METHOD GUIDE
Likert Scale Data in SPSS
Single Items, Composite Scores and Their Tests
Whether you analyze each item or a summed score decides which descriptives, tests and models can be defended. Response labels, reversed items and the scoring rule are fixed first, and each later table then follows the level the analysis works at.
IN SHORT
Items Versus Scores: the Decision Behind Every Likert Analysis
A Likert item is an ordered response to one statement, and a Likert scale combines several such items into one score, so the two are analyzed differently. A single item is ordinal, and its distribution, median and rank-based or ordinal models fit it. A composite score is often treated as approximately interval when the items hang together and the score is not strongly skewed or bounded. To analyze Likert scale data in SPSS, label the response values, reverse-code negatively worded items, set Measure in Variable View, summarize single items with Frequencies, and build a composite with Transform, Compute Variable (labels vary by version). OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, carries out and reviews this analysis through an in-house PhD-level statistician.
ANALYSIS SUPPORT
Analysis Support for This Method
Items are summed only after the scale has been checked, which is the work of Reliability Analysis (Cronbach's Alpha), and an item kept as an outcome calls for a model built for ordered categories.
Likert Scale Analysis
The analysis of survey items and scores within your study design, with the scoring rule, the scale checks, the choice of test and the reporting agreed for your questionnaire.
Ordinal Regression
A model for an ordered response with predictors, for when a single item is the outcome and the ordering should be kept without treating the categories as equally spaced.
BEFORE YOU RUN IT
Fixing What a Likert Score Means
Response codes, reversed wording and the rule for missing items are settled and recorded before the first analysis, because every later table depends on them.
Labels and Direction Each response value carries a label, and every item points the same way, with high codes meaning more agreement or more of the attribute. Reversed items are recoded into new variables and checked against the originals.
Measurement Setting Variable View lets you set Measure to Ordinal for items and Scale for composite scores. The setting guides defaults in some dialogs and charts and does not change the arithmetic of a test, so it documents your decision.
Sum or Mean Score A mean of the items keeps the original response range and compares across scales with different item counts, while a sum is also common. A rule for missing items, such as a minimum number answered, is fixed in advance.
Items That Belong Together Items are combined only if they measure one attribute. That rests on the design of the questionnaire and on checks of dimensionality and reliability in your own data, not on a shared response format.
IN SPSS
How to Analyze Likert Scale Data in SPSS
Most of the work happens under Transform in IBM SPSS, before any analysis menu is opened. Menu wording varies a little by version.
- Add value labels, then recode negatively worded items into new variables with Transform, Recode into Different Variables.
- For single items choose Analyze, Descriptive Statistics, Frequencies, and request the frequency table with the median and quartiles, and a bar chart.
- Build the composite with Transform, Compute Variable, using the MEAN function with a minimum number of valid items, or SUM.
- Check the reliability of the items and the histogram of the composite before analyzing it.
- Compare groups on a single item with a rank-based test, and on a composite with a mean-based test where its distribution supports it.
- Model a single item as an outcome with an ordinal model, and a composite as an outcome with a linear model.
READING THE OUTPUT
Which SPSS Tables Describe an Item and Which Describe a Score
Single items are read from frequency tables and charts, composite scores from descriptive tables and a histogram, and mixing the two is the commonest source of confusion.
| Output element | What it tells you | What to check |
|---|---|---|
| Frequencies: Statistics | The Valid and Missing N, the median and the percentiles requested for an item | That each item reports its own valid N, since missing responses differ across items |
| Frequencies: item table | Frequency, Percent, Valid Percent and Cumulative Percent for each response category | That Valid Percent is quoted when responses are missing, and unused categories at either end are noticed |
| Bar chart of an item | The shape of the responses across the categories | That a ceiling, a floor or two peaks is described before any mean is quoted |
| Descriptive Statistics for the composite | The N, Minimum, Maximum, Mean and Std. Deviation of the new score | That the minimum and maximum equal the possible range of the score, which exposes an unrecoded item |
| Histogram of the composite | The distribution of the summed or averaged score | That strong skew or a pile-up at the bound leads to a method suited to it, not an automatic comparison of means |
| Reliability Statistics | Cronbach’s alpha and the N of Items for the items that were combined | That it is reported for each composite, with the items listed |
| Crosstabulation: item by group | The counts and percentages of each response in each group | That percentages run within group, and sparse categories are merged on substantive grounds before a chi-square is read |
REPORTING
Reporting Single Items and Composite Scores Separately
Items and composite scores call for different statistics, so the frame below treats them in separate sentences.
For a single item, report the number of valid responses, the percentage in each category or in grouped categories such as agree and strongly agree, and the median with its interquartile range. A mean and standard deviation may be added for items with several response points when the distribution is shown. For a composite, state the items, the scoring rule, the handling of missing items and the reliability, then report M = ___, SD = ___ and N = ___, and for a comparison the mean difference with its 95% CI [___, ___] and a standardized effect size. For a rank-based comparison of one item, give the test statistic, p = ___ and a rank-based effect size with its interval. Add a table of the item wording and response labels so readers can see exactly what was asked.
COMMON MISTAKES
Likert Habits That Draw Methodological Comments
Reviewers rarely object to Likert data itself. They object to unexplained scoring rules, means given without distributions and conclusions larger than the measure supports.
One Item Averaged Blindly A mean of one five-point item can hide a split between opposing views. The distribution is shown, with a median or category percentages, before a mean is added.
Sum Without a Scale Check Adding items that were never shown to measure one attribute produces a number without a clear meaning. Dimensionality and reliability are checked in the study sample first.
Reversed Wording Forgotten A negatively worded item left unrecoded cancels part of what the other items add to the score. The range of the composite and the item correlations expose it once they are inspected.
Ordinal Treated as Distance Equal spacing between categories is an assumption. For a single item a rank-based or ordinal approach avoids it, while for a composite the assumption is argued and stated.
Missing Items Skipped Silently A sum that skips missing items lowers the score of everyone who left one blank. The rule for incomplete responses is fixed in advance and applied to every respondent.
Tiny Difference, Significant p In a large survey a test on scores can be significant for a difference too small to matter. The size of the difference is reported with its interval beside the p-value.
Scores already built? Independent Statistical Review checks the scoring rule, the test chosen for each item or score and the wording before you submit.
FREQUENTLY ASKED QUESTIONS
Questions About Likert Data in SPSS
Likert scale analysis in SPSS raises questions before any test is chosen: how variables are declared, why a score falls outside its range and what a frequency table leaves out.
How should single Likert items be set in Variable View?
Set single items to Ordinal and computed composite scores to Scale, and treat the setting as documentation of your decision. The Measure property does not change how a t-test or a correlation is computed. It guides defaults in some dialogs and charts, and it tells a reader how you regard each variable. The choice of test still follows from the question, the design and the distribution.
Why do my Likert composite scores in SPSS go beyond the possible range?
A minimum or maximum outside the possible range means some item still carries an out-of-range value. The usual causes are a code such as 99 for not applicable that was never declared as missing, or an item recoded twice or not at all. Run Frequencies on every item and compare the Descriptive Statistics table of the composite with the range the scoring rule allows.
Why is a response option missing from my SPSS frequency table?
Frequencies and Crosstabs list only the values that occur in the data, so an option nobody chose is left out of the table instead of being shown with a count of zero. A five-point item with no strongly disagree responses therefore prints four rows. The value labels confirm which options were offered, and the reported distribution names every option, with 0% for the empty one.
Can OnlineSPSS check my scoring rule and the tests for items and scores?
Yes. An in-house PhD-level statistician can review how items were recoded and combined, confirm which test or model suits each item or score, or carry out the analysis on your data. If the analysis is done and only the write-up of items and scores remains, Results Reporting & Interpretation is the matching service. Which questions the survey answers, and what to conclude, stay with you.
How do you protect my research data and confidentiality?
Your data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the dataset or the output you have, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Items or Scores:
Which Should Carry Your Analysis?
Outline the questionnaire, which items form each construct and the comparison you plan. The confirmed scope then says whether items, scores or both will be analyzed.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014