The Washington Post reports that the "lower 48" states of the USA are enjoying spring-like weather. It quotes a meteorologist as saying 1495 record high temperatures have been reached during the month of February (as against 10 record lows); among them:
[Ed Note: it is actually Mangum, OK, not Magnum. The original WaPo article is incorrect.]
(Score: 0) by Anonymous Coward on Monday February 20 2017, @05:01PM
Exactly. A single six, or even a six two times in a row doesn't mean the die is biased. But if you get ten sixes in twenty rolls, it is not rational to assume the die is fair. Even though the result is certainly consistent with a fair die; every possible result is, including a million sixes in a row.
(Score: -1, Offtopic) by Anonymous Coward on Monday February 20 2017, @05:15PM
And if a dice isn't fair, we can conclude CO2 emissions are causing the entire temperature increase, right?
This NHST-approach (where you test a strawman) needs to end before it destroys our civilization.
(Score: 1) by khallow on Monday February 20 2017, @05:30PM
This NHST-approach (where you test a strawman) needs to end before it destroys our civilization.
This is a red herring. We have a model and evidence supporting that model. It has nothing to do with null hypothesis significance testing.
And let us note that NHST is a valid scientific procedure no matter that it can be abused. I've discussed it here [soylentnews.org] and here [soylentnews.org]. For example, a classic place to use NHST in a valid way is when you have a large amount of complex data, have at best very poor models of the data, and want to quickly find possible patterns in the data (such as health care or customer data at a retail store or grocery). NHST can do that.
(Score: 0) by Anonymous Coward on Monday February 20 2017, @05:42PM
No one believes "the climate never changes"[dice is fair] null hypothesis, so rejecting it is a waste of time. It doesn't tell us anything about CO2, etc.
(Score: 1) by khallow on Monday February 20 2017, @06:26PM
No one believes "the climate never changes"[dice is fair] null hypothesis, so rejecting it is a waste of time. It doesn't tell us anything about CO2, etc.
I did say NHST was a red herring here.
(Score: 0) by Anonymous Coward on Monday February 20 2017, @07:30PM
You should have, NHST is bullshit that needs to die in the hottest fire we can find. We're rarely in a position where we don't have an established procedure, we usually have some sort of status quo that we want to evaluate other options to. The relevant question is usually whether or not there's enough evidence to support making a change in a particular direction, rather than whether or not we have sufficient evidence to support the status quo. The status quo virtually always has enough substance to justify it's use, we wouldn't be using it if it were complete BS.
There's a reason why spam filters typically run on bayesian inference rather than NHST.
(Score: 1) by khallow on Monday February 20 2017, @07:49PM
We're rarely in a position where we don't have an established procedure, we usually have some sort of status quo that we want to evaluate other options to.
I mentioned two such areas: health care and customer data for stores.
There's a reason why spam filters typically run on bayesian inference rather than NHST.
Who was claiming NHST was the only tool out there?