Scientists are trying to get rid of an elementary degree used in science


For a while now, scientists have been discussing what they need to use with one of the most famous devices used to interpret scientific evidence, the concept of concept; statistical significance; Some people think he is alright. Others are ready to make it smaller, but others still argue for extinction.

By accepting that we were dragging the side and rubbing the concept of its probability values, we would have to replace it with a better idea. This issue recently The American Statistician there are some ideas – 43 of these are true.

To revive Winston Churchill's famous phrase, the value has been the most damaging way of finding useful ideas in science… but for all other ways that have been tested periodically.

It's not a loss indeed. On its own, the value tells you just how much you have supported the wrong horse in your exam.

Normally, if the value falls below 0.05, that means that less than five per cent of the chances are that the hypothesis – interpreting for your opinions that isn't part of your idea – is certainly on the back of your results.

Why are five per cent? History is, of course, true. Gamble is better than the first 10, although it isn't as stringent as one per cent. It does not have a magical quality otherwise.

There is a collection of statistical tools to use the researchers to assess this number of importance. Difficulties arise when we try to translate this mathematical overview into something that the computer computers inside our skull can get into action.

Our brain does not deal well with probability. Perhaps this is related to the fact that we have never moved to be worried about the excitement of being fed by a bear when he is already progressing.

We deal better with a quick report of a false or misleading statement. It is therefore difficult to swallow p <0.05 easily, leaving it unattractive.

University of Georgia Nicole Lazar told NPR writer Richard Harris that, "The world is worse."

Together with the executive director of American Statistical Group Ronald L. Wasserstein and vice-president who retired from the Allen Schirm Mathematica Policy Research, Lazar edited a preliminary edition of an upcoming collection of results on how to improve our performance. ;

It is clear that there are ways in which the probability figure will benefit us, but only if we don't do it with foolish things, as we realize it's more than just a competitor. correct explanation.

"It is important to know what you will not do with values, but that is not enough," write the three.

“It is as though statisticians want consumers to keep statistics of the spells and tents to keep up the development of new scientific research without replacements. "

Articles are not fully on the issue of what these building materials would look like. But many of them share some basic elements.

Some people believe that some should be allowed to include data data and method descriptions that will bring added vitality, withdrawing from the opportunities but still arguing for t one explanation.

“We need to learn to accept uncertainty," some authors write Nature an idea.

"One practical way to do is renamed trusted differences such as compatibility rods and the interpreting in a way that avoids too much comfort."

This is not just a value-for-value update. Researchers would have to explain actively on the practical impact of values ​​within these times.

The ultimate goal is to establish practices that will avoid the barrier to real or false thinking and rather to strengthen the uncertainty that underpins the scientific basis.

Science at its heart is a conversation after everything. Guests, technologists and engineers are the guests who dislike this curse to a definitive conclusion, but for the scientists running the next step in research, not the scientists t P-value is not helpful.

Unfortunately, it has been a finish line in the race for experience, where funding for research and public awards is waiting for the winners.

Its use of cultural expression that is so strongly focused on just some editorial edits and some science papers with a good argument. The pedigree has been part of the popular science base for around a hundred years now, so it will be around for some time.

But maybe this sort of thought gives us some useful stones to get out of statistical meaning, into a place where uncertainty can be identified.

Source link