Apr 25, 2019
Happy Earth Day (week), everyone! It was back in 1970 that the first ever Earth Day was celebrated at over two thousand colleges and universities, even more primary and secondary schools, as well as hundreds of cities and towns across the United States. Today more than 190 countries take part in Earth Day activities.
Celebrate our beautiful Earth!
So, it appears the statisticians have significant concerns over the significance of significance. So much so that they have decided that our old measures of significances are no longer significant. And I have decided that this story will work much better if I significantly increase my vocabulary for the remainder of it.
Statistics has been the bane on many a student’s life. I for one studied biological sciences for 2 reasons; 1 I love it and 2 it involves much less maths than any of the others. So you can imagine my joy when I rocked up to Statistics 101 and had to contend with ANOVA, the t-test and P values. So did that mammoth slog end up being a total waste of my time? Mercifully not.
As hard as maths may be to understand, the wonderful thing about it is that it does not go out of date. Logic is logic and maths is the system of logic that we use to underpin this institution of science. It is the foundation that we build our knowledge upon. Change it and you literally have to start working from the ground up again. So the maths is going nowhere. What some 800 statisticians are proposing, in its own special edition of the American Statistician publication, is that we only bin the concept of statistical significance.
Afterall, we need statistics to make science work. In a field of academia where there are no certainties beyond maths, all this leaves us with is uncertainty and uncertainty is no use to anyone unless you can quantify it. You then, at least, know what it is that you’re dealing with; and statistics allow us to do that and there are various statistical tests that can help us with this.
In particular it is the P value that is coming under the heaviest fire. And it is easy to see why. P values have become judge, juror and executioner to papers across the scientific spectrum. This single number that interprets the probability that the results of your experiment are repeatable has become the gold standard, not just for result significance, but overall experimental design and credibility.
And here’s the best bit. The very point of statistics is to interpret raw data in a way that allow us to make informed judgment calls upon it, in the full knowledge that nothing in science is absolute. We then go and stick an absolute cut-off onto that interpretation. P=0.05, or 5%. We don’t even provide a margin of error! Sorry Dr. Watson, you study was only P=0.051, no funding for you today but congratulations Mr. Holmes your study with a P=0.049 represents stirling work, here have bucket load of cash to continue your studies! This may sound like a ridiculous over exaggeration, but shit like this really does happen. And the worse thing about it is that the difference between a P of 0.051 and 0.049 is not even statistically significant!
So where did this mystical P=0.05 even come from? Considering its level of importance, surely some rigorous work went into establishing it. Well, its origins can be traced back to 1925 and famed statistician Ronald Fisher. In a monogram he proposed this cut-off and wrote: “it is convenient to take this point as a limit in judging whether a deviation [a difference between groups] is to be considered significant or not.”
Convenient?! We use this number because it is convenient?! I’m sure that it’s not all that convenient to poor Dr. Watson! But I get his point.
If I learnt anything in Statistics 101 it’s that statistics is not an easy game. There are some truly fiddly sums in there so it’s not practical to expect everyone to pull out the most appropriate statistical test all of the time. Standardising the system would make things much easier for everyone… in 1925.
But times have changed. We carry more computational power in our pockets these days that we had sitting our desktops only a few years ago. Power that would have look like black magic to Roland Fisher, power that can work out those statistics for you. We are not constrained by the same limitations that we were a century ago, so why do continue to apply those same limitations to our science?
Mainly because we have become institutionalised. We may well know the faults but we don’t necessarily know the solution. We are talking about changing something that has become fundamental to the current scientific establishment. Whilst I hate to make the comparison, but I’m british so I can’t help it, it’s a bit like Brexit. We know what we don’t want but we don’t know what we do want as an alternative.
But when scientists like Darwin, Galileo, Boyle & Franklin managed to achieve what they did in an age before statistical significance. Surely we can find a way for science to work afterwards.
A team of Russian researchers have found the frozen carcass of a 42,000 year old foal in the permafrost of the Verkhoyanks region in Siberia. This Lena Horse foal belonged to the Lemnyaska breed which went extinct 4000 years ago.
And that’s not it. This discovery was a surprise because the foal carcass showed the presence of liquid blood and urine.
Semyon Grigoriev, the head of the Mammoth Museum at North-Eastern Federal University in Yakutsk remarks that prior to this only once was liquid blood found in an animal of the Pleistocene epoch. In 2013 Grigoreiv’s team discovered a carcass of mammoth about 11,700 years old in which liquid blood was found.
If we just think about this for a second, blood should technically coagulate or has been seen to turn into a powdered form in several well-preserved carcasses in the past. However, in the case of the foal or the mammoth the biological fluids were preserved well due to the permafrost conditions.
Grigoriev’s team has been working on cloning and resurrecting the woolly mammoth. They have been working on retrieving viable DNA from the mammoth carcass, which would then be inserted into an elephant embryo. An elephant could be then used as a surrogate to clone the Mammoth.
On similar lines, Grigoriev is working in collaboration with a South Korean team to clone the foal. They have been trying to retrieve viable DNA from the organs of the foal carcass for the past two months, with no success. If successful, modern horses can be used as a surrogate to clone the Lena horse.
It is important to make a note that the South Korean collaborator, Hwang Woo-suk was under fire and guilty of faking data on human stem-cell cloning experiments in 2004-2005. However, after laying low for several years he has cloned around 1000 dogs so far and is working on the mammoth cloning project as well.
With respect to cloning an animal from tens of thousands of year old carcasses, the biggest issue is finding good quality DNA. The degradation process starts immediately post the animal’s death and in spite of excellent preservation like permafrost, in this case, one may not find good quality DNA.
Although they have not been successful yet, Grigoriev says “we in Russia say that hope dies last.“
That concludes this episode of the Blue Streak Science Podcast.
If you have any suggestions or comments email us at email@example.com
If you have an iOS device like an iPhone or an iPad you can get the Blue Streak Science app from the App Store.
This show is produced by the Blue Streak Science team, and edited by Pro Podcast Solutions.
Our hosts today were Chris MacAlister, and Amrita Sule.
I’m JD Goodwin.
Thank you for joining us.
And remember…follow the science!