… We propose a nested em routine which guarantees monotone log-likelihood sequences and
improved convergence rates in maximum likelihood estimation of latent class models with
covariates. … The latter consider, instead, a multi-step routine which first estimates π and s from
a latent class model without covariates, and then uses the predicted classes s ˆ as responses
in ℓ 1 ( β ; s ˆ , x ) – or a modification of it – to estimate β . … In particular, for each iteration t ,
we consider R ∗ = R − 1 nested cycles which sequentially improve the conditional expected …