To Parcel or Not to Parcel: Exploring the Question, Weighing the Merits

TLDR

The decision to parcel items in SEM depends on philosophical stance and study goals, such as understanding item structure or construct nature. The study examines the controversial use of item parcels as manifest variables in SEM and recommends that investigators first assess item dimensionality before parceling. The authors advise first evaluating item dimensionality, then applying various parceling techniques—each described with strengths and weaknesses—to minimize pitfalls and optimize construct measurement. They conclude that unconsidered parcel use is never warranted, while considered parcel use remains defensible.

Abstract

Abstract We examine the controversial practice of using parcels of items as manifest variables in structural equation modeling (SEM) procedures. After detailing arguments pro and con, we conclude that the unconsidered use of parcels is never warranted, while, at the same time, the considered use of parcels cannot be dismissed out of hand. In large part, the decision to parcel or not depends on one's philosophical stance regarding scientific inquiry (e.g., empiricist vs. pragmatist) and the substantive goal of a study (e.g., to understand the structure of a set of items or to examine the nature of a set of constructs). Prior to creating parcels, however, we recommend strongly that investigators acquire a thorough understanding of the nature and dimensionality of the items to be parceled. With this knowledge in hand, various techniques for creating parcels can be utilized to minimize potential pitfalls and to optimize the measurement structure of constructs in SEM procedures. A number of parceling techniques are described, noting their strengths and weaknesses.