thereby leaving only those items
where communality was greater than 0,5. fnis stitt resulted in 10
components, but increased total variance to 62.7 percent. Exami_
nation ofthe scree diagram led to the adoption ofahve component
model that 'explained' 45 percent of the variance. Table 5 illus_
trates the results using the option provided by pASW to not show
weightings below 0.4 in accordance with a practice recommended
by some as showing a clarity of interpretaiion (Hair et al., 199g).
However, to enable the reader to better understand the nature of
the data, those items that failed the test are included in the table.