In a structural equation model in which a dichotomous variable is a mediatör, can i use bootsraping ML? ( I use AMOS) (Dichotomous variable will be represented as a single indicator latent in the model)
as far as I know, you cannot do that in AMOS. Two options:
a) use lavaan/R and use the DWLS estimator. This results in a latent continuous response variable that is connected with the binary observed variable
b) use base R and conduct two models: The x-->m logistic regression and the m-->y regression using regular OLS. Then use the mediation package an the equally names mediation function to combine both and get an indirect effect. I recently found the latter quite attractive as it is very flexible. A disadvantage is that you have to find other ways, to get a model fit.
Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2010). Causal mediation analysis using R. In D. V. Hrishikesh (Ed.), Advances in social science research using R (pp. 129-154): Springer.
I can send you a pdf of the chapter if you send me a mail/pn.