Description
Reading how much of each survey item's variance is explained by the retained factors, to identify weak items and judge whether the factor solution is adequate.
An item's communality (h squared) is the proportion of its variance accounted for by the common factors; the remainder is unique variance (specific variance plus error). Initial communalities in principal axis factoring are squared multiple correlations; extraction communalities reflect the final solution. Low communalities (commonly below about 0.40, and certainly below 0.20) suggest the item is unrelated to the other items, poorly worded or measuring something else. Communalities above 1 (Heywood cases) signal a misspecified solution. MacCallum et al. (1999) showed that high communalities reduce the sample size needed for stable solutions, so they also inform design. Rigour comes from reading communalities alongside loadings and content, not mechanically deleting items.
Attributes
- Discipline
- Survey & Psychometrics
- Loss type
- Siloed
- Stage
- Evaluation
- Data
- Quantitative
- Effort
- Hours
Steps
- Request initial and extraction communalities in the EFA output.
- Flag items with extraction communalities below about 0.40 and treat those below 0.20 as strong candidates for removal.
- Investigate any communality at or above 1.0 as a Heywood case and revisit the number of factors or extraction method.
- Review flagged items' wording and content coverage before deleting; a low communality can mean a unique but important facet.
- Remove items one at a time and re-run, monitoring changes in other communalities and loadings.
- Report mean communality; with averages above about 0.60, smaller samples (around 100-200) can give stable solutions.
Sources
- SPSS FACTOR
- RELIABILITY procedures; syntax automation; parallel analysis via Python plugin
- Field, A. (2024) Discovering Statistics Using IBM SPSS Statistics, 6th ed.
- MacCallum, R. C. et al. (1999) Sample size in factor analysis. Psychological Methods
- Fabrigar, L. R. & Wegener, D. T. (2012) Exploratory Factor Analysis
- The Inquiry Series, C7 — SPSS for Survey Analysis — EFA, Reliability & Scale Validation
Also known as
- Item communality
- h-squared
Categories
- Analytics & Behaviour
- Survey & Psychometrics