And it has grown into a full scale data analysis and possible white paper (I will post here when the draft is ready).
I have been spending the last few months pouring over the MRFSS/CHTS, which is the survey that produces the participation and effort estimates, that ultimately go into the catch models.
Under the new MRIP system NMFS is still using the outdated and what I believe is just plain wrong MRFSS/CHTS (Still "FATALLY FLAWED") telephone survey of participation and effort.
The CHTS (Coastal Household Telephone Survey) is produced every 2 months and provides NMFS with the participation and effort variables they use to produce catch analysis.
Now please keep in mind that year over year there are about 1 million recreational licenses sold in Florida and for any given year there are about 1.5 to 1.75 million licensed anglers in the state.
I pulled data from NMFS' MRFSS/CHTS and the 1991 to 2006 US Census Bureau Fishing, Hunting, and Wildlife Associated Recreation Reports (links to the full report provided and to query the NMFS data simply follow their link and query Effort and Participants from 1991 to 2010).
Since the US Census report only comes out every 5 years, I used the MRFSS/CHTS 5 year running averages to perform my comparison (this gave NMFS the benefit of a doubt, but they need all the help they can get).
The statistical magic performed paints a very different picture than the ever increasing out of control recreational effort created by NMFS and if accurate, means that our catch effort has been overstated by some 150% in some cases to well over 400% in others.
You can see that with a tiny bit of smoothing the data seems to fit a more rational patterns. Any time you see a data set showing extremely wild variations in data, you should suspect that something is wrong. Now even the smoothed data shows higher than expected results as it relies on the underlying raw data, but by dampening the high variability we can see the over all slope of the line is fairly consistent and this can be measured by other variables such as tackle and license sales, plus tourism data, all of which point to a much lower estimate than is reflected here.
I postulate a few theories on why the NMFS data does not seem to fit the mold of reality.
The first is that MRFSS/CHTS was designed in 1979 and has undergone few if any modifications to account for the changes in coastal demographics or technology.
The CHTS only uses land based lines in coastal counties.
Secondly there is too much randomization in a study like this. As I posted in a previous thread, trying to produce a study on effort and participation from a larger population without filtering out noise (like asking a room full of women about menses and then saying that a certain percentage of both men and women menstruate is a bad model).
What caught my attention the most was not the actual estimates, but moreover how those estimates changed between years and waves.
One possibility is the way in which the sampling is randomized. For instance in Pinellas county approximately 321 land line phone numbers are selected and they need 32 to be answered after at least 5 attempts.
Now if in the sample there are 32 random samples that are heavily loaded with fishing participants then that survey will show a higher than expected participation result and should they get 32 widows who have never touch a fishing pole, then you can see that the sample would have below expected results. This I believe accounts for the wild swings we see in the data and as NMFS uses fill in data the errors continue to propagate.
Notice that in the very first CHTS years, the estimate is fairly close to 1.5 million resident anglers , but that the overstatement of visiting anglers throws the whole data set out of whack. Since the errors are continually propagated one wave and one year to the next, the sampling continues to grow, until something forces it to crash (such as a severe natural disaster or the economic meltdown or a statistical correction).
I am not 100% convinced that the US Census Bureau report is solid enough to be the basis to make real estimates, but I do believe it provides a solid baseline to validate NMFS' estimates. One thing I am certain of thought is that there are not 3 to 4 million uncounted anglers in Florida and even accounting for under 16 and over 65 and active duty military, that the real estimate of annual participation can be mathematically tied to the number of active licenses int the state with a factor of about 1.75 to 2.5. The maxima/minima I use are the number of active licenses (lower bound) and 3 times the active licenses (upper bound) which returns a number between 1.5 million and 4.5 million.
To query NMFS Data
To see the 1991 to 2006 Full Census Bureau reports (I used the Florida Reports only) Tables 8, 9 and 10 provide Saltwater angler data.