Abstract:
This research was the mixed method which divided into 3 phases. Of the first phase which was a qualitative research, the objectives were to study the types and causes of errors of performance appraisal of raters. The 12 key informants comprised of raters, ratees and personnel who involves with managing performance appraisal system in state enterprises. The interview technique was used to gather data. The results showed there were 3 errors in performance appraisal: leniency, severity and recency and also found that 15, 6 and 1 causes of such errors respectively. Of the second phase which was the quantitative research, the objectives were 1) to identify each raters who had leniency, severity or did not have any errors by using the Guilford's method 2) to compare job performance categorized with lenient raters and severe raters and 3) to discriminate raters to be a group which had leniency error and a group which had severity error with 4 factors: job satisfaction of raters, rater- ratee acquaintance, rater-ratee similarity, liking to ratees. The sample group was 80 raters at state enterprises. The research instrument was questionnaires. The results showed 43 raters (53.75%) had leniency error and 37 raters (46.25%) had severity error. Regarding the job performance, the lenient raters gave significantly higher scores (x ̅ = 4.41) to the ratees than the severe raters (x ̅= 3.41). The significance level established for the statistical analysis was .01. The 4 factors can discriminate raters with a statistical significance at .01 level (Chi-square = 19.710, Df. = 4, sig, = .001). The raw score equation could be showed as follows: D' = -10.205 + 2.187 (job satisfaction of raters) + 0.039 (rater - ratee acquaintance) - 2.365 (rater rate similarity) + 1.080 (liking to ratees). Of the third phase which was a qualitative research, the objective was to seek the ways which should be used to solve the leniency error problem. The Delphi technique was used to gather data. The sample group was 26 experts in performance appraisal and the related fields. The median and interquartile range and mean were used to analyze data. The results showed 61 ways which reached the expert consensus should be used to solve the leniency error problem in performance appraisal.