[Eric de Souza]

Overnight I remembered the following draft textbooks by Bruce Hansen of the University of Wisconsin-Madison that can be legally downloaded for free:
Introduction to Econometric and Econcometrics:
https://www.ssc.wisc.edu/~bhansen/

[Angelica Morgan]

This thread still needs inputs, with no limit on time. Many thanks.

Thank you very much indeed, Eric, for your continuous help. I will dive into these textbooks.

[Nick Cox]

You can start a thread but you can't control it.

This thread still needs inputs, with no limit on time. Many thanks.

You got some great and serious answers, but what is it that is still missing so far as you are concerned? I can't speak for any participants, but my not so wild guess is that they think they answered the question, exposing some confusions along the way. If you want more you please need to spell out what is still missing.

[Angelica Morgan]

Will the writing "This thread still needs inputs, with no limit on time. Many thanks." cause pressure on anyone? Will anyone feel compelled to reply? I thought this is an open forum, where everyone feels comfortable to attend to or ignore any post. So I don't think the writing would cause anyone troubles. Did I miss any implicit forum etiquette? If so, please let me know and I can apologize. With all respect, honestly, I feel scared to use the forum now.

I know I don't have control and I don't look for it. I just expressed my wish to hear more opinions from anyone who is freely willing to join the conversation. The discussion on delta method ended beautifully, with great contributions from Jeph, Leonardo, Eric, and Jeff. Now the discussion is about consistency of estimators, because some scholars voiced counterintuitive opinions. Both I and Eric tend to agree that if betahat is a consistent estimator of beta then 1/betahat is a consistent estimator of 1/beta. We have been trying to find proof. How can it bother anyone? With all respect, I am confused.

[Nick Cox]

The advice I've already given is to be more specific about what else you seek from the thread. I think you've now done something towards that end.

The point was merely that you repeated this line twice in 24 hours:

This thread still needs inputs, with no limit on time. Many thanks.

which wasn't intended or (I guess) taken as being rude, but it's not quite the best way to be clear on what else you want. There's nothing for you to apologise for, any more than I feel the need to apologise for giving frank advice in your own best interests.

[Angelica Morgan]

I thought it is clear to anyone who follows the thread to know what the current discussion is about. I kind of reiterated it once or twice already and thought it would actually bother people if I repeat it again.

Me too, I hope I did not accidentally offended any contributor (if any, most likely our super helpful Eric, who continued helping me while I continued search, ha ha). I did thank each of them along the way, and I hope they received my gratitude. When searching for truth, we need to get every stone turned. It can be irritating, as it is out of our comfort zone. But I think most of us scholars should have already been used to it.

I perceived your earlier writing differently. This is the shortcoming of writing, as sometimes it delivers what we don't intend to deliver. Now I know you meant to help, no need for you to apologize, tough love is still love. Thank you indeed. Taking your advice, in future, maybe I can emphasize that I appreciate contributions made and that everyone shall feel comfortable to join or ignore the thread while I continue welcoming not only different but also similar opinions (to some extent, volume is important in an open discussion).

[Nick Cox]

Thanks for #21, which I do appreciate. We do have guidance on bumping and so forth at #1 of https://www.statalist.org/forums/help#adviceextras which is (indirectly) relevant.

On a different level, it seems to me that the source used in #1 is not authoritative.

If this thread arises from an assignment (given the unsourced questions in #11) there are some awkward questions there for your teachers.

[Angelica Morgan]

I think what I posted earlier is not bumping in section 1.1 but keeping a thread going in section 1.4. For example, if Eric did not reply to me, I would not post "This thread still needs inputs, with no limit on time. Many thanks." I posted the message while replying to Eric to thank him, because I am afraid other readers would think the thread is closed. The message is to indicate that the thread is still open.

#1 has two questions. One about the delta method, which is solved. The other about consistency of estimators, which is ongoing. #11 and #6 are reiterations of #1, regarding the ongoing discussion.

The question in #11 is sourced. I provided two screenshots from Zelner(2009) and made the source clear. The full citation is given in #1 as shown below.
Zelner, B. A. (2009). Using simulation to interpret results from logit, probit, and other nonlinear models. Strategic Management Journal, 30(12), 1335-1348.

Despite the paper is published in a top journal, I agree with you that it is not authoritative, in the sense that publications in top journals can be wrong. As wrote earlier, both I and Eric (so far only him) tend to disagree with the paper on a particular point, so we have been trying to find proof, so that I would have more confidence in using commands such as margins and nlcom for nonlinear models.

To clarify, the question is not from an assignment. Seeking online help to do homework to earn credits would be unethical. In the sense that we are life-time learners, I am a student. But, to clarify here, I am an independent researcher. And I am always ready to acknowledge people's significant help, whether they are aware of it or not; and when it is time, I would do so in one way or another, with their consent if in need and if appropriate.

With that said, thank you for policing the forum Nick. We need that to ensure we have a professional virtual space.

[Maarten Buis]

This sentence irritated me to: I read ""This thread still needs inputs" as "I want more information, but I am not telling you what I am missing". A thread is never closed on Statalist, so you don't need to say that it is open. But if you want something more, then it is your responsibility to tell us as exactly as possible what is still missing. We want to help you, but if we don't know what you are missing, it obviously becomes impossible to give you want you want. You need to help us help you.

[Nick Cox]

In #11 there are 5 numbered questions. Are you saying that they are to be found in the Zelner article?

They look(ed) like assignments to me. This is just to make clear why I called them unsourced. I have no interest in (further) confrontation over this thread. I just think it was a fair guess to say what I did.

[Angelica Morgan]

It certainly did not occur to me that a wish to keep the post open can be irritating. Also, I thought the quest is very clear. If it is a taboo on this forum and if anyone else was accidentally offended, my apologies.

By the way, as wrote earlier, the 5 steps in #11 describe procedures to perform simulations. They are provided as supportive information to facilitate the relevant discussions. They are not assignment questions. As wrote earlier, Zelner (2009) writes that 1/betahat is not a consistent (in his or her writing a biased) estimator of 1/beta and proposes simulation as a remedy. I am not convinced by the remedy, and thus post simulation steps to give others more details. If you follow the thread, you will see it.

I don't want to continue this discussion too. It makes me feel very uncomfortable. Feels like being attacked as the kind of person I am not.

Honestly, I would not have courage to use this forum again. Thank you for your help to all contributors. And take care in this pandemic.

[Sebastian Kripfganz]

Eric already gave the answer:

The result that the inverse of a consistent estimator is also consistent (subject to regularity conditions) is an application of what is known as Slutsky's Theorem in probability and statistics.
On Edit: Jeff Wooldridge's book does have a chapter on short chapter on Asymptotic Theory and there is a discussion of Slutsky's Theorem.

[Dimitriy V. Masterov]

Is it wrong to say that the delta-method CIs constructed in this way cover the true value 95% of the time? If it's not wrong, then why worry if they contain impossible values? You could always just clip them.

[Stephen Jenkins]

Angelica Morgan Please continue to use the Forum. In my view, you asked interesting questions and they generated interesting responses related to the topic. I would even suggest that you take it as a compliment that several Statalist "gurus" got interested in the thread and contributed. (And despite their brusque tone, I believe they were simply being direct and 'to the point' as they saw things.) Your clarification in #21 was indeed useful, as was Nick's in #22.

[Angelica Morgan]

Sebastian, thank you for echoing.
Dimitriy, thank you for looking at the issue in a different way.
Stephen, thank you very much for your good intentions.

Take care,
Angelica