Page 203 - Quantitative Data Analysis

Quantitative Data Analysis
Simply Explained Using SPSS

Example of Principal Component Analysis

The scales assessing employment-related issues consisted of 32
items in which participants responded to prompts addressing both
perceived importance of and level of satisfaction with the issues.
The participants’ unweighted response to the satisfaction level with
each issue was used for the principal components analysis. A
conceptual assessment of the items by two of the researchers
resulted in a division of the items into 6 scales. A six-factor principal
components analysis was selected for testing the model, allowing
for correlated factor loadings given that there were both research-
substantiated and hypothesized relationships between selected
scales.
A factor loading value of .40 was selected as the criterion for
retaining an item in a scale. The factor loadings ranged from .46 to
.71 on the Employer Support Scale; .43 to .70 on Program
Knowledge; .41 to .66 on External Support; .42 to .71 on Service
Provision; .61 to .69 on Work Potential; and .65 and .73 on
Prescriptions and Healthcare. Two of the 32 items were not
retained in the scales; items 56 and 63. These two items did not
cleanly load on the scales initially hypothesized by the conceptual
assessment and thus were not included for the analyses in the
study. A list of the items for each of the 6 scales is provided in Table.
The 6-factor solution was moderately effective in accounting for the
variability in individual item responses, with a range of 46% to 68%
of the variability explained by the common factors. The correlations
between factors ranged from .26 to .50. None of the items had
significant loadings greater than .40 on more than one factor in the
oblique factor solution, resulting in a relatively clean differentiation
between item groupings.

The Theory and Applications of Statistical Inferences 187