TABLE 6-3: PATTERN MATRIX WITH VARIMAX ROTATION OF STRATEGIC LEADERSHIP
From the factor pattern matrix it can be concluded that the data shows a four-factor structure, i.e. 1)
Business development, 2) Operational efficiency 3) Client centricity, 4) Organizational creativity. One
LWHP ³FRQVFLRXVO\PDNHVroom for creativLW\ ́ ORDGVRQWZRIDFWRUVKRZHYHUWKHKLJKHVWORDGLQJLV
on the factor on which it was supposed to load following the item generation method, i.e.
Organizational creativity.
6.3.3 AGGREGATION, INTERNAL RELIABILITY AND CORRELATIONS
Before internal reliability of the four dimensions and the intercorrelations between these four
dimensions were calculated, the scores were aggregated for individuals. That is, different raters scored
one and the same individual. In order to obtain single ratings for the focal manager, the scores were
averaged across raters (of a focal managers). Prior to aggregating rater' scores, Interrater agreement,
Rwg (James, 1982; James et al., 1984) was assessed using the multiple-item estimator for within-
group agreement (Rwg) as proposed by James et al. (1984) and group reliability (ICC(1) and ICC(2))^4
(Bliese, 2000) (see Table 6.4).
(^4) The formulae for ICC(1) and ICC(2) are based on Bartko, J. J. 1976. On various intraclass correlation reliability coefficients. Psychological Bulletin,
83(5): 762-765, Newman, D. A. & Sin, H. P. 2009. How Do Missing Data Bias Estimates of Within-Group Agreement? Sensitivity of SD WG, CVWG,
rWG (J), rWG (J)*, and ICC to Systematic Nonresponse. Organizational Research Methods, 12(1): 113-147. ICC(1) = one-way ANOVA intraclass
correlation and is interpreted as the proportion of total variance that can be explained by group membership. 1-ICC(1) 0 is interpreted as the %age of
variance due to the disagreement among raters. ICC(2) = intraclass correlation for the reliability of average ratings. For example, if another random
sample of raters rate the same subjects, ICC(2) is approximately the correlation between the averaged ratings from the two sets of raters. High values
indicate that aggregate measures of a group-level construct are reasonably free of idiosyncratic rater error (have relatively low within-groups variance).