Cluster analysis for test-retest Reliability

Conventional statistical model for assessing test-retest reliability of questionnaire is to compare composite scores of matched subtests or items between periods. Current study examined test-retest reliability in terms of both subtest and item wise comparison through paired t-test statistics. Data w...

Full description

Autores:
Dutta Roy, Debdulal
Tipo de recurso:
Fecha de publicación:
2010
Institución:
Universidad de San Buenaventura
Repositorio:
Repositorio USB
Idioma:
spa
OAI Identifier:
oai:bibliotecadigital.usb.edu.co:10819/6504
Acceso en línea:
http://hdl.handle.net/10819/6504
Palabra clave:
Paired t-statistic
Hierarchical clustering
Complete linkage
Dendograms
Reading motivation
Prueba t pareada
Análisis jerárquico de conglomerados
Agrupamiento por ligamiento completo
Dendogramas
Motivación para la lectura
Estadística
Motivación (psicología)
Investigación cuantitativa
Rights
License
Atribución-NoComercial-SinDerivadas 2.5 Colombia
Description
Summary:Conventional statistical model for assessing test-retest reliability of questionnaire is to compare composite scores of matched subtests or items between periods. Current study examined test-retest reliability in terms of both subtest and item wise comparison through paired t-test statistics. Data were collected from 72 students twice within 8 months intervals using reading motivation questionnaire (Dutta Roy, 2003). The questionnaire includes 3 intrinsic (rApp, rKnow,rAch) and 4 extrinsic (rAes, rRecog, rAff and rHarm) reading motivation subtests. Finding inconsistency between both subtest and item wise comparison, I propose a model of clustering items between periods. Hierarchical clustering with complete linkage shows that more than 55% of both test and retest items are assembled in the primary cluster suggesting high test retest reliability among 6 reading motivation variables.