They are claiming there is a point where money stops being important.
There is no such point.
There is no reason to favor any particular income level. If you use the happiness graph to make an argument that $50k is all anyone needs, I can use the same graph to show that $400 is all anyone needs.
If the 'up to a certain point' hypothesis was correct, with the graph leveling out, there would be a maximum happiness-from-money. You could say "oh well 85% of max is plenty, the ideal income is X". But the data says that's not true.
---
Humans intuitively understand logarithmic scales. Arguably even better than linear scales. A lot of our senses are logarithmic, after all. It's obvious that one dollar does not buy one happiness, given that millionaires aren't in ecstatic shock. But every time I double my income I gain the same number of happiness units? Sounds like money buying happiness to me.
Regarding your comment about log scales, is there a "sense" (or kind of perception, etc...) beside earing that "obviously" logarithmic ?
Also, as another "I don't get stats get me out of here" question, isn't it problematic to plot against a "perception" level that (unless I'm mistaken) is linear ? What do you do with someone that answers "I'm happy at a 10/10 level". Do you consider it an outlier by definition ?
Assuming the "happiness" self-rating is indeed linera, is it still a fallacy to think that, behind the mathematical rightness, there is value in knowing that after $x/year, y% of the population rates itself as above, say, 8/10 ? I'm under the impression that it's a different argument than say "after $x/year, everyone is happy" ; but that it can still be practical (especially if you can corelate lower levels a happiness with stuff you want / should / need to get rid of. But again, talking politics at a math cocktail, probably bad manners.)
Brightness is pretty logarithmic. Touch, smell, taste, all of those seem logarithmic to me.
People rating out of 10 distorts the numbers in multiple ways, that's a separate issue to deal with. Ideally you'd remove self-reporting in some manner...
>value in knowing that after $x/year, y% of the population rates itself as above, say, 8/10
If it's linear with a cap of 10 then all the reasoning from before goes in the trash. The graph won't actually be logarithmic and you can set a cutoff easily.
If you can calculate a linear happiness score with no cap, then it's possible to pick a point that's "happy enough", but that point will be arbitrary. It won't be based on an inflection point on the graph, because the graph has no inflection points. You can use your judgement to say that 12 happiness points is plenty, but that's not a math question.
There is no such point.
There is no reason to favor any particular income level. If you use the happiness graph to make an argument that $50k is all anyone needs, I can use the same graph to show that $400 is all anyone needs.
If the 'up to a certain point' hypothesis was correct, with the graph leveling out, there would be a maximum happiness-from-money. You could say "oh well 85% of max is plenty, the ideal income is X". But the data says that's not true.
---
Humans intuitively understand logarithmic scales. Arguably even better than linear scales. A lot of our senses are logarithmic, after all. It's obvious that one dollar does not buy one happiness, given that millionaires aren't in ecstatic shock. But every time I double my income I gain the same number of happiness units? Sounds like money buying happiness to me.