Is C‐OAR‐SE best for internet retailing service quality?

This paper aims to respond to claims by Collier and Bienstock and Rossiter that reflective measurement is wrong for internet retailing service quality (IRSQ). The research empirically assesses Rossiter's proposal that the C‐OAR‐SE procedure for index development will generate a more valid way to measure IRSQ than is otherwise available.

C‐OAR‐SE is used to develop a formative IRSQ index. The index is administered to internet shoppers in an online survey. The index is compared with an existing IRSQ scale in terms of content, parsimony, measurement scores and criterion validity.

The scale and index display parity in content, parsimony and measurement scores, while the scale shows higher criterion validity. The results contradict Rossiter's claims and foster doubt regarding the usefulness of C‐OAR‐SE's formative measurement procedures.

IRSQ can be conceptualised as reflective or formative, but C‐OAR‐SE does not necessarily generate a better way to measure the construct. Furthermore, implementing C‐OAR‐SE unearths problems with the procedure.

Multiple variations of IRSQ exist, as well as multiple views on how to measure the variations and differing degrees to which the variations are actually measured. Crucially, the situation is not as bleak as Collier and Bienstock or Rossiter suggest: the literature does offer sound, valid IRSQ measurement scales.

The paper resolves unwarranted criticisms of IRSQ scales, highlights the limitations with some scales, offers the first complete example of using C‐OAR‐SE to develop a new index and lends applied support to theoretical criticisms of C‐OAR‐SE.