Thanks for the thoughts, Kevin. A few comments about statistics and the purpose of the BBS. First, statistics. Statistics do not determine what is real, they determine the degree to which the data do or do not conform to the model of "significant difference" on which a particular test is designed. All tests have their own particular assumptions such as linearity, normality, constant variances, etc. to which real world data rarely conform precisely. A non-significant trend is not in fact "spurious by definition." A "spurious" trend is an apparent trend that is in fact not real; as statistics do not determine what is real, they do not determine what is spurious either. Apparent trends or differences that fail to be statistically significant may be real, they may not. They may in fact be highly important. The first Starling that showed up in Tennessee was not a statistically significant change from the long string of zeros that came before it; it was however extremely significant ecologically. And, of course, as Kevin points out, a "statistically significant" change is not necessarily a meaningful one. The mass media love to play up studies that show significant differences, ignoring that these differences are often tiny. This is why I presented the raw trend graphs for the species, regardless of their statistical outcomes. A trend is a trend, there are many factors that determine if it is a trend worth thinking about, the results of statistical tests being only one of them.
About confidence levels -- when you have 84% of the species showing a statistically significant change at the p = 0.05 level, that does mean something. In contrast, when I applied a similar analysis to my own data from Lewis County, using data from just three routes and only a 5 year time interval, I only had 6% of the species show a "significant" difference at the p=.05 level -- in other words about what one would expect from spurious differences. So I think it is pretty safe to conclude that there has been real, meaningful, extensive, large-scale change, affecting the majority of the 104 species considered, not just among the species that show dramatic 90% drops or 800% increases. Now any individual trend might be spurious, but the big picture of generally shifting populations is worth talking about. The REASONS why these shifts have occurred are another matter entirely, and the BBS was not designed to determine that.
Now about the purpose of the BBS. It was always intended to be a broad-scale monitoring program -- the big picture, the coarse view. It's number one job is to detect signs of population changes. It was never designed to actually measure populations themselves. Its severe limitations caused by observer effects, roadside biases, drastic changes in detectability between species and between habitats within a species, etc. were recognized from the beginning and had to be baked right in. All sampling protocols require tradeoffs and compromises, all have severe limitations, all data are flawed in some ways. The bigger your scope in space, time, and number of species, the more limitations you have to accept. The BBS serves to give an overview and sometimes to sound an alarm bell. The more extensive "disaggregated" studies provide much more info; but I can absolutely guarantee you that they will NEVER be conducted for every species across every state in every year! Thorough research on individual species only trumps more superficial research on large numbers of species if you only want to know about one species. You need the overview to select which species are of the most interest; otherwise you might be playing your trump cards up the wrong tree (how's that for a mixed metaphor?). Within the context of their limitations and intentions, the BBS data are actually rather exceptional; there is nothing comparable for the big picture, even if that picture is viewed through a tinted, blurry, smudged, constantly wiggling glass.
As the saying goes, it is better to have flawed data we can argue about than no data at all!
Bill Pulliam
Hohenwald TN
=================NOTES TO SUBSCRIBER=====================
The TN-Bird Net requires you to SIGN YOUR MESSAGE with
first and last name, CITY (TOWN) and state abbreviation.
You are also required to list the COUNTY in which the birds
you report were seen. The actual DATE OF OBSERVATION should
appear in the first paragraph.
_____________________________________________________________
To post to this mailing list, simply send email to:
tn-bird@xxxxxxxxxxxxxx
_____________________________________________________________
To unsubscribe, send email to:
tn-bird-request@xxxxxxxxxxxxx
with 'unsubscribe' in the Subject field.
______________________________________________________________TN-Bird Net is owned by the Tennessee Ornithological Society Neither the society(TOS) nor its moderator(s)
endorse the views or opinions expressed
by the members of this discussion group.
Moderator: Wallace Coffey, Bristol, TN
wallace@xxxxxxxxxxxxxxxxxxx
------------------------------
Assistant Moderator Andy Jones
Cleveland, OH
-------------------------------
Assistant Moderator Dave Worley
Rosedale, VA
--------------------------------
Assistant Moderator Chris O'Bryan
Clemson, SC
__________________________________________________________
Visit the Tennessee Ornithological Society
web site at http://www.tnbirds.org * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ARCHIVES TN-Bird Net Archives at //www.freelists.org/archives/tn-bird/ MAP RESOURCES Tenn.Counties Map at http://www.lib.utexas.edu/maps/states/tennessee3.gif Aerial photos to complement google maps http://local.live.com _____________________________________________________________
