That joke is there because the "cures" are most often based on faulty statistical inference. A closer look at much of the data will reveal the cure did not exist for mice in the first place, the results were just much more likely to occur by chance than conveyed by the literature. The issue of mice not being completely analogous to humans is an issue faced by researchers but it is being used to hide failure to correctly report and interpret the results of studies (systemic incompetence). All the evidence points towards false positive rates of 70% or higher throughout biomedical literature.
Its very hard to publish there, but the quality of publications is not that high, possibly even lower than elsewhere if you measure by false positive rate. There is a mass failure to understand the importance of the assumptions underlying statistical inference (as you mentioned), as well as the importance of completely reporting your methods and data so that it is possible for others to intelligently draw their own inferences and replicate your work. In short, those journals have a culture that encourages "sexy" and "conclusive" results at the expense of the fundamental basis for successful science that we learn in gradeschool.
Plot the temperature by month from this paper (not just averages of months). The distribution of temps suddenly jumps upwards about two degrees starting in 1989 corresponding to when a sensor was switch out. They also corrected for some calibration errors and sensor drift, starting that year. But fail to tell us how exactly they corrected for the drift.
It is more obvious if you plot only the data they actually had.
On 18 January 2011, a new CR1000 datalogger (used to record and disseminate the readings from the various AWS sensors) was installed on the Byrd AWS in replacement of the AWS-2B electronic system used since 1989. Upon inspection of the old system at the AMRC, a calibration error of 1.5 â--¦C (in excess) was identified for the temperature observations recorded since 2002. In addition, subsequent testing in a newly available cold chamber at the AMRC provided more accurate measurements of the temperature sensitivity of the AWS-2B system, which results in a negative temperature drift as the temperature decreases. As a result, corrections were made to the temperature observations recorded by the Byrd AWS between 1989 and 17 January 2011. The release of the corrected dataset on the AMRCâ(TM)s ftp server in December 2011 was followed by an update of the monthly mean temperatures from Byrd AWS available on the READER online archive10 (http://www.antarctica.ac.uk/met/READER/). The effect of the corrections on the reconstructed temperatures is illustrated in Supplementary Fig. S10. In the lower temperature range (typically -50 to -30 â--¦C), the temperature drift largely compensated for the 1.5 â--¦C error in the 2002â"01/2011 observations, whereas no compensation occurred at higher temperatures (-20 to 0 â--¦C). This explains the differences in the correctionsâ(TM) impact between summer and winter, and between the 1989-2001 and 2002-2010 periods (Supplementary Fig. S10). It is noteworthy that the temperature drift problem did not affect significantly the AWS observations from 1980â"1988, and this for two reasons: (1) Excess power from the RTG was used to keep the internal temperature of the electronics above -20 â--¦C. This extra power was no longer available when the AWS started relying on batteries charged by solar panels. (2) The central processing unit of the AWS was (paradoxically) a newer version than the one subsequently used from 1989 onward.
Also both papers were published in Nature, which is not really well known for publishing sloppy statistical papers.
I consider Nature well known for publishing incomplete, sloppy papers. This may not be true for climate science, but in other fields nature articles are a step above news articles.
The only thing that matters is "What are the odds that I would get this measurement by chance?"
Please stop propagating statistical myths. That is not the only thing that matters. What matters is how well the model fits the data relative to competing models.
I can't be sure but I suspect you are also falling into the trap of thinking at just because there is statistical significance it means that the research hypothesis is true.
I don't think it is that bad (outright fraud). It is just bias run rampant along with financial incentive to underestimate uncertainty combined with widespread failure of science education in statistics.
I don't know where I said that paying for public healthcare leads to a collapse. The current method of providing healthcare is basically stealing from future generations (which will eventually lead to collapse), if you think about how it works (at least in the US) it is a big ponzi scheme. Every generation needs to convince the ones after them to buy in to the insurance/Social Security scam. Please explain how else it could work.
A healthcare system coudln't tank an economy unless people were forced to pay into it. People wouldnt be forced to pay into it unless it had already gotten TBTF.
So..based on the metrics you provided, if the ponzi scheme collapses you will admit that the free healthcare was basically paid for by subtle enslavement of the third world?
If you disagree, then we and our children really could sustain our lifestyle forever without ripping off poor people?
Slashdot Mirror
That joke is there because the "cures" are most often based on faulty statistical inference. A closer look at much of the data will reveal the cure did not exist for mice in the first place, the results were just much more likely to occur by chance than conveyed by the literature. The issue of mice not being completely analogous to humans is an issue faced by researchers but it is being used to hide failure to correctly report and interpret the results of studies (systemic incompetence). All the evidence points towards false positive rates of 70% or higher throughout biomedical literature.
Its very hard to publish there, but the quality of publications is not that high, possibly even lower than elsewhere if you measure by false positive rate. There is a mass failure to understand the importance of the assumptions underlying statistical inference (as you mentioned), as well as the importance of completely reporting your methods and data so that it is possible for others to intelligently draw their own inferences and replicate your work. In short, those journals have a culture that encourages "sexy" and "conclusive" results at the expense of the fundamental basis for successful science that we learn in gradeschool.
Interesting stuff
I hope you aren't a scientist.
I couldn't make sense of this post...
Plot the temperature by month from this paper (not just averages of months). The distribution of temps suddenly jumps upwards about two degrees starting in 1989 corresponding to when a sensor was switch out. They also corrected for some calibration errors and sensor drift, starting that year. But fail to tell us how exactly they corrected for the drift.
Here it is from all their data (including interpolated):
http://oi50.tinypic.com/2qn8x29.jpg
It is more obvious if you plot only the data they actually had.
Sorry, that isn't what I am talking about. I am talking about the pressures scientists in all fields face to get "significant" results.
I consider Nature well known for publishing incomplete, sloppy papers. This may not be true for climate science, but in other fields nature articles are a step above news articles.
Please stop propagating statistical myths. That is not the only thing that matters. What matters is how well the model fits the data relative to competing models.
I can't be sure but I suspect you are also falling into the trap of thinking at just because there is statistical significance it means that the research hypothesis is true.
I don't think it is that bad (outright fraud). It is just bias run rampant along with financial incentive to underestimate uncertainty combined with widespread failure of science education in statistics.
Well not even scientists are saying this. That is just the news. Go read what the scientists are saying then come back informed.
Look up monetary policy. It has more influence on your life that 95% of what your government votes on.
defense.... you mean the war department?
Just forget it then.
Regarded as healthy by who?
If the government promised to people they better pay it. That will come out of someones paycheck.
I don't know where I said that paying for public healthcare leads to a collapse. The current method of providing healthcare is basically stealing from future generations (which will eventually lead to collapse), if you think about how it works (at least in the US) it is a big ponzi scheme. Every generation needs to convince the ones after them to buy in to the insurance/Social Security scam. Please explain how else it could work.
A healthcare system coudln't tank an economy unless people were forced to pay into it. People wouldnt be forced to pay into it unless it had already gotten TBTF.
So..based on the metrics you provided, if the ponzi scheme collapses you will admit that the free healthcare was basically paid for by subtle enslavement of the third world?
If you disagree, then we and our children really could sustain our lifestyle forever without ripping off poor people?
I am kind of trolling, but really?
Yes they say the words that make you happy... do they actually deliver?
Which means nothing on its own. How should we account for the quality of care, taxes paid by each person recieving care, etc?
What metrics do you use to assess a country's economy?
Australia, I've been thinking about moving there anyway.
Name one. Real economists make money off it.
It was published July 2012... what are you referring to?
This one:
http://www.nature.com/srep/2012/120712/srep00507/pdf/srep00507.pdf
?