Subjectivity Problem in Student Assessment: Theoretical and Practical Aspects
##plugins.themes.bootstrap3.article.main##
Abstract
Objective evaluation is not an easy task when assessing student attainment even if we are evaluating such a well-measured subject as knowledge of mathematics. This subject is discussed in a number of papers. Their authors are of the same opinion that even when the evaluation criteria is matching in detail, different teachers evaluate the same work differently. Subjectivity in knowledge assessment is researched in this paper. Not only teachers’’ individual characteristics in assessing student attainment are analyzed, but also problems of designing objective evaluation criteria of evaluated object, for example, solved problem. Authors analyze their own works and researches of other scientists in this field. The paper describes the authors’ original experiments and the results of their statistical analysis. As the solution of the evaluation subjectivity problem authors propose a methodology which allows dividing students to groups according to their attainment level. This methodology could be applied not only for evaluation of mathematical knowledge, but also for attainment evaluation in other disciplines.
The experiment, when 15 accidentally selected student’s works in higher mathematics were independently verified by 6 teachers, is described in the paper. The analysis of the total test result and the results of individual tasks was performed with nonparametric statistical methods. To check the compatibility of teachers assessment Kendall’s coefficient of concordance W was calculated, nonparametric Friedman’s test was applied. The conclusion of the investigation was that all 6 teachers’ estimates were rather similar when evaluating students’ attainment except the estimation of the fifth task. Page’s L test was applied to the total test result to determine trend and consequently norm-referenced estimate of tested students. Least significant differences between sums of neighbouring ranks were calculated to establish 5 groups in which students received the same marks, so the criterion-referenced evaluation of student attainment was performed. The first task is not suitable for testing of this group of testees, because it is too easy for them. Tasks that are too easy or, conversely, too hard, are not suitable, because the results of such assignments are well predicted and do not distinguish between students.
The fifth task should be recognized as inappropriate, because it does not affect the total assessment score and the criterion-referenced evaluation of the total test score is performed based on only the first four tasks.
The applied nonparametric statistical methods were described in the paper in detail. The results were summarised in inferences and recommendations. The recommendations to escape or reduce the influence of subjectivity in attainment evaluation were formulated:
• to assign more points to the task, where teachers disagree in their assessment of students knowledge,
• to use standard tests for attainment evaluation,
• to formulate more strict evaluation criteria and agreement on the most often occurring mistakes of evaluation.
The proposed methods could be applied not only for attainment in mathematical assessment, but also in other natural sciences.
The experiment, when 15 accidentally selected student’s works in higher mathematics were independently verified by 6 teachers, is described in the paper. The analysis of the total test result and the results of individual tasks was performed with nonparametric statistical methods. To check the compatibility of teachers assessment Kendall’s coefficient of concordance W was calculated, nonparametric Friedman’s test was applied. The conclusion of the investigation was that all 6 teachers’ estimates were rather similar when evaluating students’ attainment except the estimation of the fifth task. Page’s L test was applied to the total test result to determine trend and consequently norm-referenced estimate of tested students. Least significant differences between sums of neighbouring ranks were calculated to establish 5 groups in which students received the same marks, so the criterion-referenced evaluation of student attainment was performed. The first task is not suitable for testing of this group of testees, because it is too easy for them. Tasks that are too easy or, conversely, too hard, are not suitable, because the results of such assignments are well predicted and do not distinguish between students.
The fifth task should be recognized as inappropriate, because it does not affect the total assessment score and the criterion-referenced evaluation of the total test score is performed based on only the first four tasks.
The applied nonparametric statistical methods were described in the paper in detail. The results were summarised in inferences and recommendations. The recommendations to escape or reduce the influence of subjectivity in attainment evaluation were formulated:
• to assign more points to the task, where teachers disagree in their assessment of students knowledge,
• to use standard tests for attainment evaluation,
• to formulate more strict evaluation criteria and agreement on the most often occurring mistakes of evaluation.
The proposed methods could be applied not only for attainment in mathematical assessment, but also in other natural sciences.
##plugins.themes.bootstrap3.article.details##
Section
Articles
- The Author grants to the Publisher the exclusive right and licence to publish this Article without remuneration until the expiry of the economic rights: to reproduce the article in print and digital form, including its publication; to disseminate the original version of the Article or its copies in Lithuania and foreign countries; to translate the Article; to publish the article, including making it publicly available via computer networks; to reproduce and publish the Article in Lithuanian and foreign databases; to licence usage of the Article in ways described in this paragraph.
- The Author warrants that the economic rights of the Author have not been assigned or granted to third parties, that the Article has not been published so far and is not under consideration of being published elsewhere.
- The Author warrants that the Article does not violate intellectual property rights of others.
- After the Article is published in Social Technologies the Author shall have a right to make it public on personal website or on a website of an institution of affiliation, to reproduce it for non-commercial teaching or scholarly research purposes, clearly indicating that the primary source of its publication is Social Technologies.
- This consent shall be considered invalid if the Editorial Board of the Social Technologies decides not to publish the Article.
Authors contributing to Social Technologies agree to publish their articles under a Creative Commons Attribution 3.0 Unported (CC BY 3.0) Licence, allowing third parties to share their work (copy, distribute, transmit) and to adapt it, under the condition that the authors are given credit, and that in the event of reuse or distribution, the terms of this licence are made clear.
