There should be a universal human standard to define what extreme poverty is--i.e. the amount needed to secure food, shelter, and clothing--and then that amount should be assessed country-by-country (or region-by-region) by an independent body. The number of $3 per day is well above the "basic needs" threshold in some of the poorest countries, and well below it in the US, for example.
Makes you wonder what the real purpose of that number was. Must have served some agenda, because saying some people live on less than $3 (when it's not a fair statement) definitely could serve a purpose.
Does the person buy food and basics per day? Then don’t worry about what the dollar amount equates to. It’s a ridiculous metric when it comes to measuring abject poverty.
This is addressed in the article - see the section titled "Estimating comparable national distributions". (In short: income is being scaled relative to purchasing power parity.)
Usually they choose a deliberately stupid measurement such as "household income below a percentage of the median wage".
This is stupid for many reasons, including (but not limited to): non-monetary, in-kind benefits being excluded, perverse outcomes such as a decline in median wages "reducing poverty" and just about guaranteed continuation of this "poverty". So left wing politicians LOVE it. It's an everlasting cudgel that can never be fixed.
It's fairly easy to fix, as long as you are willing to do what it takes to address income inequality. Reduce the Gini coefficient and poverty decreases.
That's actually my point: if you take (e.g.) 65% of median income, in a world with a Gini coefficient of 1 - perfect inequality - the rate of poverty is 0%.
This seems sane. The real question one should ask is, how many people can earn a living that allows them to meet basic needs, without state support?
You can have a separate figure that out of the number of poor people (like defined in the last sentence), how many are no longer poor with state support?