David Baker <This is one of my big problems with the BIP survey data.>

I have recently learned a few things.  The BIP survey was approved by a Human Subjects Committee (these committees are found at just about every University).  But there are should be two parts to that review and approval: 1) Ensure privacy and confidentiality, and 2) Check that the survey does what it's designed to do - asks the proper questions to get reliable information.  Otherwise you are wasting everyone's time and the reputation of the authorizing institution.


The BIP survey reportedly was subjected to a review for item 1.  No one seems to know whether anyone addressed item 2.  The required annual renewal by an east coast university appears to be a rubber stamp.


BIP has some good people organizing and managing the database.  Unfortunately, in attempts to protect privacy, the raw data is virtually unattainable to most, if not all, of us who are not part of BIP.  The US map with click to see loss is pretty, but the data presentation is very misleading.  In addition, no one could tell me what, other than very simple data aggregations, are used to analyze the BIP data.  I did learn that pie charts were chosen for 'ease' of visualization and interpretation.


Looking at these two issues: 1) In the map of the US, for my state, Montana, has the largest area colored in of any state in the U.S. regarding bee losses.  There's a great big patch of RED in western MT.  It appears that we have the most extensive and worst loss in the nation.


There are two problems with that presentation, the patch more or less corresponds to Flathead county (which is probably either a result of using the zip code to ID the region reporting OR the data came from a single commercial beekeeper - I suspect the former, but since I can't access the data, I don't know).   So, in a year when BIP said annual national loss averaged ~ 40%, that big RED area in MT was about 12% loss, as I remember. That's 1/4 the loss reported for the nation.   I doubt anyone in our climate does much better.


Why RED?  We've all been conditioned to associate RED with an alarm.  If anything, Flathead should have been color coded BLUE or GREEN.  So our apparent bee loss problem isn't a problem.


2) As per the pie chart approach - it's pretty, simple, and there's the rub.  It's too simple to be of much use.  There's no real detail, no error bars, just big buckets of data.




Since apparently no one outside of BIP can access (or at least Randy Oliver and I haven't been able to obtain access - maybe we're a special case), I asked who was doing the stats? - other than the basic data aggregations, simplistic analysis, but no one that I talked to could name anyone who was really digging into the data analysis issue.  That person may exist and may not be ready to release the resutls.  If so, I hope she or he will let us know when we might see something that takes a more in depth look.   I did locate a BIP participant who was looking at sensors and sensor data, but for the survey data - not much.



As most of you know, I pioneered landscape analysis of pollution using bees, chemical residues, and a few hundred volunteer beekeepers.  What you may not know is that my colleagues and I also pioneered data analysis and presentation methods using advanced geospatial statistics, namely a statistic termed Kriging.  Derived from the petroleum industry for oil well placement, and used by EPA to analyze environmental data, we were the first biologists to publish (in Science in 1985) data that was analyzed and presented as 2-dimensional isopleth maps.


In addition, now days, geospatial stats employing over 20 major programs, are readily available for both 2-D and 3-D mapping (examples appear in our 2015 Biosensors article).  As an aside, a major provider of geospatial software is a graduate of my university -  the University of Montana-Missoula.


Best as I can determine based on a variety of interviews, BIP may be conducting more in depth data analyses using conventional stats, but no one is applying geospatial analysis and reporting.


I've considered making more of an issue about the data lock-out - after all, the data was collected using federal funds.  When we published our nosema and virus paper, I had to upload the entire database - all 8 GB of it, to a data archive site, as required by PLoS One.  Took me all day.   Unfortunately, the grant funded data archive site - the so called permanent repository, lost its funding and has vanished.



So how does BIP manage to withhold the raw data? - it's become common for reputable journals to require making data available.  The privacy argument doesn't wash.  To protect privacy, actual names can be hidden by assigning each reporting individual a unique identifier.  That's a common practice - in fact for classes we now have to post grades with a student number, not a name.  Precise locations can be restricted to use for analysis only, not for publishing a map sticking a pin into each reporting place.  Non Disclosure Agreements stipulating what's private among multiple researchers are common.  The privacy issue is simply a matter of setting appropriate safeguards in place.


The only reason I'm not charging into the fray concerning data acces is how the survey form is set up. I don't think the data can be analyzed properly. Too many unknowns that were never addressed by the surveys.  It may be a "garbage in, garbage out problem."  However, without seeing the data, I can't tell and there's no way of knowing.  Is it garbage, or is it a simply a lack of in depth data analysis, or a combination of both?  Maybe it's better than I think, but without access to the raw data - no one knows.


I do not mean this as any critic of everyone who conscientiously fills out their form - but they should be mad as hell that BIP didn't design a better survey, doesn't provide access to the data, or do any advanced statistical analysis of their data - or at least let others access and analyze the data.  It's a black hole - might be something there and it may have much more information of use, but only BIP knows.


Also, I do not mean to criticize the hard-working BIP folks who are trying to provide something useful.  My sense is that they have one-hand tied behind their backs as a result of: how the program is managed, the diverse and  diffuse group of individuals at multiple locations trying to work with 'their part' and coordinate activities, and too much spread in terms of effort given the funds available.    Just running down who was organizing and inspecting the data, I found a group of individuals at several institutions - all trying hard to do their best with their part.


Overall, lofty objectives, but for the $$ spent, we should ask whether the bees themselves and their managers, the beekeepers, are getting commensurate value?


Is the information generated reliable and useful or are we witnessing a misinformation effect?  Maybe Kellyanne Conway was right - there are alternative facts!  The Oxford dictionary contends that alternative facts are lies.  I don't think BIP is lying, but how their facts (aka data) is collected, analyzed, and presented may be presenting a misleading picture.


I'm not going to try to get at their data, but I highly recommend that the leadership of ABF and of AHPA should step up and demand access to the raw data.  They should stipulate that the data be made available to them as representatives of the beekeepers of the US.  They in turn should then turn the data over to trusted researchers to evaluate.  To avoid conflicts of interest, those researchers should not be associated with the BIP effort. In essence, BIP has a Big Data project.  That's a hot topic in science.  Lots of groups would love to get their hands on that data set.


Jerry


P.S.  I'm sure someone will criticize me for sugar coating this opinion.


Full disclosure, at the moment I still do contract research, but none of it is financed by the US federal government.  I teach online University classes that our students pay to take and from which my co-instructors and I get a salary - but the classes were set up as a Montana University System program - no federal funds involved.  We have some overseas research ongoing. Currently, we do not have any non-profit funded work.  I'm 71; I've failed at retirement.  Like Berkeley Breathed who resurrected his Bloom County comic with Opus and the gang, I don't have anyone editing me.  Breathed quit producing his strip over 20 years ago because of his frustration with being censored by editors.  He posts his new strip on Facebook and on a comics site - and he's having a lot of fun with the strip and with being able to do whatever he wants.


In my case, I've turned to a major new effort - you'll see it some time this summer.  We're in a beta-testing phase.  I'm having fun putting together our new venture, when I'm not being a curmudgeon.  The latter seems to be a part of becoming an old fart.  My wife refers to me as grumpy.
























             ***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software.  For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html