SIS sensitivity

[Philipp-Neubauer]

From Mike:

Just to throw out the idea - there could be the potential for a sensitivity analysis using "level of assessment" as defined in SIS. This ranges from 1-5 and represents complexity of the assessment. We could later repeat the analysis to consider only level 5 as assessed, or only levels 4-5, or only levels 3-5 (assessed stocks as we've currently defined can fall within level 2).

[Philipp-Neubauer]

I should point out that not all the stocks on our list - even assessed stocks - are in RAM. Regarding "given that we're using RAM as "assessed"...", is the assessed yes/no classification not based on the 'year of first assessment' variable, with 'no' = ['Only relative indices' | 'Minimal information' | 'No published document']? Or am I forgetting and it is based on whether or not the stock is in RAM? If it's the former, then this analysis doesn't draw at all from the RAM database so it doesn't even need mentioning in the Methods.

I now realize that my previous idea of a sensitivity analysis might not work unless we want to assign those level 0-5 classifications ourselves to the stocks in our dataset that are not managed or assessed by NOAA (i.e. are not in SIS). It looks like we have 24 assessed stocks in our dataset (with a year of first assessment value) that do not appear in the SIS dataset.

Here is the breakdown of SIS assessment levels for assessed stocks in our dataset (ones with a 'year of first assessment value') that are also in SIS:
1, n=5
2, n=2
3, n=15
4, n=120
5, n=7

I think three of the '1's are for stocks that previously had population models used but now rely on survey indices (silver hake, winter flounder). We've also been a bit flexible with assessment definitions for invertebrates (the other two '1's and the two '2's). I don't really see anything super special about the category 5 stocks (Atlantic herring - Northwestern Atlantic Coast; Butterfish - Gulf of Maine / Cape Hatteras; Red grouper - Gulf of Mexico; Gag - Gulf of Mexico; Arrowtooth flounder - Bering Sea / Aleutian Islands; Flathead sole - Bering Sea / Aleutian Islands; Yellowfin sole - Bering Sea / Aleutian Islands), they could easily be pooled with 4.
There are also 24 assessed stocks in our dataset that are not in SIS.

Mike

---

Levels of Stock Assessment Models
0—Although some data may have been collected on this species, these data have not been examined beyond simple time series plots or tabulations of catch.
1—Either:
a) time series of a (potentially imprecise) abundance index calculated as raw or standardized CPUE in commercial, recreational, or survey vessel data, or
b) onetime estimation of absolute abundance made on the basis of tagging results, a depletion study, or some form of calibrated survey.
2—Simple equilibrium models applied to life history information; for example, yield per recruit or spawner per recruit functions based on mortality, growth, and maturity schedules; catch curve analysis; survival analysis; or length-based cohort analysis.
3—Equilibrium and non-equilibrium production models aggregated both spatially and over age and size; for example, the Schaefer model and the Pella-Tomlinson model.
4—Size, stage, or age structured models such as cohort analysis and untuned and tuned VPA analyses, age-structured production models, CAGEAN, stock synthesis, size or age-structured Bayesian models, modified DeLury methods, and size or age-based mark-recapture models.
5—Assesssment models incorporating ecosystem considerations and spatial and seasonal analyses in addition to Levels 3 or 4. Ecosystem considerations include one or more of the following:
a) one or more time-varying parameters, either estimated as constrained series, or driven by environmental variables,
b) multiple target species as state variables in the model, or
c) living components of the ecosystem other than the target species included as state variables in the model.

[Philipp-Neubauer]

, is the assessed yes/no classification not based on the 'year of first assessment' variable, with 'no' = ['Only relative indices' | 'Minimal information' | 'No published document']?

Yes, that's what I used; RAM (or not) doesn't really figure. If its easy enough to assign the SIS variables to non-NOAA stocks, I guess it could be worth doing this kind of sensitivity. The main question is if we can get a first year of assessment for each category - e.g., stocks may have been assessed as category 3 at some early stage, and then later used a category 4 model? At that point, we'd want to have the cat. 3 first assessment date for a cat 3&up analysis, but the cat 4 date for a cat. 4&5 analysis? Does that make sense?

[James-Thorson]

Yeah that sounds hard! My vote is to not get stuck on any but the simplest
updates to our data (including the extra 9 from SIS sounds sufficiently
simple to me)

On Oct 25, 2016 12:46 PM, "Philipp Neubauer" notifications@github.com
wrote:

, is the assessed yes/no classification not based on the 'year of first
assessment' variable, with 'no' = ['Only relative indices' | 'Minimal
information' | 'No published document']?

Yes, that's what I used; RAM (or not) doesn't really figure. If its easy
enough to assign the SIS variables to non-NOAA stocks, I guess it could be
worth doing this kind of sensitivity. The main question is if we can get a
first year of assessment for each category - e.g., stocks may have been
assessed as category 3 at some early stage, and then later used a category
4 model? At that point, we'd want to have the cat. 3 first assessment date
for a cat 3&up analysis, but the cat 4 date for a cat. 4&5 analysis? Does
that make sense?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#3 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AHnqTbA0HSpttgbXkYnIu3vXQGtYpcaAks5q3lyFgaJpZM4KgZT_
.

[Philipp-Neubauer]

yes; just had a quick chat to Mike - seems that using the SIS definitions
would be hard in a quantitative sense. But it would probably be worthwhile
adding a qualitative link to SIS - e.g., a table or figure that links our
analysis to SIS (e.g., what category are the stocks that we assume
assessed...) and have a paragraph of discussion about this.

On Wed, Oct 26, 2016 at 8:51 AM, Jim Thorson notifications@github.com
wrote:

Yeah that sounds hard! My vote is to not get stuck on any but the simplest
updates to our data (including the extra 9 from SIS sounds sufficiently
simple to me)

On Oct 25, 2016 12:46 PM, "Philipp Neubauer" notifications@github.com
wrote:

, is the assessed yes/no classification not based on the 'year of first
assessment' variable, with 'no' = ['Only relative indices' | 'Minimal
information' | 'No published document']?

Yes, that's what I used; RAM (or not) doesn't really figure. If its easy
enough to assign the SIS variables to non-NOAA stocks, I guess it could
be
worth doing this kind of sensitivity. The main question is if we can get
a
first year of assessment for each category - e.g., stocks may have been
assessed as category 3 at some early stage, and then later used a
category
4 model? At that point, we'd want to have the cat. 3 first assessment
date
for a cat 3&up analysis, but the cat 4 date for a cat. 4&5 analysis? Does
that make sense?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<https://github.com/Philipp-Neubauer/FirstAssessment/
issues/3#issuecomment-256154964>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/
AHnqTbA0HSpttgbXkYnIu3vXQGtYpcaAks5q3lyFgaJpZM4KgZT_>
.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#3 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ACJDC1yhyg-awt4PckjPvi6fXle45SQ2ks5q3l2ngaJpZM4KgZT_
.

Phil

[Philipp-Neubauer]

I believe this is addressed qulitatively in the discussion and can be closed