The effect of nonlinear transformations on a Likert scale.

Area of Science:

• Psychometrics
• Behavioral Science
• Statistical Analysis

Background:

• Behavioral assessment instruments often use rating scales.
• Parametric statistical methods are commonly applied to these scales.
• The nature of the scale's intervals and distribution normality are often overlooked.

Purpose of the Study:

• To investigate the appropriateness of parametric statistics for behaviorally-anchored rating scales.
• To determine the impact of nonlinear transformations on the factorial structure of such scales.

Main Methods:

• Utilized data from a clinical performance evaluation instrument with a four-point behaviorally-anchored scale.
• Applied various nonlinear transformations to the collected evaluation responses.
• Analyzed the factorial structure of the fifteen items on the evaluation form before and after transformations.

Main Results:

• The factorial structure of the evaluation items remained largely unaffected by nonlinear transformations.
• This suggests that the underlying structure of the behaviorally-anchored rating scale is robust.

Conclusions:

• Parametric statistical analyses are suitable for data derived from behaviorally-anchored rating scales.
• The robustness of the scale's factorial structure supports the application of parametric methods even with potential non-linear relationships.