print this page

related pages

on our site

news

© Copyright Commonwealth of Australia - Australian Public Service Commission

Home page
> State of the Service reports
> State of the Service Report 2005-06 > Appendixes > Factor analysis
‹ Previous page

Last updated: 30 November 2006

Appendix 4: Factor analysis

A factor analysis was conducted on a broad range of the 2006 State of the Service employee survey questions, and the questions selected based on their relevance to employee engagement.¹ These included most of the general impressions questions, and the questions related to work-life balance, merit, learning and development, senior leadership, immediate supervisor and agency culture.² Factor analysis is a

‘data reduction’ technique, which statistically groups together highly related questions. The factor analysis process primarily uses high correlations between question responses to group them together. For example, a high positive correlation between two questions is where most of the responses to one question show a very similar pattern of responses as those to the other question (i.e. across individuals, high scores on one generally correspond to high scores on the other). Factors are considered to reflect underlying processes or relationships that have created the correlations among variables.³

In this case, a factor analysis was conducted to determine whether, based on the questions from the employee survey, broader groups of engagement issues or ‘factors’ existed. Therefore, each factor represents a measure of a broader construct than the individual employee engagement questions in the 2006 employee survey. The factor analysis for the employee engagement items resulted in a 10 factor structure.⁴ The factors were then named based on their item content. The 10 factors were:

1. ‘Senior Leaders/Culture’
2. ‘Diversity—Recruitment and Retention’
3. ‘Immediate Supervisor’
4. ‘Governance’
5. ‘Diversity—Barriers’
6. ‘Merit’
7. ‘Current Job’
8. ‘Work Group’
9. ‘Understanding Current Role’
10. ‘Work-Life Balance/Learning and Development’

Factor content

The following tables show the question content of each factor. In the text of the report, factor names appear in quotation marks with the first letter capitalised to distinguish them from other uses of the word(s).

Calculation of factor scores for each factor

A factor score was calculated for each of the 10 factors. Each factor score is the average (mean) of the responses to the questions contained in the factor. Factor scores will therefore range from 1 to 5 on a continuous scale (i.e. they will not necessarily be whole numbers). Factor scores can be used in a similar way to question results, for example, in cross-tabulations with other questions and other factor scores.

Each respondent to the survey has one score for each factor. For ease of interpretation, the factor scores have been recoded as such:

• 1 to 2.499 = agree/satisfied
• 2.5 to 3.499 = neutral (neither agree nor disagree)
• 3.5 to 5 = disagree/dissatisfied

Factor score results

The table below contains the APS-wide frequency results for the 10 factors based on the approach outlined above.

1. The job satisfaction index was originally included in the analysis, but was removed due to significant ‘cross-loading’ across factors (i.e. job satisfaction was highly related to multiple factors).
2. All of the items included in the factor analysis are listed in this appendix under each factor heading. Individual item results for all employee survey questions are available in the State of the Service Employee Survey Results 2005–06 publication.
3. B. G. Tabachnick & L. S. Fidell 2001, Using Multivariate Statistics, 4th edn, Allyn & Bacon, Needham Heights, MA.
4. Extraction Method: Principal Axis Factoring. Rotation Method: Varimax with Kaiser Normalisation.