Date of Award

2008

Degree Type

Thesis

Degree Name

Bachelor of Arts Honours

School

School of Psychology and Social Sciences

Faculty

Faculty of Computing, Health and Science

First Advisor

Dr Ricks Allan

Second Advisor

Craig Harms

Abstract

Historically psychological scales have used a mix of positively keyed and negatively keyed items (balanced scales) to control for the effects of response sets. While it has been established that the use of balanced scales does effectively control for response sets such as acquiescence, issues relating to the psychometric properties of these scales emerge. The following review investigated issues surrounding the reliability, validity and factor structure of balanced scales by considering whether these issues were caused by positively and negatively keyed items measuring different aspects of a construct or whether they emerged simply due to measurement error. Both these positions are supported by research with various balanced scales, though it is necessary for future research to consider the effect that negative item framing, rather than negative item keying, has on the psychometric properties of balanced scales. The Life Orientation Test (LOT) and Life Orientation Test-Revised (LOT-R) were investigated to see how item framing influenced both scales factor structure. Two modified versions of both scales were created, one with consistent positive framing and the other with consistent negative framing. In both scales the original meaning (keying) of items was maintained so that each version had a balance of positively and negatively keyed items. Confirmatory factor analysis results indicated that a two-factor model was a significantly better fit to the data from the positively and negatively framed LOT and positively framed LOT-R. It was suggested that participants do not respond to item framing but instead item-keying direction when completing both scales. Furthermore participants responded differently to negatively framed items, perhaps due to the increased semantic complexity.

Share

 
COinS