The Effect on Test Reliability and Validity of Scoring Aptitude and Achievement Tests With Weights for Every Choice

Davis, Frederick B.; Fifer, Gordon

doi:10.1177/001316445901900202

Cited by 58 publications

(41 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Nas décadas de 50 e 60 vários autores investigaram o impacto da redução do número de itens na precisão e validade de alguns testes de inteligência geral e de aptidões cognitivas (Davis & Fifer, 1959;Merenda, Hall, Clarke & Pascale, 1962;Clarke, 1960). Em alguns dos estudos comparou-se os parâmetros psicométricos (precisão e validade) do Teste de Aptidão Diferencial (DAT, Bennett e cols.…”

Section: R Primi E L S Almeidaunclassified

Estudo de validação da bateria de provas de raciocínio (BPR-5)

Primi

Almeida

2000

Psic.: Teor. e Pesq.

View full text Add to dashboard Cite

Esse estudo investiga a validade e precisão da BPR-5, versão reduzida da Bateria de Provas de Raciocínio Diferencial (BPRD) de Almeida (1988), composta por cinco subtestes: Raciocínio Abstrato, Raciocínio Verbal, Raciocínio Numérico, Raciocínio Espacial e Raciocínio Mecânico. A BPR-5 foi aplicada em 1243 alunos brasileiros (N=771) e portugueses (N=472) da sexta série do ensino fundamental até a terceira série do ensino médio. Os coeficientes de consistência interna variaram de 0,62 a 0,84 e os de precisão pelo método das metades de 0,65 a 0,87. Encontrou-se um único fator explicando aproximadamente 55% da variância representando uma medida composta de inteligência fluida, cristalizada, processamento visual e habilidade quantitativa e conhecimento prático de mecânica. As correlações da BPR-5 com as notas escolares foram, no geral, positivas chegando a atingir 0,54 (p < 0,01). Conclui-se que a BPR-5 constitui um instrumento eficiente e rápido para avaliação simultânea do raciocínio geral e das aptidões podendo ser útil aos profissionais em suas diversas áreas de atuação.

show abstract

Section: R Primi E L S Almeidaunclassified

Estudo de validação da bateria de provas de raciocínio (BPR-5)

Primi

Almeida

2000

Psic.: Teor. e Pesq.

View full text Add to dashboard Cite

show abstract

“…Davis and Fifer (1959) noted that the traditional method of grading on the basis of correct answers only, or with correcting for guessing, penalizes the student who chooses incorrect responses that are very nearly correct compared to the student who chooses incorrect responses that show very little knowledge about the question.…”

Section: Discussionmentioning

confidence: 99%

Item analysis as a basis for weighting examination questions

Yeager

1994

Clinical Anatomy

View full text Add to dashboard Cite

Not all tost questions are of equal importance and, therefore, should not be weighted equally. To determine weight, a method has been used in which the weight of each question is based on what the class as a whole learned from what was presented in lectures, laboratories, and assigned reading material. Multiple choice written tests are machine‐graded and the percentage of students correctly answering each question is used as the weighting factor. Important basic material that is emphasized in lecture or that can readily be recognized by the students as important will be weighted high if most students learn it. Questions about small details that are not stressed will be weighted low if few students learn that detail. However, questions on poorly taught material will be weighted low if the students do riot learn it, even if material is important. Students who just memorize detail but do not understand what is important will not gain from this method. Since the methods weights the question based on what the class as a whole learned, the students think it is fair. Students are more willing to learn when they believe trey are being treated fairly. © 1994 Wiley‐Liss, Inc.

show abstract

“…For another thing, technical treatments were good ways to mine the potential information from distractors without constructing a new test. To assign different weights to options was one way (Davis & Fifer, 1959). To create an augmented data matrix transformed from a raw response matrix by using some special scoring rules was another (Luecht, 2007).…”

Section: Introductionmentioning

confidence: 99%

An investigation of enhancement of ability evaluation by using a nested logit model for multiple-choice items

2017

View full text Add to dashboard Cite

Título: Una investigación de la mejora de la capacidad de evaluación mediante el uso de un modelo logit anidado para items de elección múltiple. Resumen: Los items de elección múltiple se han usado ampliamente en tests psicológicos y educativos. Este estudio investiga si los items de elección múltiple tiene ventajas sobre los items dicotómicos o sobre la evaluación de rasgo latente. Un modelo de respuesta al item, con un modelo logit anidado, logístico 2-parámetros (2PL-NKM), fue usado para ajustar los datos de elección múltiple. Los estudios de simulación y empíricos indicaron que la precisión y la estabilidad de la estimación de capacidad mejoró usando el modelo de elección múltiple en contraposición al modelo dicotómico, debido a la mayor información incluida en los items distractores de la elección múltiple. Pero la precisión y la capacidad de estimación mostró pequeñas diferencias en items de cuatro elecciones, cinco y seis elecciones. Además, el modelo 2PL-NLM puede extraer más información respondientes de bajo nivel que de los de alto nivel, debido a que tienen conductas de elección con más distractores. En el estudio empírico, los respondientes en diferentes niveles de rasgo fueron atraídos por diferentes distractrores del Test de Vocabulario chino en el primer grado, usando trazos cambiantes en la probabilidad de distractor a partir de 2PL-NLM. Esto sugiere que las respuestas de los estudiantes a diferentes niveles puede reflejar un proceso evolutivo de vocabulario en los estudiantes. Palabras clave: items de elección múltiple; modelo logit anidado; información distractora; capacidad de evaluación.Abstract. Multiple-choice items are wildly used in psychological and educational test. The present study investigated that if a multiple-choice item have an advantage over a dichotomous item on ability or latent trait evaluation. An item response model, 2-parameter logistic nested logit model (2PL-NLM), was used to fit the multiple-choice data. Both simulation study and empirical study indicated that the accuracy and the stability of ability estimation were enhanced by using multiple-choice model rather than dichotomous model, because more information was included in multiple-choice items' distractors. But the accuracy of ability estimation showed little differences in four-choice items, five-choice items and sixchoice items. Moreover, 2PL-NLM could extract more information from low-level respondents than from high-level ones, because they had more distractor chosen behaviors. In the empirical study, respondents at different trait levels would be attracted by different distractors from the Chinese Vocabulary Test for Grade 1 by using the changing traces of distractor probabilities calculated from 2PL-NLM. It is suggested that the responses of students at different levels could reflect the students' vocabulary development process.

show abstract

The Effect on Test Reliability and Validity of Scoring Aptitude and Achievement Tests With Weights for Every Choice

Cited by 58 publications

References 9 publications

Estudo de validação da bateria de provas de raciocínio (BPR-5)

Estudo de validação da bateria de provas de raciocínio (BPR-5)

Item analysis as a basis for weighting examination questions

An investigation of enhancement of ability evaluation by using a nested logit model for multiple-choice items

Contact Info

Product

Resources

About