6570141443

they are smaller and are not systematic. In this case, instead of differences, we may talk about trends in leniency of raters, which are typical of particular OKE.

Another condusion from the analysis of the rater effect is the observed disruption of result distribution that occurs at the exam cut-off score. Since the exam cut-off score is used for certification, it is desirable to increase the reliability of assessment in its area. In the case of the Polish language exams, the cut-off score is unfortunately the source of the rater effect involving increased local leniency just below the pass grade threshold. It is worth noting that even if the effect of a local increase in reliability of assessment has a positive impact on the assessment of students whose papers are in the area of the exam cut-off score, a local increase in leniency is a negative factor in the context of the exam as a whole. Because of it, some papers are assessed in a systematically dif-ferent manner than others. This effect is much stronger for the Polish language Matura exam than for mathematics. There is also a statistically significant difference in rater leniency around the exam cut-off score among the regional examination boards.

Another issue analysed in this study is the comparability of assessment of handwritten and computer-typed essays. Based on the analysis of data from comparative studies of Matura essays assessment results in the original handwritten copy and computer-typed copy, it was observed that handwritten essays are assessed higher than copies typed on a Computer. The difference in the ave-rage grades at the Matura exam assessed on a five-point scalę is morę than 2 points and is statistically significant at the significance level of p = 0.01. This result is surprising for practitioners, although a similar result was obtained in a number of studies conducted throughout the world, comparing the quality of assessment of handwritten texts and their copies written using a word processor.

In the presented study, the link between assessment and the author's handwriting traits was also the subject of analysis. The results suggest that essays written by authors in poor handwriting are assessed lower compared to essays written in a sophisticated and moderately sophisticated handwriting. This difference is large and statistically significant. Links of handwriting legibility with a rater's essay assessment are moderate. It should be noted, however, that legibility and sophistication are highly correlated traits of handwriting. Most of the barely legible essays are written in poor handwriting. These results are consistent with the results of studies conducted in other countries.

In the case of the impact of variables describing individual characteristics of raters (recorded during the study) on the degree of precision and leniency of raters for both the Polish language and mathematics, only links specific for the particular subject were ob-served. A link with the index of attitude to work In teams during the assessment may be observed only in the case of leniency of the Polish language raters. Differences in the level of internal motivation and attitude to work in teams in specific REBs were observed.

The impact of time spent assessing essays and the order in which they were assessed on the number of scores obtained in the constructed responses was also examined. In the case of the order in which essays were assessed and time spent assessing them, these effects, although statistically significant, are dose to zero (for both Polish language and mathematics papers). These results suggest that the fatigue does not impair the rater's evaluation process.

H