英语论文网

p; 4.4 Exploration of systematic rater bias patterns......... 47-49
Chapter 5 Discussion .........49-56
    5.1 Discussion on research questions......... 49-53
        5.1.1 Discussion on reliability of two rating sessions......... 49-50
        5.1.2 Discussion on rater-category bias interactions......... 50-52
        5.1.3 Discussion on rater-writer bias interactions......... 52
        5.1.4 Discussion on systematic bias patterns......... 52-53
    5.2 Discussion on severity and difficulty via mean scores......... 53-56
        5.2.1 Discussion on the mean scores assigned to writers 53
        5.2.2 Discussion on the mean scores assigned by raters 53-54
        5.2.3 Discussion on the mean scores assigned ......... 54-56

Conclusion

The research also has some limitations as follows.First, the CEPT test is used in Hunan University, and the writing tasks in one testare not the same, but theoretically equated by experienced teachers and experts.Moreover, the topics in the test are all argumentative in nature, whether similar resultswill be obtained in rating other rhetorical types needs further research.Second, the present study is a quantitative study, and no supportive qualitativestudy was conducted. Part of the reasons in the rater bias might be explained bythink-aloud and questionnaire statistics.Third, the raters may vary in experience, though this reflects the realisticsituation.It also should be noted that since this study was initially set for the CEPT test ofHunan University, and the experiment design was also trying to resemble the realisticsituation so as to produce more realistic and more useful data evidence, hence thestructure of the raters and the writing task differences were in one way not limitations.So the generalizability of the present study may be limited to a small range.
Based on the findings and limitations of the present study, further research isexpected concerning the following aspects.First, in connection with rating scales, the score steps distribution deserves furtherstudy, in order to improve the score span and provide effective and distinctive scoreinformation.Second, since raters tended to be over-fitting to the IRT model, the attempt todiversify their ratings is desired, maybe related to the modification of the rating scaleincluding the description of the different categories and the above mentioned score Based on the findings and limitations of the present study, further research isexpected concerning the following aspects.First, in connection with rating scales, the score steps distribution deserves furtherstudy, in order to improve the score span and provide effective and distinctive scoreinformation.Second, since raters tended to be over-fitting to the IRT model, the attempt todiversify their ratings is desired, maybe related to the modification of the rating scaleincluding the description of the different categories and the above mentioned score

References
[1] Alderson, C.2005. Diagnosing foreign language www.51lunwen.org/dxyylw/ proficiency. The interface betweenlearning and asse本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。