If you're a data analyst, then surely you must be familiar with (and recognize the usefulness of) proxy variables? If we can easily measure x + e, but not x, then our estimate of x is biased but comparisons of (x1 + e) - (x2 + e) are not biased, and ratios (x1 + e) / (x2 + e) are biased but proportional to the true ratio.

What I meant re: the CO2 statistic is that that's only a very, very small part of what the report is about, that unfortunately is getting a lot of attention. You're absolutely right that it's misleading to use this information as a headline, but the report and the associated UN effort is simply about how to reduce food waste, as a part of the sustainable development goals for 2030: https://sdgs.un.org/goals/goal12

Of course. A common sentence I have to say to people is "This is not a perfect number, but it's been imperfect in the same way for the last 10 years. As such, you should regard it as merely an approximation for any given year, but also that any trend you see is unlikely a result of those imperfections." This is literally a sentence that I copy & paste almost verbatim into a lot of reports.

I'm not allowed to do that sort of thing when the data is reasonably available. The UN could have obtained at least enough data to get an estimate rather than ignoring it completely. The Department of Agriculture in the US and similar organizations all over the world actually provide databases on their specifications for inedible portions of many specific foods. The UN could have at least used the US data alone, along with research showing that food waste in the US tends to be higher than other parts of the world, to place a probable upper-bound on this.

I think we can agree there's not a hard & fast rule here about when to use estimates and when to get full (or at least more) data. I sometimes have to impute values where data is available but it would take many hours of someone's time going through countless boxes of documents in cold storage and doing manual data entry with them. If the missing data is small enough or the issue is of low enough importance, I do the best I can. However a recent project of mine did actually involve having a few people go back into those files and get a random sampling of the data so that I could make reasonable estimates.

It all comes down to resources & importance. So lets look at this situation through that lens with the UN: Resources? The UN absolutely has them, and this information is neither unavailable nor unreasonably burdensome to gather enough for an estimate. Importance? It doesn't get much more important than a report intended to be used to guide global food & climate policy at the highest levels. The UN department that produced this report has an annual budget around $500 million/year [0], and this is one of their landmark publications: They have the resources, and this is important, they shouldn't have ignored the issue.

I also don't think the headlines were unreasonable in leading with that given the report's contents: Literally the first sentence of the report is about the climate impact of food waste. Media often gets science interpretation wrong, but when the first sentence of a report is:

"If food loss and waste were a country, it would be the third biggest source of greenhouse gas emissions"

Given that, I think it's reasonable for reporters to lead with a similar headline. I don't blame the reporters here, or at least it's not where I assign the majority of blame.

[0] https://www.unep.org/about-un-environment/funding-and-partne...

I don't disagree with anything you say in general or in principle... but if you actually bothered to even skim the report, you'd see that almost half of it is about data collection: existing academic research, best practices, challenges, common mistakes, the different levels of granularity / accuracy achievable depending on how much each country is willing to invest in measuring food waste, how to sample, categorization of countries depending on the quality of the data, imputation for countries that have no data available at all, and so on. Furthermore, the UN is coordinating the effort but it's up to individual countries to produce the actual statistics, so no, I suspect with $500 million/year they can do a lot of cajoling but they can't do the actual data collection because that's not their remit. You've got to admit there's more than a little irony in claiming that this is one of the most misleading reports you've ever seen, when it turns out you've not actually seen it.