Friday, July 24, 2015

The Bee Crisis Is Over, But Not the Data Crisis

Climate Change Threatens Bees: Even More Evidence : Animals : Nature World News:

The linked article tells us that it really really looks like "the bee crisis" is partially caused by Global Warming -
With the threat neonicotinoid use poses to honeybees gaining international attention, researchers are now focused on finding other factors that are contributing to a worrying decline in bees across the globe. Climate change, they say, is certainly to blame, and parasites may be one reason why.
  This article says what I've been hearing more of from  techie or "right wing" news sources the last year or so ... there isn't any bee crisis.
I have great news for honey lovers everywhere. The Canadian honeybee industry is thriving. Despite those headlines about mass die-offs and and killer pesticides, the number of honeybee colonies is at a record high. Last year, according to Statistics Canada, nearly 700,000 honeybee colonies produced $200-million worth of honey. Bee survival rates have rebounded even in Ontario, which was hard hit by unusually high winter die-offs
The purpose of this post is not to prove which position is right, although I will give my opinion. When a problem is indicated as a "crisis", we know that the title is more intended to convey emotional content and grab interest rather than provide factual information. That indication ought to have the effect of a rattlesnake rattle on our focus as to likely trouble with the assertions in the article.

Couple "crisis" with "Global Warming" -- something that has been crying wolf for over 20 years now, but for the past 15 been forced to quietly try to suppress the news that it has "paused", -- and the chances for "correctness" drop rather dramatically.

My real point here is modern data and "analysis" overload. There is so much data now being collected and so much statistical computing power available to run all sorts of "Spearman Rank Correlation", "Guttman Scaling", "Mann-Whitney U", regression, ANOVA, etc against any sort of data plucked from who knows where and get lots of REALLY impressive charts!

One can definitely show that "Women are better drivers and ice cream sales cause drownings" with no problem at all ... Who drives your car after you were out for dinner and drinks or the traffic, weather, etc are bad?  Who drives the most miles? There is a PERFECT correlation between drowning deaths and ice cream sales -- only it is because of a common third factor -- hot weather, and CORRELATION IS NOT CAUSALITY!

But our brains are wired with a bias to ASSUME that correlation IS causality -- from a primitive POV it makes all the sense in the world to believe that if A and B happen together, they are probably related. Good for survival in nature -- not good for correctness in a wired world with lots of computers. Do enough analysis on data selected with a goal in mind, and you are certain to be able to find "correlation". Add in selective and biased journalism and it is no wonder that people have all sorts of assumptions that they hold as gospel that have at best tenuous and only very specific well manicured connections to the real world.

Statistics are like gauges on your car -- they look at certain data with certain assumptions. Hook the oil pressure gauge to the water temp sensor and vice versa, you get bad readings. ANY of the data is only as good as the sensors -- the engine may have low oil pressure but the sender is not correct for example. Most likely of all, the engine is fine but the sender or gauge have failed.

Today we have bazillions of "sensors" (data collectors) hooked up to all sorts of things on which little if any data was collected before and "scientists" -- many of them little more than amature statisticians and computer software users with a HUGE pressure to "publish something impressive... maybe even about a CRISIS!" spewing "reports / studies / papers" at a rate far beyond anything seen in history. So there is almost constant "crisis".

Add to this the fact that the bulk of them are in one way or another funded by giant governments whose primary mission is the grow government power in any way they can, and it is no wonder at all that "crisis" is MUCH more common than  "boring news day when not much really happened". Remember, the goal of the government is to increase it's size and power -- so government funded research should be instantly suspect if it's result is serving that end.

The real "crisis" is an out of control government growing in power and scope every day.

No comments:

Post a Comment