SPRI library catalogue
View a record
Please note: You are viewing the legacy database of the Scott Polar Research Institute Library catalogue. It is no longer being updated, so does not reliably reflect our current library holdings.
Please search for material in iDiscover for up-to-date information about the library collection.
Record #186045:
Estimation of effective sample size for catch-at-age and catch-at-length data using simulated data from the Dirichlet-multinomial distribution / S.G. Candy.
Title: | Estimation of effective sample size for catch-at-age and catch-at-length data using simulated data from the Dirichlet-multinomial distribution / S.G. Candy. |
Author(s): | Candy, S. G. |
Date: | 2008. |
In: | CCAMLR Science. (2008.), Vol. 15 (2008) |
Abstract: | Examines performance of set of four alternative estimators of effective sample size (ESS) to assess numerically efficient approaches for determining ESS to apply within existing, commonly available, integrated assessment software. Existing method of estimating ESS for catch-at-age or catch-at-length data in integrated assessments that uses aggregate year-level data is shown to have poor accuracy when heterogeneity in class proportions is due to between-haul within-year variation. Three alternative methods of estimation based on haul-level data, including new method recently implemented for CASAL-based assessment for Patagonian toothfish fisheries around sub-Antarctic Heard and McDonald Islands, are shown to greatly improve estimation. Examination of process error shows that when systematic lack-of-fit across years is mis-specified as process error, methods which attempt to account for process error give inappropriately low values of ESS. This has important implications for determining appropriate implicit weight given to commercial catch length of age frequencies versus other datasets via their negative log-likelihood contribution to objective function in integrated assessment software. Five-step approach to estimating ESS for catch-at-age and/or catch-at-length data in integrated assessments is recommended. |
Notes: | CCAMLR Science. Vol. 15 :115-138 (2008). |
Keywords: | 639.2 -- Fisheries. 639.2.001 -- Fisheries research. 51 -- Mathematics. 519.2 -- Statistical methods. 519.673 -- Modelling. N -- Renewable resources. (*7) -- Antarctic regions. (*78) -- Sub-Antarctic islands. (*785) -- Heard and McDonald Islands. (*80) -- Southern Ocean. (*86) -- Southern Ocean, Australian sector. |
SPRI record no.: | 186045 |
LDR 02446naa#a2200000#a#4500 001 SPRI-186045 005 20231210125608.0 007 ta 008 231210s2008####xx#a##|##|s##|0||#0|eng#d 035 ## ‡aSPRI-186045 040 ## ‡aUkCU-P‡beng‡eaacr 100 1# ‡aCandy, S. G. 245 10 ‡aEstimation of effective sample size for catch-at-age and catch-at-length data using simulated data from the Dirichlet-multinomial distribution /‡cS.G. Candy. 260 ## ‡a[S.l.] :‡b[s.n.],‡c2008. 300 ## ‡ap. 115-138 :‡bill., diags., tables. 500 ## ‡aCCAMLR Science. Vol. 15 :115-138 (2008). 520 3# ‡aExamines performance of set of four alternative estimators of effective sample size (ESS) to assess numerically efficient approaches for determining ESS to apply within existing, commonly available, integrated assessment software. Existing method of estimating ESS for catch-at-age or catch-at-length data in integrated assessments that uses aggregate year-level data is shown to have poor accuracy when heterogeneity in class proportions is due to between-haul within-year variation. Three alternative methods of estimation based on haul-level data, including new method recently implemented for CASAL-based assessment for Patagonian toothfish fisheries around sub-Antarctic Heard and McDonald Islands, are shown to greatly improve estimation. Examination of process error shows that when systematic lack-of-fit across years is mis-specified as process error, methods which attempt to account for process error give inappropriately low values of ESS. This has important implications for determining appropriate implicit weight given to commercial catch length of age frequencies versus other datasets via their negative log-likelihood contribution to objective function in integrated assessment software. Five-step approach to estimating ESS for catch-at-age and/or catch-at-length data in integrated assessments is recommended. 650 07 ‡a639.2 -- Fisheries.‡2udc 650 07 ‡a639.2.001 -- Fisheries research.‡2udc 650 07 ‡a51 -- Mathematics.‡2udc 650 07 ‡a519.2 -- Statistical methods.‡2udc 650 07 ‡a519.673 -- Modelling.‡2udc 650 07 ‡aN -- Renewable resources.‡2local 651 #7 ‡a(*7) -- Antarctic regions.‡2udc 651 #7 ‡a(*78) -- Sub-Antarctic islands.‡2udc 651 #7 ‡a(*785) -- Heard and McDonald Islands.‡2udc 651 #7 ‡a(*80) -- Southern Ocean.‡2udc 651 #7 ‡a(*86) -- Southern Ocean, Australian sector.‡2udc 773 0# ‡7nnas ‡tCCAMLR Science. ‡gVol. 15 (2008) ‡wSPRI-131147 916 ## ‡aHS 917 ## ‡aUnenhanced record from Muscat, imported 2019 948 3# ‡a20231210 ‡bHS