Data Overload?
Sometimes, when information derived from scientific research is abundantly available, we take for granted that some of the finer points of the data found in that research actually represents what we think it should. That is why interpretation of that data is key. This begs the question, what exactly is data and why is it essential for research?
According to Merriam-Webster, data, is defined as:
- factual information (as measurements or statistics) used as a basis for reasoning, discussion, or calculation <the data is plentiful and easily available — H. A. Gleason, Jr.>
- information output by a sensing device or organ that includes both useful and irrelevant or redundant information and must be processed to be meaningful.
The word data is derived from the Latin word datum, and is defined as:
- plural data: something given or admitted especially as a basis for reasoning or inference
- plural datums mathematics : something used as a basis for calculating or measuring <measuring the distance between datum points> < … make things more efficient for those of us whose work requires a time datum. — Robert Steinbrunn>
Scientific research that is independently conducted, peer reviewed and published is critical for this reason because the research is conducted without bias. There are no motives behind the outcome. However, that is not the case with paid research that has been conducted to support a particular view. Such is the case with research conducted by the agro-chemical industry which manipulates data to support a particular outcome. This isn’t helping protect the honey bees much less any of our pollinators. Meanwhile, commercial migratory beekeepers are barely keeping their operations functioning, often without any profit and at the bare minimum capacity. This will not last much longer.
Isn’t it time for industry to begin working with beekeepers to sincerely protect honey bees instead of working against them to protect their profits? Apparently, this may be wishful thinking, at best. What appears to be the constant in the equation regarding the cause of the global decline of our honey bees, are the rhetorical responses that it must be either the fault of the beekeeper or that of the most convenient culprit, the Varroa mite (which subsequently, has not been found in Australia, regardless of widespread bee losses).
What the future holds remains to be seen. Only one of two outcomes are indeed possible. Either our precious pollinators will finally be protected from neonicotinoid pesticides or we will not have enough honey bees to pollinate the crops we are accustomed to. China is experiencing this already as documented in the 2013 film, More Than Honey.
After 4 Years Of Research, Was Dumbing Up The Data Effective?
In this week’s segment of The Neonicotinoid View, host June Stoyer and Colorado beekeeper, Tom Theobald talk to Dr. Robert S. Schick from Duke University who was the lead researcher on a review of what is referred to as the Pilling’s study which was conducted by Syngenta. The research consisted of a four-year field study which investigated the long-term effects of repeated exposure of honey bee colonies to flowering crops that were treated with thiamethoxam. We’re going to take a closer look at this study according to what the data is proposing to tell us. To listen to the interview, press play on the video below.
“The Neonicotinoid View”, which is produced by The Organic View Radio Show is unique, weekly program that explores the impact of neonicotinoids on the environment. Tune in each week as host, June Stoyer and Colorado beekeeper, Tom Theobald, explore the latest research and news from the beekeeping community.
Highlights From The Interview
JS: Could you explain what were the findings? Why did Syngenta conclude that the research Pilling and his colleagues concluded were acceptable?
RS: I won’t speak for them but essentially what they did, in the paper and the paper is published in PLOS One was to say well, we’re basically not going to do a formal statistical analysis and the reason they chose not to was for the simple reason was that the study lacked a lot of replicates. So, they didn’t have a lot of different fields where bees were exposed and not exposed in different places. They only had a very small set of them.
And that, so that without getting into too many sort of quantitative details, that means there is a lack of statistical power, to make inferences on differences, whether they are there or not. So, they said because of that (lack of power) we are going to sort of forgo a statistical analysis and essentially plot the data, which they did, in their paper and then said if you had a low powered study, then the kind of result you would find is something that is very strong like very obvious, big effect! In their words, they said when they look at the data, we don’t see it.
So, that’s fine. So, basically we don’t see an effect and there’s not enough replicates in the data themselves to test for an effect formally, so we’ll leave it at that. That’s essentially the guts of it.
RS: Where we felt they went wrong was that they essentially just grasped the data. I won’t speak to them about whether they were doing this deliberately or not but they said here’s what the data looked like for bees that are exposed to the treatment and what they looked like for a control and just looking at them, we don’t see much of difference and so, therefore we’re going to say there’s no real difference…and most commonly accepted analysis in the scientific literature these days, certainly in ecology, that’s not sufficient to say whether or not something going on. You can’t usually hope to send a paper to a journal and say well, we did an analysis of these data and we really didn’t see a difference and so therefore there is nothing going on here.
Certainly, in my experiences publishing, I’ve never done that and any referee would just immediately say that you can’t call this statistical analysis just by plotting the data. That is a first and necessary exploratory step you would do in any such analysis. Circling back, on there. I won’t touch on the formal design of the study. It’s just not sufficient and it’s just not modern scientific practice to just plot something and say, well, there’s definitely no difference.
TT: Could you just explain it in a little more detail, the concept of statistical power?
RS: Sure. If you wanted to compare say two groups of something. Let’s say you wanted to compare the height of one group of students in second grade vs one group of students in the eighth grade to say whether or not these are statistically different. Can we look at these data and make some inference to say, yes, these groups are definitely different in height?
If you had 2 second graders and 3 eighth graders who have their own range of height, you might say, ok, they look different but there’s not a lot of data here to say for sure. If you did that same test (I’m going to grossly inflate the numbers) with 1000 second graders and 1000 eighth graders, all of the variations within each of these groups would be swamped by the overall difference between the two groups. In most cases, kids that are six years older, are going to be significantly taller than the younger ones. So, by having many more replicates, you can start to look at the differences. If you broaden that out to eighth graders all across the country, for instance, you are getting more and more information and so you have more power to discern whether any differences are significant and matter.
TT: Am I correct in saying that there are were no replicates in this study?
RS: Well, I do believe they had two for rape and three for maize. So, essentially, they had a control field and a treatment field. Within each of those they had six hives. The exact spacial arrays weren’t given in their paper nor was the actual lat/long coordinates between them.
So, one paper that criticized their study was the Hoppe P P et al. paper from 2015. They said in terms of the spacial separation it was likely that the control and treated could have flown back and forth between these fields. So, there certainly wasn’t sufficient replicates in this study.
About Dr. Robert Schick:
Dr. Robert Schick is a quantitative ecologist with over 20 years experience—both in aquatic and marine realms. He has held a number of posts—both domestically and internationally—in non-profit research labs, in federal research labs, and in academia. His research lies at the intersection of health, space and disturbance in marine systems, and has now expanded into human healthcare—with an active project looking at the spatial patterns and behaviors of human smokers.
Dr. Robert Schick
He has over 30 publications in the peer-reviewed literature, and has garnered over $2.5 million in research support. He is a past elected board member of the North Atlantic Right Whale Consortium, and has been a member of the Population Consequences of Acoustic Disturbance (PCAD/PCOD) working group since 2010. He holds a PhD in quantitative ecology from Duke University.