Self versus assessor ratings and their classification in assessment centres: profiling the self-rater

Citation metadata

From: New Zealand Journal of Psychology(Vol. 36, Issue 2)
Publisher: New Zealand Psychological Society
Document Type: Article
Length: 5,270 words
Lexile Measure: 1360L

Document controls

Main content

Article Preview :

Self versus assessor behavioural ratings from 214 participants were investigated along with psychometric measures in a development centre. Using cluster analysis, it was shown that sub-groups of self-raters could be established without the need to use commonly criticized difference scores. Four clusters of self-rater were identified in this analysis, a result that builds on popular theory (Atwater & Yammarino, 1992). Self-ratings by cluster were plotted with psychometric measures and assessor ratings. Those who self-rated higher tended to score relatively high on personality constructs related to social dominance. However, sub-clusters of self-rater were neither differentiated by cognitive ability scores nor by assessor ratings. Strength of difference between self and assessor ratings was contingent on cluster membership. Cluster analysis is suggested as a useful technique for understanding self-rater behaviour and as a guide to identifying how feedback should be tailored to potential recipients.


The subject of self-ratings in assessment centres (ACs) has been of growing interest since Heneman (1980) suggested that ACs are ideal settings for their study. Particularly with regard to developmental assessment centres (i.e., development centres, DCs), self-ratings present an issue of importance to both participants and assessors. Halman and Fletcher (2000) emphasized that participants in development centres require feedback for the developmental event to operate effectively. This feedback is generally intended to increase a participant's awareness of performance-related competencies and later job performance (Van Velsor, Taylor, & Leslie, 1993).

Several theoretical arguments have been presented with respect to self-ratings. Notably, impression management may affect and inflate self-ratings, particularly in selection scenarios (Fahr & Werbel, 1986). However, even data from a DC, where impression management concerns might be less salient, tended to show an inflated bias in post-assessment self-ratings (Halman & Fletcher, 2000). Perhaps, then, inflated self-ratings tend to reflect a self-serving bias, as described in social psychology, whereby protection of self-esteem is sought.

Randall, Ferguson, and Patterson (2000) point out that the influence of impression management in self-ratings lead them to be unreliable as selection measures. However, it is possible that certain personality attributes could also contribute to patterns in self-perceptions of performance (Church, 1997; Fletcher, Taylor, & Glanfield, 1996). Islam and Zyphur (2005) found that individuals with a higher social dominance orientation tended to voice their opinions more openly in group situations, like those often presented in ACs.

A body of literature to date has focused on self versus assessor discrepancies on performance dimensions. Schmitt, Ford, and Stults (1986) studied eight performance dimensions prior and subsequent to the assessment event. They found small correlations between performance on each dimension and assessor judgments (<.20). Similar results were reported by Clapham (1998), who also investigated the moderating effects of cognitive abilities and gender. Again, and even with external variables controlled, small correlations were observed (< .33). Note, however, that an earlier study by Byham (1971) found higher correlations between assessor and post AC participant performance (around .60).

Participants who overestimate (or, at least, self-rate higher than assessors) their performance may, potentially, feel discouraged by assessor ratings. It appears to be...

Source Citation

Source Citation   

Gale Document Number: GALE|A169382758