3 ms·

Every single one of your points is not boolean and depends on the definition and the data you include and exclude. For each you could easily find studies and st

Every single one of your points is not boolean and depends on the definition and the data you include and exclude. For each you could easily find studies and statistics in either direction. The fact that this is apparently not obvious to you proves the point that all fact-checking is inherently biased and depends on the subjective opinions of the checking person.

mrguyorama 2y ago
People who study statistics are pretty good at saying "look, that data set was probably gamed, I would have done it <different way>", or "that conclusion does not follow from the data presented". It's no different to someone claiming on twitter that they are a great programmer who can fix twitter's search in a weekend who then has to tweet for suggestions on how to write a search feature in javascript. People familiar with the subject matter can see right through your bravado. I'm so tired of people with no expertise on anything insisting that people who have clear expertise "didn't think of trivial point A that just came to mind" as if some of these fields aren't centuries old and have been around the block a few times. It's similar to the teenager insisting "you just don't get it mom", but like, mom totally gets it, she was a teenager once too. And while there are occasions when mom might not get it, like how she didn't grow up in a world with social media so she might not be able to help you through that, but she ABSOLUTELY gets that it feels like your world is ending when your first love leaves you, and in fact it is YOU who does not "get it" that you will move on eventually.
- ramblenode 2y ago
  Most experts will not give simple answers to simple questions because they see the question itself as ill-posed. Theses could be written about "Is crime up or down?" GP's claim is that this has a simple answer that can be checked. The bigger issue isn't whether a dataset is statistically valid but which data would even be relevant to a particular underspecified and vague question.
- miramba 2y ago
  Not sure what you are trying to say - my point was that ie the question "is crime up or down" is not a yes/no answer. Depending on the input, you can easily create a statistic pointing in any direction. I think abtinf elaborated better on that here: https://news.ycombinator.com/item?id=42628198 https://news.ycombinator.com/item?id=42628198 My personal highpoint in using statistical methods was probably implementing an analysis of variance for thousands of lab values (https://en.wikipedia.org/wiki/Analysis_of_variance https://en.wikipedia.org/wiki/Analysis_of_variance).

mrguyorama 2y ago

People who study statistics are pretty good at saying "look, that data set was probably gamed, I would have done it <different way>", or "that conclusion does not follow from the data presented". It's no different to someone claiming on twitter that they are a great programmer who can fix twitter's search in a weekend who then has to tweet for suggestions on how to write a search feature in javascript. People familiar with the subject matter can see right through your bravado. I'm so tired of people with no expertise on anything insisting that people who have clear expertise "didn't think of trivial point A that just came to mind" as if some of these fields aren't centuries old and have been around the block a few times. It's similar to the teenager insisting "you just don't get it mom", but like, mom totally gets it, she was a teenager once too. And while there are occasions when mom might not get it, like how she didn't grow up in a world with social media so she might not be able to help you through that, but she ABSOLUTELY gets that it feels like your world is ending when your first love leaves you, and in fact it is YOU who does not "get it" that you will move on eventually.

ramblenode 2y ago
Most experts will not give simple answers to simple questions because they see the question itself as ill-posed. Theses could be written about "Is crime up or down?" GP's claim is that this has a simple answer that can be checked. The bigger issue isn't whether a dataset is statistically valid but which data would even be relevant to a particular underspecified and vague question.
miramba 2y ago
Not sure what you are trying to say - my point was that ie the question "is crime up or down" is not a yes/no answer. Depending on the input, you can easily create a statistic pointing in any direction. I think abtinf elaborated better on that here: https://news.ycombinator.com/item?id=42628198 https://news.ycombinator.com/item?id=42628198 My personal highpoint in using statistical methods was probably implementing an analysis of variance for thousands of lab values (https://en.wikipedia.org/wiki/Analysis_of_variance https://en.wikipedia.org/wiki/Analysis_of_variance).