 |
PSYCHOMETRICS
PSICOMETRIA PSICOMETRÍA Prof. Dr. J. L. Melià |
|
|
PSYCHOMETRICS
PSICOMETRIA PSICOMETRÍA Prof. Dr. J. L. Melià |
Libros / Books
Resumenes / Abstracts
Introducción
a la medición y análisis de datos /
An Introduction to measurement and Data Analysis
Josep Lluís Melià
Este texto presenta una introducción al concepto de medición
en psicología en relación con el contexto metodológico
de diseño y análisis de datos en que cobra sentido e importancia
para el profesional y el investigador. Se presenta desde una perspectiva
integradora las diversas concepciones y niveles de medición en psicología,
los conceptos de diseño experimental y correlacional y el análisis
de datos. Entre las aportaciones originales, el texto introduce el concepto
de error de datos en análisis de datos, los tipos de error de datos,
los modos de enfrentamiento del error de datos, incluida la prevención
depuración y estimación del mismo.
An
Introduction to Measurement and Data Analysis
Josep Lluís Melià
This book presents an integrated view of the main concepts of the principal
areas in the methodology of psychology, including measurement, design and
data analysis.
Usually students of psychology learn about data analysis, psychometrics
or research design in a non-integrated method. Consequently, some of them
find it difficult to see what the place of methodology is in the professional
work of the psychologists. In “An Introduction to Measurement and Data
Analysis” an integrated view of the methodology research and professional
work is emphasized.
The book presents an ordered view of the stages in data analysis and
introduces the concept of data error. Introduced are classes of data error
and ways to cope with it, as well as statistics and procedures to estimate
the remaining data error.
La Construcción
de la Psicometría como ciencia teórica y aplicada /
The Construction of Psychometrics as a theoretical and Applied
Science
Josep Lluís Melià Disponible
"on line"
Este manual presenta una visión propia, integrada y estructurada
del campo de la Psicometría, sus relaciones con otras disciplinas
metodológicas, con la Psicología y con otras ciencias afines,
y presenta una aproximación a la historia de la disciplina alejada
de los tópicos habituales.
A través de la presentación de la estructura del campo
psicométrico y de los principales hitos de su historia se van introduciendo
los conceptos básicos en cada una de sus ramas de modo que el texto
constituye, a su vez, una primera introducción a la psicometría
desde una perspectiva sistemática, histórica y conceptual.
The
Construction of Psychometrics as a theoretical and Applied Science
Josep Lluís Melià Available
"on line"
This book presents an integrated and structured view of the field of
Psychometrics as a science, its main concepts, and the relationship to
psychology and other adjacent sciences, as well as an approximation to
its history.
Métodos de Escalamiento Unidimensional /
Unidimensional Scaling Methods
Josep Lluís Melià
Este es un manual para estudiantes y profesionales de Psicología
y Ciencias Sociales interesados en aprender como se construyen escalas
de medida. El texto presenta los métodos clásicos de escalamiento,
los más utilizados y aquellas aportaciones recientes que pueden
resultar más prácticas y útiles en el trabajo profesional.
La presentación es extraordinariamente didáctica y se enmarca
en una visión integrada, analítica, del papel y las fases
de los métodos de escalamiento que produce una nueva comprensión
estructurada del campo que permite no solo aprender sino entender el significado
de los distintos métodos y el modo en que sus fases pueden ser adaptadas
a las necesidades profesionales específicas.
El libro permite una comprensión útil y profunda de los
métodos que permite al lector adaptarlos a sus propias necesidades.
Se introduce un nuevo método de escalamiento unidimensional
que permite escalar items tipo Likert e items dicotómicos en un
solo paso sin utilizar tareas de juicio.
Unidimensional Scaling Methods
Josep Lluís Melià
This is a handbook for students and professionals of psychology and
social sciences who are interested in learning how to construct a scale
of measurement
The book presents the classical methods (Thurstone, Likert, Guttman…)
as well as more recent methods that are useful for professional work in
schools, attitudes and behavior surveys, marketing research, political
research and other subjects.
The book facilitates a sound and useful understanding of the scaling
methods, which allows the reader to adapt methods to his/her own needs.
Also introduced is a new unidimensional scaling method, which permits
scaling Likert and dichotomous scales in one step, without use judgment
tasks.
Apuntes sobre
teoría clásica de tests /
Notes on Classical Test Theory
Josep Lluís Melià
Se trata de un manual para estudiantes y profesionales interesados en
conocer los principales conceptos y métodos prácticos en
teórica clásica de test, aquella que ha dado lugar a la gran
mayoría los tests y otros instrumentos de medida utilizados en psicología
y otras ciencias afines.
El libro combina una exposición muy didáctica y sistemática,
donde toda la teoría aparece como consecuencia de una elección
inicial, con otra reflexiva preocupada por desvelar supuestos, limitaciones
y contradicciones.
Para el especialista la principal aportación del libro en una
nueva visión crítica que cuestiona muchos de los aspectos
supuestamente bien establecidos y comúnmente trasmitidos de manual
en manual sin una reflexión crítica. Se ofrece así
una mirada distinta, inquisitiva, inquieta y escéptica de la teoría
clásica.
Notes on Classical Test
Theory
Josep Lluís Melià
This is a handbook for students and professionals who need to understand
and apply the methods of test construction, reliability and validity, developed
in Classical Test Theory (CTT).
CTT has provided methods and resources to construct and develop all
the most important psychometric tests in intelligence, abilities, personality,
attitudes and others. Nowadays, CTT is still useful for at least three
reasons. Firstly it helps define the basic problems of reliability, validity
and test construction. Secondly, it is necessary to interpret the majority
of the results and research about tests. Finally, it is helpful and unavoidable
to analyze small samples and to introduce the analysis of big samples with
other more sophisticated psychometric models.
The book combines a very clear and systematic exposition, where all
of the CTT is presented as a consequence of a simple original decision,
and a sound exposition showing assumptions, limitations and contradictions.
For the specialist, the principle contribution of the work is a systematic
critique of the Classical Test Theory, the introduction of some new formulaes
and a restrictive definition of validity. The book offers a different view,
inquisitive, reflective, and skeptical, of the CTT.
Teoría de la Fiabilidad
y la validez /
Theory of Reliability and
Validity.
Josep Lluís Meliá (2000)
Valencia. Ed. Cristobal
Serrano
ISBN: 84-89596-73-5
DL: V-931-2000
Páginas/ Pages: 482.
Available "on line" - Disponible
"on line -
Psychometrics has been subject to rapid change in the last few decades. Classical Test Theory (CTT) has been substituted by Item Response Theory (IRT) for the most part of research and major institutional testing programs. Computerized Adaptive Testing has increased its presence and many new models and computer programs allow new applications and theoretical developments.
Nevertheless, CTT remains important for four reasons:
This book presents a didactic new look to the CTT combining
the
most important classical concepts and an unusual critical view of foundations
and procedures. Theory of Reliability and Validity introduces also new
concepts, formulas and procedures.
The book is divided into three parts. The first and second chapters introduce the concepts of psychometrics and psychological measurement. The central chapters, chapters 3 to 10, develop in a systematic and pedagogical way the main concepts, procedures and results of CTT. The final two chapters are devoted to the validity theory.
The firsts chapters discuss the traditional idea of psychometrics and introduce a new view based on a psychological modeling standpoint. The traditional conception of psychometrics needs to be changed to adjust the new developments and challenges. The traditional transverse concept of psychometrics is substituted in a new matrix concept. This matrix concept sustains the mathematical modeling of human behavior as the main objective of psychometrics. The new matrix concept integrates psychometrics in psychology, subordinates assumptions to the object, and brings us a structured 4 x 4 bidimensional vision of psychometrics crossing four defined levels and four content areas.
The central chapters help the reader to get a structured view of CTT contributions. All CTT is presented as a logical answer to the ontological dilemma born from the conjunction of the entity axiom and the empirical fact of find several measurement results for the same entity. The style is primarily based on introducing a question an only when the reader needs a solution, one or more answers. Solutions commonly accepted as obvious reveal their hidden assumptions and some traditional misunderstandings around reliability and its estimation are dispelled.
Chapter six is devoted to a criticism of the classical theory of reliability. The criticism is structured in four main areas:
1) The design used to estimate the reliability coefficient,
2) the non-contrastability of parallel measurements,
3) the problems of the reliability coefficient with linear change, and
4) the limitations of the congeneric tests in the estimation of the reliability coefficient.
1) All the classical procedures to estimate the reliability coefficient
have the same design based on a first and second measurement. This design
used to estimate the reliability coefficient is analyzed from the point
of view of its internal validity. Factors that affect the reliability coefficient
and factors that can introduce some changes in the second measurement are
described. This analysis shows that spurious effects are mixed inseparably
from the effects of the error of measurement. The design used to estimate
the reliability coefficient is unable to establish the main sources of
variation that affects the second measurement, and therefore, is a pre-experimental
design without any internal validity.
2) The second main criticism is about the concept of parallel measurements. This concept is accepted as the cornerstone of the estimation of reliability. The text shows how the concept of parallel measurements was constructed precisely to allow the estimation of the unobservable terms of CTT. But the concept of parallel measurement is itself unobservable. The book reviews the several procedures that have been proposed to contrast the observable consequences of the parallel measurement definition, and concludes that parallel measurements remain still as an uncontrastable concept.
3) The third problem is related to the inability of the reliability coefficient to detect linear change. If two measures are parallel then their correlation is the coefficient of reliability, even if this correlation is close to zero. But if two measures are not parallel their correlation is not the coefficient of reliability, even if this correlation equals the unity. Defined as a correlation, the coefficient of reliability can not show the effect of linear changes. Any change between measurements that can be described by a linear function will be ignored by the reliability coefficient, but if there is a linear change we can not estimate a reliability coefficient.
4) In the early seventies, Jöreskog introduced a new procedure
to estimate reliability coefficients based on a structural equation method
applied over congeneric tests. If some tests are parallel, then the Jöreskog
procedure estimates the same reliability coefficients that CTT postulates.
Nevertheless, the Jöreskog procedure does not support the classical
methods to estimate a reliability coefficient. Moreover, in the Jöreskog
procedure, the estimation of reliability coefficients depends on a latent
trait, but the definition of the latent trait depends on the election of
the tests included in the procedure.
The book discusses also the relationship between reliability and
precision and shows how they are different concepts and, furthermore,
how the CTT idea and estimation of the reliability coefficient can be opposed
to the estimation of the precision of a measurement.
The central chapters also introduce some new formulas and deductions. For example, chapter seven introduces new formulas for the standard error of measurement. For example is introduced a simple formula for the standard error of measurement that has the same assumptions as the classical Spearman-Brown prophecy, and can be used in the same situations and for the same purposes.
Chapter nine discusses the problem of interpreting and comparing punctuations in an unusual integrated way. This chapter covers the problem of the normative population, the problem of the meaning of punctuations and introduces a new general procedure to contrast differences between individual punctuations, including a Fisher like procedure to find the small significant difference.
May be the most important contribution of the book is the concept of
restrictive
validity, introduced in chapter eleven.
The restrictive validity of a test is based on the quality of the meaning
of its punctuations. The meaning of the punctuations is decided by some
empirical procedures avoiding the traditional use of unobservable latent
concepts.
Chapter index:
1. Psychometrics
2. Psychological measurement
3. Error of measurement
4. Parallel measurements
5. The estimation of the reliability coefficient
6. A criticism of the classical theory of reliability
7. Internal consistency
8. True scores
9. Comparing scores
10. Item analysis
11. Validity
12 The estimation of validity