You can find related literature in the SEM - or classical test theory - side of things. One of the concerns with longitudinal models (from the CTT perspective) is measurement invariance; in other words, while you may have developed a measurement tool that measured what you wanted during the first administration it might not measure the same latent over time. On the IRT side, these types of discussions happen regularly in the context of parameter drift (although there are other circumstances other than time that could lead to parameter drift). Todd Little has a well written book on longitudinal SEM if you need a place to start.
↧