4 ms·

I think it's telling that none of the authors are geophysicists, and they definitely aren't statisticians either. The causality they're claiming just does not h

by mjhay 3y ago

I think it's telling that none of the authors are geophysicists, and they definitely aren't statisticians either. The causality they're claiming just does not have any physical plausibility. > Dr. Homola states, “In the scientific world, it is accepted that a discovery can be said to have been made when the statistical confidence level of the corroborating data reaches five sigma, or standard deviations.” > Homola continues, “For the observed correlation, we obtained more than six sigma, which means a chance of less than one in a billion that the correlation is due to chance. We therefore have a very good statistical basis for claiming that we have discovered a truly existing phenomenon. The only question is, is it really the one we were expecting?” This is not what p-values mean! P-values are not at all the chance of a result being erroneous. It's an unfortunately widespread misconception among people who should know better. https://www.amstat.org/asa/files/pdfs/p-valuestatement.pdf https://www.amstat.org/asa/files/pdfs/p-valuestatement.pdf

mjburgess 3y ago
No statistical analysis at all applies if the relevant premises are false. People defend the statistical analysis using statistical metrics, rather than the premesis by which the stats are used (eg., the model type, the causal model, etc.). Eg., if there is no causal connection between X, Y then any analysis of association between X, Y is not a causal one. Entire fields of people have been taught that stats can justify science -- and hence the heaps of pseudoscience. It is science which justifies the use of stats; not vice versa.
refibrillator 3y ago
I agree with you regarding p-values. But why are you appealing to credentials and dismissing the idea with your own boldly unsupported claim? > The causality they're claiming just does not have any physical plausibility. Survey the literature and you’ll find plenty of evidence that there is some credence to the idea.
- mjhay 3y ago
  Geomagnetism is easily measurable and has been monitored for a long time. The central claim is that changes in cosmic ray frequencies can predict earthquakes. The proposed causal mechanism is changes in the geodynamo affecting the propagation of cosmic rays. The effect on cosmic rays is secondary to the changes in geomagnetism. Despite decades of people looking for evidence, changes in geomagnetism are not associated with earthquakes[0]. If this paper's hypothesis is true, geomagnetic fluctuations should co-occur with fluctuations of cosmic ray counts - thus geomagnetic fluctuations should also predict earthquakes. But they don't. Claims of being able to predict earthquakes have had a long and not great history. Earthquake prediction is the faster-than-light travel of geophysics. Any claims that it is possible should be viewed with extreme scrutiny. [0] https://www.usgs.gov/faqs/are-earthquakes-associated-variations-geomagnetic-field https://www.usgs.gov/faqs/are-earthquakes-associated-variati...
  - refibrillator 3y ago
    You're incorrect or confused on several aspects so let me break it down for you. Frequency of detection of secondary cosmic radiation is very sensitive to geomagnetic conditions. That is not up for dispute. The paper in question provides statistical evidence that "variations of secondary cosmic ray detection rates are periodically correlated with future global earthquake magnitude sum". The paper contains no claims about precisely predicting seismic activity in terms of geographic location or time of occurrence. The authors make this abundantly clear. The USGS page you linked has no citations or even a timestamp. I'd wager it hasn't been updated in ages. It's not the decisive source you want it to be. Here's the paper in question if you want to actually read it: https://arxiv.org/ftp/arxiv/papers/2204/2204.12310.pdf https://arxiv.org/ftp/arxiv/papers/2204/2204.12310.pdf Take a look at this research from NASA, it is one of the many papers that completely contradicts your claim that "changes in geomagnetism are not associated with earthquakes": https://www.scirp.org/journal/paperinformation.aspx?paperid=82421 https://www.scirp.org/journal/paperinformation.aspx?paperid=...
    - SiempreViernes 3y ago
      The linked article fails many test of credemtialing, but most egregious is the entirely fale presemtation of "Geo Cosmo Science and Research Center" (really this: https://www.geocosmo.net https://www.geocosmo.net) as being a NASA lab.
nearbuy 3y ago
Unless I'm missing something, that statement appears to be correct. They don't say their result isn't erroneous. They say there is statistically only a very tiny chance that the correlation is due to chance. If the correlation is caused by an error in their experiment, their statement is still correct, since the correlation was not due to chance.
- whimsicalism 3y ago
  Did they hypothesize this effect beforehand? If not, this was a data collection and analysis exercise - not an experiment. How many comparisons did they do before stumbling on this result? Unreported.
  - nearbuy 3y ago
    They could be p-hacking, though getting 6 sigmas through p-hacking without outright fraudulent data is actually some pretty impressive level of p-hacking. I'd say it's more likely they either made an error or there really is a correlation.
- mustacheemperor 3y ago
  At the top of the second page of that pdf: >P-values do not measure the probability that the studied hypothesis is true, or the probability that the data were produced by random chance alone. It was very informative for me. Worth reading.
  - nearbuy 3y ago
    Without getting nitpicky to the point where it's impossible to say anything at all without complaining it's not a perfectly complete and true statement, their statement still seems correct to me. Especially given that this is a statement for non-statisticians and they can't exactly go on a 5 minute tangent to explain exactly how p-values work. Perhaps it would have been more accurate if they phrased it: "which means a chance of less than one in a billion of seeing a correlation this strong if it's due to chance", but I think few readers would even notice the difference.
8bitsrule 3y ago
May not be causal, but a six-sigma correlation is sure-as-hell interesting. Edit: I just remembered that today is the 229th anniversary of a meteor fall in Siena, Italy, so big that it established tan undeniable correlation between falling stars and hundreds of chunks of iron lying on the ground. (Putting Newton's prejudice to rest.)

by mjhay 3y ago

mjburgess 3y ago

No statistical analysis at all applies if the relevant premises are false. People defend the statistical analysis using statistical metrics, rather than the premesis by which the stats are used (eg., the model type, the causal model, etc.). Eg., if there is no causal connection between X, Y then any analysis of association between X, Y is not a causal one. Entire fields of people have been taught that stats can justify science -- and hence the heaps of pseudoscience. It is science which justifies the use of stats; not vice versa.

refibrillator 3y ago

I agree with you regarding p-values. But why are you appealing to credentials and dismissing the idea with your own boldly unsupported claim? > The causality they're claiming just does not have any physical plausibility. Survey the literature and you’ll find plenty of evidence that there is some credence to the idea.

mjhay 3y ago
Geomagnetism is easily measurable and has been monitored for a long time. The central claim is that changes in cosmic ray frequencies can predict earthquakes. The proposed causal mechanism is changes in the geodynamo affecting the propagation of cosmic rays. The effect on cosmic rays is secondary to the changes in geomagnetism. Despite decades of people looking for evidence, changes in geomagnetism are not associated with earthquakes[0]. If this paper's hypothesis is true, geomagnetic fluctuations should co-occur with fluctuations of cosmic ray counts - thus geomagnetic fluctuations should also predict earthquakes. But they don't. Claims of being able to predict earthquakes have had a long and not great history. Earthquake prediction is the faster-than-light travel of geophysics. Any claims that it is possible should be viewed with extreme scrutiny. [0] https://www.usgs.gov/faqs/are-earthquakes-associated-variations-geomagnetic-field https://www.usgs.gov/faqs/are-earthquakes-associated-variati...
- refibrillator 3y ago
  You're incorrect or confused on several aspects so let me break it down for you. Frequency of detection of secondary cosmic radiation is very sensitive to geomagnetic conditions. That is not up for dispute. The paper in question provides statistical evidence that "variations of secondary cosmic ray detection rates are periodically correlated with future global earthquake magnitude sum". The paper contains no claims about precisely predicting seismic activity in terms of geographic location or time of occurrence. The authors make this abundantly clear. The USGS page you linked has no citations or even a timestamp. I'd wager it hasn't been updated in ages. It's not the decisive source you want it to be. Here's the paper in question if you want to actually read it: https://arxiv.org/ftp/arxiv/papers/2204/2204.12310.pdf https://arxiv.org/ftp/arxiv/papers/2204/2204.12310.pdf Take a look at this research from NASA, it is one of the many papers that completely contradicts your claim that "changes in geomagnetism are not associated with earthquakes": https://www.scirp.org/journal/paperinformation.aspx?paperid=82421 https://www.scirp.org/journal/paperinformation.aspx?paperid=...
  - SiempreViernes 3y ago
    The linked article fails many test of credemtialing, but most egregious is the entirely fale presemtation of "Geo Cosmo Science and Research Center" (really this: https://www.geocosmo.net https://www.geocosmo.net) as being a NASA lab.

nearbuy 3y ago

Unless I'm missing something, that statement appears to be correct. They don't say their result isn't erroneous. They say there is statistically only a very tiny chance that the correlation is due to chance. If the correlation is caused by an error in their experiment, their statement is still correct, since the correlation was not due to chance.

whimsicalism 3y ago
Did they hypothesize this effect beforehand? If not, this was a data collection and analysis exercise - not an experiment. How many comparisons did they do before stumbling on this result? Unreported.
- nearbuy 3y ago
  They could be p-hacking, though getting 6 sigmas through p-hacking without outright fraudulent data is actually some pretty impressive level of p-hacking. I'd say it's more likely they either made an error or there really is a correlation.
mustacheemperor 3y ago
At the top of the second page of that pdf: >P-values do not measure the probability that the studied hypothesis is true, or the probability that the data were produced by random chance alone. It was very informative for me. Worth reading.
- nearbuy 3y ago
  Without getting nitpicky to the point where it's impossible to say anything at all without complaining it's not a perfectly complete and true statement, their statement still seems correct to me. Especially given that this is a statement for non-statisticians and they can't exactly go on a 5 minute tangent to explain exactly how p-values work. Perhaps it would have been more accurate if they phrased it: "which means a chance of less than one in a billion of seeing a correlation this strong if it's due to chance", but I think few readers would even notice the difference.

8bitsrule 3y ago

May not be causal, but a six-sigma correlation is sure-as-hell interesting. Edit: I just remembered that today is the 229th anniversary of a meteor fall in Siena, Italy, so big that it established tan undeniable correlation between falling stars and hundreds of chunks of iron lying on the ground. (Putting Newton's prejudice to rest.)