Comparison Of The Reliability Co-Efficients  Obtained According To Classical Test Theory  And  Factor Analytic  Approach

Ece Bülbül; Önder Sünbül

Makale detayları

Comparison Of The Reliability Co-Efficients Obtained According To Classical Test Theory And Factor Analytic Approach

Yazarlar:

Ece Bülbül | Önder Sünbül

Yayıncı:

Mersin University

Ana Başlıklar:

Eğitimde ve Psikolojide Ölçme Teorileri ve Uygulamaları

Dergi:

Mersin Üniversitesi Eğitim Fakültesi Dergisi Vol.20 Issue 1, pp.112-127

Belge Tipi:

Makale

Yayın Tarihi:

4/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001315442

pISSN:

1306-7850

Anahtar kelimeler:

Özet:

 The purpose of this study is the examination of the omega reliability coefficients used in KR-20 and factor analytic approach under the factors affecting reliability such as test length, sample size and factor loadings variance in parallel and tau-equivalent measurement structures.Data has been generated according to various factors sample size, test length and factor loadings levels. 100 replications have been performed for each state in the generated data. The data have been generated seperately in parallel and tau-equivalent measurement structures in compliance with each state consisted as a result of crossing the designated conditions. The data production has been carried out by using the codes written by the researcher and the psych package in R 3.0 program. The average and standart deviation values of KR-20 and omega coefficients have been calculated according to different alternation criterias for parallel and tau-equivalent measurement structures.No significant effect of sample size by itself on coefficient average and standart deviation in parallel and tau-equivalent measurement structures has been seen. However, it has been observed that the material number and factor loading levels have caused significant changes on the coefficient average and standart deviation. The omega and KR-20 coefficient averages obtained in different material numbers, different sample sizes and different factor loadings in parallel measurement structure has been, contrary to expectations, lower compared to the ones obtained in tau-equivalent and congeneric measurement structures.

 The purpose of this study is the examination of the omega reliability coefficients used in KR-20 and factor analytic approach under the factors affecting reliability such as test length, sample size and factor loadings variance in parallel and tau-equivalent measurement structures.Data has been generated according to various factors sample size, test length and factor loadings levels. 100 replications have been performed for each state in the generated data. The data have been generated seperately in parallel and tau-equivalent measurement structures in compliance with each state consisted as a result of crossing the designated conditions. The data production has been carried out by using the codes written by the researcher and the psych package in R 3.0 program. The average and standart deviation values of KR-20 and omega coefficients have been calculated according to different alternation criterias for parallel and tau-equivalent measurement structures.No significant effect of sample size by itself on coefficient average and standart deviation in parallel and tau-equivalent measurement structures has been seen. However, it has been observed that the material number and factor loading levels have caused significant changes on the coefficient average and standart deviation. The omega and KR-20 coefficient averages obtained in different material numbers, different sample sizes and different factor loadings in parallel measurement structure has been, contrary to expectations, lower compared to the ones obtained in tau-equivalent and congeneric measurement structures.

 The purpose of this study is the examination of the omega reliability coefficients used in KR-20 and factor analytic approach under the factors affecting reliability such as test length, sample size and factor loadings variance in parallel and tau-equivalent measurement structures.Data has been generated according to various factors sample size, test length and factor loadings levels. 100 replications have been performed for each state in the generated data. The data have been generated seperately in parallel and tau-equivalent measurement structures in compliance with each state consisted as a result of crossing the designated conditions. The data production has been carried out by using the codes written by the researcher and the psych package in R 3.0 program. The average and standart deviation values of KR-20 and omega coefficients have been calculated according to different alternation criterias for parallel and tau-equivalent measurement structures.No significant effect of sample size by itself on coefficient average and standart deviation in parallel and tau-equivalent measurement structures has been seen. However, it has been observed that the material number and factor loading levels have caused significant changes on the coefficient average and standart deviation. The omega and KR-20 coefficient averages obtained in different material numbers, different sample sizes and different factor loadings in parallel measurement structure has been, contrary to expectations, lower compared to the ones obtained in tau-equivalent and congeneric measurement structures.

 Bu çalışmada paralel ve eş biçimli ölçme yapılarında güvenirliği etkileyen faktörlerden, test uzunluğu, örneklem büyüklüğü ve faktör yükü dağılımları etkisi altında KR-20 ile faktör analitik yaklaşımda kullanılan omega katsayısının incelenmesi amaçlanmıştır. Çeşitli faktörlere örneklem büyüklüğü, test uzunluğu ve faktör yük düzeyleri göre veri üretilip, üretilen verilerde her bir durum için 100 yineleme replikasyon yapılmıştır. Veriler, paralel ve eşbiçimli ölçme yapılarında ayrı ayrı olarak, belirtilen koşulların çaprazlanması sonucu oluşan her bir duruma uygun olarak üretilmiştir. Veri üretimi R 3.0 programında psych paketi ve araştırmacı tarafından yazılan kodlar kullanılarak gerçekleştirilmiştir. Paralel ve eşbiçimliölçme yapıları için farklı değişimleme ölçütlerine göre KR-20 ve Omega katsayılarının ortalama ve standart sapma değerleri hesaplanmıştır. Farklı ölçme yapılarında tek başına örneklem büyüklüğünün, katsayı ortalamaları ve standart sapmaları üzerinde anlamlı bir etkisi görülmemiştir. Ancak madde sayısı ve faktör yükü düzeylerinin katsayı ortalamaları ve standart sapmaları üzerinde anlamlı değişikliklere neden olduğu gözlenmiştir. Paralel ölçme yapısında farklı madde sayılarında, örneklem büyüklüklerinde ve farklı faktör yüklerinde elde edilen Omega ve KR-20 katsayı ortalamaları, eşbiçimli ve konjenerik ölçme yapılarında elde edilen katsayı ortalamalarına göre, beklenenin aksine, daha düşük çıkmıştır.

Makale detayları

Comparison Of The Reliability Co-Efficients Obtained According To Classical Test Theory And Factor Analytic Approach

Yazarlar:

Ece Bülbül | Önder Sünbül

Yayıncı:

Mersin University

Ana Başlıklar:

Eğitimde ve Psikolojide Ölçme Teorileri ve Uygulamaları

Dergi:

Mersin Üniversitesi Eğitim Fakültesi Dergisi Vol.20 Issue 1, pp.112-127

Belge Tipi:

Makale

Yayın Tarihi:

4/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001315442

pISSN:

1306-7850

Parallel Measurement Structure

Tau-Equivalent Measurement Structure

Omega Coefficient

Reliability.Paralel Ölçme Yapısı

Eşbiçimli Ölçme Yapısı

Omega Katsayısı

Güvenirlik

Sayı

Makale detayları

Yazarlar: Ece Bülbül | Önder Sünbül

Yayıncı: Mersin University

Ana Başlıklar: Eğitimde ve Psikolojide Ölçme Teorileri ve Uygulamaları

Dergi: Mersin Üniversitesi Eğitim Fakültesi Dergisi Vol.20 Issue 1, pp.112-127

Belge Tipi: Makale

Yayın Tarihi: 4/2024

Belge Dili: Türkçe

Document ID (Org No.): 2001315442

pISSN: 1306-7850

Sayı

Yazarlar:

Ece Bülbül | Önder Sünbül

Yayıncı:

Mersin University

Ana Başlıklar:

Eğitimde ve Psikolojide Ölçme Teorileri ve Uygulamaları

Dergi:

Mersin Üniversitesi Eğitim Fakültesi Dergisi Vol.20 Issue 1, pp.112-127

Belge Tipi:

Makale

Yayın Tarihi:

4/2024

Belge Dili:

Türkçe

Document ID (Org No.):

2001315442

pISSN:

1306-7850