I am testing the fit of my mediation model in AMOS, and the model fit statistics are very strange. For example, the CFI is .000 and the RMSEA is way too big (14.774). It seems as though there has been some error with the data or the calculation, but there is no indication of what this might be. What do these strange resutls usually suggest?
One possible explanation is that you have constrained a parameter to a particular value that should not be constrained or you have applied constraints that are grossly inaccurate (.e.g., constraining error variances and error regression coefficients to inappropriate values).