In a reflective measurement model, each indicator has some level of unexplained variance (error). Normally, it is drawn as a circle pointing to the observed indicator. Using Smart PLS, do I have to put the error variable in my model, or does Smart PLS assume it’s existence?