Expert versus statistical scales in the diagnosis of adolescent psychopathology with the behavior assessment system for children Self-Report of Personality-adolescent /

This study compares two methods of constructing scales from a

Bibliographic Details
Main Author: Stanton, Harrison Chester, 1956-
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1995.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=742164131&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD
Description
Summary:This study compares two methods of constructing scales from a
common item pool; one method based on clinical judgment and
the other based on a strictly statistical approach. A
convenience sample of fifteen licensed, clinical
psychologists were asked to sort the items of the Self-Report
of Personality for Adolescents (SRP-A) into discrete
categories representing various forms of adolescent
psychopathology. The clinicians were asked to "free sort"
the items - meaning that no restrictions were placed on the
number or kind of categories they could create. The method
of Latent partition Analysis (LPA) was used to code the
clinician's sortings into a joint proportion matrix. The
joint proportion matrix was factor analyzed and the results
of this analysis were used to develop a set of expert scales.
The expert scales were evaluated on the basis of their
ability to add new, non-redundant assessment information.
The discriminant power of the statistical and expert scales
was evaluated by comparing how previously formed clinical
samples performed on the scales. Correlations were computed
between the expert and statistical scales to determine scale
overlap and whether the expert scales added non-redundant
assessment information. The results of the comparison of the
expert and statistical scales were both surprising and
compelling. The clinical and statistical approach to scale
construction produced nearly identical scales with nearly
equal ability to discriminate between clinical groups. The
results of this study suggest that clinical and statistical
models of prediction may share more in common than was
previously thought and that a union between the two methods
might promote the construction of tests that are both
clinically and statistically sound.
Item Description:Vita.
"Major Subject: School Psychology".
Physical Description:ix, 109 leaves ; 28 cm.
Issued also on microfiche from University Microfilms Inc.
Bibliography:Includes bibliographical references.