Philip
Bulldust. You showed graphs in your book withourt any measure of
significance and now accuse me of incompetence! I do not normally
respond to your messages - and will not in future - since it is not in
the best interests of the GBS but this one is rude beyond belief.
Martin
On Sun, Mar 23, 2008 at 5:27 PM, Philip Veerman <> wrote:
>
> Further on this: Martin wrote:
>
> "The main thing that can be said about the long term trend for Pied
> Currawong abundance in the GBS results is that there is no trend. The
> values of the correlation coefficients for linear and polynomial trends for
> 21 and 26 year results for the species range from just over 0.1 to nearly
> 0.3. These are well below the values that would indicate a trend
> significantly different to zero.
>
> The facts are that the trends are obvious and strong. There are two trends
> occurring and they need to be teased apart and seen separately, especially
> because the biological impacts are different. We should be more concerned
> about the biological impacts than the statistics. It is during the summer
> that the PC is a major predator of other birds' nesting and during the
> winter that the PC spreads seeds of woody weeds. We need to see the GBS data
> in context. There are plenty on long standing bird surveys around the world
> that are not continuous and only comprise summer or winter counts. If we did
> that, then the results would be very different.
>
> I have just done a regression analysis for the first 21 years of GBS data.
> Bearing in mind that the analysis of the GBS does not exist beyond the first
> 21 years. Although I discounted the odd months out of year 1 and 21 because
> of the split, in that the survey started in July and ended in June). Winter
> includes June, July and August and Summer includes December, January and
> February. Therefore each analysis includes 63 data points. The analysis of
> variance from the regression is also presented (below).
>
> Just imagine that we did a GBS in only summer or winter. Two separate
> regression analyses for these data show: for winter, a probability value
> (significance of the ANOVA regression F test) of 2.0769E-05 in other words,
> the chance that the observed decrease is due to chance factors is
> 0.000020769. Likewise for summer a probability value of 3.59055 E-14 in
> other words, the chance that the observed increase is due to chance factors
> is 0.0000000000000359055. Normally biologists accept as reasonable evidence,
> anything less than 0.05 (meaning 95% probability). These two trends are way
> off the scale for any suggestion of no trend! Yet Martin says no trends,
> just on the simple basis of missing that the two trends are running in
> opposite ways (an increase and a decrease). The possibility of damage done
> to the GBS and COG's data by such incompetent analysis and comment is of
> great concern.
>
> It is simply the fact that summer abundance has increased to a similar
> margin as winter numbers have decreased that appear to confuse those not
> clear in the biological meaning of the results or capacity of the survey).
> It was not by accident that I chose this species as one of the three to
> present the full graph of abundance by month throughout the 21 years of the
> GBS (on page 90).
>
>
>
> SUMMARY OUTPUT
>
> Of regression analysis of Pied Currawong Winter values of A
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Regression Statistics
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Multiple R
>
> 0.508654885
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> R Square
>
> 0.258729792
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Adjusted R Square
>
> 0.246577822
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Standard Error
>
> 1.426770386
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Observations
>
> 63
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> ANOVA
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> df
>
> SS
>
> MS
>
> F
>
> Significance F
>
>
>
>
>
>
>
> Regression
>
> 1
>
> 43.34189562
>
> 43.3419
>
> 21.29118
>
> 2.0769E-05
>
>
>
>
>
>
>
> Residual
>
> 61
>
> 124.1760978
>
> 2.035674
>
>
>
>
>
>
>
>
>
>
>
> Total
>
> 62
>
> 167.5179934
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Coefficients
>
> Standard Error
>
> t Stat
>
> P-value
>
> Lower 95%
>
> Upper 95%
>
> Lower 95.0%
>
> Upper 95.0%
>
>
> Intercept
>
> 7.934359474
>
> 0.363835294
>
> 21.80756
>
> 1.71E-30
>
> 7.206825636
>
> 8.661893313
>
> 7.206825636
>
> 8.661893313
>
>
> X Variable 1
>
> -0.04561298
>
> 0.009885272
>
> -4.61424
>
> 2.08E-05
>
> -0.06537981
>
> -0.025846155
>
> -0.06537981
>
> -0.025846155
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> SUMMARY OUTPUT
>
> Of regression analysis of Pied Currawong Summer values of A
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Regression Statistics
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Multiple R
>
> 0.782564696
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> R Square
>
> 0.612407504
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Adjusted R Square
>
> 0.606053528
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Standard Error
>
> 0.337899565
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Observations
>
> 63
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> ANOVA
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> df
>
> SS
>
> MS
>
> F
>
> Significance F
>
>
>
>
>
>
>
> Regression
>
> 1
>
> 11.00449822
>
> 11.0045
>
> 96.38179
>
> 3.59055E-14
>
>
>
>
>
>
>
> Residual
>
> 61
>
> 6.964743096
>
> 0.114176
>
>
>
>
>
>
>
>
>
>
>
> Total
>
> 62
>
> 17.96924131
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Coefficients
>
> Standard Error
>
> t Stat
>
> P-value
>
> Lower 95%
>
> Upper 95%
>
> Lower 95.0%
>
> Upper 95.0%
>
>
> Intercept
>
> 1.479013121
>
> 0.086166484
>
> 17.1646
>
> 4.95E-25
>
> 1.306712539
>
> 1.651313703
>
> 1.306712539
>
> 1.651313703
>
>
> X Variable 1
>
> 0.022983684
>
> 0.002341112
>
> 9.817423
>
> 3.59E-14
>
> 0.01830234
>
> 0.027665027
>
> 0.01830234
>
> 0.027665027
>
> Also Martin wrote:
> I have attached a graph showing A values by week (months are difficult
> since the weeks don't all start and finish at the bginning/end of
> months) for 21 years and later years. This clearly shows a flattening
> of the seasonal distribution in the recent past with more birds in the
> warmer months and less in the colder months. There has been previous
> correspondence on this list (last year some time) about the reduced
> frequency of the huge flocks seen (particularly in the Weston area) in
> the early years of the Survey.
>
> As for "months are difficult since the weeks don't all start and finish at
> the beginning/end of months". What a non issue! Weeks and months are
> arbitrary human constructs and dividing lines. (Indeed none of the weeks and
> months start on the same day apart from on 1 January but it doesn't matter
> one iota). In contrast days and seasons are real. However collecting data by
> week happens to be convenient to our brains, our lifestyles and constraints
> of designing and printing a chart. The arrangement of the GBS Calendar was a
> brilliant way to approach the problem, invented by Henry Nix and it and the
> connection of weeks and months are fully explained in the GBS Report.
>
> Looking at the graph presented by weeks tells us almost nothing more that
> looking at the graph presented by months. It simply provides more data
> points. Each with a smaller sample size, thus increasing the importance of
> random sample error for the rarer species. Thus weeks makes it harder to
> provide the information in a graph and certainly much harder to describe in
> narrative. Believe me, having written all the text I did for the GBS Report
> it would have been much more difficult to describe every change in terms of
> week numbers rather than months, because everyone understands months.
>
> Lastly, not only has there "been previous correspondence on this list ...
> about the reduced frequency of the huge flocks seen ... in the early years
> of the Survey. True but I point out that this has been fully described in
> the GBS Report and the results Martin has now provided since Year 21 are
> simply a continuation of the same trend that I described therein (which is
> hardly surprising).
>
> Philip
*******************************************************************************************************
This is the email announcement and discussion list of the Canberra
Ornithologists Group.
List-Post: <>
List-Help: <>
List-Unsubscribe: <>
List-Subscribe: <>
List archive: <http://bioacoustics.cse.unsw.edu.au/archives/html/canberrabirds>
List manager: David McDonald, email
<>
|
