within the relevant cutoff criteria. Both of the other scales had significant Chi-Square
values, which some authors would argue renders their factor models invalid (e.g. Hayduk,
Cummings, Boadu, Pazderka-Robinson, & Boulianne, 2007). However, when the other fit
indices were considered, only RMSEA was unsatisfactory for Interpersonal Facilitation,
but Job Dedication had unsatisfactory fit as assessed by both CFI and RMSEA, indicating
that it did not have simple structure. This may seem to be inconsistent with the high alpha
value for this scale, but this incongruity is apparent only — alpha does not directly reflect
the factorial integrity of a scale (Miller, 1995). Examination of modification indices for the
Job Dedication model showed that four pairs of error terms needed to be allowed to
correlate before an acceptable level of fit was achieved, suggesting that the scale may be
measuring more than one construct. The results of these CFAs show that the unifactorial
Citizenship Performance scale had the best fit to the data gathered in Study 2 suggesting
that it had a more valid internal structure.
Discussion
Study 2 provided further evidence for both the validity and the reliability of the
unifactorial Citizenship Performance scale. The unifactorial scale was strongly correlated
with the other two measures of Citizenship Performance and was also correlated with
other aspects of performance in a manner consistent with theoretical expectations,
further confirming the validity of the scale. When compared with the other Citizenship
Performance scales used in Study 2, it also had the best psychometric qualities having
demonstrated the best factor structure fit and the best convergent and divergent validity.
GENERAL DISCUSSION
This article described the development and validation of a unifactorial scale for
measuring Citizenship Performance. This scale appears to be the first unifactorial
Citizenship Performance scale to have been successfully tested with CFA using
contemporary criteria. The relationship of the unifactorial Citizenship Performance
scale with other variables was consistent with theoretical expectations as well as
previous empirical findings, providing both convergent and divergent validation. Finally,
although the unifactorial Citizenship Performance scale was shown to be highly
correlated with previously developed measures of Citizenship Performance, it had a
better internal structure and external validity than these other scales. Therefore, the
unifactorial Citizenship Performance scale appears to be a valid measure of the latent
Citizenship Performance construct.
One of the major justifications for the idea that Citizenship is a single latent construct
is the fact that most correlations with other variables do not vary between Citizenship
Table 4. Confirmatory factor analysis of measures used in Study 2a