Issues and Strategies for Reducing the Length of Self‐report Scales

Abstract

Greater understanding of the complex interrelationships among workrelevant constructs has increased the number of constructs on organizational surveys. Good psychometric practice also dictates the use of multiple items per construct. The net result has been longer surveys. Longer surveys take more time to complete, tend to have more missing data, and have higher refusal rates than short surveys. Arguably, then, techniques for reducing the length of scales while maintaining psychometric quality are worthwhile. Little guidance exists on how to reduce the length of a multi-item scale and we argue that the most common technique, maximizing internal consistency, is problematic and should be avoided. We present a set of item "quality indices" to help conceptualize the competing issues that influence item retention decisions. Statistical analysis of an example case using these indices suggested that there are 3 key aspects of item quality to consider when reducing a scale. We describe strategies that can assist scale developers in using these 3 aspects of item quality when making scale reduction decisions.

Eco-Friendly Brands to Drive Sustainable Development: Replication and Extension of the Brand Experience Scale in a Cross-National Context

“…The negative items were not applicable for the BXS in this case, and thus they were excluded from the revised version of the model. According to the psychometric literature, negatively worded items can be excluded from scales when the content is already covered by affirmative statements [142]. It has also been argued that alternating positively and negatively worded items could distort the respondents' response styles [136,143,144].…”

“…Most brand measurement scales do not have negative items [14]. Concise measurement models, such as the revised BXS model, have been called for because they can be applied and extended more easily than longer and more complicated models [142,145].…”

Guidelines for choosing between multi-item and single-item scales for construct measurement: a predictive validity perspective

“…Given that, in practical applications one needs to select a "good" item before data collection, and given that one would not normally collect data on the other items comprising the scale, it is important to know ex ante whether the selected SI represents a good choice or not. A comparison of the relative effectiveness of alternative item selection methods therefore constitutes an important avenue for future research; the literature on scale reduction (e.g., Moore et al 2002;Smith and McCarthy 1995;Stanton et al 2002) may provide a useful starting point for such research endeavors.…”

A Single-Item Measure Approach to Consumer-Based Brand Equity Based on Evolutionary Psychology and Neuroscience

“…If someone wants to measure a CBBE index in an industry with five competitive brands and he or she draws on the OBE model with the four indicator questions, then the number of questions is four times higher than that of the suggested BE_AB model. This overload of questions is likely to lead to problematic responses while at the same time decreasing the validity of the results (Dolbier et al, 2005;Stanton, Sinar, Balzer, & Smith, 2002). If this issue is ignored, the probability of the results being arbitrary is high despite the use of validated measurement models.…”