Science
20 Mar 11
29 Apr 11
1 edit
Originally posted by KostenuikSo now you agree that you disagree with the USGS. A start.
4) Earthquake clustering and human psychology. While the average number of large earthquakes per year is fairly constant, earthquakes occur in clusters. This is predicted by various statistical models, and does not imply that earthquakes that are distant in location, but close in time, are causally related. But when such clusters occur, especially when the aller periods of about 10-15 years with MANY earthquakes. THIS IS not RANDOM! It's perfection.
Reputable source for your interpretation?
29 Apr 11
4 edits
I flipped a 10 pence coin a minute ago (I actually did), flipped it 10 times and here's my result:
t,h,t,t,h,t,t,t,h,t
Now I can't remember what the results were of any other coin flipping trials in the past; and since it would clearly be unscientific to just estimate what the results where, given I didn't record or document them in any way, I'll just draw my inferences based on this sample.
Looks like I have a biased coin! 7 tails vs 3 heads. Also, the more I flip, the more bunched up the tails become!
29 Apr 11
Originally posted by PalynkaNo. It is unclear what they are meaning. Is a cluster something that happens over months? If so then they are talking about something different to the decade period that I am looking at and are talking about something different..
So now you agree that you disagree with the USGS. A start.
Reputable source for your interpretation?
29 Apr 11
Originally posted by KostenuikIt's very clear what they mean.
No. It is unclear what they are meaning. Is a cluster something that happens over months? If so then they are talking about something different to the decade period that I am looking at and are talking about something different..
"While the average number of large earthquakes per year is fairly constant, earthquakes occur in clusters. This is predicted by various statistical models, and does not imply that earthquakes that are distant in location, but close in time, are causally related."
29 Apr 11
1 edit
Originally posted by PalynkaEarthquakes happen in runs like that over a period of weeks in the size they are refering to 7+. Maybe 6 + but that isn't clear to mean what I am looking at over the entire record of earthquake data and at a size where a relationship is whole lot more likely.
It's very clear what they mean.
"While the average number of large earthquakes per year is fairly constant, earthquakes occur in clusters. This is predicted by various statistical models, and [b]does not imply that earthquakes that are distant in location, but close in time, are causally related."[/b]
29 Apr 11
Originally posted by AgergNow throw an uneven proportion of heads for 3-4 times as long. Repeat this two and a half times.
I flipped a 10 pence coin a minute ago (I actually did), flipped it 10 times and here's my result:
t,h,t,t,h,t,t,t,h,t
Now I can't remember what the results were of any other coin flipping trials in the past; and since it would clearly be unscientific to just estimate what the results where, given I didn't record or document them in any way, I'll just dr ...[text shortened]... iased coin! 7 tails vs 3 heads. Also, the more I flip, the more bunched up the tails become!
29 Apr 11
1 edit
Originally posted by KostenuikAs my knowledge of stats goes, and I've done myself the disservice of avoiding such modules (first two compulsary courses badly taught), it wouldn't be statistically remarkable if I got such a run. Indeed a bias towards one or the other is more likely in a small sample size. The law of large numbers kicks in after a *large* number of trials.
Now throw an uneven proportion of heads for 3-4 times as long. Repeat this two and a half times.
The same goes for your earthquake clusters, it doesn't seem statistically remarkable that the run of earthquakes is as you describe it. Simply put you need far more evidence to draw the conclusions you have made so far.
29 Apr 11
Originally posted by AgergYes well all we have is ~110 years to work out if the largest of earthquakes can have a relationship to others and all go off at once. From the quakes >8.5 we only have 15 and they all fall perfectly into place.
As my knowledge of stats goes, and I've done myself the disservice of avoiding such modules (first two compulsary courses badly taught), it wouldn't be statistically remarkable if I got such a run. Indeed a bias towards one or the other is more likely in a small sample size. The law of large numbers kicks in after a *large* number of trials.
The same goes f ...[text shortened]... cribe it. Simply put you need far more evidence to draw the conclusions you have made so far.
29 Apr 11
4 edits
Originally posted by KostenuikWell if we don't have enough data to make a valid call either way on this relationship (or otherwise), i.e. we have only scant statistical evidence, then why should we accept your conclusions? I could apply the same reasoning to 1 coin flipping trial and assert that given this is the only evidence I have to go on, my coin is biased!
Yes well all we have is ~110 years to work out if the largest of earthquakes can have a relationship to others and all go off at once. From the quakes >8.5 we only have 15 and they all fall perfectly into place.
Are you able to produce a physical explanation for this 'phenomenon', backed up by professionals and academics in this area, or is a short supply of statistical data all you have to work with?
29 Apr 11
Originally posted by AgergNot conclusions... hypothesis. Your coin flipping trail is only showing one trend. The earthquakes show one for x time then revert for y time. It repeats this 2.5 times EXACTLY the same.
Well if we don't have enough data to make a valid call either way on this relationship (or otherwise) then why should we accept your conclusions? I could apply the same reasoning to 1 coin flipping trial and assert that given this is the only evidence I have to go on, my coin is biased!
29 Apr 11
2 edits
Originally posted by KostenuikLet's put it another way, suppose I fire up code::blocks and get it to run 500,000,000 trials of 10 coin flips and I treat each flip as having been performed each year.
Not conclusions... hypothesis. Your coin flipping trail is only showing one trend. The earthquakes show one for x time then revert for y time. It repeats this 2.5 times EXACTLY the same.
If I extracted 11 contiguous trials where the data had the same relationship which supports your hypothesis should I then conclude there is a particular trend overall? Because that's what you seem to be asking of us with your earthquake hypothesis.