JediKnight5

Hi @jhelvy, as you have preferred, I am continuing the discussion here instead of following up in the Stackexchange thread. Regarding your last comments in the Stackexchange thread: - All...

@jhelvy: Regarding your Stata question: Yes, your assumption is correct. I exported the data set I generated within my R code to a csv file (per iteration of the Monte-Carlo...

I had a look at the codes of ``logitr`` and ``mlogit`` to better understand how to specify the respective ``standardDraws`` and the ``random.nb`` options. My understanding is that ``logitr`` is...

I actually cannot see how your described procedure of drawing standard normal draws for fixed coefficients and then replacing them with repeated means should adversely affect the estimation. After having...

> Okay, I think we may be getting down to something here. First, the `makePartials` function is essentially storing a pre-computed part of the gradients that doesn't change across iterations...

> Btw, I built this all based on Train's book. You can see from section 6.6 of [chapter 6](https://eml.berkeley.edu/books/choice2nd/Ch06_p134-150.pdf) where he describes the estimation process. In 6.7 on panel data...

First of all, sorry for the delayed response. Unfortunately, I have been super busy in recent days. I used 100 draws in the simulation above. To double check the finding...

> Based on these latest results, I'm thinking we may have a convergence issue. When you say "This displays which estimator achieves the maximum likelihood in the 100 iterations", what...

Hi @jhelvy , sorry, I have been very busy in the last two weeks. I'll definitely reply to your most recent post, hopefully in the next couple of days.

Hi @jhelvy, my apologies again, I have been very busy in the last two weeks. Regarding this post: > Okay, I've gone back and attempted to simulate my own experiment...