(1) Don’t make absolutist statements without understanding the nature of the data; (2) Don’t abuse statistical jargon; (3) Don’t assert that a conspiracy exists just because the data doesn’t fit your preferred narrative.
First, consider a Comment On the death toll from Hurricane Maria:
this [assertion that thousands of American citizens have died] Absolutely wrong, Menzie. According to statistics bureau records, the excess death toll in PR was only 654 at the end of the year. Most of this happens in the last ten days of September and throughout October. While the state ownership of the PR utility exacerbated the outages there, given the terrain and the strength of the hurricane, most of the additional deaths likely occurred anyway. So, 300-400 extra deaths could have occurred regardless of who took steps to restore power. The rest can basically be attributed to the state ownership of the power company.
I would point out that the number of excess deaths fell in half in December. Thus, the data suggest that hurricanes accelerate the death of the sick and dying rather than kill them directly. I would expect a total of 200-400 excess deaths in a year (say 1st October 2018). Still a notable number, but certainly not 4,600.
See analysis: https://www.princetonpolicy.com/ppa-blog/2018/5/30/reports-of-death-in-puerto-rico-are-wildly-exaggerated
i will notice Official death toll at 2,975 in Commonwealth of Puerto Rico-commissioned GWU reportsee estimate discussion here.
second, one After 2018 Uncertainties About Statistical Inferences.
Question from Mr. Steven Kopits The point estimates (4645) and confidence intervals (798, 8498) of excess deaths in Puerto Rico after Maria from the Harvard School of Public Health-led study are thus:
Did Harvard support the research?
That is, does Harvard SPH believe that the central estimate of 12/31 excess deaths is 4645? Does it lag behind the confidence interval? If the death toll exceeds 4600, is there still a possibility of 50+? If there are, then the PR people need to start looking for 3,250 missing persons, otherwise the media needs to assume that the PR authorities are lying. These are implied action items.
Or should we just take any future numbers released by HSPH and divide by 3 to get an actual realistic estimate?
Let’s show the details of the graph shown earlier (in the this post):
figure 1: estimated from Santos-Lozada and Jeffrey Howard (November 2017) September and October (calculated as the difference between midpoint estimates), and Nashant Kishore et al. (May 2018) December 2017 (blue triangle), and Roberto Rivera and Wolfgang Rolke (February 2018) (red squares), and Santos-Lozada’s estimates based on administrative data released on June 1 (large dark blue triangles), month-end data, all on a logarithmic scale. + indicates the upper and lower bounds of the 95% confidence interval.The orange triangle is Steven Kopits’ estimate for the end of the year June 4Cumulative figures for Santos-Lozada and Howard October figures were calculated by the authors based on reported monthly figures.The middle paragraph (highlighted in red) shows a misunderstanding of the confidence interval. The true parameter is or is not within the confidence interval. Instead, this would be a better characterization with 95% CI:
“If this process is repeated over a large number of samples, the proportion of calculated confidence intervals (which will vary from sample to sample) that contain the true population parameter will tend towards 95%.”
In other words, it is wrong to say that there is a 50% chance that the actual number will be higher than the point estimate. But that’s exactly what Mr. Kopits means by a confidence interval. He is incorrect about this.from political facts:
Roberto Rivera, a statistician at the University of Puerto Rico and his colleague Wolfgang Rolke, who used death certificates to estimate the death toll was much lower, said indirect estimates should be interpreted with caution .
“Note that, based on this study, Maria’s true death toll could be anywhere between 793 and 8,498: 4,645 is no more likely than any other value in that range,” Rivera said.
Again, I think those wishing to comment on estimates would do well to be familiar with statistical concepts.
this is the most recent duplicate postal.
reader steve kopitz Writing about the employment numbers debate:
At the same time, I think it’s possible that both surveys are actually correct but give a misleading impression because we’ve misinterpreted the data. That still seems possible, although I’ve read others that CES was rigged to provide a more rosy picture going into the election.
The statement joins a long list of such allegations, such as, Senator Barrasso, Jack Welch, former congressman allen west, zero hedge, Mick Mulvaney,etc. All I can say is that if there was a conspiracy, they didn’t do it well.With the benefit of the January baseline revision, we can renew Our assessment of how poorly the alleged co-conspirators performed their jobs.
figure 1: Nonfarm Payrolls released Jan 2023 (red), Nonfarm Payrolls released Oct 2022 (blue), 000’s, sa Source: BLS via FRED.
Now, it may finally turn out (following another benchmark revision, the results of which are due in February 2024) that NFP in Q2 will be lower than indicated in CES. But in order to deceive voters in November 2022, this seems like a bad practice.
Anyway, before people start complaining that the data is manipulated, I would like them to read the BLS Technical Note on (1) Revision and Mean Absolute Revision, (2) Baseline Revision, (3) Calculation of Seasonal Adjustment Factor, (4) Apply population controls in CPS. Before they start citing the various series, I want them to understand the informative content (relative to business cycle fluctuations) of the CPS employment series vs. the CES employment series. This understanding can be gained by reading the work of people who understand the characteristics of macro data (Furman (2016); China Eastern Airlines (2017); Goto et al. (2021)).
From a sociological perspective, I do wonder why conspiracy theories are so attractive to some people.This is a scientific american article List some of the personality traits associated with holding on to conspiracy theories.
f
