A problem with standard errors estimated by many structural equation modeli
ng programs is described. In such programs, a parameter's standard error is
sensitive to how the model is identified (i.e., how scale is set). Alterna
tive but equivalent ways to identify a model may yield different standard e
rrors, and hence different Z tests for a parameter, even though the identif
ications produce the same overall model fit. This lack of invariance due to
model identification creates the possibility that different analysts may r
each different conclusions about a parameter's significance level even thou
gh they test equivalent models on the same data. The authors suggest that p
arameters be tested for statistical significance through the likelihood rat
io test, which is invariant to the identification choice.