I just came to post the same question. You beat me by a few minutes. I think we should reverse this, because I don't think Trump is actively claiming 6-18 deaths, but that the official (September) toll of 2,975 is grossly exaggerated.

I changed it. I hope you don't mind, but I think it is only fair to put the claimant's position in the most generous light. Under this wording, if it turns out that there were, say, 100 deaths, Trump would be substantially right, not wrong.

Is the claim that about 3000 died while Maria was going on or that 3000 died due to Maria?

I believe the latter, but if the official death toll uses unusual standards that increases the count far more than is expected (compared to similar natural diasters), I think that would make an excellent answer. [To be clear: I have no reason to believe that is the case.]

@ff524 I think that should be an important part of any good answer to this question. My understanding is that the true immediate death toll was in fact very high (morgues overflowing, etc), but they remained "officially" uncounted due to general breakdown of infrastructure and people just trying to survive. Surely other deaths followed in short order (injuries, illness, no power to run medical equipment), some of those should clearly be counted, others may be more debatable.

@Oddthinking yes, we cant determine the motivations of those who calculated the official toll, but determining what metrics where used to create the final count would certainly clarify the reality of the situation. Someone here is either being grossly misleading or is terribly mistaken, and I wouldn't be surprised either way. Love the fact that this site exists.

One question to ask is if a death that is caused by a lack of power (critical medical equipment unable to run) and the power is lose due to the hurricane does that count as a death due to the hurricane or not. I am sure there where plenty of deaths due to medical devices failing or medicine spoiling due to lack of refrigeration and those deaths would continue to happen well after the hurricane passed.

As with almost every claim that "Trump is being dishonest" or "Trump is wrong" etc...it all comes down to what definition one chooses to apply to what Trump said as to what the correct answer will be. Nobody knows what definition Trump means when he talks about "people that died in the hurricane". Because of that, this question is meaningless.

@Dunk It's unequivocally false. Either he means in the storm, in which case he is misrepresenting the studies which say 3000, or he means because of the storm in which case his claimed low numbers are wrong. Either way, the statement is wrong. Trump is, intentionally or not, conflating two things to make a political statement.

@Dunk The issue with that reasoning is that it makes all questions meaningless if you don't know the specific definition the speaker was going for (which is almost always the case). To illustrate: "As with almost every claim about X, it all comes down to what definition one chooses to apply to what the claimant said as to what the correct answer will be. Nobody knows what definition the claimant means when he talks about X. Because of that, this question is meaningless." The way around this is to assign some responsibility to the claimant for badly-constructed claims.

@Oddthinking NPR and PBS had a recording of Trump talking about 6 - 18 people dead at a time when there was 45 recorded deaths, though it is actually possible he made an honest mistake there.

en.wikipedia.org/wiki/… seems highly relevant to @Oddthinking's point. Note that the death tolls for other disasters - like, say, Hurricane Katrina, or the Peshtigo fire - consist of direct fatalities only. An answer with deeper knowledge and research could flesh this out more than I can, but at a glance, the approach of designating the "death toll" of a disaster based upon an "excess mortality" calculation doesn't seem to have been used prior to Maria.

@MarkAmery death tolls for other hurricanes DO include indirect deaths. The death toll from the wikipedia list you linked references the wikipedia article on hurricane Katrina, which says many of the deaths are indirect

@DeNovo Yeah, I realised that after my comment; the ~1200 deaths estimate for Katrina is made up of "direct" fatalities only while the ~1800 estimate includes "indirect" fatalities. I don't know what exactly went into that "indirect" estimate or whether it can be reasonably used in a like-for-like with the alleged 3000 "excess mortality" from Maria.

@MarkAmery how very refreshing :) Thanks for admitting your realization. Re: excess mortality, it's a common statistic used to attribute deaths to rare events, and solve the challenging problem of interpreting death certificates. An individual who reads a death certificate and attempts to code whether the death would have occurred without the storm is subject to a great deal of variability, especially when there is limited information on the ground.

@MarkAmery how do you know, for example, that an electrocution was indirectly due to a hurricane or not. How many power lines go down in Puerto Rico under non hurricane conditions? Normalizing using a good statistical model to predict and subtract expected electrocutions, and quantifying the uncertainty around that number, prevents a post hoc ergo propter hoc problem with coding a death as attributable to a hurricane or not. This is especially important when bodies sat in morgues for as long as they did, and death certificates had so many "garbage codes".

LMAO, 26 people upvoted a false claim by TemporalWolf that 3000 people died in the storm. NOBODY, ANYWHERE is making that claim other than TemporalWolf in his comment. Certainly, the study is not making that claim.

@probably_someone - If the question asks "is Trump correct" then the question is 100% totally meaningless unless one knows specifically what Trump is correct or incorrect about. You are correct, all questions of this type are indeed meaningless without knowing relevant details like 'what did they mean'. However, that does not mean ALL questions are meaningless. If the OP had simply asked "Is the 3000 death toll figure grossly exaggerated?". Then the question has meaning. Adding the "Is Trump correct.." to that same question is what makes it meaningless without getting clarification from Trump.

@Dunk You're either misreading or intentionally misrepresenting what I said: If he means in the storm then he's misrepresenting the 3,000 because the 3000 isn't deaths in the storm, it is deaths because of the storm. He's conflating the two.

@Dunk The question "Is the 3000 death toll figure grossly exaggerated?" is equivalent to the question "Is it correct that the 3000 death toll figure is grossly exaggerated?". If the 3000 death toll figure is grossly exaggerated, then "the 3000 death toll figure is grossly exaggerated" is a correct statement. Likewise, if the 3000 death toll figure is not grossly exaggerated, then "the 3000 death toll figure is grossly exaggerated" is not a correct statement. "Is it correct that X?" is also equivalent to "Is [person] correct when they claim X?" when X is a falsifiable claim.

@Dunk Since you assert that "the 3000 death toll figure is grossly exaggerated" is a falsifiable claim (by calling "Is the 3000 death toll figure grossly exaggerated?" a meaningful question), it doesn't matter whether the OP asked about it or if the OP asked about Trump claiming it, as long as Trump actually did make that claim (you haven't disputed this, and the OP provides good evidence, so we seem to be agreed). So both questions are meaningful. If you intend to renege on your logic and dispute the definition of "grossly exaggerated", then both questions are meaningless.

This conversation seems to be devolving. I think we are now all clear on the facts of this case. The issue of conflicting definitions can be (and has been) resolved in the answers explaining the differences, ideally without political point-scoring.