Ergül DEMİR
In this study, it is aimed to examine the relationships between items and test parameters estimated using different missing data methods in multiple choice tests, in the presence of missing data, and to determine the lost data methods that are suitable for use in such tests. In this study, which was designed as a research in the basic research type, in the relational screening model, the analyzes were conducted on the data of the SBS (Placement Exam) 2011 Mathematics Test A Booklet for 527517 respondents. The 'index deletion method' from the methods based on deletion in data analysis, '0 assignment', 'series means assignment', 'observation unit mean assignment', 'near points mean assignment', 'near points median assignment', 'linear interpolation' and 'linear trend point assignment' methods, 12 different loss data methods were used, namely 'regression assignment', 'expectation-maximization algorithm' and 'data replication' methods among the most likelihood methods, and 'Markov chains Monte Carlo' method among multiple data assignment methods. The findings show that if the missing data are not negligible, it is necessary to use an appropriate loss data method in statistical estimation for multiple choice tests. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data. Twelve different loss data methods were used, including 'expectation-maximization algorithm' and 'data replication' methods, and 'Markov chains Monte Carlo' method among multiple data assignment methods. The findings show that if the missing data are not negligible, it is necessary to use an appropriate loss data method in statistical estimation for multiple choice tests. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data. Twelve different loss data methods were used, including 'expectation-maximization algorithm' and 'data replication' methods, and 'Markov chains Monte Carlo' method among multiple data assignment methods. The findings show that if the missing data are not negligible, it is necessary to use an appropriate loss data method in statistical estimation for multiple choice tests. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data. Twelve different lost data methods are used, one of the multiple data assignment methods, the 'Markov chains Monte Carlo' method. The findings show that if the missing data are not negligible, it is necessary to use an appropriate loss data method in statistical estimation for multiple choice tests. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data. Twelve different lost data methods are used, one of the multiple data assignment methods, the 'Markov chains Monte Carlo' method. The findings show that if the missing data are not negligible, it is necessary to use an appropriate loss data method in statistical estimation for multiple choice tests. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data. shows that an appropriate loss data method should be used in statistical estimates for multiple choice tests in case the missing data are not negligible. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data. shows that an appropriate loss data method should be used in statistical estimates for multiple choice tests in case the missing data are not negligible. Deletion-based methods and the Assign 0 method are not suitable methods for this type of data. Simple assignment methods are likely to produce biased estimates. The most likelihood and multiple data assignment methods are considered to be the most suitable loss data methods to be used in such data.
