Evaluation and Research in Education
Editor: Professor Keith Morrison, Inter-University Institute of Macau
Associate Editor: Professor Stephen Gorard, University of York
Statistical Adviser: Professor Colin Baker, University of Wales Bangor
Reviews Editor: Dr. Emma Smith, University of York
|
Volume: 15 Number: 2 Page: 84–94
|
|
|
|
|
An Examination of the Validity of Positive and Negative Items on a Single-scale Instrument
|
|
G.R. Schott and W. Bellin
|
|
It has become common practice for test constructors to balance the use of positively and negatively worded item statements within rating scales in order to disrupt possible response sets. This practice
is based upon an assumption that positively and negatively worded items function in the same way in their contribution to the composition of scale constructs. This paper describes a strategy developed to
account for the impact of item presentation upon ensuing constructs, during the development of a self-report measure termed the Relational Self-Concept Scale. The salience of item content over item presentation
was determined by administering two versions of the scale to a sample of adolescent school pupils (n = 978). Each version of the scale contained of a balanced mixture of positive and negative items, identical
item content, but differed in terms of item presentation. Items positively worded on the one version were negatively worded on the other version, and vice versa. Factor analyses conducted on both versions
of the scale confirmed six consistent sub-scales that were determined by item content rather than item presentation. The benefits of developing two versions of the scale are summarised with reference to
protection against presentational bias and the process of item selection and sub-scale composition.
© Multilingual Matters 2001
Full text PDF
|
