Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Counter balance:

“According to the 2023 Global Hunger Index, India ranks 111th out of 125 countries, with a score of 28.7, indicating a “serious” level of hunger. Although this is a slight improvement from its 2015 score of 29.2, the progress has been slow compared to countries with similar economic growth rates. India also has the highest child wasting rate in the world at 18.7%. And 35.5% of children under five are medically listed as having their growth stunted, a condition that leads to long-term developmental complications.”

https://outreach-international.org/blog/malnutrition-in-indi...



I would urge you to watch a statistical debunking of these so-called "global goodness" metrics. Sanjeev Sanyal does a good job of it [1],[2],[3] Multiple organization(s) that manage such indices are, turns out, funded by the same cabal who want to concoct a narrative. They are not unbiased.

[1] https://www.youtube.com/watch?v=3Lgf2WHstBY

[2] [PDF] https://eacpm.gov.in/wp-content/uploads/2023/03/Reversing-Ga...

[3] https://www.businesstoday.in/india/story/sanjeev-sanyal-blas...


You quote a ruling party cheerleader to claim that anyone showing the real truth about India is just part of Soros gang :)


Three significant figures for a somewhat rough measurement of poverty over a sample of a population as diverse as India should have one's BS alarms blaring.

Actually both 28,7 and 29,2 = 30.

Social workers playing at science.


I fail to see how diversity in the country would change those metrics.


It makes it harder (impossible even) to take precise measurements deep into the decimal places if you have massive contours in the very thing (wealth) you are trying to measure with just one scalar number.

Diversity is not meant in the modern political interpretation.

As an less triggering exercise: try record in a single number the temperature - and change - of a large geographically diverse place like Australia - which has deserts and forests and is also subject to large periodic fluctuations like el Niño/la Niña.

I submit that you would be embarrassed to say it went from 28,7°C to 29,2°C and think it meant anything.

The misuse (and misunderstanding) of significant figures (let alone averages) is rampant in the social sciences. To the disadvantage of the folk we are all trying to help.


If the mean rolling average of all tempreture readings across Australia moved by 0.5 across a twelve month period that would indeed be a significant change that would certainly mean something.


Indeed, but not if your thermometer could only measure in 2°C increments.

You don't gain accuracy by averaging a bunch of rough measurements, especially if the micro-climate has a lot of diversity.

Statistics is tough.


Luckily the Australian BOM Weather Station specifications require that air and soil temp sensors have a -60° C to +60° C range with a 0.1° C resolution and a ±0.1° C accuracy.

The specs are in attachment 2 of the BoM Observation Specification manual.

Spurious comments are easy, looking stuff up is tough.


That is also something really hard to measure correctly. There are many sources of error that are hard to correct for.


The Central Limit Theorem smooths any errors. Decades of observation data make instrument errors very easy to spot.

In recent decades daily MODIS (satellite) polar orbit overpasses provide air column and ground temp data that infills station data (from 1900 onwards).

There's a science to ground truthing, error detection, and data correction.

Here's the gross single number continent wide average over a century: http://www.bom.gov.au/cgi-bin/climate/change/timeseries.cgi

It's not hard to find raw data, finer grids, methodology, etc.


Only sometimes. It is easier to measure in cities, and thus introduce bias based on whatever pavement is populate.

There is a lot of data, and the people who measure this (or at least some of them) are aware of the issues and so correct for them. However you cannot just blindly apply the central limit theorem.


> It is easier to measure in cities, and thus introduce bias based on whatever pavement is populate.

That's a myth.

* MODIS sat readings are available under the precessing polar orbit path and are indifferent to urbanisation or not.

* National BoM networks have always included stations outside of cities and great care is taken to position stations and to normalise readings.

* Urban Heat Island Bias is a myth of long standing; asked and answered three decades ago, two decades ago, a decade ago.

Why is it _still_ being raised in 2025?

See: https://skepticalscience.com/urban-heat-island-effect-interm...

https://www.youtube.com/watch?v=pidRKmq2iGI


Who said anything about blindly?

Apart from yourself, implying that happens as a matter of course?


Can you elaborate on that? Why would it be significant and what could it mean?


it would mean it was warmer.


Really? That's pretty horrible understanding of stats you have there. By your logic, it makes no sense to say that the fertility rate is 1.81 and has changed by 0.8 to now be 1.73 because a woman can either give birth to 1 or 2 kids.

We do measure Earth's average temperature for what it's worth and while different metrics might give different results, the absolute value is not really useful. What is useful is the change in that metric. And if our global temperatures change by 0.1C by any one metric, it is a very significant change given that the metric is well designed. And yes, it is possible to design a good metric.


The median is not the message!


Economic class is highly correlated with social, religious, ethnic, tribal class. If there are more social, religious, ethnic, tribal classes in a single country (i.e. the country is diverse) then in another, then you do need to sample more of the groups to get sample means free from systematic bias.

Simple example: image you are asked to compute the mean diameter of all balls in each of two bags. You open the first bag, and it is all blue balls, and they all look very close in radius. So you measure a few and call it a day. You open the second bag, and there are balls of about 50 different colors, and it seems like all the colors have different radius. So, then to get a good result, you need to measure almost the diameter of most of the colors.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: