44 fish out of 1000+ might be extremely significant.
If we’re going to say “what is the likelihood of pulling up a fish with 180x cesium levels?” then the number 44/1000+ might seem like we should say “It’s not very significant.” And for some phenomena, that’s an okay assessment. It’s the same as 4.4/100+, which is a less than 5% chance. Of course, having a 5% chance of getting shot on the way to work is a bit high. It probably means your work is being in active combat.
You also have to measure downstream effect. If there were 44/1000 and 1000 was the average catch size, that means that for a single catch, 44 fish that I can only assume are poisonous or at least medically dangerous. How many people would those 44 fish serve? Could we expect 50-60 downstream deaths or cases of cancer? How many shipments of 1000 fish, with 44 having 180x cesium levels, are going to make their way into the coastal cities and food distribution in the area?
But even beyond all of that, you cannot look at a statistical number intended to characterize something like this and make an interpretation as to its significance without more data. In this case, the number we’re measuring significance against is “Out of x fish, how many do we expect to have 180x cesium?” If the answer is “40,” then 44 is likely not significant. If the answer is “less than 1,” then I suspect 44 is extremely significant. It’s multiple orders of magnitude from baseline, which means that something very very different and quite possibly unexpected and extremely bad is going on.
Basically - and I’m not accusing you of this in any way - some people will use this kind of non-analysis to deliberately mislead their readers about health or safety or environmental issues. It’s nothing against you, OP. It’s just that I taught and still sometimes teach this kind of thing, and I want to make sure that people are aware when they read this sort of thing.
The 1000+ number was just a random number. It was simply to highlight that the article never mentioned the total numbers sampled, just the total numbers found to have the high levels.
I don’t doubt it was 44 out of 44, or that 44 out of 1000 is a lot as well, it simply wasn’t the point that I was trying to make.
My point isn’t about 1000 or 10000. It’s that we shouldn’t make assumptions as to the interpretation of statistical characteristics without sufficient additional data.
Depends, I didn’t see the total number of fish sampled.
If it was 44 fish out of 44 sampled, than 100% being over the limit is significant.
If it was 44 fish out of 1000+, than not so much.
44 fish out of 1000+ might be extremely significant.
If we’re going to say “what is the likelihood of pulling up a fish with 180x cesium levels?” then the number 44/1000+ might seem like we should say “It’s not very significant.” And for some phenomena, that’s an okay assessment. It’s the same as 4.4/100+, which is a less than 5% chance. Of course, having a 5% chance of getting shot on the way to work is a bit high. It probably means your work is being in active combat.
You also have to measure downstream effect. If there were 44/1000 and 1000 was the average catch size, that means that for a single catch, 44 fish that I can only assume are poisonous or at least medically dangerous. How many people would those 44 fish serve? Could we expect 50-60 downstream deaths or cases of cancer? How many shipments of 1000 fish, with 44 having 180x cesium levels, are going to make their way into the coastal cities and food distribution in the area?
But even beyond all of that, you cannot look at a statistical number intended to characterize something like this and make an interpretation as to its significance without more data. In this case, the number we’re measuring significance against is “Out of x fish, how many do we expect to have 180x cesium?” If the answer is “40,” then 44 is likely not significant. If the answer is “less than 1,” then I suspect 44 is extremely significant. It’s multiple orders of magnitude from baseline, which means that something very very different and quite possibly unexpected and extremely bad is going on.
Basically - and I’m not accusing you of this in any way - some people will use this kind of non-analysis to deliberately mislead their readers about health or safety or environmental issues. It’s nothing against you, OP. It’s just that I taught and still sometimes teach this kind of thing, and I want to make sure that people are aware when they read this sort of thing.
The 1000+ number was just a random number. It was simply to highlight that the article never mentioned the total numbers sampled, just the total numbers found to have the high levels.
I don’t doubt it was 44 out of 44, or that 44 out of 1000 is a lot as well, it simply wasn’t the point that I was trying to make.
My point isn’t about 1000 or 10000. It’s that we shouldn’t make assumptions as to the interpretation of statistical characteristics without sufficient additional data.