your results — it only changes how future output displays. Just rerun the analysis to see the new labels in fresh output; existing output tables do not update retroactively.

Why do my tables still show numbers instead of labels?

Usually one of three causes: the variable has no value labels yet, the output display option is set to show values rather than labels, or the variable is being treated as scale rather than nominal. Check the labels in Variable View first, then Edit > Options > Output.

Do labels transfer if I share my data file?

Yes, if you share the SPSS .sav file — labels are stored inside it. If you share a CSV or Excel export instead, labels are lost, so share the .sav file (or the syntax that recreates the labels) alongside any export.

Should value labels match my questionnaire wording exactly?

As closely as practical. If participants saw “Strongly agree”, label it “Strongly agree” — not a paraphrase. Exact wording keeps your output, your Method section, and your questionnaire visibly consistent.

Need this done on your data? Get a free quote → /services/