Cwd article on outdoornews

Discussion in 'Whitetail Deer Disease' started by Groundsize, Feb 2, 2019.

  1. G20man

    G20man Banned

    Messages:
    1,157
    Likes Received:
    1,698
    Location:
    Detroit

    So you support the ban within the zone.
    I read that as you do not support the ban outside the zone.
    Am I interpreting that correct?
     
  2. rork

    rork

    Messages:
    329
    Likes Received:
    426
    Location:
    Dexter
    Those questions are actually from a different thread I think, which is not very polite.
    I was estimating prevalence of CWD in northern counties of IL with data, saying it was not growing very fast. I do not have time right now to get the data from WI and test if it grew faster there than in IL, cause I wanna kill some tumors, but I'll bet it did. If you have time you can get the data and do the tests yourself and tell us if it is interesting or not. Mergerovic paper estimated 0.63% increase per year in WI, and I'll be it increased significantly more there than in counties I looked at.

    You seem fixated on power unnecessarily perhaps, since once I have data I can just model and test a hypothesis. If I find a interesting difference I do not need to compute the power. It is only if I fail to find a difference (or before I do the experiment) that the power is interesting. If I have 2 sample of 1000 deer, and proportions are .03 and .06, there's 88% power to obtain p<.05, so I think doing a test has chances. That's modelling it as binomial, and sure, fancier things are possible. I'll see if I can find time. If someone wants to point me to the easiest raw data for WI, that would speed it along.
    upload_2019-3-19_13-19-17.png
     

  3. swampbuck

    swampbuck

    Messages:
    20,754
    Likes Received:
    10,768
    Location:
    Majinabeesh
    Correct outside the core and buffer zone, but if they want to remove other vectors, such as APR's, statewide. Then I would agree with the statewide baiting ban.

    So how about you ?

    Did you notice yearling buck harvest was listed BEFORE baiting

    Screenshot_20190101-094600.jpg
     
    Last edited: Mar 19, 2019
  4. otcarcher

    otcarcher Banned

    Messages:
    3,697
    Likes Received:
    3,250
    You completely missed the point of what I said again.......

    Numbers matter in this case. It's not simply about you putting random numbers you gleaned in to test a hypothesis here. It's about specificity, and what numbers matter. The number that matters here, that you've missed yet again....is population within the test. That is the answer. You need to know the deer population from within the area.

    Is 70 deer tested from a population of nearly 5000 deer enough to reliably determine prevalence and significance at .06? No, it is not. But your math assumed that it was.

    Done with this.....
     
  5. rork

    rork

    Messages:
    329
    Likes Received:
    426
    Location:
    Dexter
    If you could be even close to fluent it would help. As it is you seem unable to express yourself clearly. Your "is 70 deer enough" sentence is a statement about power I think. There is far more than 70 deer in the IL surveys if I use 4 counties, and if I get to use data from multiple years, sample sizes are over 1000. You are claiming there's no power without giving a calculation. I gave a calculation. See the difference?
     
    Liver and Onions likes this.
  6. Lightfoot

    Lightfoot

    Messages:
    1,048
    Likes Received:
    1,845
    Location:
    Above the bridge EUP
    If you expand the survey area to 4 counties to increase sample size would you not be increasing the size of the population tested as well? Kind of like chasing your tail.

    Using data from multiple years makes no sense to me, perhaps I'm missing something. If you stretched it out over a long enough period could you in effect have a higher sample size than actual deer numbers for a given area? You need a snapshot, not a combination of a picture and movie clip.
     
    Dish7 and otcarcher like this.
  7. otcarcher

    otcarcher Banned

    Messages:
    3,697
    Likes Received:
    3,250
    Fluent? Ha, ha. Ok.

    No, 70 deer is the number they tested in Boone County, out of a deer population of ~5000. You cannot just add deer from random counties and say it has power. It doesn't, without knowing deer were tested at significant enough levels based on that counties deer population and expected CWD prevalence rate.
     
  8. otcarcher

    otcarcher Banned

    Messages:
    3,697
    Likes Received:
    3,250
    Correct.
     
  9. Lightfoot

    Lightfoot

    Messages:
    1,048
    Likes Received:
    1,845
    Location:
    Above the bridge EUP
    You do know it wasn't really a question. ;)
     
    otcarcher likes this.
  10. otcarcher

    otcarcher Banned

    Messages:
    3,697
    Likes Received:
    3,250
    Ha, ha. Yes.
     
  11. G20man

    G20man Banned

    Messages:
    1,157
    Likes Received:
    1,698
    Location:
    Detroit
    Ok good.

    I assume that means you are ok with aprs outside of the core and buffer areas such as the NW12 and thumb area.

    I would agree outside of disease areas those things should be allowed.
     
  12. rork

    rork

    Messages:
    329
    Likes Received:
    426
    Location:
    Dexter
    By using 4 counties rather than one the sample size increases which increases power. Example: If I see 10 of 100 here and 5 of 100 over there I get P=.18 (Z test of two proportions, the cheap one), but if I see 100 out of 1000 here and 50 out of 1000 there (same proportions, just bigger samples) I get a P=.000022. I am much more certain the two places are different. P=.18 means that even if there is no difference I will get P less than .18 about 18% of the time.

    If I'm trying to estimate the increase over 10 years, I can average 2005-7 and 2015-2017. The difference is like averaging the 2015 vs 2005 change, the 2016 vs 2006 change, and 2017 vs 2007 change. I don't think averaging 3 consecutive years is too many if the curve is not too steep and you are trying to estimate a 10 or 15 year change. Could go to just two. If we want to work hard we'd need a model for what the curve looks like and use all the data to estimate the curve (what a science paper would do probably), however we might find that straight lines work pretty good in IL, but are completely unrealistic in WI.
    Ah, it has dawned on me that since we jumped threads the original question may have been lost - it was whether rates went up faster in WI than IL. I wasn't asking if they went up in this or that individual year.
     
    Last edited: Mar 20, 2019
  13. rork

    rork

    Messages:
    329
    Likes Received:
    426
    Location:
    Dexter
    Those are neighboring counties not random ones. If I use the same sampling method every year it doesn't much matter how many deer are in the county. For example imagine a random sample of 100 deer. It matters very little if there are 1000 or 100000 deer in the county. Your intuition is what you are using, and it might not be that informed.
    I realize we expect hunters to get more CWD animals than the average for the population though. But that's probably fairly similar everywhere. It's likely a reason some studies just use hunter killed deer though.
     
  14. otcarcher

    otcarcher Banned

    Messages:
    3,697
    Likes Received:
    3,250
    Rork, it doesn't increase any power if you don't know whether enough deer were tested in each particular county to accurately detect CWD at it's current prevalence rate respectively. All it does is combine numbers and bias overall results, much like what Illinois did in testing more in non-CWD areas.

    Lightfoot also touched on that in his post. Combining counties without knowing that information does absolutely nothing to increase strength or power of the statistical sampling because as you increase county samples you also increase the deer population and thus number of samples needed to accurately discern prevalence.
     
  15. otcarcher

    otcarcher Banned

    Messages:
    3,697
    Likes Received:
    3,250
    As an example, given a county (A) with a deer population of 5000 deer, current prevalence hypothetically calls for the testing of 250 deer but only 70 were tested. In an adjoining county (B) with a deer population of 5000, current prevalence also hypothetically calls for testing of 250 deer and 275 were tested.

    In your mind, we combine the two counties to gain power in the sample. We get a total testing of 345 deer and think prevalence rate would be accurate. The problem is that it is not because you were 180 tests short in county A and have no clue if combining results did anything.
     
    LabtechLewis likes this.