Countries have incentives to manipulate population data. Most error that I’m aware of is not attributable to poor data quality. For example, if you have a real estate bubble you have a strong incentive to show population growth.
>For example, if you have a real estate bubble you have a strong incentive to show population growth.
That's one source of bias that is present at a specific time. Mostly you would have competing incentives. There is usually more than one agency that runs does the counting. Vital records registration, voter rolls and tax payers lists, for example are separate agencies in some countries. Not every tax payer is a voter and not everyone who was born still lives in the country. The sources are sometimes cross-referenced too. Then there is usually a place that needs to do macroeconomic forecasting and needs to have some numbers to do it's job.
King Louis XIV lost a bunch of his land to astronomers able to more accurately measure said land. This is the sort of thing that can happen when you want to turn your country into a world leader in science.
This study published in Nature [0] says that rural populations in particular are typically UNDERCOUNTED (exactly like the Papa New Guinea in the OP's article), and that this happens at similar rates across poorer and wealthier countries: "no clear effect of country income on the accuracies of the five datasets can be observed."
