點數對評定量表信度與效度影響之模擬研究（I）
Date Issued
2001
Date
2001
Author(s)
DOI
892413H002048
Abstract
The Monte Carlo method was proposed
to examine the factors that influence the
reliability of Likert-type rating scales. The
independent variables manipulated included
number of response categories, inter-item
covariance, number of items, respondents’
judgment accuracy (or reliability of
individual item), and item score distributions
of the rating scales. Dependent variables
included three estimates of test reliability:
the internal consistency reliabilityα , the
test-retest reliability, and the squared
correlation between true scores and observed
scores. The research is the first part of a
two-year project. Results indicated that
internal consistency reliabilityαwas mainly
affected by inter-item covariance and
number of items. Respondent judgment
accuracy and number of items affected
test-retest reliability. The squared correlation
between true scores and observed scores was
affected by judgment accuracy, number of
items, number of response categories, and
item score distributions of the rating scales.
It is anticipated that the results of the
simulation study will assist developers of
Likert-type rating scales in choosing
appropriate design to achieve optimal
reliability for measurement in their research.
Subjects
rating scales
Likert-type scales
number of response categories
item distributions
test-retest reliability
internal consistency reliability
Publisher
臺北市：國立臺灣大學心理學系暨研究所
Type
report
File(s)![Thumbnail Image]()
Loading...
Name
892413H002048.pdf
Size
43 KB
Format
Adobe PDF
Checksum
(MD5):268e83e5815aa5e9073414ad884011f5