Sample Size Requirement for Index Construction (Measurement Model)
Posted: Fri Jun 07, 2019 9:15 am
My aim is to construct a formative measure (index) on the basis of the following hierarchical component model (HCM): one second-order construct (SOC), four first-order constructs (FOCs), and a total of 23 indicators (grouped into the 4 FOCs). (There is no structural model.)
I am using the repeated indicator approach to model the SOC and follow the assessment procedure of formative measurement models by Hair, Hult, Ringle, & Sarstedt (2017) on the first and second order (collinearity assessment etc.).
My question concerns the minimum required sample size. If I calculate this model, the R2 of the SOC results by definition in a value of 1. When Hair et al. (2017) address sample size requirements, they refer to "minimum R2 values [...] in any of the endogenous constructs in the structural model..." (p. 20).
Does this mean that I necessarily need a structural model to determine the minimum required sample size? Or do I go about assessing sample size in a measurement model like mine? (If I consider R2 = 1, required sample size becomes virtually 0.)
Thanks!
I am using the repeated indicator approach to model the SOC and follow the assessment procedure of formative measurement models by Hair, Hult, Ringle, & Sarstedt (2017) on the first and second order (collinearity assessment etc.).
My question concerns the minimum required sample size. If I calculate this model, the R2 of the SOC results by definition in a value of 1. When Hair et al. (2017) address sample size requirements, they refer to "minimum R2 values [...] in any of the endogenous constructs in the structural model..." (p. 20).
Does this mean that I necessarily need a structural model to determine the minimum required sample size? Or do I go about assessing sample size in a measurement model like mine? (If I consider R2 = 1, required sample size becomes virtually 0.)
Thanks!