You have just finished collecting survey data. Ten questionnaire items all measure the same thing — say, anxiety, job satisfaction, or attitude toward online learning. Each item is scored 1 to 5. Now what? You cannot run a t-test or a regression on ten separate items and expect a clean answer. You need one number per person that captures the whole scale. That number is a composite scale score, and computing one in SPSS takes about two minutes once you know the steps.
In this tutorial you will learn exactly what a composite score is, the three checks to run before you compute one, how to create it with the menus and with syntax, and how to report it in APA style.
What Is a Composite Scale Score?
A composite scale score combines several questionnaire items that measure the same underlying construct into a single variable. If your anxiety scale has 10 items scored from 1 (strongly disagree) to 5 (strongly agree), each participant gets one composite score instead of ten separate answers.
There are two common ways to combine items:
- The mean score: the average of the items. A participant who answered 4, 3, 4, 5 across four items gets a mean of 4.00. Mean scores stay on the original response scale, so a mean of 4.00 on a 1–5 scale is easy to interpret: it means the person generally agreed with the items.
- The sum score: the total of the items. The same four answers give a sum of 16. Sum scores are common when a scale has published cut-off points — for example, a depression inventory where a total above a certain value signals clinical concern.
For most student research, the mean score is the better default. It keeps the score on the familiar 1–5 (or 1–7) scale, and it stays comparable even if different participants answered a different number of items. Use a sum score when your scale’s manual, your supervisor, or the published literature reports totals and cut-off values.
Before You Compute: 3 Checks First
Computing a composite score is easy. Computing a correct one requires three quick checks first. Skip them and your score can be quietly wrong.
1. Reverse-score any negatively worded items
Most scales mix positively and negatively worded items to catch careless responding. An item such as “I feel calm and relaxed” sits on an anxiety scale alongside “I feel tense and nervous.” A 5 on the calm item means low anxiety, while a 5 on the tense item means high anxiety. If you average them as they are, the items cancel each other out and your composite score is meaningless.
Reverse-score every negatively worded item before computing the composite. If you have not done this yet, see our step-by-step guide on reverse-scoring items in SPSS, and come back once the reversed variables are ready.
2. Decide how to handle missing data
Real data has gaps. A participant may skip one item out of ten. You have two choices: drop anyone with any missing item (listwise deletion), or compute the score from the items they did answer. Dropping people wastes data and shrinks your sample; using the items they answered is usually better.
A sensible rule is to require a minimum number of valid items — for example, at least 8 out of 10. SPSS lets you build this rule directly into the calculation, as shown in the syntax below. If a participant answered fewer than 8 items, their composite score stays missing, which is the honest result.
3. Check reliability first
A composite score only makes sense if the items actually hang together as one scale. Run a reliability analysis (Cronbach’s alpha) on the items before combining them. As a rough guide, an alpha of .70 or above is usually treated as acceptable for research purposes, .80 or above is good, and below .60 the items probably do not belong in a single score. If alpha is low, look at the item-total statistics: removing one bad item sometimes fixes the problem. Do not average items into a scale that the reliability analysis says do not belong together.
Method 1: Compute a Mean Scale Score (Menus)
This is the method most students should use. It creates a new variable holding each participant’s mean across the items.
- Click Transform > Compute Variable.
- In the Target Variable box, type a clear name for the new score, for example Anxiety_Mean. Variable names cannot contain spaces, so use underscores.
- In the Numeric Expression box, type the MEAN function with your item variables separated by commas, for example: MEAN(Anx1, Anx2, Anx3, Anx4, Anx5, Anx6, Anx7, Anx8, Anx9, Anx10). You can also select MEAN from the Function group list and double-click each variable from the list on the left to add it.
- To require a minimum number of valid items, change the function to MEAN.8(…). The .8 means SPSS needs at least 8 valid answers out of 10 before it calculates a score; anyone with fewer gets a missing value.
- Click OK. SPSS creates the new variable at the end of your data file. Switch to Data View and scroll to the last column to check it.
One detail worth knowing: the plain MEAN function without a number uses every valid value it finds, even if a participant answered only one item. That is why adding the .n rule is good practice — it stops a single answered item from standing in for the whole scale.
Method 2: Compute a Sum Score (Menus)
The steps are identical, except that you use the SUM function instead of MEAN:
- Click Transform > Compute Variable.
- Type a Target Variable name, for example Anxiety_Sum.
- In the Numeric Expression box, type SUM.8(Anx1, Anx2, Anx3, Anx4, Anx5, Anx6, Anx7, Anx8, Anx9, Anx10).
- Click OK.
Remember that a sum score changes meaning if participants answered different numbers of items, so the .8 minimum-valid rule matters even more here. A participant who answered 8 items cannot be fairly compared with one who answered all 10 on a raw total. If missing data is common in your file, prefer the mean score.
Paste-Ready SPSS Syntax
If you prefer syntax — or want a record of exactly what you did — paste the block below into a syntax window and run it. Replace the variable names with your own item names, and adjust the .8 to whatever minimum you chose.
* Compute mean scale score, requiring at least 8 of 10 valid items.
COMPUTE Anxiety_Mean = MEAN.8(Anx1, Anx2, Anx3, Anx4, Anx5, Anx6, Anx7, Anx8, Anx9, Anx10).
EXECUTE.
* Compute sum score, requiring at least 8 of 10 valid items.
COMPUTE Anxiety_Sum = SUM.8(Anx1, Anx2, Anx3, Anx4, Anx5, Anx6, Anx7, Anx8, Anx9, Anx10).
EXECUTE.
* Quick check of the new scale score.
FREQUENCIES VARIABLES=Anxiety_Mean Anxiety_Sum
/STATISTICS=MEAN STDDEV MINIMUM MAXIMUM
/HISTOGRAM NORMAL.To open a syntax window, click File > New > Syntax, paste the code, select it, and click the green Run button. Using syntax has a real advantage for a thesis: you can save the file and re-run the exact same steps later, which makes your analysis reproducible.
How to Interpret the Output
The FREQUENCIES command at the end of the syntax gives you a quick health check on your new score. Here is what to look for:
- Mean and standard deviation: the mean tells you the average level of the construct in your sample. On a 1–5 scale, a mean near 3 is middling; a mean near 4.5 means most participants scored high. The standard deviation tells you how spread out the scores are.
- Minimum and maximum: check these against what is possible. A mean score on a 1–5 scale cannot be below 1 or above 5, and a 10-item sum score cannot fall outside 10–50. A value outside the possible range means something went wrong — usually a missed reverse-scored item or a data-entry error.
- The histogram: look at the shape. A roughly bell-shaped distribution is what most parametric tests assume. Strong skew is not a disaster, but it is something to note before you run a t-test or ANOVA on the score.
- Missing values: the output shows how many participants have no composite score. If that number is large, revisit your minimum-valid rule or check whether particular items were skipped by many people.
If the numbers look wrong, do not press on. Check the reverse scoring first — it is the most common cause of a composite score that makes no sense.
Reporting in APA Style (7th ed.)
When you write up a composite score in a thesis or paper, report the scale, how the score was formed, its descriptive statistics, and its reliability — all in one or two sentences. Here is a template you can adapt:
Anxiety was measured with a 10-item scale (1 = strongly disagree, 5 = strongly agree). Items were averaged to form a composite score (M = 2.84, SD = 0.71), and the scale showed acceptable internal consistency (Cronbach’s α = .82).
APA conventions to follow: italicize M, SD, and α; report means and standard deviations to two decimal places; report Cronbach’s alpha to two decimal places with no leading zero before the decimal point (α = .82, not 0.82). If you used a sum score instead, say so plainly — “Items were summed to form a total score (possible range 10–50)” — so readers can interpret the mean correctly. If any items were reverse-scored, mention that too: “Three negatively worded items were reverse-scored before computing the composite.”
Key Takeaways
- A composite scale score turns several items measuring one construct into a single variable — usually a mean, sometimes a sum.
- Reverse-score negatively worded items before you compute anything, or the items will cancel each other out.
- Check Cronbach’s alpha first: a composite only makes sense if the items hold together as one scale (roughly .70 or above).
- Use MEAN.n or SUM.n in SPSS so a score is only computed when a participant answered enough items — for example, MEAN.8 for at least 8 of 10.
- Always check the minimum and maximum of the new score against the possible range; out-of-range values point to a missed reverse-score or a data-entry error.
- Report the scale, the scoring method, M and SD, and Cronbach’s alpha in APA style.
Frequently Asked Questions
Should I use the mean or the sum for my composite score?
Use the mean unless you have a specific reason to use the sum. The mean stays on the original response scale (for example, 1–5), so it is easy to interpret and stays comparable when participants answered different numbers of items. Use the sum when the scale has published totals or cut-off values, or when your supervisor or the literature reports totals.
How many missing items can I tolerate per participant?
A common rule is to allow up to 20% missing — for example, at least 8 valid answers out of 10 items. Build this into the calculation with MEAN.8 or SUM.8 so SPSS applies the rule automatically. If a participant answered fewer items than your minimum, leave their score missing rather than guessing.
Do I need to reverse-score items before computing the composite?
Yes, always. Any negatively worded item must be reverse-scored first. If you skip this step, high scores on positive items and high scores on negative items pull the average in opposite directions, and the composite score will not measure what you think it measures.
What should I do if Cronbach’s alpha is too low?
First, check the item-total statistics in the reliability output — one bad item is often the cause, and removing it can lift alpha into the acceptable range. Also double-check that all reverse scoring was done correctly, because a single missed reversal can destroy alpha. If alpha stays low after that, the items may measure more than one thing, and you should reconsider whether a single composite score is appropriate.
Can I combine items that use different response scales?
Not directly. If some items are scored 1–5 and others 1–7, averaging the raw scores gives the 1–7 items more weight. Standardize the items first (Transform > Descriptives > Save standardized values as variables) and average the z-scores, or rescale all items to a common range before computing the composite.
Need this done on your data? Get a free quote → /services/