October 09, 2012

Thinking Science 1: Useful Data

     Talk radio is one of those areas that once you start listening, you never know what information you might gather.  I was listening to how the state of Oregon fish and wildlife department is setting fines for hunters for not reporting their hunt to the state.  The fines will be tacked to the following year's tags - the state will use it's power of coercion to force hunting into extinction, by making it too expensive a hobby.  Political ends achieved through pocketbook regulation.
     What interested me is that the ODFW wants to use the data to manage wildlife populations, now that predator hunting with dogs is outlawed.  If the cats, cougars and the like, take a deer a week, and there are say a dozen cats per watershed - then there may not be enough deer to go around.  Or elk.  Nature, of course, has managed populations of animals for oh, a few years - so why is this bureaucracy attempting this agenda?
     The problem lies in the data collection process here.  The observer has to report accurately, but has no means of sight to the information.  The hunter, who is already strapped to hunt, is now responsible for providing factual information from memory to people coercing the information - er - which red flag are you screaming at.
     Data collection is serious work.  The first question is whether you use a standard protocol, or a specific protocol for your process.  The protocol is set up to allow even level comparison - apples to apples.  The challenge is nuance - every different situation is unique and one size fits all measuring according to a protocol is extremely limiting.  You can change a protocol on the fly - just make a note of what you did.
     There is another level of measurement that addresses this challenge - the meta-data level.  Meta-data is recording things like noting the weather when scoring a football game.  The result of the game doesn't change, but knowing the game conditions allows a frame of reference for understanding the strategy within the game.  The reason for collecting the data should be clear, not a fishing expedition - otherwise you waste people's time, like the ODFW is wasting the hunters time (and collecting money - hmm).
     Measurement relies on both accuracy and precision.  To measure accuracy and precision, we embed quality control and quality assurance.  The net affect of these big words on quality is that we measure a few extra samples - duplicates, blanks, spikes and unknowns.  This generates the units of variance and the limits of the measurement.  We also look at the data after the fact (never during the measurement) - to be able to throw away bad data - based on being over two standard deviations from the mean.
     When collecting data - you should not have preconceived notions.  The idea is to use the data to measure validity of a theory - if you look at the data and change things to make the answer that you want, then you need to work in the pharma industry.  It doesn't help you to answer the question that you pose, when you go through the work of collecting good data.  The observer is de facto part of the experiment, so scientists are really as embedded in their work as journalists - we tend to trust what we personally measure and have a need to constantly vette the information that we receive against what we think we know.
     It is absolutely necessary to have opening and closure to each experimental measurement.  To really have enough information for statistical significant, there must be enough repetitions of the measurement to allow statistical significance - using standard measures like the chi test or the student's t-test.  If you have never heard of these tests, then maybe you wish to find someone else to make the measurement observations for you to interpret.
     If this interests you, please contact Doc through the Northwest Education and Training Institute at the Oregon Natural Resources Research Institute.  Soon to be associated formally with the Organization for the Advancement of Knowledge.  The game theory for the next game is being developed on the fly - all we can do is all we can do.  Find what you can do and plan to do it well - cooperative community is a team sport.

No comments: