Cronbach's Alpha Reliability Analysis

Testing the internal consistency of a measurement scale — an essential foundational step before running Exploratory Factor Analysis (EFA), Confirmatory Factor Analysis (CFA), or Structural Equation Modeling (SEM) in SPSS.

1 Introduction to Cronbach's Alpha

Cronbach's Alpha is a statistical measure that evaluates the reliability of a scale (Internal Consistency) — the degree of internal consistency among the items within the same scale. This coefficient ranges from 0 to 1; the closer it is to 1, the higher the reliability of the scale.

A good scale doesn't need too many items — what matters most is that each item reflects an independent aspect without overlapping in content with one another. Overlap creates data redundancy and reduces the accuracy of quantitative estimates.

To understand this more clearly, consider an example: the construct GSA (Graduate Soft Skills) is a research concept measured through 5 observed variables (items) coded from GSA1 to GSA5 below. This set of 5 items is called a scale.

Code Detailed Item Content Rating Level (Likert)
1 2 3 4 5
GSA1 I can communicate ideas clearly and effectively.
GSA2 I work effectively as a member of a team.
GSA3 I am confident solving problems in unfamiliar situations.
GSA4 I can adapt quickly to new environments and tasks.
GSA5 I am confident in my overall competence in the workplace.
1 — Strongly Disagree
2 — Disagree
3 — Neutral
4 — Agree
5 — Strongly Agree
Key distinction to note: a "scale" in Cronbach's Alpha refers to a set of interdependent items measuring a single research construct — completely different from the "rating scale" (Likert 1–5) shown above. Testing scale reliability means checking the consistency among the component items, not evaluating the width of the Likert range.
2 SPSS Testing Criteria

To conclude that a scale meets scientific standards, the analyst must sequentially check it against the 2 testing criteria below:

Criterion 1 — Overall Scale Alpha Coefficient

The overall Cronbach's Alpha coefficient (found in the Reliability Statistics table) must reach a minimum value of > 0.7 (in some new exploratory studies, a threshold of > 0.6 may still be considered acceptable) — according to Hair et al. (2019).

0.8 ≤ α < 1.0 Excellent
0.7 ≤ α < 0.8 Good
0.6 ≤ α < 0.7 Acceptable
α < 0.6 Poor / Unacceptable

Criterion 2 — Corrected Item-Total Correlation

At the same time, the corrected item-total correlation coefficient (Corrected Item–Total Correlation) of every item in the scale must reach a minimum value of ≥ 0.3 — according to Cristobal et al. (2007).

Minimum requirement
If an item scores < 0.3, it must be removed from the model.
≥ 0.3
3 Illustrative Example of Real SPSS Results

Below is a sample results table for the Cronbach's Alpha test of the GSA scale, taken directly from SPSS output:

Overall Alpha Coefficient
.884
Number of Items (N)
5 Items
MEETS SCIENTIFIC STANDARD (> 0.7)

Item-Total Statistics Table

Item Code Scale Mean if Item Deleted Scale Variance if Item Deleted Corrected Item-Total Correlation Cronbach's Alpha if Item Deleted
GSA1 12.06 16.139 .679 .869
GSA2 12.06 16.221 .731 .857
GSA3 12.08 16.483 .686 .867
GSA4 12.38 14.526 .791 .842
GSA5 12.22 15.522 .722 .859

Analysis of the Real Results

The overall Cronbach's Alpha coefficient reaches 0.884 > 0.7; every item-total correlation coefficient (4th column) meets the > 0.3 standard; and every value in the Cronbach's Alpha if Item Deleted column is lower than the overall Alpha (0.884). Therefore, the GSA scale achieves excellent reliability and is kept in full to move on to the next step.

A Classic Mistake to Avoid

There is absolutely no such thing as "an individual Cronbach's Alpha coefficient for each item." The Cronbach's Alpha if Item Deleted column is the new Alpha coefficient for the whole group AFTER that item has been removed. Many students take the value in this column and compare it independently against the 0.7 threshold to decide whether to keep or drop the item — a conclusion that is completely wrong from a statistical standpoint.

4 Precise Steps to Follow in SPSS

The testing procedure consists of the 7 specific steps below, helping you easily follow the correct technical process:

1

Open the Test Dialog Box

On the main toolbar of SPSS, follow this path: Analyze > Scale > Reliability Analysis...

2

Add the Component Items to the Test

Find and select the items belonging to one factor (e.g. GSA1–GSA5) in the left-hand column, then click the arrow button to move them into the Items box on the right.

3

Configure Additional Statistics

Click the Statistics... button in the top-right corner of the dialog box. Under "Descriptives for", check the "Scale if item deleted" box, then click Continue.

4

Run the Analysis

Make sure the Model field shows the default option, Alpha. Click OK to start the calculation and display the data in the Output window.

5

Check for Items That Violate the Rules

Review the "Corrected Item-Total Correlation" column: filter out any items with a coefficient below 0.3, or items whose removal would noticeably raise the overall Alpha.

6

Rule for Removing a Violating Item

If the scale contains a violating item, remove only the single worst-offending item — never remove multiple items at the same time — then repeat the process from step 1 to re-run the test.

7

Finalize the Reliability Conclusion

Repeat this loop until every remaining item meets the requirement (item-total correlation ≥ 0.3) and the overall Alpha coefficient reaches ≥ 0.6 (or ideally ≥ 0.7).

Pro tip from A2Z Supporter: always remove the item with the negative or lowest Corrected Item-Total Correlation first. Once that item is excluded, the internal consistency of the remaining items will improve noticeably right away.
5 Summary Flowchart of the Real Process

The complete decision logic for testing scale reliability is visually summarized by the A2Z Supporter expert team in the diagram below:

A

CASE A — Group Alpha DOES NOT yet meet the standard (< 0.7)

B

CASE B — Group Alpha ALREADY meets the standard (≥ 0.7)

Is there any item with Alpha if Item Deleted > the group Alpha?
YES NO
Remove that violating item from the scale and re-run the reliability test. Repeat until all items meet the standard.
After completing the removal, has the scale's Alpha coefficient reached ≥ 0.6?
MET NOT YET MET
Provisionally acceptable: Scale meets the minimum threshold for new exploratory research. The items should be improved in future studies.
Discard the scale: Scale lacks sufficient scientific reliability to continue the analysis and must be discarded entirely from the study.
Is there any item with Alpha if Item Deleted > the group Alpha?
NO YES
Perfect: Keep the entire set of items in the scale. The scale directly meets the scientific standard.
Item-total correlation: Does that item have an item-total correlation coefficient ≥ 0.3?
YES
Accept: No need to remove this item. The scale still meets the theoretical reliability standard.

Important Supplementary Rule

In every case, if the gap between the overall Alpha and the Alpha-if-Deleted value of an item is greater than or equal to 0.3, removing that item is entirely justified to strengthen the robustness of your research model.