The purpose of the present investigation was to provide a clearer understanding of how various levels of perceptual correspondence are associated with one another and with performance appraisal. Specifically, the present study focused on the variable of accuracy in an attempt to discover its relative importance to performance appraisal. Also of interest was the relationship between accuracy and agreement.

The purpose of the present study was to determine what, if any, relationship existed between the correspondence of perceptions between superior-subordinate work dyads and the superior's rating of the subordinate's work performance.

A research report submitted to the Faculty of Commerce, Law and Management, University of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Management (in the field of Public and Development Management) 31 March 2014 / Non Profit Organisations (NPOs) are not exempt from the demands of employee attraction, retention and motivation. As NPOs seek to sustain themselves, the need to manage the performance of employees will continue to be a critical human resource management issue. Performance Appraisals (PAs) are used as tools that help manage the performance of employees; however, there is little understanding by those who participate in their use in NPOs. The purpose of this research is to explore how PAs are used at the Centre for Education Policy Development (CEPD) and how the employees perceive their use. Using qualitative research methods, primary data was collected through interviews and document analysis. The main findings of the research were that, the CEPD was unclear about its objectives for conducting PAs due to poor implementation of performance management systems and that, employees’ perceive the performance appraisal process as ineffective and irrelevant. There are serious managerial implications for defining the process of conducting appraisals and these include; training, selection of appropriate tools and clarifying the roles and responsibilities of each stakeholder in the process.

This study tests whether criteria variance in five job-facet satisfactions and performance rating explained by selected demographic and personality variables is significantly greater than by the selected demographic variables. This study offers a useful procedure for treating job-satisfaction data and predictor variables and measures the significance in predictability between criteria variance accounted for by variables not generally found in personnel files and those usually found there. It was found that job satisfactions and performance rating differ both in terms of predictability and functional relationships with Predictor Sets.

Prior research has shown that accountability influences decision-making processes (e.g., Tetlock, 1985a). Tetlock's work has found when accountable decision-makers know the view of the person to whom they are accountable, decision-makers will shift their decision towards that known view. In addition, these decision-makers may more frequently rationalize a previous decision. Accountable decision-makers who do not know the view of the person to whom they are accountable have been found to exhibit vigilant and complex cognitive processing. The goal of this study was to apply Tetlock's model of accountability to performance evaluations. Undergraduate raters in accountability conditions expected to meet with one of their actual professors to justify performance ratings. To further investigate the nature of accountability, this study contrasted with Tetlock's work in utilizing two different types of accountability and a decision shift control group. Results of this decision shift control group raised questions about Tetlock’s conclusions about accountability's effects on decision shifts. In addition, reward aspects of accountability may subsume Tetlock's regular operationalization of interpersonal accountability. Overall, raters appeared to focus on the possibility of giving unfavorable ratings. This focus may be of primary importance to accountability effects in performance evaluation. Suggestions for future research and the generalizability of these processes were discussed. / Ph. D.

Previous research exploring the assumptions of process invariance (Walker, 1989) found discrepancies between process and rating outcomes when rater and ratee populations were crossed. Divergent results were attributed to differences in the ratee stimulus performance tapes. The present study attempted to explore how levels of task complexity would moderate the relationship between job knowledge and both information processing and rating accuracy. 123 male subjects were measured on their knowledge of football, and viewed the performance of either offensive tackles (complex task) or running backs (simple task) under directions to either form an impression or remember as much detail as possible. It was expected that observational purpose would moderate the type of information recalled, rating accuracy, and accuracy in recording behavior frequencies in the complex task condition only. Results supported hypotheses only for Cronbaoh-. (19S5) elevation measure pertaining to behavior frequency ratings, and consistent task complexity main effects among accuracy measures for behavior frequency ratings. Some possible limitations and explanations for the present results and some implications for future research are offered. / Ph. D.

Because the act of supervisory feedback can critically affect a subordinate's performance, it is imperative to explicate the various conditions in which the character of feedback is determined. The purpose of the following research was to model the conditions under which supervisors adopt a negative feedback bias (NFB). This bias was first described by Kahneman and Tversky (1973), when they noted that Israeli flight instructors found that praise of exceptionally good piloting was often followed by poorer performance, while criticism of exceptional poor flying was usually followed by improved performance. Thus, the flight instructors came to believe that negative feedback motivated people effectively, while positive feedback appeared ineffective. Of course, supervisors had erred by failing to recognize the natural variation of their students' performance. In general, this error applies primarily to the conditions under which supervisors acquire and interpret information. Two factors were hypothesized as responsible for the emergence of a NFB: (a) limitations caused by opportunity-based supervision, where only a certain amount of subordinate behavior can be sampled at any given moment, and (b) supervisors find it difficult to recognize the natural variation, random fluctuations, and regression to mean processes characteristic of performance governed by common causes (cf. Deming, 1982; Hogarth, 1980 and Kahneman & Tversky, 1973). Results indicated that NFB was an emergent process occurring over time and under conditions where (a) supervisors managed highly inconsistent subordinate performance and (b) supervisors had limited information regarding a subordinate's performance per evaluation episode. Since this experimental approach and set-up is relatively novel, the results are discussed from several conceptual perspectives. Finally, a discussion regarding the ecological approach to feedback research, and the importance of model building and testing is offered. / Ph. D.

This field study investigated the relationship among behavioral and perceptual measures of the Leader-Member Exchange (LMX) relationship and organizational outcome variables. The major purposes of this study were to determine: (1) the extent to which the dyad is the appropriate level of analysis for the examination of manager-subordinate relationships, (2) which objectively observed managerr behaviors are important in defining negotiating latitude, the most commonly used measure of LMX, and (3) whether these behaviors or traditional self-reports of negotiating latitude better predict organizational outcome variables. Two hundred ninety-two manager-subordinate dyads from 18 work groups in a manufacturing plant were observed by trained research assistants and completed questionnaires. Measures included observed manager behaviors defined by the Operant Supervisory Taxonomy and Index (OSTI), subordinates' absence rates, and managers' and subordinates' perceptions of subordinates' satisfaction (general satisfaction, satisfaction with supervisor, growth satisfaction, work satisfaction, coworker satisfaction), subordinate's intent to withdraw, and negotiating latitude (NL). Within and between analysis I (WABA I) indicated the manager-subordinate dyad as the appropriate level of analysis for NL, general satisfaction, satisfaction with supervisor, growth satisfaction, work satisfaction, coworker satisfaction, and turnover intent. WABA II revealed significant dyadic relationships between NL and all measures of satisfaction as well as turnover intent. Bivariate regression analysis revealed a significant relationship between negotiating latitude and absenteeism. Multiple regression analysis showed manager behaviors (i.e., antecedents) predicted negotiating latitude. However, multiple regression failed to reveal that manager behaviors predicted measures of satisfaction, turnover intent, or absenteeism. Finally, hierarchical regression revealed that manager behaviors (i.e., consequences) added to the predictive ability of negotiating latitude for general satisfaction. However, the addition of manager behaviors to negotiating latitude using hierarchical regression failed to produce a significant change in R-square for any other outcome variables. / Ph. D.

This research attempts to identify factors which may influence instrumentality beliefs in a merit pay environment. Specifically, the purpose was to investigate the relationship of: (1) of perceived performance appraisal accuracy, merit increase satisfaction, trust in top management, and trust in supervisor to instrumentality beliefs, (2) of performance appraisals, deviations between self and supervisor ratings of performance, and trust in supervisor to perceived performance appraisal accuracy, and (3) merit pay increases and perceived relative size of merit pay increase to merit increase satisfaction. The research site for this study was a large transit authority on the West Coast. Results of this study of 1,260 managerial, professional, and clerical employees suggest that individuals are more likely to believe that pay is tied to performance if they are satisfied with their merit pay increase and they trust top management. Findings also suggest that for individuals who receive a lower than expected performance appraisal, higher levels of perceived performance appraisal accuracy are likely to be associated with stronger beliefs that pay is tied to performance. Findings further suggest that individuals are more likely to perceive their performance appraisal to be accurate if the supervisor rating of performance is consistent with their self-rating of performance and they trust their supervisor. Also, the higher the performance appraisal, the more likely an individual is to perceive it to be accurate. Finally, individuals are more likely to be satisfied with their merit pay increase if they perceive it to be large relative to the average pay increase. In addition, the larger the merit pay increase, the more likely an individual is to be satisfied with it. Study findings in total suggest that a large number of employees are likely to believe that pay is not tied to performance. To the extent successful merit pay programs require strong beliefs that pay is tied to performance, findings cast doubt about the ability of merit pay to elicit improved job performance. Implications for compensation practice are discussed and suggestions for future research are presented. / Ph. D.

Recent work in performance appraisal has focused on the cognitive operations involved in a rating task, but has neglected the influence of affect. The purpose of the present study was to examine the process by which affect influences evaluations and to examine the impact of affeet on outcome variables. A categorization-based model of the rating process was developed that included the influence of affect. According to the model, affect was postulated to influence the categorization process at the superordinate level. The model also postulated that the classification would then influence the processing of subsequent information. To test predictions derived from the model, the present study simulated a work situation. Subjects were required to learn materials, train confederates, and then observe and evaluate confederate's videotaped performance. The impact of affect and item type was examined on process measures and the impact of affect and performance was examined on outcome measures. The results showed that affect biased leader evaluations in the form of leniency and severity. Signal detection analysis indicated that response bias may have been the cognitive mechanism underlying this effect. Leaders attributed non-performed behaviors to confederates in a category consistent manner. Path analysis showed that affect has direct and indirect effects on performance evaluations. The theoretical and applied considerations of the study's findings are discussed and future research directions are highlighted. / Ph. D.