The direction of the response scale matters – accounting for the unit of measurement

Purpose
Psychometric analyses of self-administered questionnaire data tend to focus on items and instruments as a whole. The purpose of this paper is to investigate the functioning of the response scale and its impact on measurement precision. In terms of the response scale direction, existing evidence is mixed and
inconclusive.
Design/methodology/approach
Three experiments are conducted to examine the functioning of
response scales of different direction, ranging from agree to disagree versus from disagree to agree. The response scale direction effect is exemplified by two different latent constructs by applying the Rasch model
for measurement.
Findings
The agree-to-disagree format generally performs better than the disagree-to-agree variant with spatial proximity between the statement and the agree-pole of the scale appearing to drive the effect. The difference is essentially related to the unit of measurement.
Research limitations/implications
A careful investigation of the functioning of the response scale
should be part of every psychometric assessment. The framework of Rasch measurement theory offers unique opportunities in this regard.
Practical implications
Besides content, validity and reliability, academics and practitioners utilising
published measurement instruments are advised to consider any evidence on the response scale functioning
that is available.
Originality/value
The study exemplifies the application of the Rasch model to assess measurement
precision as a function of the design of the response scale. The methodology raises the awareness for the unit of measurement, which typically remains hidden.