[AMOS] - measurement model not great - is this the end of SEM for my data?

#1
Hello there,

This is my first attempt at SEM and I got as far as running my measurement model, with mixed results regarding goodness of fit. I've "tweaked" the model about as much as I can (removed a few items, covaried some error terms etc). Some of my fit stats are good, some awful. I've seen so much literature arguing that certain stats are inconclusive in SEM - not sure if this set is sufficient to proceed or whether I have to give up on SEM at this point - some opinions would be great (with references if you can). Many thanks.

PS: I have run with and without bootstrapping and got the same results (there is a multivariate kurtosis problem).

Using ML estimation with and without bootstrapping - same in each case
Chi-square p=.000 (one attempt at bootstrapping got as high as .006)
CMIN/DF 3.9
CFI .963
GFI .918
RMSEA .074
PCLOSE .000
TLI .952

Using ADF estimation (n=530 - not really enough for ADF)
Chi-square p=.000 (one attempt at bootstrapping got as high as .006)
CMIN/DF 2.798
CFI .782
GFI .876
RMSEA .058
PCLOSE .052
TLI .713

Please gurus - can I continue with SEM (and if so how do I justify that well enough for publication?). I don't even know if I can use regression with a multivariate kurtosis problem - next step is to investigate that.

Thanks.