News for hackers. Stuff that matters.
Re:Naked mole rats have more than one weapon against aging by dekhn on Tuesday October 01, 2013 @02:26PM (#6478097)
In general, when you read statements like this in an article: "lthough we cannot directly test whether the unique 28S rRNA structure contributes to the increased fidelity of translation, we speculate that it may change the folding or dynamics of the large ribosomal subunit, altering the rate of GTP hydrolysis and/or interaction of the large subunit with tRNA during accommodation, thus affecting the fidelity of protein synthesis."

it's a strong suggestion that the authors are searchng for impact- that is, they found some mildly interesting data from an experiment that is replicable. The rest of it- the speculation- can just be tossed out. Their data is not strong enough to support it, so it's just speculation.

I once had a prof who seemed really arrogant to me. He said, "when I read a paper, I skip to the methods. if the methods make sense (IE, they actually know how to run a basic experiment) then I look at the results. That's all."

I thought that was arrogant, but in retrospect he was giving sage advice: the only content in most papers is in their data sections, not in the introduction, conclusion or discussion. Spend your time looking at the data, not what the authors think the data means, and make up your own mind.

Re:Naked mole rats have more than one weapon against aging by phaemon on Tuesday October 01, 2013 @03:56PM (#6478684)
That's fine if you happen to only read papers that are dealing with data you understand. In short, only papers in your own field.

You can't understand the statistics, and therefore the conclusions, if you don't understand the data. Your Professor was certainly arrogant (and foolish) if he didn't comprehend that.

Re:Naked mole rats have more than one weapon against aging by dekhn on Tuesday October 01, 2013 @07:24PM (#6479740)
Disagree completely. I can easily read papers well outside of my area (biophysics), such as climate research, and follow most if not all of the analysis. my field along includes hundreds of subspecialities, and I must frequently read and follow papers in another specialty.

Statistics itself is indepedent of the study details; it's just the application of probability to hypothesis testing and I know of no science which cannot express its models in the form of probabilistic analysis. That's why people who are physicists can move to medicine and produce useful criticisms of study design (cf Stan Glantz).

The one big issue with studies outside of my domain is that it takes more time and attention to read because I have to refer to citations or online resources to understand the technical details of their domain.

Re:Naked mole rats have more than one weapon against aging by phaemon on Wednesday October 02, 2013 @05:29AM (#6481475)
A while back, I picked up an old introduction to statistics book from the 1950s, just out of interest to see how things were done before computers, and I remember that near the beginning, there was a cautionary tale.

It started out with a simple bimodal distribution and explained that it was the results of some student tests. The reason it was bimodal was simply that it was the results from two different classes, the "smart" class and the "others". And talked about how this is a classic way to spot that you are dealing with two populations.

Then it related the tale of First World War marksmen. It turns out that when they were training their sharpshooters, the results, when plotted, had a bimodal distribution where they were expecting a normal one. They took this to mean that there were such things as "natural shots": men with a natural ability for shooting. If they could identify early which ones were the natural shots, they could concentrate their training much more effectively.

But it didn't work. After they separated the groups, it turns out that not only did the "natural" group also have a bimodal distribution, but they actually didn't perform any better than the "bad shots". To cut a long story short, they discovered that the bimodal distribution was actually an artifact of the scoring system they were using. When they recalculated the data with a different scoring system, they had a regular normal distribution.

It's a simple story with a simple moral: you can't interpret your results if you don't understand your data. Maybe I'm wrong though; it's been a while since I did stats and I've driven most of it from my memory. :-)

Oh, and it's not at all sad that I buy second-hand statistics books for fun reading. It's just not.

Re:Naked mole rats have more than one weapon against aging by dekhn on Wednesday October 02, 2013 @09:58AM (#6482379)
The scoring system would be described in the methods section and anybody with a decent data analysis and experimental design background would ferret that out.